FINAL REPORT: PHASE 2 EVALUATION OF THE EFFICIENCY VERMONT RESIDENTIAL PROGRAMS

FINAL REPORT: PHASE 2 EVALUATION OF THE EFFICIENCY VERMONT RESIDENTIAL PROGRAMS Prepared for Vermont Department of Public Service Montpelier, Vermont Prepared by KEMA, Inc. Burlington, MA December 2005

TABLE OF CONTENTS SECTION ES EXECUTIVE SUMMARY... E 1 E.1 Introduction...1 E.1.1 Program Overview...1 E.1.2 Evaluation Objectives...2 E.2 Methods...2 E.3 Key Findings and Recommendations...2 E.3.1 Efficient Products: Lighting Component...2 E.3.2 Efficient Products: Appliance Component...4 E.3.3 Residential New Construction...6 E.3.4 Potential for New Program Initiatives: Appliance Recycling...8 E.3.5 Process Evaluation...10 SECTION I INTRODUCTION...I 1 I.1 Overview... I 1 I.1.1 Program Overview... I 1 I.1.2 Evaluation Objectives... I 2 I.2 Methods... I 3 SECTION 1 LIGHTING PROGRAM ASSESSMENT...1 1 1.1 Introduction...1 1 1.1.1 Evaluation Objectives...1 1 1.1.2 Background and Motivation...1 1 1.1.3 Overview of Data Collection and Analysis...1 2 1.1.4 Structure of this Report...1 3 1.2 Review of EVT Program Activity to Promote CFLs...1 4 1.2.1 Objectives and Operations...1 4 1.2.2 Patterns of Program Activity and Participation...1 5 1.3 Estimates of CFL Sales...1 9 1.3.1 CFL Sales in Vermont: 2004...1 9 1.3.2 Comparison to CFL Sales Studies of Other Geographic Regions...1 13 1.4 Segmentation of the Residential CFL Market...1 21 1.4.1 Overview...1 21 1.4.2 Factors Associated with CFL Awareness and Purchase...1 22 1.4.3 Remaining Market Potential...1 22 1.5 Installation Rates...1 23 1.6 Conclusions and Recommendations...1 24 1.6.1 Key Findings...1 24 1.6.2 Recommendations...1 26 i

TABLE OF CONTENTS SECTION 2 APPLIANCE PROGRAM ASSESSMENT...2 1 2.1 Introduction...2 1 2.1.1 Objectives...2 2 2.1.2 Data Collection and Analysis...2 2 2.1.3 Key Findings and Recommendations...2 3 2.2 Program Activity Summary...2 5 2.2.1 Objectives...2 5 2.2.2 Key Findings...2 5 2.2.3 Patterns of Customer Participation...2 6 2.2.4 Rebates by Store Type...2 6 2.3 Estimation of Appliance Sales and Energy Star Market Share...2 8 2.3.1 Summary...2 8 2.3.2 Clothes Washer Sales and ENERGY STAR Market Share...2 9 2.3.3 Refrigerator Sales and ENERGY STAR Market Share...2 13 2.3.4 Dishwashers Sales and ENERGY STAR Market Share...2 14 2.3.5 Room Air Conditioners Sales and ENERGY STAR Market Share...2 16 2.4 Assessment of Net Program Effects...2 17 2.4.1 Approach...2 17 2.4.2 Specification of Independent Variables...2 19 2.4.3 Model Selection...2 21 2.4.4 Model Enumeration: Estimation of Program Effect on ENERGY STAR Market Share in Vermont...2 22 2.4.5 Net Unit Sales and Energy Savings...2 24 2.4.6 Comparison to Massachusetts Analysis...2 24 2.4.7 Explanations for Apparent Reduction in Net Program Effects...2 26 2.5 Recommendations...2 27 SECTION 3 APPLIANCE SATURATION SURVEY...3 1 3.1 Overview...3 1 3.1.1 Project Background...3 1 3.1.2 Report Overview...3 1 3.1.3 Methodology...3 2 3.2 Space Heating...3 3 3.2.1 Primary Heating System...3 3 3.2.2 Heating Systems and Controls...3 6 3.2.3 Heating System Efficiency...3 7 3.2.4 Supplemental Heating Systems...3 8 3.3 Air Conditioning and Ventilation...3 11 3.3.1 Air Conditioning...3 11 3.3.2 Ventilation...3 15 3.4 Water Heating...3 15 ii

TABLE OF CONTENTS 3.5 Laundry Equipment...3 18 3.5.1 Clothes Washers...3 20 3.5.2 Clothes Dryers...3 21 3.6 Food Preparation...3 21 3.6.1 Penetration...3 21 3.6.2 Fuel Shares of Cooking Equipment...3 22 3.6.3 Dishwashers...3 23 3.7 Refrigerators and Freezers...3 24 3.7.1 Penetration and Saturation...3 24 3.7.2 Age of Equipment...3 25 3.7.3 Equipment Features...3 26 3.8 Home Office Equipment...3 28 3.9 Entertainment Equipment...3 30 3.10 Miscellaneous Appliances...3 30 3.11 Lighting...3 32 3.11.1 Lighting Fixture Inventory...3 32 3.12 Building Shell Features...3 32 3.12.1 Windows...3 32 3.12.2 Insulation...3 33 3.13 Energy Efficiency...3 34 3.13.1 Lighting...3 34 3.13.2 ENERGY STAR Awareness...3 35 3.13.3 Efficiency Vermont Awareness...3 36 3.14 Customer Demographics...3 36 SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT...4 1 4.1 Introduction...4 1 4.1.1 Objectives...4 1 4.1.2 Overview of Data Sources and Approach...4 1 4.1.3 Key Findings and Recommendations...4 3 4.2 Market Characteristics...4 5 4.2.1 Overview...4 5 4.2.2 Refrigerator and freezer units currently in Vermont households...4 5 4.2.3 Usage Factors...4 6 4.2.4 Characteristics of units in service...4 7 4.2.5 Recently Discarded Units...4 9 4.3 Unit Energy Consumption (UEC) Analysis...4 10 4.3.1 UEC Estimates Developed by Modeling RASS Data...4 10 4.3.2 UEC Estimates from Refrigerators Replaced by EVT LISF Program...4 14 iii

TABLE OF CONTENTS 4.4 Net-to-Gross (NTG) Analysis...4 16 4.4.1 Details of the California Analysis...4 16 4.5 Effective Useful Life (EUL) of savings...4 18 4.6 Conclusion and Recommendations...4 18 4.6.1 Conclusions...4 18 4.6.2 Recommendations...4 19 SECTION 5 VERRESIDENTIAL NEW CONSTRUCTION...5 1 5.1 Overview...5 1 5.1.1 Evaluation Objectives...5 1 5.1.2 Key Findings...5 1 5.1.3 Program Overview...5 3 5.2 New Construction Program Activity and Participation...5 5 5.3 Market Description...5 14 5.3.1 Demand Side Overview...5 14 5.3.2 Supply Side Overview...5 25 5.3.3 Detailed Builder Characteristics...5 29 5.4 Baseline Construction Practices...5 31 5.4.1 Characteristics of Homes Built...5 31 5.5 Energy Efficiency in Marketing and Construction Practices...5 34 5.5.1 Energy efficiency in the home sales and planning process...5 34 5.5.2 Energy-Efficient Features...5 35 5.5.3 Energy-Efficient Options...5 38 5.5.4 Energy Efficiency as a Business Proposition...5 40 5.6 Awareness of Energy Ratings and Standards...5 41 5.7 Vermont Energy Star Homes Program...5 42 5.7.1 Reported Program Awareness and Participation...5 42 5.7.2 Participating Builder Response to Vermont ENERGY STAR Homes...5 43 5.7.3 Nonparticipating Builder Response to Vermont ENERGY STAR Homes...5 47 5.8 Builder Survey Methods...5 50 SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS...6 1 6.1 Insulation Contractors...6 1 6.1.1 Survey Objectives and Methods...6 1 6.1.2 Sample Characteristics...6 2 6.1.3 Awareness of Statewide Energy Standards...6 3 6.1.4 Awareness and Knowledge of EVT Programs...6 3 6.1.5 Marketing Energy Efficiency Features...6 4 iv

TABLE OF CONTENTS 6.1.6 Energy Efficiency Market Opportunities...6 5 6.2 HVAC Contractors...6 6 6.2.1 Survey Objectives and Methods...6 6 6.2.2 Respondent Characteristics...6 7 6.1.3 2004 HVAC Installations...6 8 6.2.4 Awareness and Installation of ENERGY STAR-qualifying Equipment...6 9 6.2.5 Sizing and Installation Methods...6 11 6.2.6 Awareness and Assessment of EVT Programs...6 12 6.3 Remodelers...6 13 6.3.1 Survey Objectives and Methods...6 13 6.1.2 Sample Characteristics...6 14 6.1.3 Awareness of Statewide Energy Standards...6 15 6.1.4 Awareness and Knowledge of EVT Programs...6 16 6.1.5 Marketing Energy Efficiency Features...6 18 SECTION 7 PROCESS EVALUATION...7 1 7.1 Internal Organization and Administration...7 1 7.2 Marketing: Customer Awareness and Perceptions...7 2 7.3 EVT Performance Metrics Compared to Other States...7 3 SECTION A QUESTIONNAIRES... A 1 A.1 RASS Telephone Survey...A 1 Space Heating...A 3 Cooling...A 7 Water Heating...A 9 Refrigerators...A 10 A.2 Builder Questionnaire...A 26 A.3 Remodeler Questionnaire...A 47 A.4 HVAC Contractor Questionnaire...A 59 A.5 Insulation Contractor Questionnaire...A 77 v

ES EXECUTIVE SUMMARY E.1 INTRODUCTION This Executive Summary presents an overview of the Phase 2 Evaluation of Efficiency Vermont s programs to promote the adoption of energy-efficient products and construction practices among residential customers and the building professionals who serve them. Efficiency Vermont (EVT) delivers energy efficiency programs to electric customers statewide, with the exception of customers served by the Burlington Electric Department. EVT initiated operation in November 2000, under contract to the Vermont Department of Public Service (DPS). Previous to that, electric and gas utilities delivered energy efficiency services to their customers under the regulatory supervision of DPS. KEMA, Inc. (formerly XENERGY Inc.) completed the first evaluation of EVT s residential programs in 2002. E.1.1 Program Overview EVT delivers three main residential programs. Efficient Products. The Efficient Products Program (EPP) encourages and facilitates the purchase of ENERGY STAR qualified home lighting products and appliances through a combination of customer incentives, merchandising support to retailers, advertising, and public relations. The appliance component focuses primarily on promoting ENERGY STAR qualified clothes washers; the lighting component primarily on promoting compact fluorescent light bulbs (CFLs). New Construction. The Residential New Construction (RNC) Program promotes the use of energy-efficient construction methods and components in new single-family homes. This is accomplished primarily by providing a combination of technical assistance, financial incentives, and marketing support to builders, along with Home Energy Ratings (which include on-site inspection and testing) to certify to buyers and owners that participating units meet high energy performance standards. Existing Homes. The Existing Homes Program consists of a number of initiatives designed to capture energy efficiency opportunities in existing homes through retrofit projects addressing building envelope, heating, and cooling systems. In some cases, these programs also aim to provide training and incentives for adoption of energyefficient practices to contractors who typically serve those markets. The Efficient Products and New Construction Programs have been part of EVT s portfolio since the organization s inception, and most Vermont utilities had operated predecessor programs for as many as 10 years prior to that. The Existing Homes Program was initiated in 2003 and is still in its early phases of development. As directed by DPS, KEMA did not evaluate the Existing 1 VT DPS

SECTION ES EXECUTIVE SUMMARY Homes Program. We did, however, capture responses to the program by vendors in the market they address, and those responses are reported in the relevant sections of the report. E.1.2 Evaluation Objectives The principal objectives of the evaluation were as follows. Estimate the net effects of EVT on purchases of compact fluorescent light bulbs (CFLs). Estimate the net effects of EVT on purchases of ENERGY STAR appliances. Assess the effect of EVT on new construction practices. Identify and characterize opportunities for new program efforts, particularly appliance recycling. Identify opportunities for improving performance of major program components. Assess overall program performance vis-à-vis similar programs. E.2 METHODS Table ES-1 summarizes the data collection and analysis efforts undertaken to support the evaluation. E.3 KEY FINDINGS AND RECOMMENDATIONS The following paragraphs present key findings and recommendations in regard to individual program components and opportunities for new initiatives. E.3.1 Efficient Products: Lighting Component Key Findings EVT s CFL promotion program showed strongly improved results in 2004. The number of CFLs purchased through the program through the combined coupon, ITP, and catalog channels grew to 178,669 in 2004 from 72,791 in 2003, and increase of 144 percent in one year. Over the same time period, sales of CFLs in the U. S. as a whole decreased by 6 percent. The recently introduced manufacturer buydown initiative (ITP) contributed significantly to overall program growth. The manufacturer buydown or ITP, which was implemented on a full scale in 2004, yielded sales of 34,430 CFLs. That is 19 percent of total program volume and 32 percent of the increase in volume from 2003 to 2004. 2 VT DPS

SECTION ES EXECUTIVE SUMMARY Table E-1 Summary of Phase 2 Residential Program Evaluation Research Activities Activity / Summary of Topics Covered Sample Size and Other Details Review of tracking system database and other documentation. Interviews with DPS, EVT and Program Contractor Staff. In-depth interviews with 15 individuals. 1. Lighting Program Net Effects Study. Collect and analyze CFL sales data from a sample of Vermont locations. Corroborate findings with data from the Appliance Saturation Survey. Assess program net effects through comparison to similar data from other states. 2. Appliance Sales Data Collection and Net Effects Analysis. Collect appliance sales data from a sample of independent appliance dealers: volume of sales, model numbers for estimation of ENERGY STAR market share. Compile information on state-level appliance sales and ENERGY STAR market share from DOE and AHAM. 3. Appliance Saturation Survey: Telephone Component. Collect information on appliance holdings, age, efficiency, CFL holdings and purchase, plug loads, heating and cooling equipment, basic housing and demographic characteristics, recruitment for on-site component. Appliance Saturation Survey: On-Site Component. Verify numbers and location of CFLs installed, appliance efficiency, appliance age (via model numbers), basic housing characteristics. 4. Refrigerator Life Cycle Analysis. Estimate potential gross savings, net savings, and persistence of savings using a combination of primary and secondary data. 5. New Construction Update: Builder Survey. Assess builders knowledge of energy efficiency techniques and benefits; use of efficient construction techniques and equipment; opinions of business value of energy efficiency; current energy efficiency promotion practices; energy-related construction and specification practices prior to and since participation 6. Other trade ally interviews Assessment of program marketing and administrative effectiveness, customer response, and trade ally response. 7. Process Evaluation. Comparison of EVT performance to that of other similar organizations. Dimensions of comparison will focus on market penetration and cost of conserved energy. Collected data from 5 retailers accounting for 70 80 percent of program volume. Survey of 100 non-participating retailers to estimate CFL sales. Sales data obtained from 48 retail locations, accounting for 50 percent of all program rebates 600 random digit dial sample stratified by region 83 nested sample recruited from the telephone panel Data collected through the Appliance Saturation Survey, plus interviews with program staff, local appliance recyclers, and retailers 61 builders, random sample stratified by size and region 30 HVAC contractors, insulation contractors, appliance and lighting retailers, and remodelers. Data to be derived primarily from EVT records and reports and publicly available information on other programs. Vermont recorded the highest level of CFL sales per household of any state for which sales data (as opposed to program activity records) were available. Using a combination of program records, retailer survey data, and retailer sales records collected by EVT, KEMA was able to develop a robust estimate of total CFL sales in the state in 2004 of 271,170 units or 1.08 units per household. Among areas with active promotional programs, the Pacific Northwest recorded sales of 1.01 units per household estimated 3 VT DPS

SECTION ES EXECUTIVE SUMMARY using similar methods. Analysis of point-of-sale scanner data conducted for California utilities found that 2004 sales in California totaled 0.43 units per household and 0.29 units at the national level. The Net-to-Gross ratio estimated for EVT s promotion of CFLs ranges from 1.22 to 1.36. The Net-to-Gross ratio estimated for EVT s promotion of CFLs ranges from 1.22 to 1.36. That is, 2004 CFL sales attributable to the program s influence range from 217,088 to 243,844 units versus the 178,669 units sold or subsidized through the program. This estimate was not sensitive to wide variations in the estimate of key input variables. It is also significantly higher than the current estimate of 1.19 used for planning and savings tracking purposes. At the household level, installation rate was not strongly related to the number of CFLs purchased. Thus, we do not believe that attempts to more closely enforce unit limits for individual customers will yield higher net program savings. Recommendations The findings of this evaluation clearly indicate that EVT has done a very good job with the lighting program, particularly in the past two years. KEMA suggests the following steps to sustain and enhance the progress that has been made. Expand the ITP manufacturer buydown component of the program. We believe further development of this approach will help reduce unit costs of the program and diversify the base of retailers and customers participating in the program. Require that EVT attempt to collect CFL sales information from all participating retailers. EVT s success in obtaining the sales data used for this evaluation and the usefulness of the analysis these data supported suggests that the effort should be incorporated into ongoing operations. E.3.2 Efficient Products: Appliance Component Key Findings Vermont continues to have high Energy Star market shares for all four appliances. When compared to national and regional averages, Vermont s ENERGY STAR market share in chain stores for each of the four major appliance categories is consistently among the highest estimated levels for individual states. The weight of evidence suggests that EVT s appliance program and its utility-based predecessors, as well as other long-standing utility programs had a profound impact on the national market success of energy-efficient clothes washers. A study commissioned by the Consortium for Energy Efficiency (CEE) found that utility energy efficiency program managers, primarily from California and the Northwest states, played a crucial role in initiating the development of the infrastructure required for effective market transformation programs: contacts with manufacturers, technology assessments, and 4 VT DPS

SECTION ES EXECUTIVE SUMMARY common equipment specifications. 1 These efforts were organized on the national level by CEE, beginning informally in 1992. The Vermont utilities joined the initiative in 1996. This work in turn supported a major national effort involving manufacturers and over 200 local program sponsors by the year 2000. Since then the national market share of ENERGY STAR clothes washers has grown from 9 percent to 27 percent (2004). In 2007, the federal minimum standard will be revised to match the current ENERGY STAR specification. The net effects of the program on the market share of ENERGY STAR clothes washers sold in a given year have declined significantly in recent years. In 2001, the net unit sales attributable to the program were estimated at 1,045 versus total program-subsidized sales of 2,719 for a net-to-gross ratio of 0.38. In 2004, the maximum number of net unit sales attributable to the program was 724 versus program volume of 4,179 units, for a net-to-gross ratio of 0.17. This is a maximum estimate that reflects the inclusion in the analysis of a variable that captures the effect of past program efforts as well as other factors that accounted for past growth in Vermont s ENERGY STAR clothes washer market share. Findings of decreased net effect in the current year are consistent with recent increases in national ENERGY STAR appliance market shares, as well as with other regional studies. These findings are consistent with a similar analysis conducted in 2003 for a consortium of Massachusetts utilities, and with the overall perceptions of appliance program managers around the country. Recommendations In the short run, we believe that EVT should continue to administer rebates for ENERGY STAR clothes washers. This is necessary to maintain good relationships with retailer channels that have been built up over many years and to prevent potential selloffs of non-qualifying models prior to implementation of the next round of federal minimum energy efficiency standards in 2007. Assess and implement changes to the program that can increase cost effectiveness. These expedients may include: Reduce incentive amounts. Limit the duration of appliance incentive promotions to specific months. This approach has been tried in the Pacific Northwest. Market share of ENERGY STAR appliances remain high in that region. Restrict eligibility to models that qualify for the Consortium for Energy Efficiency s Tier 3 standards. In the longer term it may advance efficiency goals to redirect incentive and business development resources away from appliance promotions to other opportunities. EVT 1 Feldman, Shel, Mitchell Rosenberg, Jane Peters. 2001. The Residential Clothes Washer Initiative: A Case Study of a Collaborative Effort to Transform a Market. Boston: Consortium for Energy Efficiency. 5 VT DPS

SECTION ES EXECUTIVE SUMMARY should initiate discussions with appliance retailers to develop procedures for an exit strategy. E.3.3 Residential New Construction Findings The portion of single-family new homes that enroll in the program is very high compared to participation rates for similar programs elsewhere. In the years for which data were available (2001 through 2003), the program developed leads on 1,551 to 1,950 projects, which corresponds to 61 to 74 percent of the typical volume of single-family homes permitted each year (around 2,500). The program also does a good job converting leads to enrollments. This ratio ranged from 39 to 52 percent in the three years covered. The percentage of permitted single family homes that enrolled in the program remained stable in the range of 29 to 32 percent from 2001 to 2003. EVT has reduced the portion of projects that drop out of the program prior to certification and the pace at which projects are processed. Portion of projects completed in same year as enrollment increased from 14% in 2001 to 36% in 2003. Number of builders with projects enrolled and number of builders participating for the first time increased steadily from 2001 through 2003. As of the end of 2003, 200 builders had completed projects through the program. The number of builders completing multiple projects in one year increased to 30 in 2003 v. 12 in 2002. Repeat participation is key to the execution of EVT s strategy to transform the residential new construction market. The depth and quality of energy efficiency measures in participating homes increased significantly from 2002 through 2004. The portion of completed projects with 6+ enduses addressed increased from 38% in 2002 to 69% in 2004. The portion of completed projects meeting Energy Star rating criterion (86.0) increased from 77% in 2002 to 92% in 2004. Baseline energy efficiency practices improved between the Phase 1 evaluation (2001) and the Phase 2 study (2005), particularly for measures promoted by the EVT program. Inclusion of selected energy efficiency measures, including Energy Star appliances, energy efficiency fluorescent hard-wired lighting fixtures, and use of blower door tests to measure air infiltration, has increased since the Phase 1 study. These increases are likely a result of program influence. KEMA uncovered additional evidence of market transformation in interviews with builders. Key findings in this regard were as follows. Builder awareness of the non-energy benefits of energy-efficient equipment (including increased comfort and lower equipment maintenance costs) increased by statistically significant margins since 2001. 6 VT DPS

SECTION ES EXECUTIVE SUMMARY Customer requests for Energy Star rated high-efficiency heating and cooling equipment as priced options have increased dramatically since the Phase 1 evaluation: builders representing 39 percent of the new construction market in Vermont indicated that customers request the equipment as compared with only 8 percent in 2001. Builder perceptions of the importance of energy efficiency to the success of their businesses have increased, another likely demonstration of the program s success in transforming the new construction market. Recommendations EVT has significantly improved the quality of marketing and delivery for the Residential New Construction program since the Phase 1 evaluation. KEMA has no recommendations for improving program operations or design. Over the course of the evaluation, DPS and KEMA raised questions concerning the cost effectiveness of the program, prompted by its relatively low share of total annual electric savings as a portion of total annual electric savings achieved by EVT s residential programs. Upon review of cost and benefit results published in EVT s 2004 Annual Report 2 the program appears to be cost-effective on a total resources basis. Specifically: In 2004, the RNC program accounted for 25 percent of EVT s total budget for residential energy efficiency programs. First year MMBtu savings, which take into account fossil fuel energy as well as the energy content of delivered electricity accounted for 21 percent of total MMBtu savings. The program as a whole was not subjected to a formal total resource benefit-cost analysis in the 2004 Annual Report. However, cost and benefit estimates included in the report suggest that the program was cost-effective. The discounted value of the lifetime savings generated by the 2004 RNC program totaled $2.615 million (constant 2003 dollars), without accounting for the environmental and energy market risk adders that have been incorporated into cost-effectiveness calculations in Vermont. The total cost of the program, also in 2003 dollars, was estimated at $1.901 million. The observations concerning electric savings that led to the initial concern about the cost effectiveness of the program are accurate. The RNC accounts for only 3 percent of estimated annual electric savings from residential programs, and 6 percent of the lifetime savings. The more favorable total resource results stem from the program s emphasis on longer-lived measures and on measures that save primarily fossil fuel energy. KEMA recommends that EVT in conjunction with DPS take steps to clarify the rationale for the program. For example, it seems likely that some of the improvement in baseline construction practices is attributable to the program. RNC is also an important for reducing overall energy costs of consumers and a vehicle for leading the market towards future increases in the 2 Efficiency Vermont. November 2005. Year 2004 Annual Report and Annual Savings Claim. Burlington, VT. Tables 2.1.16 and 3.1.12. 7 VT DPS

SECTION ES EXECUTIVE SUMMARY stringency of the Energy Code. Moreover, the RNC is an important channel for the delivery of energy-efficient lighting fixtures. During the period between evaluations, we recommend that the EVT Business Development and Planning staffs seek to characterize these program contributions more fully than was possible in this evaluation and to develop potential changes to the program that will enhance these effects. E.3.4 Potential for New Program Initiatives: Appliance Recycling The DPS requested that KEMA assess the potential cost-effectiveness of an initiative to provide incentives to customers to turn in used refrigerators and freezers. KEMA used data collected through the 2005 Residential Appliance Saturation Survey (RASS), EVT data, and evaluations of appliance turn-in programs in other jurisdictions to estimate: Likely volume and characteristics of appliances that would be turned in; Average unit annual energy consumption (UEC) for units likely to be collected by the program (a measure of gross savings per unit); A net-to-gross ratio (NTG) that reflects the net effect of a turn-in initiative on a customer s decision to recycle an appliance, versus retaining it in use or transferring it to another user; A measure effective useful life (EUL) that reflects how long the appliance would have remained in service in the absence of a turn-in initiative. The DPS and the Energy Efficiency Utility can use these findings, along with cost information from appliance recycling service providers, to assess the cost-effectiveness of pursuing an appliance turn-in initiative. Key Findings Market of Available Units Of the 600 RASS respondents, 128 (21 percent) reported that they had had a refrigerator and/or freezer removed from their home in the past three years. The majority of all reported discarded units (62 percent) were 15 years or older. The vast majority (98 percent) of respondents who disposed of one refrigerator in the past 3 years said that the refrigerator had been their primary refrigerator (not an extra or a back-up unit). 64 percent of the discarded primary units were reported to be working units (just old or replaced). Twelve percent were reported to be working, but in need of repair, and nearly 24 percent were characterized as not working when they were discarded. 8 VT DPS

SECTION ES EXECUTIVE SUMMARY Savings Estimates Unit energy savings. KEMA estimated potential unit gross savings from a turn-in program by applying data from the RASS on the age, usage patterns, and means of disposal of refrigerators and freezers to Unit Energy Consumption (UEC) data developed by other studies. This approach yielded estimates of potential gross savings per unit of 1,899 kwh per year. Potential net energy savings. KEMA developed estimates of potential net program energy savings using the results of recent evaluations of refrigerator turn-in programs in California. Applying findings regarding participation and net-to-gross ratios from these studies, we arrived at an estimated net savings of 1,824 MWh/year. Table E-1 displays the results of this estimate. By way of comparison, the planned level of savings for the clothes washer component of the Appliance program was 1,238 MWh/year in 2004. Table E-2 Preliminary Estimate of Net Program Electricity Savings Potential Number of Households 249,450 Annual Participation Rate Annual unit gross savings (UEC) Net-to-Gross Ratio Annual Gross Savings 1.1% x 1,899 kwh/year x 0.35 x 1,824 MWh/Year Recommendations It is clear that an appliance recycling program offers significant savings. We are confident that the estimate of annual unit gross savings represents a reasonable estimate of savings that can actually be achieved with the collection of each unit. We are less confident about the assumed participation rate and NTG ratio. Any attempt to project a participation rate for Vermont on the basis of observations in other states is somewhat speculative. In the case of the NTG ratio, we note that this performance parameter has varied in unpredictable ways within single jurisdictions. The results of this analysis strongly suggest that further efforts to assess the cost-effectiveness of a prospective appliance recycling program are warranted. We recommend taking the following steps. Approach vendors such as Recycling North and Appliance Recycling Centers of America to obtain non-binding estimates of the costs of operating a program that involves recycling of 1,500 to 2,500 units. Conduct cost-effectiveness screening using the net and gross savings estimates developed for this analysis. 9 VT DPS

SECTION ES EXECUTIVE SUMMARY Conduct sensitivity analysis by varying the average gross savings figure as high as 2,000 kwh per year, and the NTG ratio as high as 0.50. Based on the results of this analysis, assess the likelihood of the program becoming cost effective. E.3.5 Process Evaluation Generally, KEMA found EVT s residential programs to be very well-managed. In particular, we found the organization s marketing and program tracking procedures and results to be excellent in comparison with peer programs. One of the most useful ways to assess the overall performance of EVT is to compare it to that of peer programs using a consistent set of indicators. In this case, peer programs would consist of long-established public benefit-funded energy efficiency programs that address all or most of the customers in a given state. Unfortunately, regulators in the individual states do not collect program performance information in uniform ways. Thus direct comparisons between programs need to be treated as very general in nature. Overall, Vermont tied with three out of eight other peer programs considered for the highest level of electric savings as a portion of sales with 0.8 percent. EVT s residential programs captured savings equal to 0.9 percent of total residential electric sales in Vermont. The cost per first year savings provides a very rough measure of cost-effectiveness for a portfolio of programs. It would be much more appropriate to consider the combined results of the portfolio using the Total Resource Cost Test, or similar measures that take measure lifetime savings, customer costs, and environmental benefits into account. However, such an analysis would require a great deal more data than is readily available from other programs. With this caveat in mind, we see that EVT ranked third among states with peer programs in terms of cost per first year kwh savings. For all of EVT s programs combined, this figure was $0.281/kWh. The range for this indicator ran from $0.231/kWh in Wisconsin to $0.580/kWh in New Jersey. Thus, EVT s performance in this regard was very close to the best of the range presented by the eight peer programs. 10 VT DPS

I INTRODUCTION I INTRODUCTION I.1 OVERVIEW This is the Final Report of the Phase 2 Evaluation of Efficiency Vermont s programs to promote the adoption of energy-efficient products and construction practices among residential customers and the building professionals who serve them. 1 Efficiency Vermont (EVT) delivers energy efficiency programs to electric customers statewide, with the exception of customers served by the Burlington Electric Department. EVT initiated operation in November 2000, under contract to the Vermont Department of Public Service (DPS). Previous to that, electric and gas utilities delivered energy efficiency services to their customers under the regulatory supervision of DPS. KEMA, Inc. (formerly XENERGY Inc.) completed the first evaluation of EVT s residential programs in 2002. I.1.1 Program Overview EVT delivers three main residential programs. Efficient Products. The Efficient Products Program (EPP) encourages and facilitates the purchase of ENERGY STAR qualified home lighting products and appliances through a combination of customer incentives, merchandising support to retailers, advertising, and public relations. The appliance component focuses primarily on promoting ENERGY STAR qualified clothes washers; the lighting component primarily on promoting compact fluorescent light bulbs (CFLs). New Construction. The Residential New Construction (RNC) Program promotes the use of energy-efficient construction methods and components in new single-family homes. This is accomplished primarily by providing technical assistance and marketing support to builders, along with Home Energy Ratings to certify to buyers and owners that participating units meet high energy performance standards. Existing Homes. The Existing Homes Program consists of a number of initiatives designed to capture energy efficiency opportunities in existing homes through retrofit projects addressing building envelope, heating, and cooling systems. In some cases, 1 Executives of Efficiency Vermont prefer to discuss their activities in terms of markets rather than programs. Energy-efficient equipment markets, new construction, and renovation services market overlap extensively with each other, and synergies can be captured in planning and implementing program activities by taking those overlaps into account. On the other hand, the initiatives to address efficient products, new construction, and renovation opportunities are organized and delivered as more or less free standing programs. There is no short substitute for the word program. Thus, to avoid extra words and confusion among readers accustomed to program evaluations, we will refer to Efficiency Vermont s service delivery activities as programs. I 1 VT DPS

SECTION I INTRODUCTION these programs also aim to provide training and incentives for adoption of energyefficient practices to contractors who typically serve those markets. The Efficient Products and New Construction Programs have been part of EVT s portfolio since the organization s inception, and most Vermont utilities had operated predecessor programs for as many as 10 years prior to that. The Existing Homes Program was initiated in 2003 and is still in its early phases of development. As directed by DPS, KEMA did not evaluate the Existing Homes Program. We did, however, capture responses to the program by vendors in the market they address, and those responses are reported in the relevant sections of the report. I.1.2 Evaluation Objectives The Phase 1 Evaluation 2 was structured primarily to provide a detailed characterization of the markets in which the EPP and RNC operated. This work addressed market structure, segmentation, baseline market share for energy-efficient products and practices, market actors motivations for adopting energy-efficient measures, and the barriers inhibiting such behavior. The DPS structured the Phase 2 Evaluation to address a number of specific issues identified in the Phase 1 Evaluation and through on-going supervision of EVT programs. The project objectives arising from these issues were as follows. Estimate the net effects of EVT on purchases of compact fluorescent light bulbs (CFLs). CFLs account for 74 percent of estimated savings from all EVT residential programs. The Phase 1 Evaluation included an analysis of CFL sales data from a small sample of stores which suggested that the program could be having a very large effect on CFL purchases outside the program : that is, without price subsidies provided through coupons or catalog sales. DPS charged KEMA to expand the sample from which CFL sales data were obtained and to use this larger data set to assess net program impacts on CFL sales. Estimate the net effects of EVT on purchases of ENERGY STAR appliances. In the Phase 1 Evaluation, KEMA estimated the net effect of EVT programs on the market share and number of units sold for ENERGY STAR qualified clothes washers, refrigerators, dishwashers, and room air conditioners. This was accomplished by using state-level market share, program, and demographic data to estimate a regression model of net program effect on market share for each appliance. KEMA implemented a similar assessment for Phase 2, incorporating refinements developed in similar studies conducted since the Phase 1 evaluation. Assess the effect of EVT on new construction practices. The Phase 1 Evaluation included a survey of a large sample of builders to characterize their current practices in regard to energy-efficient construction, use of energy efficiency in marketing their 2 XENERGY Inc. 2002. Final Report: Phase 1 Evaluation of the Efficiency Vermont Efficient Products Program and Final Report: Phase 1 Evaluation of the Efficiency Vermont Residential New Construction Program. Both available on the DPS Website. I 2 VT DPS

SECTION I INTRODUCTION services, and perceptions of the importance of energy efficiency to the overall success of their businesses. The Phase 1 Evaluation also included an on-site survey of a sample of 159 recently built homes. We found that builders characterization of their practices closely matched the results of the on-site inspections. We therefore felt the most efficient use of evaluation resources would be to repeat the builder survey and to compare the results from Phase 2 to those of Phase 1. KEMA supplemented this assessment with an analysis of the extensive program data maintained in EVT s program tracking database. Identify and characterize opportunities for new program efforts, particularly appliance recycling. KEMA carried out a broad Residential Appliance Saturation Survey of the entire state. The telephone portion collected information on energy-related home features, systems, and appliance holdings, as well as demographic and attitudinal items needed for analysis of program effects. The on-site portion collected more detailed information on home construction and CFL holdings. We used this information in the program impact evaluations and in an assessment of the potential savings achievable through a refrigerator recycling effort. Identify opportunities for improving performance of major program components. KEMA addressed this objective on a program-by-program basis, using the results of the analyses described above, as well as evaluations of similar programs elsewhere for support. Assess overall program performance vis-à-vis similar programs. KEMA conducted a high-level comparison of the performance of EVT versus programs of similar scope and standing, using metrics such as participation rates among targeted customer groups and cost per kwh saved. I.2 METHODS Table I-1 summarizes the data collection and analysis efforts undertaken to support the evaluation. Table I-2 depicts the relationship between the research and analysis tasks on the one hand and the broad objectives of the evaluation on the other. I 3 VT DPS

SECTION I INTRODUCTION Table I-1 Summary of Phase 2 Residential Program Evaluation Research Activities Activity / Summary of Topics Covered Review of tracking system database and other documentation. Review to support development of flow charts and process descriptions. Assess capability of the system to support planned operations; identify potential bottlenecks, sources of inaccuracy, and areas for potential process improvements. Develop participant samples. Interviews with DPS, EVT and Program Contractor Staff. Gather information on understanding of program goals, value propositions (customer and trade ally), effectiveness of program marketing and administration, assessment of program contractor performance, assessment of customer and trade ally response, areas for improvement, issues to be explored by program contractor, trade ally and customer interviews. 1. Lighting Program Net Effects Study. Collect and analyze CFL sales data from a sample of Vermont locations. Corroborate findings with data from the Appliance Saturation Survey. Assess program net effects through comparison to similar data from other states. 2. Appliance Sales Data Collection and Net Effects Analysis. Collect appliance sales data from a sample of independent appliance dealers: volume of sales, model numbers for estimation of ENERGY STAR market share. Compile information on state-level appliance sales and ENERGY STAR market share from DOE and AHAM. Compile state-level information on demographic, market, and program characteristics that may affect ENERGY STAR market share. Conduct net effects analysis similar to previous evaluation 3. Appliance Saturation Survey: Telephone Component. Collect information on appliance holdings, age, efficiency, CFL holdings and purchase, plug loads, heating and cooling equipment, basic housing and demographic characteristics, recruitment for on-site component. Appliance Saturation Survey: On-Site Component. Verify numbers and location of CFLs installed, appliance efficiency, appliance age (via model numbers), basic housing characteristics. 4. Refrigerator Life Cycle Analysis. Estimate potential gross savings, net savings, and persistence of savings using a combination of primary and secondary data. 5. New Construction Update: Builder Survey. Knowledge of energy efficiency techniques and benefits; use of efficient construction techniques and equipment; opinions of business value of energy efficiency; current energy efficiency promotion practices; energyrelated construction and specification practices prior to and since participation; decision-making on energy efficiency components, information sources, and perception; estimates of costs to implement efficiency measures and other measures in typical home; potential effect on practices in work done outside the program. Sample Size and Other Details Review included tracking system, documentation, and program management reports In-depth interviews 3 DPS staff 10 EVT staff 2 Program Contractors Collected data from 5 retailers accounting for 70 80 percent of program volume. Survey of 100 non-participating retailers to estimate their CFL sales volume. 48 retail locations, accounting for 50 percent of all program rebates Compilation of state-level appliance market share data compiled by US DOE Compilation of state-level demographic and housing data from the US Census 600 random digit dial sample stratified by region 83 nested sample recruited from the telephone panel Data collected through the Appliance Saturation Survey, plus interviews with program staff, local appliance recyclers, and retailers 61 builders, random sample stratified by size and region I 4 VT DPS

SECTION I INTRODUCTION Table I-1 (Continued) Proposed Research and Analysis Activities Activity / Summary of Topics Covered 6. Other trade ally interviews Assessment of program marketing and administrative effectiveness, customer response, and trade ally response. 7. Process Evaluation. Comparison of EVT performance to that of other similar organizations. Dimensions of comparison will focus on market penetration and cost of conserved energy. Supplemental Builder Survey: Open-ended survey of builders who registered but did not complete residential new construction projects through the program. The objectives of the survey will be to assess the reasons for project attrition. Sample Size and Other Details 30 HVAC contractors, insulation contractors, appliance and lighting retailers, and remodelers distributed per advice from DPS. HVAC contractor results will also be used in Business Program Evaluation. Data to be derived primarily from EVT records and reports and publicly available information on other programs. 20 builders I 5 VT DPS

SECTION I INTRODUCTION Table I-2 Relationship of Research Tasks to Evaluation Objectives EVALUATION OBJECTIVES Lighting Savings Parameters Lighting Program Net Effects Appliance Program Net Effects New Const. Program Effects New Const. Baseline Updates Refrigerator Life Expectancy Program Design & Improvement Review of tracking system database and other documents Review of secondary data and reports Interviews with DPS, EVT and Program Contractor Staff X X X X X X X X X X X X X X Lighting Sales Data Collection X X Appliance Saturation Survey: Telephone Component Appliance Saturation Survey: On- Site Component X X X X X X X Appliance Sales Data Collection X X Builder Survey X X X Other trade ally interviews Process Evaluation X X X I 6 VT DPS

1 LIGHTING PROGRAM ASSESSMENT 1 LIGHTING PROGRAM ASSESSMENT 1.1 INTRODUCTION 1.1.1 Evaluation Objectives The objectives of the lighting program assessment are to: Develop reliable estimates of the volume of sales of compact fluorescent light bulbs (CFLs) to Vermont households. Characterize residential customers awareness, holdings, and installed stock of CFLs. Characterize the extent of retailer involvement in the sale of CFLs. Assess the net effect of the program on the current level of CFL sales. 1.1.2 Background and Motivation The Department of Public Service (DPS), Efficiency Vermont (EVT), and KEMA all identified development of reliable CFL estimates and the assessment of net program effects they could support as highest priority tasks for this evaluation. The assignment of such high priority was based on the following observations. CFLs account for a very large portion of total estimated savings and costs for EVT s residential programs. Over the course of EVT s operations, CFLs have accounted for over 42.7 GWH of estimated energy savings. That is 67 percent of all estimated savings from the Efficient Products program. In 2004, savings from CFLs accounted for 74 percent of estimated savings from all residential programs. Thus, it is clear that any useful assessment of the impacts of EVT s residential programs will require a defensible estimate of the impacts of EVT s efforts to promote CFLs. Uncertainty over key elements of the net impact estimate. The Phase 1 Evaluation 1 included findings from the collection of CFL sales data from a small sample of Aubuchon and True Value hardware stores in Vermont (10 stores) and Maine (7 stores). Analysis of these data suggested that customers were purchasing a very large volume of CFLs without using program rebates and that CFL sales per household in Vermont were very much higher than they were in Maine or in the U. S. as a whole. Moreover, these findings were consistent with more comprehensive analysis of check-out scanner data and retailer sales records conducted in California and the Northwest, two other areas 1 XENERGY Inc. December 2002. FinalReport: Phase 1 Evaluation of the Efficiency Vermont Efficient Products Program. Montpelier, VT: Vermont Department of Public Service. 1 1 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT served by long-standing residential lighting programs. Unfortunately, the Vermont data were too sparse to support an estimate of total sales in the state, the foundation for analyses of net program effects. Shortcomings of prior evaluation methods. Energy efficiency program sponsors in the Northeast have conducted many market assessment and evaluation studies of programs to promote compact fluorescent bulbs. Most of these studies have relied heavily on customer surveys to estimate market size and market penetration. However, all of these studies have vastly overestimated CFL sales, based on comparisons with national or regional scanner data. These problems are largely attributable to sample bias. Customers who are interested in energy efficiency and who have adopted energy efficiency measures are more likely to complete the survey than those who are not. On-site surveys eliminate the problem of misreporting CFL holdings, but do nothing to address sample bias. Thus, it has become clear that some form of sales data collection and analysis is the only reliable method to assess customer acceptance of CFLs. Improvement of program management. Given the ongoing importance of CFLs in EVT s overall energy savings, it will be useful to monitor CFL sales on an ongoing basis. Thus DPS viewed the sales data collection effort as a pilot for potential inclusion of sales data collection in the scope of work for the Energy Efficiency Utility 1.1.3 Overview of Data Collection and Analysis To address the evaluation objectives listed above, KEMA collected and analyzed the following sets of data. KITT database of program transactions. EVT s KITT database contains records of each CFL for which coupons were redeemed. These records contain identification of the customer and retailer, as well as customer contact, measure savings, and measure cost information. Additional program records. EVT also provided to KEMA other information on program activities, including the number of bulbs subsidized via buy-downs and catalog sales. CFL sales records from Vermont stores. The Business Development staff of EVT was able to convince managers at retail locations that represented 69 percent of all rebates in 2004, as well as 92 percent of manufacturer write-downs to make 2004 CFL sales data available to the evaluation effort. These sales data took a variety of forms, ranging from printouts of monthly computer records to copies of paper records. KEMA staff compiled the records and verified Energy Star designation of products where possible. We describe these data in greater detail below. CFL sales records from Maine stores. A major hardware chain provided sales CFL sales data for all of its stores in Maine. Survey of nonparticipating retailers. Using the KITT database records and the imarket database of Dun & Bradstreet records, KEMA developed a sample frame of over 900 1 2 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT retailers that were likely to carry lighting products but who did not participate in the program. These included grocery, drug, hardware, and other small merchandisers. KEMA completed interviews with over 200 of these establishments to determine whether they sold CFLs and, if so, how many they had sold in the past year. Customer telephone survey data. The Residential Appliance Saturation Survey (RASS) conducted as part of this evaluation contained a number of items on compact fluorescent lamps, including awareness of CFLs, current number of CFLs installed, awareness and use of incentives to purchase CFLs, and influence of the incentives on the purchase decision. On-site site survey data. KEMA conducted on-site surveys in the homes of 83 customers who responded to the telephone survey. Our inspectors collected data on the number of lighting fixtures of various types in the home, the number of CFLs currently in place, the number of sockets that could be retrofitted with CFLs, and the number of CFLs in storage. Survey of overlimit customers. Customers were limited to purchasing six bulbs with coupons in any one transaction. However, given the instant rebate nature of the program, it was not practicable to limit customer purchases through multiple transactions. From program inception through late 2004, 1,327 customers redeemed coupons or received rebates for more than 20 CFLs. That is, 1.9 percent of total participants accounted for 10.5 percent of the units supported by the program. KEMA interviewed 16 customers who had purchased more than 20 bulbs in the past year to assess the disposition of those bulbs. Secondary studies. KEMA reviewed and used information on CFL sales and customer market segmentation developed for the U. S. as well as a number of regions, including Massachusetts, Long Island, Wisconsin, California, and the four states of the Pacific Northwest. Using these data we were able to develop a reasonably accurate estimate of the number of CFLs sold by Vermont retailers in 2004. We were also able to estimate a defensible estimate of baseline CFL sales, that is: the number of CFLs that would have been sold in Vermont in 2004 if EVT had not implemented its residential lighting program. In addition to this net effects analysis, we analyzed the results of the telephone RASS and the on-site survey to explore a number of issues identified by DPS. The first was segmentation of the market for CFLs. We were motivated to explore segmentation to obtain practical information for marketing CFL promotions. The second was the relationship between bulbs installed and bulbs purchased at the customer level. We explored the hypothesis that customers who purchased a large number of bulbs in a short period were likely to hold a large portion of those units as spares, thus reducing the net savings of the program. 1.1.4 Structure of this Report The remainder of this report is arranged in the following sections. 1 3 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Review of program activity. In this section we review program operations from inception to the present and analyze trends in the volume and nature of program activity. We base this analysis primarily on the KITT database, supplemented by information from program managers. Sales estimates and comparisons. This section presents our estimate 2004 CFL sales in Vermont along with the methods and input data used to produce it. It also presents estimates for CFL sales we developed for Maine, and estimates for other areas from secondary sources. The estimates from other regions provide a number of useful comparisons to market trends in Vermont and enable us to narrow down on a reasonable range of estimates for net program effects. This section presents the results of that process, along with a recommended net-to-gross ratio based on the sales analysis. Customer market characterization and program response. In this section we present the results of the customer surveys. We use these data to assess the level of customer awareness of CFLs, patterns of installation, and reported influence of the program. 1.2 REVIEW OF EVT PROGRAM ACTIVITY TO PROMOTE CFLS 1.2.1 Objectives and Operations The objectives of the lighting component of the Efficient Products Program (EPP) are to encourage and facilitate the purchase of ENERGY STAR qualified compact fluorescent light bulbs, lighting fixtures, and ceiling fans. The program accomplishes these objectives through the following means. Retailer promotion and merchandising support. EVT solicits retailers and supports their participation in the program by providing point-of-purchase materials, training to sales staff in program operations and product attributes, and facilitation of instant rebate coupon processing. From program inception through December 2004, this function had been fulfilled by Applied Proactive Technologies, Inc. Beginning in January 2004, EVT staff will be carrying out the retail representation role. General ENERGY STAR marketing. EVT participates in a Northeast regional effort to promote ENERGY STAR qualified products through media advertising, public relations campaigns, and other outreach efforts. EVT has also organized and sponsored a number of special promotional events involving individual communities or retailers. Instant Rebate Coupons. EVT provides instant rebate coupons for ENERGY STAR qualifying CFL bulbs, hard wired fixtures, and table and floor lamps. During most of the evaluation period, the coupons were valued at $3 per piece for CFLs, $15 for fixtures, and $20 for torchieres and table lamps. As of December 1, 2004, EVT implemented a standard $15 rebate for all light fixtures, torchieres, desk lamps, floor lamps, and ceiling fans with light kits. 1 4 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Manufacturer Buydowns. In order to gain program access to retailers that weren t prepared to process instant rebate coupons, EVT initiated a manufacturer buydown offering in 2003. Through this mechanism, EVT uses program funds to subsidize high volume purchases of qualifying units from manufacturers. These savings are passed on to participating retailers, who are then able to maintain required margins while selling to the public at a discount. The Energy Federation, Inc. handles the logistics of this operation. The buydown is attractive to mass merchandisers such as Wal-Mart and Costco and to smaller retailers such as supermarkets where the use of rebates had been resisted. Mail Order. EVT also offers a range of ENERGY STAR qualified lighting products at discounted prices through a mail order catalogue. 1.2.2 Patterns of Program Activity and Participation Key Findings Participation in CFL promotion programs in 2004 was strong. The number of customers using coupons to purchase CFLs increased by 25 percent from the prior year to 24,664. This count does not include customers who purchased reduced-price units through cooperative promotions of manufacturer buydowns. The volume of bulbs purchased through the program increased sharply to 178,669 from 72,791 in the previous year. Units purchased through the manufacturer buydown initiative accounted for nearly 19 percent of the program volume. In the first four years of operation 69,649 unique customers or 23.7 percent of all Vermont households redeemed coupons for CFLs through the program. Fifteen percent of these customers have participated in more than one year. This pattern of participation is consistent with the experience of other states that have operated strong CFL promotion programs. Business customers purchased at least seven percent of the units sold through the program. An additional 8 percent were purchased by customers who did not mark their coupons. Participants in the program tend to purchase a large number of CFLs. Among all participants, the median number of coupons redeemed was 5 over the full time span of EVT operation. Moreover 23.5 percent of the participants who purchased more than 6 bulbs were responsible for 53 percent of the total number of coupons redeemed. This pattern is typical of residential lighting incentive programs. Table 1-1 summarizes participation by retailers and customers from program inception through the end of 2004. With the exception of a dip in program-supported CFL and fixture sales in 2003, the figures show an uninterrupted increase in CFL sales. CFL purchases through the program totaled 178,669 units, more than double the volume in the first year of the program. 2 2 EVT assumed responsibility for the program in March 2000. The 2000 sales figures in Table 2-1 represent 10 months of activity. 1 5 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT On the other hand program-supported sales of fixtures have declined steadily since 2000. It is not clear why this should be the case. Other trends in lighting activity are as follows. Customer participation. The numbers of unique customers that participate in the program increased sharply between 2000 and 2001, then declined steadily by roughly 27 percent between 2001 and 2003. The number of customers using coupons to purchase CFLs increased to 24,664 in 2004, slightly below the 26,887 customers who used coupons in 2001. Customer participation counts do not include those who bought CFLs through reduced-price promotions at Costco and Wal-mart. These promotions accounted for 18 percent of program volume. Purchases through the program. After fluctuating in the range of 56,000 to 96,000 units over the past four years, the number of CFLs purchased through the program shot up to 178,669 in 2004. Customers used coupons to purchase 141,875 CFLs, the highest number in any of the five program years. In addition, customers bought 34,430 CFLs through the manufacturer buydown component and 2,364 through the program catalog. Participation Table 1-1 Summary of EPP Lighting Component Activity Units Rebated/Sold YEAR Retailers Customers Bulbs Fixtures Program and PR Notes 2000 90 13,793 56,511 22,887 13 special events, including Torchiere Turn-in: 3,000 halogen torchieres exchanged; 5,300 bulbs and 400 fixtures sold. 2001 108 26,877 86,353 22,294 23 Special Events: Torchiere turn-ins, home show booths, in-store promotions. Rebate levels reduced. 2002 125 22,744 95,517 15,522 31 Special Events thru October: includes 16 instore promotions plus a variety of home shows. Bulb rebate levels reduced to $3. 2003 123 19,783 72,791 11,492 Village of Poultney Change a Light challenge. Introduction of manufacturer buydowns. 2004 112 24,664 178,669 13,034 Coupon redemption processes simplified. Large volume of manufacturer buy-down sales. Total varies n/a 489,841 85,229 57,292 unique residential customers through 2003. Customer id numbers not available for 2004. Retailer participation. Retailer participation grew from 90 to 125 locations between 2000 and 2002. The number of locations participating in the program leveled off after that and declined to 112 in 2004. However, the introduction of the manufacturer buydown approach has attracted the participation of mass merchandisers that had previously stayed out of the program. Program-supported sales are highly concentrated in a few chains. Aubuchon Hardware, with over 40 locations in Vermont, accounted for 1 6 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT 54 percent of program-supported sales in the first 10 months of 2004. Home Depot, with 4 Vermont locations, accounted for an additional 16 percent, and 25 True Value affiliates accounted for 9 percent of program-supported sales. Repeat Participation. Given the relative ease of participation in the lighting program and the opportunity to install multiple CFLs and fixtures, it is important to estimate the number of unique customers who participated in the program and the number of program years in which they participated. This provides a more meaningful view of participation than the raw number of participants per year. Table 1-2 displays the results of KEMA s analysis of rebate coupon records from program inception through the end of 2003. The table shows that 69,849 unique customers have participated in the program since its inception in 2000. Of these 57,292 were residential customers. In 2004, there were 249,950 households in Vermont. Thus the program has attracted the participation of roughly one-quarter of all Vermont households. Table 1-2 shows that 15.5 percent of all participants made purchases through the program in more than one year. Most of these (12.3 percent) made purchases in two years. Participation in EVT s residential lighting program has been strong compared to other jurisdictions for which similar kinds of information are available. For example, the Long Island Power Authority has operated a residential lighting program using the NEEP template since 1999. From inception through 2002, 16.8 percent of residential customers participated in the program. Eleven percent of those participants purchased bulbs through the program in more than one year. 3 By contrast, EVT attracted the participation of 23 percent of residential customers, of whom 15 percent purchased bulbs through the program in more than one year. Table 1-2 Lighting Program Participants by Number of Years in Which They Participated Number of Years w/ Purchase # of Total Customers % of Total Customers 1 year 59,038 84.5% 2 years 8,621 12.3% 3 years 1,843 2.6% 4 years 347 0.5% Total 69,849 Bulb purchases per customer. Table 1-3 shows the distribution of customers by number of bulbs purchased. It also shows the percentage of total CFLs purchased by customers in each group defined by number of bulbs bought. Nearly 80 percent of participants in the program bought more than one bulb. The most frequent multiple purchased was six, which corresponds to the maximum number of coupons that can be redeemed at one 3 KEMA-XENERGY Inc. 2004. Final Report: Evaluation of LIPA s Residential Lighting and Appliance Program. Melville, New York: Long Island Power Authority. 1 7 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT time. Twenty-three percent of customers have purchased more than six bulbs over the program period. This group accounts for 53 percent of the total number of bulbs for which coupons were redeemed. Customers with 20 or more units rebated account for over 10 percent of the total number of units rebated. Thirty-three customers who purchased 60 or more bulbs (including one who purchased 606 in 2001 and 2002) accounted for 1 percent of total sales. KEMA conducted short interviews with twenty of these customers. While not all customers were characterized as businesses in the database, they all reported using the bulbs in business establishments. Table 1-3 Distribution of Lighting Program Participants by Number of Coupons Redeemed Participating Customers Coupons Redeemed # of Coupons Redeemed Percent Cumulative Percent Percent Cumulative Percent 1 20.8% 20.8% 3.8% 3.8% 2 13.5% 34.3% 5.0% 8.9% 3 8.3% 42.5% 4.6% 13.4% 4 9.1% 51.7% 6.8% 20.2% 5 4.5% 56.2% 4.1% 24.4% 6 20.5% 76.6% 22.8% 47.1% 7 8 7.6% 84.2% 10.5% 57.6% 9-12 9.0% 93.1% 17.5% 75.0% 13-20 5.0% 98.1% 14.5% 89.5% >20 1.9% 100.0% 10.5% 100.0% Total 69,849 376,979 Residential versus Business Customer Purchases. Several years ago, EVT established a goal to increase the number of businesses purchasing efficient lighting through the retail program. Customers who used rebate coupons were asked to mark whether they were purchasing bulbs and fixtures for use in a residence or business. Residential customers purchased 85 percent of all units sold through the program. Businesses purchased 7 percent of total units. The remaining rebate coupons were not marked. The share purchased by businesses is significantly higher than the 3 percent estimate from the previous evaluation. That estimate was based on imputation of customer type using the customer name, and likely underestimated the portion of purchases by businesses. The split of residential versus business customers was the same for all kinds of lighting equipment supported by the program: CFL bulbs, hard-wired fixtures, and table and floor lamps. Composition of fixture sales. Hardwired interior fixtures accounted for 72 percent of all fixture sales; torchieres for 22 percent. These proportions have held stable over the program years. 1 8 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT 1.3 ESTIMATES OF CFL SALES 1.3.1 CFL Sales in Vermont: 2004 KEMA estimated that Vermont retail outlets sold 271,170 CFLs in 2004, or 1.08 CFLs per household. This was the highest level of sales per household that we could identify in our search of the secondary literature. A 2004 analysis of national scanner data resulted in sales estimate of 0.29 CFLs per household for the U. S. as a whole. Using a sales monitoring process similar to the one employed by KEMA, the Northwest Energy Efficiency Alliance found estimated per household sales of 1.01 in 2004. That region has had CFL promotion programs similar in scope and duration to Vermont s. We examine these comparisons and others in more detail below. Data Sources and Methods The following paragraphs describe the sources and quality of the sales data we received and the methods we used to process these data into sales estimates. Under the terms of the confidentiality agreements we and EVT made with retailers, we do not mention retail establishments by name, only by type. Coupons and manufacturer write-down data. The KITT database records can be used to calculate the volume of coupons redeemed by store. EVT maintains similar store-level records of bulbs sold with manufacturer writedowns as well as records of catalog sales. Sales data provided and estimation methods. The Business Development staff of EVT, working in consultation with KEMA, was able to convince participating companies representing 35 retail locations to provide sales data. Altogether, facilities for which actual sales documentation were available accounted for 69 percent of all coupons redeemed by the program and 92 percent of the sales of bulbs with manufacturer buydowns. The following paragraphs provide further detail about these data. Participating Retailers who Provided Sales Data Hardware stores. Representatives of two hardware stores with a total of 47 Vermont locations agreed to make sales data available. KEMA received 2004 sales data for 32 of these locations. We estimated total CFL sales for the remaining locations by multiplying the number of coupons they redeemed by the ratio of total sales to coupons redeemed for the locations with sales data. Home Centers. There are two retailers who operate stores classified as home centers in Vermont. There are four participating home center locations in the state. We received sales data for one of them. Total CFL sales for the remaining locations were estimated using the same method described for participating hardware store chains. Building Supply Store. KEMA received sales data for one building supply store that participated in the program. 1 9 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Participating Retailers who did not Provide Sales Data. We estimated total sales for retailers who did not provide sales data by multiplying the number of coupons redeemed by the ratio of total documented unit sales to coupons redeemed for stores of that type. For stores that participated in the manufacturer buydown initiative, we assumed that the bulbs sold through that initiative represented total CFL sales. This was the case for one large mass merchandiser that provided sales data. Nonparticipating Retailers. Based on the results of previous studies, KEMA identified 9 types of stores that might sell CFLs. Using the imarket database of Dun & Bradstreet records, we developed a sample frame of the full population of those stores in Vermont. We then eliminated from the sample frame any locations that appeared in the KITT database of retailers participating in the program. The resulting sample frame listed 937 retail locations. See Table 1-4 for the distribution of these establishments among store types. Research America conducted a telephone survey with a random sample of 209 of the nonparticipating establishments. KEMA staff recalled over 50 of these respondents to verify answers and address potential anomalies in the data. After confirming that the location did not participate in the EVT program, we asked whether the location sold CFLs, and, if so, how many it had sold in the past 12 months. Sixteen of the 209 surveyed establishments reported that they sell CFLs. Fourteen of these reported selling fewer than 100 units in the 12-month period ending March 2005. Of the 16 respondent stores selling CFLs: 6 had been selling CFLs for 1 year or less; 4 had been selling CFLs for 2 or 3 years; 3 had been selling CFLs for 4 or more years; and 3 were not sure how long they have been selling CFLs. Based on an analysis of survey data, KEMA estimated that nonparticipating retailers sold approximately 2,940 CFLs in 2004. More than half those (59 percent) were sold in larger grocery and drug stores. 1 10 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Table 1-4 summarizes the non-participant retailer population, number of surveys conducted, number of surveyed stores selling CFLs, and an estimate of the sales volume, all broken out by type of retail establishment. Table 1-4 Estimated Annual CFL Sales by Non-Participant Vermont Retailers 4-Digit SIC Code Store Type # Stores in Population # Stores Surveyed # Stores Surveyed selling CFLs Estimated Annual Sales of CFLs % Sales by Store Type 5411 Grocery stores 510 84 7 1,155 39% 5912 Drug stores and proprietary stores 105 25 2 596 20% Home furnishing stores: Lighting, 5719 e lamps, accessories 10 7 3 650 22% 5211 b Lumber and other building materials 73 20 3 444 15% 5251 c Hardware stores 22 9 1 94 3% 5399 a Miscellaneous general merchandise stores 99 25 0 0 5331 Variety stores 68 20 0 0 5311 Department stores 35 10 0 0 5722 d Household appliance stores 16 10 0 0 Total 937 209 16 2,939 100% Specific SICs: a 5399-0000, 5399-9901, 5399-9903, 5399-9906 b 5211-00, 5211-0100, 5211-0301, 5211-9906, 5211-9908 c 5251-00, 5251-99 d 5722-0000 e 5719-02 Results Estimate of CFL Sales. Table 1-5 consolidates the results of the various sales estimation steps described above. We estimate that all retailers in Vermont sold a total of 271,170 CFLs, or 1.08 CFLs per non-institutional household. Nonparticipating retailers accounted for less than one percent of those sales. By way of comparison, a study of Northwest retailers found that nonparticipating retailers were responsible for roughly seven percent of regional CFL sales. 4 4 EcoNorthwest. 2004. Residential Lighting Program. Final Market Progress Evaluation Report. Portland, OR: Northwest Energy Efficiency Alliance. 1 11 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Table 1-5 Estimate of CFL Sales in Vermont: 2004 Number of Stores Number of CFLs w/ Incentives* Total CFLs Sold Sales/Store Participating Stores that provided Sales Data Home Center 3 8,496 16,126 5,375 Hardware Stores 47 68,364 144,411 3,073 Mass Merchants 1 53,164 53,164 53,164 Other 1 48 125 125 Participating Stores that did not provide Sales Data Home Center 1 245 465 465 Hardwares Stores 17 3,159 6,673 393 Mass Merchants 9 5,274 5,274 586 Other 33 18,181 8,138 247 Total: Participating Stores 112 141,875 234,376 Nonparticipating Stores Grocery Stores 510 0 1,155 2 Drug Stores 105 0 596 6 Home Furnishings 10 0 650 65 Lumber & Building Mat'l 73 0 444 6 Hardware Stores 22 0 94 4 Total: Nonparticipating Stores 2,939 Manufacturer Buydown Sales 34,430 34,430 Program Catalog Sales 2,364 2,364 Vermont Totals 178,669 271,170 Sales per Household* 1.08 * Number of Households: 249,590. US Census, American Community Survey, 2004 Assessment of the Quality of the Estimate. Like most sales estimates based on market research, our estimate of CFL sales has required us to make a number of assumptions. Perhaps the assumption with the greatest potential impact on the overall outcome is participating stores in a given retailer category (e.g. large home centers) achieve roughly equal ratios of total sales to rebated sales. However, we note that KEMA was able to obtain sales, coupon, or catalog order documentation for 229,000 units, or 85 percent of the total sales estimate. Thus, even if our assumptions concerning undocumented sales are inaccurate, the overall estimate of statewide sales would not change substantially. 1 12 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT 1.3.2 Comparison to CFL Sales Studies of Other Geographic Regions Energy efficiency program sponsors in California, Wisconsin, and the four-state Northwest region (Oregon, Washington, Idaho, and Montana) have recently undertaken efforts to estimate CFL sales in the areas they serve. These efforts used a variety of data sources and covered different periods. However, taken together, they provide an important set of references for assessing the effect of EVT s CFL promotion programs. Summary of Study Methods and Results The following paragraphs briefly review the results of the above-mentioned studies. California. As part of their market-share tracking efforts, the three large investor-owned utilities in California sponsor periodic studies to estimate CFL sales and market share (as a percent of screw-based bulbs) for California and for the nation as a whole. 5 The most recent of these studies used Point-of-Sale (scanner) data purchased from two market research organizations to estimate sales of CFLs through the following channels: food stores, drug stores, mass merchandisers, hardware stores, and home centers. These data have certain limitations. First, they are collected from stores in a relatively small number of primary sampling areas across the nation, e.g. 52 substate regions for food stores; 11 regions for mass merchandisers and hardware stores. Sample results must be projected to areas of interest such as states and the nation as a whole. The sampling plans for the scanner data services are generally set up to produce reliable results only at the national and large regional levels. Production of state-level estimates from these data requires the use of many assumptions. Second, large home centers such as Home Depot and Lowe s stopped authorizing release of scanner data in December 2002. Thus, sales estimates for later periods require forecasts of sales by these outlets based on projection of past performance. Finally, scanner data is collected only from stores with more than $1 million in sales. Wisconsin. The Energy Center of Wisconsin used a method similar to that described above to develop estimates of CFL sales for Wisconsin. Northwest Energy Efficiency Alliance. The Alliance operates a program that provides merchandising and mass marketing support for sales of CFLs across the Northwest region as part of its Energy Star Consumer Products program. Ecos Consulting, the program contractor collects CFL sales data from a large portion of the participating retailers and consolidates coupon and other incentive data from local program operators. The Alliance and its evaluation contractors use these data, as well as Dun & Bradstreet information and surveys of nonparticipating retailers to generate sales estimates for the region. The method is largely the same as the one KEMA used to develop sales estimates for Vermont. Results. Table 1-6 displays the findings from these studies. To aid comparison, we have also presented the results in terms of CFL unit sales per household for the areas covered. We have 5 Itron, Inc. 2004. California Lamp Report. Rosemead, CA: Southern California Edison. 1 13 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT chosen the number of households in the region (as estimated by the US Census American Communities Survey) as the divider for purposes of developing an indicator that can be used to compare findings across areas of vastly different size. We considered other potential dividers, such as the total number of housing units (including vacant units) and population. However, we believed that the number of households (which in most cases equals the number of occupied housing units) best represented the number of eligible consumer decision-makers. Vermont is something of a special case on this issue because the portion of vacant housing units in the state is 19.9 percent, compared to 10.3 percent for the country as a whole. This reflects the relatively high proportion of vacation homes in the state. However, given the range of conditions affecting CFL sales, we believed it was most appropriate to use the number of households to standardize results from one area to another. Annual Unit Sales Table 1-6 Comparison of CFL Sales and Sales per Household 2000 2001 2002 2003 2004 United States 7,040,165 14,830,120 27,036,000 33,517,000 31,362,000 California 973,307 1,256,552 4,462,000 4,702,000 5,236,000 Pacific Northwest 1,614,257 4,599,596 4,353,703 4,773,093 Wisconsin 1,317,162 1,271,373 Vermont 271,170 CFLSales/Household United States 0.067 0.141 0.252 0.309 0.289 California 0.103 0.109 0.381 0.391 0.437 Pacific Northwest 0.355 1.000 0.934 1.011 Wisconsin 0.615 0.589 Vermont 1.084 Vermont (program only) 0.093 0.236 0.359 0.301 0.715 The results in Table 1-6 show the following. National sales of CFLs increased rapidly from 2000 to 2003 but declined in 2004. National sales of CFLs more than doubled between 2000 and 2001, driven largely by public awareness of energy efficiency driven by the California and Northwest power crises and associated short-term information and incentive programs. In 2001, the full effect of those programs was felt. Between 2001 and 2002, national CFL sales grew by over 12 million units or 82 percent, with California and the Northwest accounting for nearly 34 percent of national sales. As the market effects of the crisis abated, growth in California slowed considerably in 2003, and sales of CFLs actually decreased slightly in the Northwest. Results of the analysis of scanner data published in the California Lamp 1 14 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Report 2004 6 show that national sales fell by 9.4 percent from 2003 levels. However in California, sales grew by 11 percent and by 9.6 in the Northwest. Regions with programs that combine incentives with merchandising support show significantly higher CFL sales per household than other regions. The programs in the Northwest and Vermont (as well as much of New England and Long Island) have featured extensive services to recruit and support retailers who participate in the incentive programs. These services are provided by program representatives or circuit riders who visit the participating locations on a regular basis, train sales staff, ensure that pointof-purchase and rebate materials are in place and properly displayed, and handle any logistical problems the retailer may be having. Vermont and the Northwest both show extremely high levels of CFL sales per household. Wisconsin and California programs do not include this high level of retailer support. Moreover, the coverage of incentive programs in these two states has not been as universal or consistent as it has been in Vermont. In California CFL purchases per household were 0.39 in 2003 and in Wisconsin 0.59 in 2004. These are considerably below estimated 2004 sales of 1.08 CFLs per household in Vermont. Vermont has the highest level of CFL purchases per household of any area for which data are available. Only the Northwest states approach Vermont on this measure. Estimate of Baseline CFL Sales Operational Definition. Conceptually, the baseline level of CFL sales is defined as the number of CFLs that would have been sold in Vermont in 2004 if the EVT program were not in operation. For purposes of this analysis, we use the level of CFL sales per household in states that historically have not had vigorous CFL promotion programs as an estimate of the baseline sales. The Estimation Process. We developed the baseline sales estimate through the following steps. Identification of states with programs. KEMA scanned the websites of utilities across the country that had appliance rebate and other residential programs to identify those with CFL promotion programs. The states with programs available to more than half of their population included: California, Oregon, Washington, Idaho, Montana, Vermont, New Hampshire, Massachusetts, Connecticut, New Jersey, and Wisconsin. We also included Long Island as a state in this analysis. 7 6 Itron, Inc. 2005. California Lamp Report 2004. Rosemead, CA: Southern California Edison. 7 We are aware of CFL promotion programs operated by a number of Midwest Utilities including Minnesota Power and Interstate, as well as the program operated by Efficiency Maine. We have not included these among the states with programs because these efforts are relatively new or because they reach only a small percentage of customers in the state. 1 15 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Estimation of CFL sales nationally and in the states with programs. We estimated 2004 CFL sales for the states with programs using the following sources and methods. United States and California. Used estimates from the California Lamp Report 2004 which were developed using scanner data from the five major retail channels. Northwest States: Oregon, Washington, Idaho & Montana. KEMA obtained quarterly reports for the Northwest Energy Efficiency Alliance Energy Star Home Products program covering the period Q1 2002 through Q3 2004. We estimated sales for the full calendar year 2004 on the basis of these data, using the seasonal sales pattern from previous years as a guide. Wisconsin. We used a simple trend of the estimated sales in 2002 and 2003 to estimate sales for 2004. Massachusetts, Connecticut, Rhode Island, New Hampshire, and Long Island Utilities serving these areas all participate in the regional program facilitated by the Northeast Energy Efficiency Partnerships, Inc. (NEEP). EVT participated in this program until the end of 2004, at which point it took over local retailer support operations. EVT continues to participate in other elements of the NEEP lighting program. In 2003, the NEEP utilities (not including Vermont) supported program supported sales of roughly 800,000 CFLs. New Jersey. New Jersey utilities initiated vigorous statewide promotion of CFLs in 2003. In 2004, the statewide manufacturer buydown program supported the sale of 2,000,500 CFLs in New Jersey. Thus, we believe it is justified to include New Jersey among the group of Northeast states with well established programs. We are aware of no sources of CFL sales data for the areas covered by the NEEP program and New Jersey. To address this situation in the analysis we conducted a sensitivity analysis, using a range of assumptions concerning CFL sales per household. At the high end, we assumed that sales per household were considerably higher than they are in Vermont at 1.3. At the lower end, we assumed that the sales per household were 0.3 CFLs per household, well below the level observed in California. As described below, the results of the analysis were not particularly sensitive to variations in this assumption. 1 16 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Calculation of Baseline Sales per Household. We calculated the baseline level of sales using the following formula: Sales b = (Sales US Sales PS ) HH NS Where: Sales b = Baseline sales Sales US = Total national CFL sales, 2004 Sales PS = Total CFL sales, 2004, in states with promotional programs HH NS = Total number of households in states with no programs. Calculation of the Net-to-Gross Ratio. We calculated the Net-to-Gross ratio using the following formula: NTG = (CFL/HH VT Sales b ) * HH VT Program Sales Where: Sales b = Baseline sales CFL/HH VT = CFL sales per household in Vermont, 2004 HH VT = Number of Households in Vermont 2004 Program Sales= Total number of CFLs sold through the program, including coupons, manufacturer buydowns, and catalog sales Results. Table 1-7 displays the results of the baseline sales estimates and Net-to-Gross calculations under the three assumptions concerning sales levels in the NEEP states. Table 1-7 Results of Baseline Sales Estimates and Net-to-Gross Calculations Assumed CFL Sales/HH in NEEP States Estimated Sales in States without Programs Baseline CFL Sales/Household Net-to-Gross Ratio 1.3 8,461,431 0.10 1.36 1.1 10,214,014 0.12 1.33 0.9 11,966,598 0.15 1.30 0.7 13,719,181 0.17 1.27 0.5 15,471,764 0.19 1.24 0.3 17,224,347 0.21 1.22 1 17 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Key conclusions to be drawn from Table 1-7 are as follows. The Net-to-Gross ratio estimated for EVT s promotion of CFLs ranges from 1.22 to 1.36. That is 2004 CFL sales attributable to the program s influence range from 217,088 to 243,844 units versus to the 178,669 units sold or subsidized through the program. The estimate of the Net-to-Gross ratio is not sensitive to large changes in the assumed level of sales in the NEEP states. A change of over 300 percent in the assumed level of sales in the NEEP states and New Jersey results in a change of less than 12 percent in the estimate of the Net-to-Gross ratio. Our estimate of the Net-to-Gross ratio is higher than the planning estimates found in the Technical Resource Manual. The most recent revision to the Residential Lighting Measure resource parameters sets the free ridership adjustment at 6 percent, and the spillover adjustment at 1.25, for a combined multiplier of 1.19 times gross savings. Our midpoint estimate is 9 percent higher than the planning estimate. Interpretation of Results The relatively high Net-to-Gross estimate invites scrutiny if not skepticism. In the first part of this section we review potential threats to the validity of the findings. In the second we assess the consistency of the Net-to-Gross estimate with related findings from this and other studies. Threats to validity. The following paragraphs discuss potential threats to validity as they pertain to the Net-to-Gross analysis. Accuracy of input sales data. In the course of preparing this study we considered purchasing Itron s services to estimate CFL sales using scanner data. 8 We carefully reviewed the methods they and their data providers use to estimate sales in the U.S. and in California and interviewed their project manager regarding potential shortcomings in the data. On the basis of this work, we believe the national sales estimates are accurate within a margin well below 10 percent. The California and Wisconsin estimates require more data manipulation and modeling, but are likely to underestimate actual sales by a 15 20 percent margin, based on comparison of scanner-based estimates to documented program sales. If actual sales in California and Wisconsin are higher than the estimates used here, then the estimates of program net effects presented here would be somewhat low. However, as in the case of the Northeast states, this effect would be relatively modest. The sales estimates for the Northwest are built up through the same process that we used in Vermont. Despite some attrition in retailer participation in the reporting process, the Northwest Alliance program contractors continue to receive documentation of a large portion of total sales. 8 Ultimately we decided not to pursue this method because the sampling procedures used by the data providers were unlikely to support good estimates of sales in small states such as Vermont. 1 18 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Face Validity. Under this rubric we consider whether the indicator we have selected actually measures program influence. From a formal standpoint, we believe our measure of Net-to-Gross does meet this criterion. One question on this point has to do with cumulative effect. Vermont has had programs in more or less continuous operation since 1992. Examination of trends in program volume suggest that these earlier programs had had some effect on the market but that EVT significantly accelerated trends in CFL adoption. Specifically, the bottom row of Table 1-6 displays program volume in terms of CFLs per household. In the year EVT assumed management responsibility, program activity (as opposed to total sales) roughly equaled total sales per household at the national level. With the exception of 2003, program volume was considerably higher than the national benchmark. In 2004, program volume was more than twice the national benchmark (CFLs purchased per household). In the same year, overall national sales fell by over 10 percent. For these reasons, we believe our approach provides a basically fair characterization of program influence. Unaccounted Variables. Many studies have found associations between socioeconomic status, housing characteristics, and age on one hand and adoption of CFLs on the other. Differences among states on those characteristics may have some influence on the outcomes of our analysis. However, there was no effective method of controlling for these influences, given the general absence of sales data at the state level. Consistency with other Vermont Findings. As discussed earlier, the RASS telephone and onsite surveys captured information on customers awareness, purchase, and holdings of CFLs. The following paragraphs report findings relevant to assessing program effects. Awareness of CFLs. Awareness of CFLs is very high in Vermont. Nearly 71 percent of the 600 customers surveyed reported that they were aware of compact fluorescent bulbs. An additional 20 percent reported that they were aware of the product after hearing a description. There was little variation between regions in this finding. Purchases and program participation. Table 1-8 displays results from the RASS telephone survey in regard to CFL purchases and use of rebates. Table 1-8 Results of Vermont RASS: CFL Purchase and Program Participation Purchased CFLs in Past 2 Years Used a Rebate to Purchase CFLs Would have Purchased Without a Rebate Yes 33.7% 12.2% 72.6% No 64.8% 83.8% 21.9% DK 1.5% 4.0% 5.5% n = 600 200 73 Volume of purchases. Reported purchases of CFLs are consistent with the sales estimate developed above. Roughly one-third of RASS respondents reported purchasing CFLs in the past two years. Those respondents to the RASS on-site surveys who purchased CFLs 1 19 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT in the past two years reported that they had purchased an average of 7.9 bulbs. Customers using instant rebate coupons purchased an average of 5.2 lamps with the rebates. If we assume that the typical CFL purchaser bought 6 bulbs, we come up with an estimate, based on the survey data, of roughly 253,000 bulbs purchased per year. Using sales data, we estimated 2004 purchases at 271,170 units, which included 25,000 35,000 bulbs purchased by businesses. Program participation. The RASS survey estimated the level of program participation fairly accurately. According to the KITT database, roughly 40,000 customers residential or 16 percent of all households used instant rebate coupons over the past 2 years. By way of contrast 12.2 (+/- 2.2) percent of respondents reported participating. Because it takes relatively little effort to use the instant rebate coupons, it is possible that some respondents simply did not remember using them. Likelihood of purchase without the rebate. Nearly 73 (+/- 8.6) percent of the respondents who reported using rebates to purchase CFLs opined that they would have purchased those units without the rebate. At first glance this result appears inconsistent with the high Net-to-Gross ratio computed using sales data. However, we need to remember that the Net-to-Gross ratio we developed reflects the cumulative effects of the program over a period of five to ten years. According to our estimates 34 percent of all CFLs sold in 2004 were purchased without the use of coupons. This suggests that there is a relatively large group of customers who accept the value of CFLs and do not require financial incentives to purchase them. This group is sufficiently large to drive the high level of observed purchases. We believe the high level of reported free ridership reflects the strong segmentation of the customer market between customers who have and who have not purchased CFLs. We discuss findings on segmentation from this and other studies below. 1 20 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT 1.4 SEGMENTATION OF THE RESIDENTIAL CFL MARKET 1.4.1 Overview Recently, energy efficiency program sponsors in a number of jurisdictions have undertaken market and evaluation studies for programs that promote CFLs. These studies have yielded remarkably similar estimates of the extent of customer awareness and adoption of CFLs. Table 1-9 summarizes the results from relevant items in the Residential Appliance Saturation Study (RASS) conducted for this evaluation, along with similar results from studies conducted in California, Massachusetts, and the Pacific Northwest. Table 1-9 Key CFL Market Segmentation Indicators (self-reported): Vermont and Other States with Programs Vermont California 9 Massachusetts 10 4 NW States 11 Year of Survey 2005 2003 2004 2004 % Aware of CFLs (no description) 69% 45% % Aware of CFLs (with description) 91% 82% 71% 68% % with CFLs in home or ever purchased 54% 56% 43% 49% Purchased CFLs in Past 2 Years 34% 32% n/a 34% Mean # installed (among those with bulbs) 5.9 7.4 6.5 The key points illustrated in Table 1-9 are as follows. Customer Awareness. The percentage of customers who report being aware of Compact Fluorescent Lamps (after hearing a description of them) ranged from 68 to 91 percent, with Vermont recording the highest level of prompted awareness. The percentage of customers who report being aware of CFLs has risen slowly over the past five to seven years. For example, a survey conducted by New England Electric System in 1998 found that 70 percent of respondents were aware of CFLs. Ownership of CFLs. The percentage of customers who reported having purchased at least one CFL in the past or, alternatively, reported having one installed ranged from 43 to 56 percent, with Vermont recording a level of 54 percent. The percentage of 9 KEMA, Inc. 2003. Evaluation of the 2002Statewide Crosscutting Residential Lighting Program. San Diego: San Diego Gas & Electric Company. 10 Megdal Associates & Opinion Dynamics. 2003. Evaluation of the Massachusetts ENERGY STAR Residential Lighting Program. Westborough, MA: National Grid USA. 11 KEMA, Inc. 2005. 2004-2005 Consumer Products Program Evaluation: Market Update Report. Portland, OR: Northwest Energy Efficiency Alliance. 1 21 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT customers with CFLs installed in their homes appears to have increased steadily over the past five years after holding relatively constant around 30 35 percent during the nineties. 12 Annual Purchases. The percentage of customers who reported purchasing CFLs in the past two years ranged from 32 to 34 percent, a remarkably narrow clustering. This market indicator also appears to have increased in areas with vigorous promotion projects. Studies conducted in the late 1990s typically yielded results in the 18 25 percent range for this indicator. Number of bulbs purchased. Customers in areas with strong promotional programs generally purchase 5 7 bulbs when they purchase them at all. This pattern has been present in virtually all studies of CFL promotions going back many years. It is likely related to program rules that limit the number of coupons redeemable at one time. 1.4.2 Factors Associated with CFL Awareness and Purchase KEMA analyzed the association between respondents reported CFL awareness and purchase behavior on the one hand and attributes that might affect such behavior: homeownership, income, education, and regional location. Briefly, we found that customers with college educations or above were more likely to be aware of and to have purchased CFLs. However, this association was not particularly strong as measured by the Chi Square test and other statistical indicators. There were no clear patterns of association between awareness or purchase on one hand and customer income, homeownership, or regional location. KEMA recently conducted a study for the Northwest Energy Efficiency Alliance in which we used non-linear regression modeling to analyze segmentation of the CFL market. We arrived at similar findings. Awareness and purchase were found to be most strongly associated with education. Other factors with weaker associations included home ownership, frequency of shopping at home centers and mass marketers where CFLs were heavily promoted, and location in areas in which media campaigns had been run. A recent study in Massachusetts found a statistical association between male purchasers and the likelihood of purchasing CFLs. The data we collected for this study did not support examination of that particular association. 1.4.3 Remaining Market Potential The on-site portion of the RASS contained an inventory of all lighting fixtures in the sample homes. We used the data generated by this lighting inventory to assess the extent of the remaining market for screw-in CFLs. Specifically, the RASS collected information on the following: Total number of lighting fixtures in the home; 12 See XENERGY Inc. 2001. Starlights Program Evaluation. Consortium of New England Utilities for a review of findings from older studies. 1 22 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT Number of fixtures currently fitted with incandescent, compact fluorescent, and tube fluorescent bulbs; The number of sockets that could be retrofitted with CFLs. These latter included recessed cans. The criteria applied here included the size and shape of the fixture and whether it required specialty type bulbs (in chandeliers for example). Table 1-10 summarizes these results. On average, we found that the full sample of respondent homes could accommodate an additional 24.6 CFLs. Among customers who already had one or more CFL installed, the corresponding figure was 20.3. We thus conclude that, even among current users of CFLs, there is still substantial room for further purchases. Table 1-10 Results of On-Site Surveys: Lighting Fixture & Bulb Inventory Full Sample n = 83 At Least 1 CFL n = 52 No CFLs n = 31 Number of Fixtures 36.6 35.7 38.3 Number of Bulbs 49.7 48.0 52.5 Number of Fluorescent Tubes 6.3 5.9 7.0 Number of CFL Bulbs Installed 4.7 7.5 0.0 Number of Incandescent/Other Bulbs 38.7 34.6 45.5 Number of Bulbs w/ Potential CFL Retrofit 24.6 20.3 31.8 1.5 INSTALLATION RATES DPS requested that KEMA assess the hypothesis that the portion of CFLs that customers held in storage varies directly with the number of bulbs they purchase. If this hypothesis were true, it might be possible to increase the installation rate of bulbs purchased through the program (and thus net program savings) by more vigorously enforcing rules about how many bulbs individual customers may purchase. KEMA included a number of items in the on-site portion of the RASS to assess this hypothesis. Specifically, we asked customers who reported that they had purchased CFLs in the last two years how many CFLs they purchased and how many were currently in storage. The key results from this line of inquiry were as follows. Installation rate. We found that the 52 respondents who reported purchasing CFLs in the past two years had bought a total of 413 units, of which 122 or 27 percent were in storage. This finding is very close to the planning assumption of a 75 percent installation rate used in estimating program savings. Relationship between number of bulbs installed and stored. KEMA estimated the correlation coefficient between the number of bulbs purchased and the percentage of 1 23 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT bulbs installed at the customer level. The resulting coefficient was ρ=.09, which indicates that there was no statistical correlation between the two variables. Figure 1-1 shows the two-way plot of this data. Figure 1-1 Plot of CFLs Purchased v. Percent Stored: n = 52 CFLs # Purchased and % Stored 100% 75% % Stored 50% 25% 0% 0 5 10 15 20 25 # Purchased 1.6 CONCLUSIONS AND RECOMMENDATIONS 1.6.1 Key Findings EVT s CFL promotion program showed strongly improved results in 2004. The number of CFLs purchased through the program through the combined coupon, ITP, and catalog channels grew to 178,669 in 2004 from 72,791 in 2003, and increase of 144 percent in one year. Over the same time period, sales of CFLs in the U. S. as a whole decreased by 6 percent. The recently introduced manufacturer buydown initiative (ITP) contributed significantly to overall program growth. The manufacturer buydown or ITP, which was implemented on a full scale in 2004, yielded sales of 34,430 CFLs. That is 19 percent of total program volume and 32 percent of the increase in volume from 2003 to 2004. Vermont recorded the highest level of CFL sales per household of any state for which sales data (as opposed to program activity records) were available. Using a combination of program records, retailer survey data, and retailer sales records collected by EVT, KEMA was able to develop a robust estimate of total CFL sales in the state in 2004 of 1 24 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT 271,170 units or 1.08 units per household. Among areas with active promotional programs, the Pacific Northwest recorded sales of 1.01 units per household, estimated using similar methods. Analysis of point-of-sale scanner data conducted for California utilities found that 2004 sales in California totaled 0.43 units per household and 0.29 units at the national level. The Net-to-Gross ratio estimated for EVT s promotion of CFLs ranges from 1.22 to 1.36. The Net-to-Gross ratio estimated for EVT s promotion of CFLs ranges from 1.22 to 1.36. That is, 2004 CFL sales attributable to the program s influence range from 217,088 to 243,844 units versus the 178,669 units sold or subsidized through the program. This estimate was not sensitive to wide variations in the estimate of key input variables. It is also significantly higher than the current estimate of 1.19 used for planning and savings tracking purposes. Awareness of CFLs among Vermont households is very high. Ninety-one percent of RASS respondents reported being aware of CFLs once the product was described to them. Among four regions with strong programs and recent studies, Vermont recorded the highest fraction of aware customers. Over one-half of Vermont households currently have CFLs installed in their home. Fifty-four percent of RASS respondents reported having at least one CFL installed in their home. The percentage of customers with CFLs installed in their homes appears to have increased steadily over the past five years. Among those with CFLs installed, the average number found in their home during the on-site portion of the RASS was 5.9. Roughly one-third of Vermonters reported purchasing CFLs in the past two years. These results, combined with the reported number purchased yielded an estimate of total purchases that was within 10 percent of the estimate generated through analysis of sales figures. Segmentation in the CFL market persists but is not strongly related to customer demographic or housing attributes. Despite over a decade of strong utility promotion and sharp declines in retail prices, only half of Vermont customers (and fewer elsewhere) have CFLs in their homes. Progress in opening the other half of the market would be greatly supported by a consistent portrait of the customers who have not so far purchased CFLs. However, we found that the difference between purchasers and non-purchasers in terms of customer attributes is not particularly clear. Among the variables examined, only education showed a strong association with CFL awareness and purchase. This was not the case for home ownership, income, or regional location. At the household level, installation rate was not strongly related to the number of CFLs purchased. Thus, we do not believe that attempts to more closely enforce unit limits for individual customers will yield higher net program savings. Among program purchasers in the on-site sample we found the installation rate to be 73 percent, very close to the 75 percent figure used for planning and program tracking. 1 25 VT DPS

SECTION 1 LIGHTING PROGRAM ASSESSMENT 1.6.2 Recommendations The findings of this evaluation clearly indicate that EVT has done a very good job with the lighting program, particularly in the past two years. KEMA suggests the following steps to sustain and enhance the progress that has been made. Expand the ITP manufacturer buydown component of the program. A number of findings and observations suggest that EVT should put business development effort into expanding the ITP component, in terms of the number of participating retailers, the types of retailers involved, and the volume of units supported. These include the following: The full implementation of the component appears to have contributed to the rapid increase in the number of CFLs supported by the program between 2003 and 2004. The per unit costs for the ITP were $2.11 in 2004 versus $4.43 for coupon sales. Further expansion of the segment of CFL users may depend on diversifying the kinds of retail venues in which the product is found. Specifically, there are many customers who rarely if ever visit hardware stores or home centers, the two most important channels in Vermont. The ITP approach appears to be more acceptable to mass merchandisers than coupons and may constitute a useful vehicle for getting CFLs into more grocery and drug stores. Require that EVT attempt to collect CFL sales information from all participating retailers. EVT s success in obtaining the sales data used for this evaluation and the usefulness of the analysis these data supported suggests that the effort should be incorporated into ongoing operations. If manufacturer buydowns grow as a portion of total program volume, analysis of participants identified by coupons will provide progressively less insight into the effect of the program. 1 26 VT DPS

2 APPLIANCE PROGRAM ASSESSMENT 2 APPLIANCE PROGRAM ASSESSMENT 2.1 INTRODUCTION The appliance component of the Efficient Products Program (EPP) was designed to encourage and facilitate the purchase of ENERGY STAR qualified clothes washers, refrigerators, dishwashers, and room air conditioners. In order to accomplish this, EVT performs the following key program activities. Retailer promotion and merchandising support. EVT solicits retailers and supports their participation in the program by providing point-of-purchase materials, training to sales staff in program operations and product attributes, and rebate processing. From program inception through December 2004, Applied Proactive Technologies, Inc had fulfilled this function. Beginning in January 2005, EVT staff began carrying out the retail representation role. EVT is currently working with a number of retailers to develop special ENERGY STAR zones within their stores that bring together all ENERGY STAR products in to a unified merchandising environment. This work is partially funded by the U.S. Environmental Protection Agency. Customer Rebates. EVT mail-in customer rebates are available for qualifying clothes washers. The current rebate is $50 per unit. EVT also has operated short-term rebate promotions for qualifying refrigerators, freezers, and room air conditioners. Dealer Incentives. The program also paid dealers small incentives (spiffs) per qualified refrigerator and freezer sold. Table 2-1 below shows the number of retailers participating in the Appliance Component of the EPP, along with the number of units rebated. Table 2-1 Summary of EPP Appliance Component Activity Year Retailers Enrolled Clothes Washers Refrigerators Room Air Conditioners 2001 N/A 2,719 0 0 2002 70 2,772 546 1,413 2003 58 3,813 1,125 3,773 2004 75 4,174 1,137 2,233 Total 12,300 2,808 7,419 2 1 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT In terms of market share, appliances rebated through the program accounted for 33 percent of all clothes washers sold in Vermont in 2004, 10 percent of all refrigerators, and 19 percent of all room air conditioners. 2.1.1 Objectives The EPP Appliance Analysis Section has two primary objectives: Market Share Estimates. Estimate state-level market share for 4 ENERGY STAR appliances clothes washers, refrigerators, dishwashers, and room air conditioners. Program Effects. Assess the effects of EVT programs on market shares for the years 2002, 2003 and 2004. 2.1.2 Data Collection and Analysis KEMA used the following data sources in compiling estimates of appliance sales and ENERGY STAR market share. KITT Database. KEMA used the KITT database provided by EVT to develop estimates of the total number of clothes washers for which incentives were paid each year. Monthly store sales records collected by EVT. Since taking over program support activities from Applied Proactive Technologies, EVT has continued to collect monthly totals of clothes washer unit sales for ENERGY STAR and conventional models for a subset of independent retail stores participating in the program. In 2002 records were collected from 48 stores, in 2003 records were collected from 49 stores, and in 2004 records were collected from 46 stores. These were primarily independent appliance stores that accounted for roughly 50 percent of program rebates from 2002 to 2004. National ENERGY STAR Appliance Sales Tracking System. Through its national ENERGY STAR Appliance program support contractor, D&R International, the U.S. Department of Energy collects sales information from six national appliance and department store chains. Only one of these chains Sears has locations in Vermont. Due to the sensitivity of the information collected, and the limited number of establishments reporting, the DOE system only reports the market share for ENERGY STAR appliances and not the total number of units sold. D&R International also furnished estimates of the percentage of AHAM shipments accounted for by total unit sales recorded in their sales database by state. We used this factor to develop a weighted average ENERGY STAR market share that reflected the relative proportion of total sales accounted for by chains (DOE tracking system) and the independents (KEMA data collection analysis). Association of Home Appliance Manufacturers (AHAM) distributor sales estimates. AHAM is the industry association of home appliance manufacturers. It collects data on its members shipments to distributors by appliance type, state and ENERGY STAR designation. The 2 2 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT correspondence between total units shipped and sales at the state level appears to be fairly close, but not exact. There is some lag between the time units are shipped to retailers warehouses and the time they are actually sold. Also, the use of regional wholesale buyers and warehouse facilities requires AHAM to make source estimates in allocating shipments to specific states. Thus, the unit shipments reported by AHAM may slightly under or overstate sales in a given year. KEMA analysis of appliance sales data from independent retailers in Vermont. As part of the evaluation research, KEMA collected and analyzed monthly sales records from nine independent retail stores in Vermont. These stores furnished records of over 10,000 transactions dating from 2002 through 2004. The records included model numbers and dates for each transaction. KEMA quality controlled all model numbers to ensure that they appeared in manufacturers catalogs and that the ENERGY STAR designation was correct. The main objective of this data collection effort was to compile information on ENERGY STAR market share for clothes washers, refrigerators, dishwashers, and room air conditioners from stores that were not covered by the DOE national effort. Sample retailers sold 44 percent of all clothes washers sold by independents in the state, 32 percent of all refrigerators, 18 percent of all dishwashers, and 8 percent of all room air conditioners. 2.1.3 Key Findings and Recommendations The following paragraphs summarize key findings and recommendations from the EPP Appliance Program analysis. Findings Vermont continues to have high ENERGY STAR market shares for all four appliances. When compared to national and regional averages, Vermont s ENERGY STAR market share in chain stores for each of the four major appliance categories is consistently among the highest estimated levels. For example, Vermont chain stores in 2004 had the highest estimated market share for clothes washer sales of any state in the U.S: 49 percent versus 27 percent for the U. S. as a whole. Furthermore, only New Hampshire had a higher estimated market share for refrigerators and room air conditioners that same year. Dishwasher sales were fourth highest, slightly behind Minnesota, Delaware and Connecticut. Differences between chain and independent retailers in ENERGY STAR market share. There are pronounced differences between chain and independent retailers in ENERGY STAR market share, but these differences are not consistent across various appliance types. For example, chain stores sold a much higher proportion of ENERGY STAR refrigerators than independent locations, but had very similar market shares for clothes washer sales. These differences in market shares between independents and chain stores result from the systematic pursuit of different merchandising strategies. According to representatives of manufacturers and other individuals familiar with appliance retailing, chain stores make best use of their size and buying power by attempting to sell to the 2 3 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT broad middle of the market. Effectively, this means displaying a large number of units that cover as many price/feature/efficiency combinations as they can, while avoiding very high and very low end products. Independents serve a smaller area and have limited display space. They therefore display (and sell) models that they believe will appeal to their customers, who generally occupy a narrower economic niche (high or low) than chain store customers. The net effects of the program on the market share of ENERGY STAR clothes washers sold in a given year have declined significantly in recent years. In 2001, the net unit sales attributable to the program were estimated at 1,045 versus total program-subsidized sales of 2,719 for a net-to-gross ratio of 0.38. In 2004, the maximum number of net unit sales attributable to the program was 724 versus program volume of 4,179 units, for a net-to-gross ratio of 0.17. This is a maximum estimate that reflects the inclusion in the analysis of a variable that captures the effect of past program efforts as well as other factors that accounted for past growth in Vermont s ENERGY STAR clothes washer market share. Findings of decreased net effect in the current year are consistent with recent increases in national ENERGY STAR appliance market shares, as well as with other regional studies. These findings are consistent with a similar analysis conducted in 2003 for a consortium of Massachusetts utilities, and with the overall perceptions of appliance program managers around the country. The weight of evidence suggests that EVT s appliance program and its utility-based predecessors, as well as other long-standing utility programs had a profound impact on the national market success of energy-efficient clothes washers. A study commissioned by the Consortium for Energy Efficiency (CEE) found that utility energy efficiency program managers, primarily from California and the Northwest states, played a crucial role in initiating the development of the infrastructure required for effective market transformation programs: contacts with manufacturers, technology assessments, and common equipment specifications. 1 These efforts were organized on the national level by CEE, beginning informally in 1992. The Vermont utilities joined the initiative in 1996. The CEE specification, already used by 30 electric and water utilities at the time, provided the basis for the ENERGY STAR specification. This in turn supported a major national effort involving manufacturers and over 200 local program sponsors by the year 2000. Since then the national market share of ENERGY STAR clothes washers has grown from 9 percent to 27 percent (2004). In 2007, the federal minimum standard will be revised to match the current ENERGY STAR specification. As the national market acceptance of ENERGY STAR washers grows and the effective date of the new national standard nears, it makes sense that the apparent effect of local incentive programs such as Vermont s will decrease. 1 Feldman, Shel, Mitchell Rosenberg, Jane Peters. 2001. The Residential Clothes Washer Initiative: A Case Study of a Collaborative Effort to Transform a Market. Boston: Consortium for Energy Efficiency. 2 4 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Recommendations In the short run, we believe that EVT should continue to administer rebates for ENERGY STAR clothes washers. This is necessary to maintain good relationships with retailer channels that have been built up over many years and to prevent potential selloffs of non-qualifying models prior to implementation of the next round of federal minimum energy efficiency standards in 2007. Assess and implement changes to the program that can increase cost effectiveness. These expedients may include: Reduce incentive amounts. Limit the duration of appliance incentive promotions to specific months. This approach has been tried in the Pacific Northwest. Market share of ENERGY STAR appliances remain high in that region. Restrict eligibility to models that qualify for the Consortium for Energy Efficiency s Tier 3 standards. In the longer term it may advance efficiency goals to redirect incentive and business development resources away from appliance promotions to other opportunities. EVT should initiate discussions with appliance retailers to develop procedures for an exit strategy. 2.2 PROGRAM ACTIVITY SUMMARY 2.2.1 Objectives The analysis of the clothes washer rebate database (KITT) has several objectives, including: Analyze the distribution of rebates by stores. This analysis highlights the types of stores that are participating in the program and which rebate the most qualifying clothes washers. Assess trends in customer participation over time. In order to assess the trends in customer participation over time, rebate volumes were compared for the years 2002, 2003 and 2004. 2.2.2 Key Findings The following paragraphs summarize the key findings from the clothes washer rebate distribution. Sears and independent retail appliance stores dominate rebate distribution. Sears and independent appliance stores together accounted for over 80 percent of all rebates 2 5 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT submitted, or 54 and 29 percent respectively. The twelve Sears stores average over five times as many rebates per store as do the 32 independent appliance retailers. Clothes washer rebates continue to grow. The volume of rebates processed in 2004 was 53 percent higher than those processed in 2002. 2003 experienced a 38 percent increase in rebate volume over 2002. Distributions by store type have remained steady. Volumes of rebates submitted by the different store types have remained steady over the past three years. Only minor increases or decreases (less than 5 percent) have been experienced by any of the five major categories since 2002. 2.2.3 Patterns of Customer Participation Table 2-2 shows the number of clothes washers rebated by the program between 2002 and 2004. The annual number of units rebated steadily increased from 2,772 in 2002 to 4,174 in 2004. In 2004, Vermonters purchased approximately 12,700 clothes washers. Thus, participants in the program accounted for nearly one-third of all clothes washers purchased in the state. Table 2-2 Summary of EPP Clothes Washer Component Activity, 2002-2004 Year Retailers Participating Participant Rebates Non Participant Rebates 2 2002 70 2,699 73 2003 58 3,727 86 2004 75 4,129 45 Total 10,555 204 2.2.4 Rebates by Store Type In order to determine the volume of rebates processed by stores of different types, stores listed in the KITT database were grouped into one of five major categories: Sears, Independent Appliance Store, Service Contractor, Major Appliance/ Home Center, and Other. 2 Retailers in this category have not signed on to participate in the program and are therefore not listed in EVT s participant database. 2 6 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Sears stores were given their own category because they are a major channel for appliance sales in Vermont and represent over 50 percent of the units rebated between 2002 and 2004. The independent appliance store category includes participating independent appliance retailers with one of more locations in Vermont, such as Bouchard-Pierce Appliances and Bailey s Maytag. Service contractors include hardware stores that sell and service various appliances (such as Bisbee s Hardware) as well as gas, oil, and electrical contactors who sell and/or service appliances (such as Dundon Plumbing & Heating and Northern Petroleum). The major appliance/ home center category includes stores such as Home Depot, Best Buy, Home & Farm Center, and Total Home Center. The other category contains all appliance retailers that do not fall into any of the previous categories, including furniture stores and miscellaneous others. Table 2-3 summarizes program participation by store type between the years 2002 and 2004. Over this three-year period the 12 Sears locations in Vermont and New Hampshire accounted for 54 percent of all units rebated through the program. Independent appliance stores were the largest group in the database (32 stores), but only represented 29 percent of clothes washer rebates. The remaining three categories service contractors, home centers, and others accounted for only 17 percent of units rebated through the program. Sears locations averaged 479 rebates per store; major appliance/home centers averaged 152; independent appliance stores averaged 95; service contractors 24; and others 19. We should note that all but 8 of the independent appliance stores listed in the Dun & Bradstreet database participated in the program. All Vermont stores in the other major categories participated. Table 2-3 Summary of Participating Locations and Rebates Processed by Store Type, 2002 2004 Stores Rebates Store Type Number % of Total Number % of Total Average # Rebates per Store Sears (8 VT, 4 NH) 12 15% 5,748 54% 479 Independent Appliance Store 32 39% 3,033 29% 95 Major Appliance/Home Center 7 9% 1,066 10% 152 Service Contractor 22 27% 535 5% 24 Other 9 11% 173 2% 19 Total 82 100% 10,555 100% 129 Trends in distribution of program sales by store type over time. Table 2-4 shows the number and percentage of clothes washer rebates processed for each of the five major store types by year. It is clearly evident from this table that Sears and independent appliance retailers dominate 2 7 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT rebate distribution in the state over 80 percent each year. This table also shows that the annual distribution of rebates by store type has remained fairly constant over the past three years. Table 2-4 Number of Rebates Processed by Store Type and Year 2002 (n=70) 2003 (n=58) 2004 (n=75) Store Type # Rebates % of Total # Rebates % of Total # Rebates % of Total Sears (8 VT, 4 NH) 1,545 57% 1,985 53% 2,218 54% Independent Appliance Store 758 28% 1,085 29% 1,190 29% Major Appliance/Home Center 200 7% 414 11% 452 11% Service Contractor 153 6% 184 5% 198 5% Other 43 2% 59 2% 71 2% Total 2,699 100% 3,727 100% 4,129 100% 2.3 ESTIMATION OF APPLIANCE SALES AND ENERGY STAR MARKET SHARE 2.3.1 Summary Comparison of Market Shares. Table 2-5 displays the comparison of ENERGY STAR market shares by chain and independent stores in Vermont for clothes washers, refrigerators, dishwashers, and room air conditioners. It is clear from this table that the direction of differences between the chain and independent store market shares for each appliance are quite different. 2 8 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Table 2-5 Comparison of Market Shares: Chain vs. Independent Clothes Washers Year 2001 2002 2003 2004 Vermont Chain (DOE) 25.0% 34.0% 53.0% 49.0% Vermont Independent 37.0% 37.0% 47.0% 51.0% Weighted Average 32.3% 35.7% 49.5% 49.6% Refrigerators Vermont Chain (DOE) 14.9% 24.8% 33.1% 46.0% Vermont Independent 8.0% 4.8% 7.3% 26.2% Weighted Average 11.2% 17.3% 27.4% 36.1% Dishwashers Vermont Chain (DOE) 14.8% 27.5% 54.2% 86.3% Vermont Independent 64.0% 64.5% 77.4% 86.6% Weighted Average 45.6% 45.5% 65.1% 86.4% Air Conditioners Vermont Chain (DOE) 19.8% 61.3% 63.7% 61.4% Vermont Independent 3.0% 34.5% 77.7% 70.9% Weighted Average 10.7% 57.3% 70.6% 67.8% The following sections describe specific results for each appliance. 2.3.2 Clothes Washer Sales and ENERGY STAR Market Share Table 2-6 summarizes estimates of clothes washer unit sales, ENERGY STAR market share, rebate totals, and unrebated sales, for 2002, 2003, and 2004. 2 9 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Table 2-6 Clothes Washer Unit Sales and ENERGY STAR Market Share Year 2000 2001 3 2002 2003 2004 AHAM Distributor Sales Estimates 12,500 9,375 11,700 12,200 12,700 EVT Monthly Sales Reports: Participating Independent Retailers ENERGY STAR Clothes Washer Unit Sales 2,201 2,121 2,407 3,039 2,979 Standard Efficiency Clothes Washer Sales 4,933 3,623 4,020 3,477 2,920 Total Independent Retailer Unit Sales 7,134 5,744 6,427 6,516 5,899 ENERGY STAR as % of Total Independent Sales 31% 37% 37% 47% 51% National Chain Sales as Remainder from AHAM Distributor Sales Estimates Total National Chain Sales as Remainder 5,366 3,631 5,273 5,684 6,801 % ENERGY STAR Unit Sales (D&R Estimate) 23% 25% 34% 53% 49% National Chain ENERGY STAR Unit Sales 1,213 911 1,766 3,001 3,326 ENERGY STAR Unit Sales and Market Share Total ENERGY STAR Unit Sales 3,414 3,032 4,173 6,040 6,305 VT ENERGY STAR Market Share Weighted Average 27.3% 32.3% 35.7% 49.5% 49.6% U.S ENERGY STAR Market Share (per D&R/DOE) 9.3% 10.3% 16.0% 23.1% 27.2% Sales of ENERGY STAR Clothes Washers Outside the Program Units for Which Incentives were Paid (Participant Retailers) 2,476 1,922 2,699 3,727 4,129 Rebates Paid to Non Participant Retailers N/A N/A 73 86 45 ENERGY STAR Sales without Rebates 938 1,110 1,401 2,227 2,131 % of ENERGY STAR Sales without Rebates 27% 37% 34% 37% 34% The key findings to be derived from Table 2-6 are as follows. Over time, ENERGY STAR market share for chain stores has caught up with independent market shares. The 2001 study found that independent stores averaged 30 to 40 percent above market shares realized in national chains. This analysis shows how national chain market shares for ENERGY STAR clothes washers have caught up with independent store market shares. In 2000, national chain ENERGY STAR market share, as estimated by D&R, was 23 percent. In 2002 that percent had risen to 34 percent, and in 2003 to over 50 percent. This result reflects the general upward trend in market share for ENERGY STAR clothes washers. Independent market share steadily increases from 2000 to 2004 with major jump from 2002 to 2003. Independent retail store market shares for clothes washers rose from 3 2001 estimates only represent data collected between January 2001 and October 2001. November and December data were not available. 2 10 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT an estimated 31 percent in 2000, to 51 percent in 2004. There was a ten percent jump in market share (37 percent to 47 percent) between 2002 and 2003. Throughout the period covered, the share of ENERGY STAR washers sold in Vermont continued to be dramatically higher than the national average. Vermont s ENERGY STAR market share grew from almost 36 percent in 2002 to almost 50 percent in 2004. That is nearly twice the national market share. Percent of ENERGY STAR units sold without rebates held fairly steady between 2001 and 2004. The percent of total ENERGY STAR Clothes washer sales that were not supported by rebates grew ten percent between 2000 and 2001 (27 percent to 37 percent), but remained fairly steady from 2001 to 2004. To put the trend of Vermont s ENERGY STAR clothes washer market share in perspective, we grouped states into two sets: those that had active incentive programs to support ENERGY STAR washer sales during 2002 to 2004; and those that did not. The states that had active incentive programs were: California, Connecticut, Indiana, Illinois, Iowa, Maine, Maryland, Massachusetts, Michigan, Minnesota, Montana, New Hampshire, New Jersey, New York, Ohio, Oregon, Rhode Island, Vermont, Washington and Wisconsin. We relied on program reports to DOE to characterize the states for program purposes. We then compiled the annual ENERGY STAR clothes washer market share figures for each of the states from the DOE database for the years 2001 to 2004. Table 2-7 summarizes the comparison of the market shares for Vermont and the two groups of states. 2 11 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Table 2-7 Comparison of ENERGY STAR Clothes Washer Market Share (Retailers reporting to DOE Market Tracking System Only) Year 2001 2002 2003 2004 Vermont Market Share ENERGY STAR 25.1% 33.5% 52.8% 48.9% Other NEEP States Connecticut 20.0% 25.5% 36.7% 43.5% Massachusetts 20.0% 24.9% 39.5% 39.4% New Hampshire 19.5% 30.4% 50.8% 48.2% States With Incentives: ENERGY STAR Market Share Median 20.0% 25.7% 34.9% 35.7% Maximum 26.9% 45.0% 52.8% 48.9% Minimum 12.5% 22.8% 19.6% 23.6% States with No Promotions or Incentives Median 10.2% 14.3% 19.8% 24.1% Maximum 4 33.6% 45.0% 51.5% 39.4% Minimum 3.4% 6.6% 9.0% 14.6% US Average 10.3% 16.3% 23.1% 27.2% The key findings illustrated in Table 2-7 are as follows. Even among states that have strong incentive programs, Vermont continues to have the highest market share of ENERGY STAR washers. Only Oregon in 2002 had a higher market share than Vermont over the three-year period covered in the analysis. Comparison to no-program states. The median market share among the 30 states with no program could be viewed as an approximate baseline market share in the absence of any programs. From 2002 to 2004, the difference between Vermont s market share and that of the no-program states was between 20 and 30 percentage points. That is, Vermont s ENERGY STAR market share was more than twice that of the median value for the non-program states. While sales of ENERGY STAR qualifying models continues to increase across the country, Vermont continues to have a market share nearly twice the national average. Patterns of market share growth. Even though Vermont continues to have the highest market share of ENERGY STAR clothes washers in the country, other states, particularly those with no program incentives, have recently experienced substantially higher growth 4 Alaska had the highest market share for no program states. 2 12 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT rates (versus a 2000 base year), and are catching up with growth in ENERGY STAR market share in the U.S. as a whole. 2.3.3 Refrigerator Sales and ENERGY STAR Market Share Comparison of Market Shares. Table 2-8 displays the comparison of refrigerator ENERGY STAR market shares by Vermont, other NEEP states, and states with and without program promotions. Table 2-8 Refrigerator Shipments and ENERGY STAR Market Share Year 2001 2002 2003 2004 MARKET SHARE ENERGY STAR (CHAIN RETAILERS TRACKED BY DOE) Vermont 14.9% 24.8% 33.1% 46.0% Other NEEP States Connecticut 20.0% 25.9% 32.9% 41.8% Massachusetts 20.0% 25.4% 31.4% 39.3% New Hampshire 19.5% 26.8% 38.8% 48.7% States with Promotion Programs Median 16.6% 22.9% 29.5% 38.7% Maximum 25.5% 26.8% 38.8% 48.7% Minimum 12.4% 18.5% 21.1% 29.9% States with No Programs Median 15.9% 18.3% 23.1% 31.0% Maximum 20.8% 27.0% 32.9% 43.9% Minimum 7.9% 12.7% 17.2% 25.4% US Average 17.3% 20.7% 25.7% 33.2% ESTIMATE OF VT MARKET SHARE: ALL RETAILERS VT: Independents, n=9 8.0% 4.8% 7.3% 26.2% VT: Weighted Average 11.2% 17.3% 27.4% 36.1% Annual Shipments (AHAM) 13,100 13,300 11,900 11,800 The key findings to be drawn from Table 2-8 are as follows. Changes in standards have initial impact on market share. Following changes in specifications (January 2001) the market share in Vermont for ENERGY STAR qualified 2 13 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT refrigerators dropped from over 21 percent in 2000 to 11 percent in 2001 5. However, in less than three years, Vermont s ENERGY STAR market share for refrigerators has rebounded back - exceeding both previous highs and national levels. 2004 estimated market share in Vermont was over 36 percent. Comparison of chain to independent retailers. The estimated market share of ENERGY STAR refrigerators sold by chain stores in Vermont is substantially higher than that for independents. One possible reason is that refrigerators differ from dishwashers and clothes washers in that there are feature laden, expensive models that do not qualify for the ENERGY STAR label. In fact, some very high-end features, such as cabinet depth design, use of two compressors, ice and water dispensers make it difficult for models to meet ENERGY STAR qualifications. Thus, retailers who serve very high-end markets may not push ENERGY STAR models. Similarly, independents serving lower income areas will cover much lower-end models, which also do not meet ENERGY STAR standards. Chain stores have more space to stock ENERGY STAR models and do not run the risk of losing sales by doing so. Comparison to other NEEP states. Unlike Vermont s position in ENERGY STAR clothes washer market share, ENERGY STAR refrigerator market share by chain stores is similar to those for the other three NEEP states. Comparison to national trends. The market share of ENERGY STAR refrigerators has roughly tripled since the introduction of new Federal standards and ENERGY STAR specifications, growing from 15 to 46 percent. By comparison, the national market share doubled over the same period. 2.3.4 Dishwashers Sales and ENERGY STAR Market Share Comparison of Market Shares. Table 2-9 displays the comparison of dishwasher ENERGY STAR market shares by Vermont, other NEEP states, and states with and without program promotions. 5 Data from KEMA s previous analysis of Vermont s refrigerator market share. November 2002. 2 14 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Table 2-9 Dishwasher Shipment and Market Share Year 2001 2002 2003 2004 MARKET SHARE ENERGY STAR (CHAIN RETAILERS TRACKED BY DOE) Vermont 14.8% 27.5% 54.2% 86.3% Other NEEP States Connecticut 14.2% 37.8% 62.0% 87.3% Massachusetts 17.0% 24.1% 54.0% 80.1% New Hampshire 15.3% 21.9% 54.5% 86.2% States with Promotion Programs Median 14.8% 33.8% 56.5% 83.9% Maximum 22.7% 39.3% 64.5% 88.0% Minimum 13.1% 21.9% 49.6% 76.1% States with No Programs Median 12.3% 34.4% 57.0% 78.9% Maximum 17.3% 44.5% 66.7% 86.5% Minimum 2.9% 18.0% 47.4% 69.5% US Average 19.9% 36.4% 56.9% 78.2% ESTIMATE OF VT MARKET SHARE: ALL RETAILERS VT: Independents, n=9 64.0% 64.5% 77.4% 86.6% VT: Weighted Average 45.6% 45.5% 65.1% 86.4% Annual Shipments (AHAM) 8,500 8,500 9,100 9,700 The key findings to be drawn from Table 2-9 are as follows. Difference between chain and independent stores. Independent retailers in Vermont have historically had very high ENERGY STAR dishwasher market shares (peaking at over 86 percent in 2004). Chain stores on the other hand, up until 2004 at least, have had significantly lower market shares. It was only in 2004 that chain stores caught up to the estimated market share of ENERGY STAR dishwashers sold by independent retail stores and themselves reached over 86 percent. Comparison to other NEEP states. The estimates for Vermont s market share of ENERGY STAR dishwashers sold through chain stores was historically lower that the shares in other NEEP states. However, that trend changed in 2004 when Vermont s chain store market share (86.3 percent) for ENERGY STAR dishwashers surpassed both Massachusetts and New Hampshire, 80.1 and 80.2 percent respectively. Only Connecticut at 87.3 percent had a higher market share in 2004. 2 15 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Comparison to states with no programs. There is little difference in market shares between chain stores in states with incentive programs and those without. Both experienced significant market share growth from 2002 to 2004, and had a median market share value in 2004 near 80 percent (83.9 percent for program states, and 78.9 percent for no program states). 2.3.5 Room Air Conditioners Sales and ENERGY STAR Market Share Comparison of Market Shares. Table 2-10 displays the comparison of room air conditioner ENERGY STAR market shares by Vermont, other NEEP states, and states with and without program promotions. Table 2-10 Room Air Conditioner Shipment and Market Share Year 2001 2002 2003 2004 MARKET SHARE ENERGY STAR (CHAIN RETAILERS TRACKED BY DOE) Vermont 19.8% 61.3% 63.7% 61.4% Other NEEP States Connecticut N/A 50.7% 42.1% 52.1% Massachusetts N/A 45.9% 45.6% 55.3% New Hampshire N/A 50.5% 63.2% 64.4% States with Promotion Programs Median 21.5% 47.4% 41.2% 44.7% Maximum 31.1% 61.3% 65.8% 64.4% Minimum 15.7% 26.6% 15.0% 25.7% States with No Programs Median 17.7% 30.5% 19.4% 29.0% Maximum 24.7% 55.4% 41.5% 43.3% Minimum 3.0% 20.8% 7.4% 21.8% US Average 11.5% 35.7% 28.8% 35.4% ESTIMATE OF VT MARKET SHARE: ALL RETAILERS VT: Independents, n=9 3.0% 34.5% 77.7% 70.9% VT: Weighted Average 10.7% 57.3% 70.6% 67.8% Annual Shipments (AHAM) 6,100 7,300 14,100 11,800 The key findings to be drawn from Table 2-10 are as follows. 2 16 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Difference between chain and independent stores. ENERGY STAR market share for room air conditioner sales by chain stores in Vermont has been fairly level since 2002, hovering around 61 percent. Independent store market share on the other hand has risen dramatically from less than 35 percent in 2002 to over 70 percent in 2004. Comparison to other NEEP states. Unlike the trends in dishwasher market share for Vermont chains mentioned above, room air conditioner shares have historically been higher in Vermont than in any other NEEP state. Growth patterns for the other NEEP states however, have increased at substantially higher rates than Vermont over recent years. New Hampshire for example has seen a room air conditioner market share rise from slightly over 50 percent in 2002 to almost 65 percent in 2004. Vermont on the other hand did not see any increase in market share over the same time period. Comparison to states with no programs. Historically, states with no incentive programs for room air conditioners have had much lower market share levels than those that do. However, states with no programs did see a much more significant increase in market share during 2004 (19 percent to 29 percent) than did those states with programs (41 percent to 44 percent). This is likely attributed to the change in ENERGY STAR criteria for room air conditioners (October 2003) to include models without louvered sides, casement models, and slider casement models. Increasing the number of eligible ENERGY STAR room air conditioners would likely have a greater impact on states with lower market share levels due to the relative nature of their inventory. 2.4 ASSESSMENT OF NET PROGRAM EFFECTS 2.4.1 Approach KEMA estimated the net effects of the program on the market share of ENERGY STAR units purchased in Vermont using a cross-sectional regression approach first developed for the Phase 1 evaluation. The basic approach proceeds in the following steps. 1. Estimate a regression model of the ENERGY STAR market share of a given appliance in a given year. The model takes the general form: MS ays = α + β1m s + β2p s where: MSays = the ENERGY STAR market share of appliance a in state s in year y as measured for chain outlets reporting to the DOE tracking system. M s = a vector of variables describing the state level market for the appliance in question. These variables include socioeconomic indicators that have been hypothesized to affect purchase of energy-efficient appliances: e.g. indicators of education, income, homeownership, race, and extent of urbanization. 2 17 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT P s = a vector of variables describing state-level program efforts to promote energy-efficient appliances. These include the average level of financial incentives available to customers who purchase ENERGY STAR appliances, the availability of tax credits, the presence of various kinds of advertising and merchandising efforts, and so forth. The basic concept is assess the effects of various kinds and levels of program effort on ENERGY STAR market share while controlling statistically for variation in market share that may be associated with differences between states in terms of their populations and markets. 2. Estimate what Vermont s ENERGY STAR market share would have been in the absence of the program. This involves enumerating the model using actual data on Vermont s market indicators (M s ), setting the value of the indicators of program activity to zero. 3. Estimate the net effect of the program as the difference between the actual market share and the model-derived share with the program indicator variable set to 0. 4. Adjust the net program effect on market share to account for differences between the chain retailers and independents in the percentage of ENERGY STAR appliances sold, by appliance type and year. This was accomplished by multiplying the difference between observed market share in chain outlets and the model estimate with the program variable set to zero by the ratio of chain store market share to the weighted average share for chains and independents. Since the completion of the Phase 1 Evaluation, Nexus Market Research (NMR) completed two impact evaluations of ENERGY STAR appliance programs for a consortium of Massachusetts utilities using a similar approach, but with the following important changes. 6 Multiple or composite program variables. In the evaluation of the 2002 program, the Massachusetts utilities adopted an approach to characterizing program efforts that took into account state-level variations in levels of incentives, the proportion of customers eligible for incentives, and different approaches to marketing and publicity. In the Phase 1 Evaluation, KEMA used an indicator variable to represent the availability of financial incentives to at least 1/3 of the customers in the state to characterize the program. In the current Vermont analysis, KEMA has implemented the more detailed approach to program characterization. Expanded range of market descriptors. In the Phase 1 Evaluation, KEMA used a narrow set of market descriptors namely measures of education and income to 6 Nexus Market Research, RLW Analytics, Inc., Shel Feldman Management Consulting, and Research Into Action, Inc. 2003. Market Progress and Evaluation Report (MPER) For the 2002 Massachusetts ENERGY STAR Appliances Program. Appendix C. Cambridge, MA: NMR. and Wilson-Wright, L., S. Feldman, L. Hoefgen and A. Li. 2005 Front-loading Marketing: Assessing Cumulative Effects of ENERGY STAR Appliance Promotions on State-by-State Market Penetration Levels. In the Proceeding of the 2005 International Energy Program Evaluation Conference. New York, NY. 2 18 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT characterize differences between states in customer characteristics. In the evaluation of the 2002 Massachusetts programs, NMR expanded the range of demographic variables to include those mentioned above in the model description. The models used in that study also included a set of market context variables such as electric prices and the presence of specific retailers in the state. Inclusion of a variable to capture cumulative effects of past program efforts. The authors of the Massachusetts 2003 7 study hypothesized not unreasonably that the current ENERGY STAR market share in a state that has had active programs for a number of years is influenced by the extent of prior program activity. To operationalize this concept, they included in the model a cumulative effects variable defined as the percentage change in ENERGY STAR market share from 1998 through 2002 for the appliance in question. Inclusion of this variable does significantly increase the portion of state-to-state variation explained by the model. However, as the authors themselves observe, this approach introduces some ambiguity in the interpretation of results. Specifically, there are many variables that affect past changes in ENERGY STAR market share. Some may be attributable in part to the programs. These may include proliferation of qualifying models in response to increased demand in the states with active promotional initiatives. Others, such as variations in distribution patterns and market share among chain versus independent retailers are not related to the programs. Ultimately we decided to include this variable in our model, and identified the contribution it made to market share as an element of the maximum impact the program may have had. We explain this approach in further detail below. 2.4.2 Specification of Independent Variables Market Descriptors Table 2-11 displays the definitions and sources for the market descriptor variables we examined for the various models. We also show variable transformations that were examined for possible inclusion in a search for the best fit. 7 The full report of the 2003 program evaluation had not been filed at the time of this writing. We relied on a conference paper that briefly presented methods and selected findings for information on the approach and outcomes of the study. 2 19 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Variable Table 2-11 Market Indicator Variables Examined in Regression Analysis of State-Level ENERGY STAR Appliance Market Share Transformations Examined Label Definition None Stdized Log EP_04 WH_02 INC_03 ED_03 INC_ED_03 Electricity Proxy: Average residential revenue per MWh (2004), Energy Information Administration Percentage of population Caucasian: US Bureau of the Census, ACS (2002) Median Household Income: US Bureau of the Census, ACS (2003) Percentage of population over 25 with a college degree: US Bureau of the Census, ACS (2003) Combined Income/Education variable: sum of standardized values of INC_03 and ED_03. x x X X x x x X indicates the form of the variable used in the models. x indicates other forms that were examined. x x X Program Variables KEMA used a number of sources to identify and characterize ENERGY STAR appliance programs. These included review of lists maintained by the Consortium for Energy Efficiency and the Northeast Energy Efficiency Alliance, review of websites for utilities in states we knew to have programs, interviews with program coordinators and administrators, and discussions with other consultants in the field. Through these methods we identified over 150 programs in 23 states. Many of these are operated by relatively small electric coops in the Pacific Northwest. In our search for the best model fit, we combined the information from the program research into a variety of different state-level variables. In some cases we examined the effect of variables describing financial incentives and marketing efforts separately. Table 2-12 displays the name and definition of these variables. 2 20 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Variable Name Program index Table 2-12 Program Variables Developed and Examined in Regression Analysis of State-Level ENERGY STAR Appliance Market Share Definition presence (1) or absence (0) of a program (incentives or marketing) Program Score Composite of Marketing and Incentive scores (range, 0 4) Incentive index Incentive score Marketing Support index Mass Marketing index In-Store materials index Field Support index Marketing Support Score Mass Marketing score In-Store materials score Field Support score presence (1) or absence (0) of an incentive statewide average, annualized, rebate dollar amount (taking into account bounties and tax credits) presence (1) or absence (0) of any marketing programs presence (1) or absence (0) of a Mass Marketing Program presence (1) or absence (0) of an In-store materials Program presence (1) or absence (0) of a Field Support Program Sum of Mass Marketing, In-store materials, Field support scores % households covered by Mass Marketing Program % households covered by In-store materials Program % households covered by Field Support Program Cumulative Effects Variable We defined the cumulative effects variable as the percentage change in ENERGY STAR market share for the longest period for which consistent data were available for the individual appliances. These intervals encompassed 1999 2002 for clothes washers and dishwasher and 2000 2002 for air conditioners. We began the interval for refrigerators at 2001, when both Federal minimum standards and ENERGY STAR specifications were revised. 2.4.3 Model Selection We sought to create a meaningful, parsimonious model for each appliance which maximized adjusted R 2 (i.e., explained the maximum amount of variation in the data) and included only meaningful variables with statistically significant coefficients (p<.05). As previously noted, many different models were estimated for each ENERGY STAR appliance. Selecting a best model for each appliance involved examining the sign and statistical significance of the coefficients and the portion of total variation in ENERGY STAR market share that the model accounted for (R 2 ). The selected models only include independent variables with coefficients that were statistically significant at the five percent probability level and had the 2 21 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT expected signs (+/-). In addition, the F statistic for the model as a whole exceeded the critical value. Table 2-13 displays the details of the selected models for 2004 market share of clothes washers, room air conditioners, and refrigerators. This approach did not yield a model of dishwasher market share with sufficient explanatory power to use in the analysis. The three models all explained a large portion of the state-to-state variation, especially given the small number of observations (50) available. All coefficients took the expected sign and were statistically significant (probability p <.05). We also estimated models for 2003 market share using a similar approach. Table 2-13 Regression Model Coefficients and Statistics: 2004 State Level ENERGY STAR Appliance Market Share Appliance Clothes Washers Room Air Conditioners Refrigerators Regression Statistics R 2 0.854 0.736 0.659 Adjusted R 2, n = 50 0.834 0.699 0.621 Standard Error 0.037 0.062 0.036 Model Coefficient t-statistic Coefficient t-statistic Coefficient t-statistic Intercept -0.025-0.343-0.075-0.591 0.204 2.900 Electricity Price Proxy 0.018 6.621 0.016 2.999 0.011 4.001 Percent Caucasian 0.269 5.171 0.427 4.784 0.135 2.822 Percent Urban -0.187-4.005-0.147-1.887-0.096-2.184 Income/Education Score 0.023 5.629 0.022 3.176 0.016 4.138 Incentive Score 0.00067 4.595 n/a n/a 0.001 2.458 Program Score n/a n/a 0.018 2.163 n/a n/a Prior Cumulative Change 0.056 3.619 0.038 2.190 n/a n/a 2.4.4 Model Enumeration: Estimation of Program Effect on ENERGY STAR Market Share in Vermont Table 2-14 displays the estimates of Vermont s ENERGY STAR market share by appliance and year, with and without the effects of the program and the cumulative effects variable. We have also included the results of the Phase 1 analysis of clothes washer and room air conditioner market share for comparison. 2 22 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Appliance/Year Table 2-14 Model Estimates of Market Share With and Without Program and Cumulative Effects Actual VT Model VT 90% Model Estimated Share Market Share Market Share Confidence Interval. No Program No Program + No Cum. Effect Clothes Washer 2001 22.6% 20.1% +/- 5.2% 12.9% n/a Clothes Washer 2003 52.6% 52.6% +/- 9.5% 46.8% 43.4% Clothes Washer 2004 48.9% 49.5% +/- 6.0% 46.1% 43.4% Room A/C 2001 19.8% 21.1% +/- 5.2% 19.4% n/a Room A/C 2003 63.7% 68.2% +/- 11.7% 62.5% 43.3% Room A/C 2004 61.4% 66.4% +/- 10.1% 58.8% 52.0% Refrigerator 2004 46.0% 48.3% +/- 5.9% 45.8% n/a Clothes Washer Results. The analysis results summarized in Table 2-14 show that the net influence of the program on sales of ENERGY STAR clothes washers has declined significantly since the inception of EVT. The Phase 1 analysis of 2001 program results found that the program was responsible for an incremental 9.7 percent of market share (Actual Market Share Model Estimate/No Program), or 43 percent of all washers sold in the state. Even if we include the cumulative effects variable in the analysis of 2004 results, we find that an incremental 5.5 percent market share can be attributed to the program. As discussed, the difference between actual market share and model estimated share with the current program and cumulative effects variables set to zero represents the maximum estimate of program effects. The minimum in this case would be the difference between actual market share and the model estimate with the cumulative effect variable included and the current program variable set to zero. In 2004, this minimum estimate of market share attributable to the program is 2.8 percent. The corresponding estimates for 2003 fell between those of 2001 and 2004. The estimate of maximum program contribution to market share is 9.2 percent. The minimum is 5.8 percent. Room Air Conditioner and Refrigerator Results. In the case of room air conditioners, the estimate of effects on market share for the current program alone is small: 1.2 percent in 2003 and 2.8 percent in 2004. If the cumulative effects variable is included, the incremental ENERGY STAR market share attributable to the program is 20.4 percent in 2003 and 9.4 percent in 2004. While the current program effects variable in the 2004 refrigerator equation is statistically significant, its effect on market share is very small only 0.2 percent. 2 23 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT 2.4.5 Net Unit Sales and Energy Savings Clothes Washers. Table 2-15 translates the regression results in Table 2-14 into number of qualifying unit sales attributable to the programs. In so doing, we accounted for the differing share of total and ENERGY STAR sales accounted for by chain and independent retailers in Vermont. (See discussion in Section 2-3.) Table 2-15 makes the decline in program influence on clothes washer sales clear. In 2001, the net unit sales were estimated at 1,045 versus total program-subsidized sales of 2,719 for a net-to-gross ratio of 0.38. In 2004, the maximum number of net unit sales attributable to the program was 724 (cumulative effects included) versus program volume of 4,179 units, for a net-to-gross ratio of 0.17. Applying these figures to gross savings estimates in the Technical Resource Manual, annual electric savings attributable to the 2004 clothes washer promotion range from 95.5 MWH/Year to 183.9 MWH/Year. Water and fossil fuel savings are also attributable to these unit sales. Room Air Conditioners. Applying the same methods, energy savings attributable to the room air conditioner promotions in 2004 ranged from 13.4 MWH/Year to 48.5 MWH/Year. Table 2-15 Net Program Effects and Net-to-Gross Ratios: Adjusted for Relative Market Shares of Chain and Independent Retailers ENERGY STAR Unit Sales Net Units Attributed to Program Net-to-Gross Ratio Appliance/Year Total VT Program Low High Low High Clothes Washer 2001 4,064 2,719 1,045 n/a 0.38 n/a Clothes Washer 2003 6,071 3,813 284 1,100 0.07 0.29 Clothes Washer 2004 6,332 4,179 376 724 0.09 0.17 Room A/C 2001 769-14 n/a n/a n/a Room A/C 2003 9,949 3,773 183 3,183 0.05 0.84 Room A/C 2004 8,007 2,233 339 1,226 0.15 0.55 Refrigerator 2004 4,260 1,137 19 n/a 0.02 n/a 2.4.6 Comparison to Massachusetts Analysis Our intention in conducting this analysis was to develop regression models that met standard criteria for goodness of fit and statistical significance, using some of the approaches explored by other analysts as guides. We did not set out to reproduce those analyses. Some reviewers of the preliminary presentation expressed concern that our findings, especially those concerning the apparent decrease in clothes washer program influence, were very much at odds with those of the 2 24 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT analysts in Massachusetts. In fact, the findings of the two studies, to the extent they can be compared, are very similar. As mentioned above, KEMA had access only to a conference paper reporting the results of the 2003 Massachusetts analysis. The paper contained a good description of the methods and basic results of the analysis. Direct comparison of the regression equations between the Vermont and Massachusetts studies is not appropriate for a number of reasons. Development of program variables. Both sets of analysts were obliged to develop program variables independently from roughly the same fund of program information. Moreover, we were not able to obtain data series developed for the Massachusetts study. We found that development of the program variables required a considerable amount of judgment in estimating the portion of a state s households who were eligible for various programs offered by local utilities, the combinations of incentives and program marketing present in various jurisdictions, and so on. Thus, it was unlikely that both studies were using the same values for this important independent variable. Differences in modeling approach. KEMA intentionally made a number of decisions in the modeling that varied from those made by the Massachusetts analysts. First, we included data from all 50 states in the analysis while the Massachusetts analysis included only the 48 contiguous states. This decision had the effect of excluding Alaska, which has consistently ranked within the top 5 states in ENERGY STAR clothes washer market share despite having no programs. Editing out the outlier results in higher R 2 and t- statistics for the program variables. However, we could discover no good theoretical reason to do this. Second, the Massachusetts analysis retained in the model variables that did not meet significance tests for inclusion. We did not. Despite these differences in approach, the models yielded remarkably similar results in terms of net changes in market share attributed to the program. Table 2-16 illustrates the similarity, using the results of the 2003 clothes washer analyses for Vermont and Massachusetts. Over the past 5 7 years, Massachusetts utilities have fielded clothes washer promotion programs roughly equal scope and design to EVT and its predecessors. Table 2-16 Comparison of Program Effects Estimates, 2003 Massachusetts and Vermont a. # of Units Massachusetts b. % of Total Sales c. # of Units Total CW Unit Sales, 2003 126,954 12,200 Vermont d. % of Total Sales ENERGY STAR Unit Sales, 2003 50,196 39.5% 6,417 52.6% Sales attributable to cumulative effects 13,226 10.4% 415 3.4% Sales attributable to 2003 Program 5,161 4.1% 708 5.8% Total sales attributable to program 18,387 14.5% 1,122 9.2% 2 25 VT DPS

1999 2000 2001 2002 2003 2004 SECTION 2 APPLIANCE PROGRAM ASSESSMENT The figures in Column a. of Table 2-16 are taken directly from the conference paper on the Massachusetts study. They show the estimated number of units associated with 2003 program and cumulative effects once the model has been enumerated for Massachusetts demographic and program variables. Column b renders the same information in terms of percent of total clothes washer sales in 2003. The Massachusetts study found that the sales of ENERGY STAR units equal to 14.5 percent of total 2003 clothes washer purchases could be attributed to the program, with 4.1 percent of that associated with current year efforts. Using the same calculation approach with the model produced for this study, we estimated the incremental market share associated with 2003 program efforts at 5.8 percent of total unit sales and cumulative effects of 3.4 percent for a total of 9.2 percent. All of these results are well within the 90 percent confidence intervals for the overall market share estimate and for the estimates of the individual coefficients. 2.4.7 Explanations for Apparent Reduction in Net Program Effects The low levels net program effects estimated for 2004 beg the question of what is happening in the appliance markets or in the mechanics of the analysis to contribute to this result. Here are some likely possibilities. Increase in national market share. We believe that the regression results reflect to a large extent the recent increase in the market share of ENERGY STAR clothes washers in the nation as a whole and in states without programs in particular. This has led to a reduction of the percentage difference in average market share for states with and without programs. Figure 2-1 illustrates this trend for ENERGY STAR clothes washers. The reduced level of program variables in crosssectional regression analyses is a mathematical consequence of the trend. Figure 2-1 Market Share for ENERGY STAR Clothes Washers 60.00% 50.00% 40.00% 30.00% 20.00% No Program US Average Program VT 10.00% 0.00% 2 26 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Match of analysis technique to the research question. Even with the addition of the change variable, the cross-sectional model used here purports only to estimate program effects in the current year. However, market transformation is a phenomenon that occurs over many years. And, as discussed above, one can plausibly argue that local incentive programs have helped contribute to the growth of ENERGY STAR clothes washer market share even in non-program areas by hastening the adoption of national specifications and by focusing manufacturer attention on the high-end washer market. This spillover effect if it indeed occurred would reduce the value of the program variable coefficients in cross-sectional models. One possible way to address this mismatch of cross-sectional modeling techniques to historical processes would be to use combined time series/cross-sectional modeling approaches, such as those typically applied in billing analysis of program energy effects. Such an approach would require painstaking year-byyear reconstruction of historical program offerings in the various states a level of effort far beyond what was available for this study. Multicollinearity. Some reviewers believed that covariation between market descriptor variables on the one hand, and the political/regulatory decisions to fund appliance incentive programs on the other may have reduced the level of program effects calculated using the crosssectional model. Specifically, they argued that programs were most likely to be implemented in states where energy prices were high and where well-educated consumers advocated for energy efficiency programs. Be that as it may, neither statistical theory nor the results of this particular modeling exercise support that line of argument. To begin with, collinearity among independent variables does not result in biased estimates of regression coefficients. 8 Therefore, the fact that the coefficients for indicators of current program activities and past changes in market share were low had nothing to do with their potential statistical association with various market descriptor variables. Collinearity does result in reduced statistical significance for the individual coefficients. However, for the clothes washer model, the t-statistic for the program indicator (Incentive Score) coefficient was very high (4.595; p < 0.001). Even the t-statistic for the marketing activities (Program Score) variable used in the Refrigerator Model was quite high (2.163; p < 0.02). 2.5 RECOMMENDATIONS EVT s ENERGY STAR appliance program has been very successful in elevating and maintaining the market share for ENERGY STAR clothes washers and other appliances. However, changes in the underlying market for these devices have led to their broad acceptance among consumers, as well as merchandising support from manufacturers and retailers. EVT and its counterparts in other states with strong appliance programs along with the federal ENERGY STAR program, support organizations such as the Consortium for Energy Efficiency, advocacy organizations, progressive retailers, and manufacturers may take credit for of these developments. However 8 See, among many others Welch, Roy and Edwin Kuh. 1979. Linear Regression Diagnostics. New York: John Wiley & Sons, Inc. and Green, Paul E. 1978. Analyzing Multivariate Data. Hinsdale, IL: The Dryden Press. 2 27 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT that credit is distributed, it does not change the fact that state-level programs, as currently constituted, are decreasing in net effect, despite high levels of participation and rate-payer expense. In discussing the findings of this study with representatives of EVT and appliance promotion managers around the country, we found that virtually all these individuals believed that the market share of ENERGY STAR appliances in their areas had reached a plateau and that the net effects of their programs were declining. Many of these programs have introduced changes to their programs to respond to these developments. Some of our recommendations reflect the experience of these program coordinators. For the time being, EVT should retain financial incentives for ENERGY STAR clothes washers, but assess the potential effects of program changes that could lead to increased cost-effectiveness in the face of low net-to-gross ratios. These include: Reduce incentive amounts. Limit the duration of appliance incentive promotions to specific months. This approach has been tried in the Pacific Northwest. Market share of ENERGY STAR appliances remain high in that region. Restrict eligibility to models that qualify for the Consortium for Energy Efficiency s Tier 3 standards. We base this recommendation on the following observations. First, clothes washers continue to offer by far the highest unit energy savings of any ENERGY STAR appliance: roughly 254 kwh to move from the current federal standard to the ENERGY STAR specification, plus water savings and fossil fuel savings for water heating and drying. Second, despite the steady rise in the number of qualifying models available, the incremental cost of ENERGY STAR clothes washers remains high. Finally, the federal minimum efficiency standard for clothes washers will increase from its current level of 0.817 MEF 9 to 1.04 MEF. See Table 2-17. This will cut the unit electricity savings associated with purchase of an ENERGY STAR qualifying model from 35 percent to 17 percent when compared to the current federal minimum standard. If historical pricing patterns prevail, incremental prices are unlikely to decline. In fact, they may increase as manufacturers and buying groups seek to sell down inventories of non-complying models. Under these conditions, rebates are likely to be necessary to continue the gains achieved to date in ENERGY STAR clothes washer market share. 9 Modified Energy Factor: equals kwh required per cubic foot of laundry processed for the entire laundry cycle washing and drying. 2 28 VT DPS

SECTION 2 APPLIANCE PROGRAM ASSESSMENT Table 2-17 Residential Clothes Washers Efficiency Standards, Specifications, and Model Availability Modified Water % Electric Savings v. Federal Standard # Models Available Efficiency Standard/Specification En. Factor Factor v. 2004 v. 2007 @ 8/05 Current Federal Minimum Standard (1994) 0.817 13.3 2004 Federal Minimum Standard 1.04 n/a 2007 Federal Minimum Standard /Current E Star Specification 1.26 11.0 17% 50 CEE Tier 2 Specification 1.42 9.5 27% 11% 50 CEE Tier 3 Specification 1.60 8.8 35% 21% 104 CEE Tier 4 Specification 1.80 7.5 42% 30% 18 CEE Tier 4b Specification 1.80 5.5 42% 30% 11 As Table 2-17 shows, many clothes washer models are currently available that provide significantly higher energy savings than those that just meet the ENERGY STAR specification. In fact, 57 percent of all qualifying models meet or exceed the Consortium for Energy Efficiency (CEE) Tier 3 specifications. These models use between 21 and 30 percent less electricity than models that meet the current ENERGY STAR specification. As manufacturers revise their product offerings to meet the 2007 standards, it will be necessary to motivate customers to purchase higher-efficiency products in order to achieve savings beyond baseline conditions. Planning discussions for such a restructuring of incentives should start now in order to achieve an orderly transition. According to the distribution of available models by efficiency rating, 79 percent of qualifying models exceed the minimum ENERGY STAR efficiency specification. In order to claim credit for the full level of savings achieved, savings estimates generated by the program tracking system should take the efficiency ratings of models sold explicitly into account, if this is not being done already. In the longer run, EVT may wish to withdraw incentive support for washers entirely and release those financial and business development resources for other uses. EVT should discuss this eventuality with participating retailers and program coordinators in other jurisdictions so that a coordinated strategy can be taken vis-à-vis this market. 2 29 VT DPS

3 APPLIANCE SATURATION SURVEY 3 APPLIANCE SATURATION SURVEY 3.1 OVERVIEW 3.1.1 Project Background This report presents findings from the 2005 Residential Appliance Saturation Study (RASS) conducted by KEMA, Inc. The 2005 RASS consisted of telephone and onsite surveys with Vermont residents living in one to four unit buildings in Vermont. The key objective of the RASS was to produce reliable and accurate estimates of appliance saturation and energy-related housing characteristics to support resource planning, program planning, and improved characterization of the residential markets for energy efficiency programs. Results of the RASS were also used to assess the size and segmentation of the residential market for compact fluorescent bulbs (See Section 1) and the potential savings associated with a prospective appliance recycling program (See Section 3). To accomplish these objectives, the RASS collected information on appliance holdings, fuel shares, and other energy related characteristics from a sample of residential customers in Vermont. The survey data were used to develop penetration and saturation rates for heating and cooling equipment, appliances, and other plug loads. 3.1.2 Report Overview Key findings are presented on the full range of survey topics covered, including heating, cooling, fuel shares, appliance and equipment holdings, and energy efficiency. The RASS findings are supplemented with data from the 2005 and 2002 onsite surveys. The RASS findings include detailed data on appliance saturation and penetration. For the purposes of this document, penetration is defined as the number of households that have a particular appliance or piece of equipment out of the total households in the sample, while saturation represents the quantity of a particular appliance or piece of equipment per household in the sample. While it is possible for saturation to exceed 1.0 or100 percent, penetration cannot. 1 This report includes the following sections: Major End Uses Separate sections for space heating, air-conditioning and water heating discuss fuel shares, equipment types and other equipment and usage characteristics. 1 To calculate the average number of appliances among households that have that particular appliance (e.g., the average number of personal computers per household among households that have personal computers) divide saturation by penetration. 3 1

SECTION 3 APPLIANCE SATURATION SURVEY Refrigerators and Freezers This section includes refrigerator and stand-alone freezer saturation and equipment characteristics. Food Preparation This section provides an overview of fuel shares and other characteristics of ranges, ovens, and other food preparation equipment. Home Office In addition to generating small increases in electric use, increased home office equipment usage such as computers and on-line services provides new ways to communicate with the customer. Entertainment Equipment and Miscellaneous Appliances These sections profile home entertainment equipment used by Vermont customers and provide saturations for more than twenty residential appliances. Energy Efficiency This section discusses lighting technologies as well as ENERGY STAR awareness and influence on purchasing decisions in the residential market. Current market shares of compact fluorescent lamps are included along with detail on compact fluorescent purchases. 3.1.3 Methodology The 2005 RASS consisted of telephone surveys and onsite surveys. Both of these components are explained below. Telephone Surveys. The telephone component of the RASS consisted of phone surveyswith 600 Vermont residents. The primary objective of these surveys was to provide an unbiased estimate of heating and cooling system and appliance holdings in the state. To accomplish this objective, the telephone survey format was chosen to reduce non-response bias and maximize ease of response among potential respondents. The telephone surveys focused on items on which customers can generally report easily and accurately; for example, whether or not they own a particular piece of equipment and how frequently they use it. Each telephone survey was approximately 20 minutes in length. Given the goals of the RASS identified in the RFP, sampling procedures were designed to yield a representative random sample of all dwelling units in 1 to 4 unit buildings in Vermont. This objective was achieved through the use of two techniques: Random Digit Dialing (RDD). RDD ensures that all Vermont residents have an equal likelihood of being contacted to complete the survey, including those with unlisted numbers; and Regional Stratification. A regional stratification scheme ensured that all areas of the state were represented adequately. Regions were defined as follows: o Northwestern Vermont: Chittenden, Franklin, Lamoille, Grand Isle, and Washington Counties; 3 2

SECTION 3 APPLIANCE SATURATION SURVEY o Southwestern and South Central Vermont: Bennington, Rutland, and Addison Counties; o Northeastern Vermont: Essex, Orleans, and Caledonia County; and o Southeastern Vermont: Windham, Windsor, and Orange Counties. Onsite Surveys. The objective of the onsite survey component of the 2005 RASS was to get detailed information on appliances and equipment. The onsite surveys focused on gathering information that would take too long to obtain during a telephone interview or that other studies have shown customers do not accurately report (for example, the number of CFL bulbs installed). In addition, the onsite surveys collected data for comparison with the results of the 2002 onsite surveys. The telephone RASS wasused to recruit survey participants into the onsite surveys. Experience from past studies of this kind has shown that approximately 20 to 30 percent of telephone respondents can successfully be recruited into on-site panels; in the 2005 RASS, 36 percent of respondents expressed interest in the on-site surveys when a $35 incentive was offered. Ultimately, onsite surveys were completed with 84 customers (14 percent of the RASS participants). Results from the onsite surveys were generally consistent with those from the RASS and supplement the RASS results herein. Onsite surveys provide the accuracy of data collected by experienced home auditors. However this gain in accuracy is purchased at the price of a smaller sample and a higher likelihood of selfselection bias. People willing to allow inspectors into their homes for one to two hours are more likely to be interested in energy efficiency than those who decline the onsite survey. Thr $35 incentive was offered to offset this tendency; however, an incentive is unlikely to overcome entirely the self-selection problem. The saturation estimates described below are derived from the telephone RASS. For other topics, the most appropriate selection of source data was made. Data from the telephone and onsite surveys is occasionally compared to explore the consistency of data from the two sources; however, onsite and phone results are not expected to match exactly for the reasons described above. 3.2 SPACE HEATING 3.2.1 Primary Heating System The vast majority (94 percent) of Vermont customers living in 1 to 4 unit dwellings pay to heat their own homes, while 6 percent have the cost included in their rent or condominium fees. Less than one percent of households in Vermont mostly seasonal vacation homes do not have space heating systems. Nearly half (47 percent) of customers who live in buildings with two or 3 3

SECTION 3 APPLIANCE SATURATION SURVEY more units indicated that heat is included in their rent or condominium fees. 2 The results that follow are for households that pay space heating costs. Vermont customers who pay for their own space heating rely more on fuel oil to heat their homes than on any other heating fuel (54 percent). Approximately 15 percent of customers use propane, bottled, or tank gas; 11 percent use natural gas or wood; 7 percent use kerosene; 2 percent use electric heat; and less than half of one percent use other fuels (including wood pellets and coal; see Figure 3-1). Figure 3-1 Primary Heating Fuel Shares Among Those Who Pay for Heating Electricity 2% Other <1% Wood 11% Kerosene 7% Natural Gas 11% Fuel Oil 54% Bottled gas 15% n = 593. Six percent of homes have natural gas fireplace inserts; of these, 63 percent are in homes built during or after 1995. Across all homes, 21 percent of those built during or after 1995 have natural gas fireplace inserts, compared with only 4 percent of homes built prior to 1995. Data from the RASS show that fuel oil is decreasing in prevalence over time as the primary heating fuel in Vermont homes, while use of bottled gas has been increasing. As shown in Figure 3-2, 71 percent of homes built before 1930 use fuel oil for primary space heating and only 9 percent use bottled gas, while only 40 percent of newer homes (those built since 1990) use oil as their primary space heating fuel and 28 percent use propane. Kerosene is predominant among the other fuels used for space heating in homes built since 1990, and both kerosene and wood comprise the majority of other fuels for homes built during the 1970s and 1980s. 2 Note that customers living in buildings with 5 or more units were excluded from the survey sample. 3 4

SECTION 3 APPLIANCE SATURATION SURVEY The trends of decreased use of oil and increased use of bottled gas as primary heating fuel in the RASS data are echoed in the onsite data; space heating fuel share for oil in 2002 was 60 percent, and in 2005 fell to 54 percent. For propane, the 2002 heating fuel share was 10 percent and in 2005 rose to 16 percent. Figure 3-2 Primary Space Heating Fuel by Dwelling Age Fuel Oil Propane or Bottled Gas Natural Gas Other 100% 75% 50% 13% 14% 7% 9% 17% 10% 27% 7% 14% 18% 14% 28% 25% 71% 59% 51% 40% 0% Before 1930 (n=121) 1940-1969 (n=118) 1970-1989 (n=193) 1990 or later (n=100) Year Built 3 5

SECTION 3 APPLIANCE SATURATION SURVEY On average, fuel oil space heating systems tend to be older (23 percent at least 20 years old) than bottled gas systems (37 percent less than 5 years old; see Figure 3-3). Figure 3-3 Primary Space Heating Fuel by Heating System Age 100% 75% 50% 25% 0% 16% 12% 23% 9% 16% 12% 18% 9% 20% 24% 23% 23% 25% 27% 17% 6% 9% 10% Fuel Oil (N=284) Natural Gas (n=55) Bottled Gas (n=77) Primary Space Heating Fuel 20+ yrs 15-19 yrs 10-14 yrs 5-9 yrs 1-4 yrs <1 yr 3.2.2 Heating Systems and Controls Overall, the majority of customers who pay their own space heating costs in 1 to 4 unit dwellings in Vermont have standard thermostats (85 percent) for their heating systems, while only 15 percent use programmable thermostats. Programmable thermostats are most common in newer homes and among homes with natural gas systems. (Figure 3-4). Figure 3-4 Type of Thermostat by Primary Space Heating Fuel Type 3 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Programmable Non-Programmable 86% 89% 72% 28% 14% 11% Fuel Oil (N=312) Natural Gas (n=65) Bottled Gas (n=88) Primary Space Heating Fuel 3 Results not reported for other fuel types because of small sample sizes. 3 6

SECTION 3 APPLIANCE SATURATION SURVEY Fifty percent of residential customers use a hot water or steam boiler regardless of heating fuel. Thirty seven percent of customers have oil-fueled hot water or steam boilers, and 17 percent have oil-fueled central forced air furnaces. Figure 3-5 below provides detail on all primary heating system equipment by heating fuel. Figure 3-5 Primary Heating System by Fuel Type 60% 55% 50% 17% 40% 30% 20% 10% 0% 37% Fuel Oil (n=314) 11% 5% 6% 1% Natural Gas (n=66) 15% 5% 6% 2% 1% Bottled Gas (n=85) Primary Space Heating Fuel 19% 5% 1% 1% 9% 4% Other (n=111) Central forced air furnace Hot water or steam boiler Direct Vent Space heaters Wood Stove Other n = 576. 3.2.3 Heating System Efficiency During the onsite inspections, auditors recorded AFUE ratings for boilers and furnaces for equipment on which these ratings were visible. They also noted whether the burners were sealed, which supports use of outside air for more efficient combustion. They found that the combustion chamber was sealed in 22 (31 percent) of the 71 systems for which observations could be made. As Table 3-1 shows, the auditors were able to record AFUEs for relatively few systems. For oil systems, they found that 6 of 19 units for which ratings were available met or exceeded the ENERGY STAR specification of AFUE 85 and above. AFUE ratings were visible for only 2 of the 10 gas units in the sample, and neither met the ENERGY STAR specification of AFUE 90. 3 7

SECTION 3 APPLIANCE SATURATION SURVEY Table 3-1 Primary Heating Equipment by Fuel and Efficiency Rating FUEL AFUE Oil Natural Gas LPG Other Less Than 70 2 70 to 75 3 76 to 80 3 81 to 85 5 1 86 to 90 5 2 1 Greater than 90 1 Don't Know 10 8 5 5 Total 29 10 7 5 Source: Onsite Survey, n = 51 3.2.4 Supplemental Heating Systems Supplemental Heating System Penetration. Fifty percent of respondents use at least one form of supplemental heating (including portable heaters). Wood or coal stoves are the most widely used form of supplemental heating (used by 24 percent of respondents); additional detail is provided in Table 3-2. Table 3-2 Supplemental Heating System Penetration Penetration Supplemental Heating Type (n = 598) Wood or coal stove 24% Wood burning fireplace 13% Portable electric heater or Quartz heater 10% Electric baseboard, wall units, or radiant ceiling panels 9% Gas fireplace insert 6% Fixed gas space heater, freestanding stove 5% Kerosene heater 3% Portable gas heater 2% Frequency of Supplemental Heating Use. Respondents who indicated that they used at least one form of supplemental heating were asked to describe the frequency of supplemental heating use during the heating season. As shown in Table 3-, frequency of supplemental heating use varies by primary space heating system fuel and primary space heating system type. Nearly three out of four respondents with bottled gas heating systems use supplemental heating at least one day per week during the heating season (74 percent), as do two out of three respondents with 3 8

SECTION 3 APPLIANCE SATURATION SURVEY oil heating systems (67 percent). Respondents with natural gas heating systems use their supplemental heating systems less frequently than respondents who use other heating fuels. Table 3-3 Frequency of Supplemental Heating Use by Primary Space Heating Fuel and System Type Frequency of Supplemental Heating Use Only a Overall Primary Heating System Fuel & Type Every Day 3-5 Days/Wk 1-2 Days few Days/Yr % n Oil 36% 18% 13% 33% 100% 144 Central forced air furnace 34% 25% 14% 27% 100% 44 Hot water or steam boiler 37% 15% 12% 36% 100% 100 Natural Gas 17% 20% 20% 43% 100% 30 Central forced air furnace 29% 29% 14% 29% 100% 7 Hot water or steam boiler 11% 17% 17% 56% 100% 18 Direct vent space heater 20% 20% 40% 20% 100% 5 Bottled Gas 31% 26% 18% 26% 100% 39 Central forced air furnace 25% 38% 6% 31% 100% 16 Hot water or steam boiler 22% 17% 33% 28% 100% 18 Direct vent space heater 67% 33% - - 100% 3 Other 35% 8% 20% 37% 100% 51 All Kerosene 42% - 8% 50% 100% 12 All Electric 40% - 20% 40% 100% 5 All Wood 32% 12% 24% 32% 100% 34 Extent of Supplemental Heating Use. Respondents who use supplemental heat were asked to indicate the number of rooms in which they use supplemental heat during the heating season. As shown in Table 3-4, respondents who use kerosene or wood as their primary heating fuels use supplemental heating in a larger mean number of rooms in their homes than customers with other heating fuel types. Overall, customers with oil heat use supplemental heating in more rooms than customers with natural gas heating systems or bottled gas heating systems. Table 3-4 Extent of Supplemental Heating Use (Mean Number of Rooms) by Primary Space Heating Fuel and System Type Number of Rooms Primary Heating System Fuel & Type Minimum Maximum Mean n Oil 1 20 3.5 302 Central forced air furnace 1 20 3.8 94 3 9

SECTION 3 APPLIANCE SATURATION SURVEY Hot water or steam boiler 1 14 3.2 203 Natural Gas 1 7 2.5 65 Central forced air furnace 1 7 3.0 28 Hot water or steam boiler 1 7 1.9 31 Direct vent space heater 1 7 3.6 6 Bottled Gas 1 8 2.5 83 Central forced air furnace 1 5 2.5 29 Hot water or steam boiler 1 6 2.2 35 Direct vent space heater 1 8 3.7 10 Other 1 20 4.3 60 All Kerosene 1 20 4.5 41 All Electric 1 5 2.7 11 All Wood 1 9 4.4 60 3.3 AIR CONDITIONING AND VENTILATION 3.3.1 Air Conditioning Air conditioning is a key end use for Vermont s current and future energy concerns. As shown in Figure 3-6, approximately 41 percent of Vermont customers have some form of space cooling in their homes. Room air conditioning units are the main method of space cooling. Thirty-six percent of customers use room air conditioners, while only 4 percent have central air conditioning. Approximately one percent of Vermont customers have both central and room air conditioning. 3 10

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-6 Air Conditioning Use 4 80% 60% 61% 40% 40% 36% 20% 0% 4% No Cooling Central or Room AC Central AC Room AC Both 1% n = 600. The penetration rates for cooling systems are 4 percent for central AC and 36 percent for room air conditioners. Saturation is 58 percent for room AC units 5. Forty-three percent of homes with room air conditioning have multiple units. 4 May not total 100% because of rounding. 5 Saturation is a representation of the total number of units in the market and can account for more than one unit per dwelling. Penetration is the percentage of homes with a particular item or characteristic and includes a single count per household. 3 11

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-7 Penetration and Saturation of Cooling Systems Penetration Saturation 80% 60% 58% 40% 36% 20% 0% 4% 4% Central Air Conditioners (n=600) Room Air Conditioners (n=597) Newer homes are slightly more likely to have one or more central air conditioning systems than older homes, and homes in the northwest are slightly more likely to have central air conditioners than homes in other regions. Use of room air conditioners is spread fairly evenly across home ages. Regionally, 45 percent of homes in the southwest/south central region of the state have one or more room air conditioners, compared with 38 percent in the northwest, 32 percent in the southeast, and only 19 percent in the northeast. Nearly nine out of ten primary room air conditioners are less than 10 years old (89 percent), and three in five are less than 5 years old (Figure 3-8). Room air conditioners tend to be oldest in the southeast (mean age 6.4 years; see Table 3-5). 3 12

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-8 Age of Primary Room Air Conditioner 6 60% 53% 50% 40% 30% 29% 20% 10% 0% 7% 6% 1% 3% <1 yr 1-4 yrs 5-9 yrs 10-14 yrs 15-19 yrs 20+ yrs n = 201 Table 3-5 Mean Age of Primary Air Conditioner by Region Region Mean Age n Southwest/South Central 4.4 56 Northwest 4.8 87 Northeast 5.8 13 Southeast 6.4 45 All Regions 5.1 201 3.3.2 Ventilation None of the homes inspected for the onsite component of the RASS featured mechanical ventilation systems that would satisfy the Residential Building Energy Standards. Roughly twothirds had only a bathroom exhaust fan. The remainder mostly had kitchen exhaust fans. Virtually all of these units were controlled by simple on/off switches. 3.4 WATER HEATING The majority of the Vermont respondents surveyed (95 percent) indicated that their water heaters serve only their home or apartment, while 5 percent indicated that their water heaters serve more 6 May not total 100 percent because of rounding. 3 13

SECTION 3 APPLIANCE SATURATION SURVEY than one home or apartment. The results that follow are for customers whose water heaters serve only their home or apartment. Thirty-seven percent of customers whose water heaters serve only their home or apartment rely on electricity as their primary water heating fuel, while 29 percent rely on oil, 18 percent on bottled gas or propane, 14 percent on natural gas, and 2 percent on other fuels (primarily wood and kerosene; see Figure 3-9. Figure 3-9 Primary Water Heating Fuel Among Customers Whose Water Heaters Heat Only Their Home or Apartment Natural Gas 14% Electricity 37% Oil 29% Bottled Gas 18% Other 2% N = 245. As expected, there is a strong correlation between space heating fuel and water heating fuel: o 80 percent of those with natural gas heat also use gas for water heating; o 80 percent of electrically heated homes use electricity for water heating; o 58 percent of those with bottled gas/propane heat use the same for water heating; o 52 percent of oil heated homes use oil for water heating as well; 35 percent use electricity and 9 percent use propane/bottled gas. Electricity is nearly twice as common as a water heating fuel in homes built prior to 1990 than in homes built during or after that year (39 and 21 percent, respectively). Natural gas and bottled gas/propane have both increased in their use as water heating fuels in newer homes (Figure 3-10). The onsite surveys also showed a decrease in use of electricity for water heating between 2002 and 2005 accompanied by an increase in propane use. Use of electricity for water heating dropped from 35 percent in 2002 to 23 percent in 2005, while use of propane increased from 20 percent to 26 percent in the same time period. 3 14

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-10 Water Heating Fuel by Age of Home (Pre-/Post-1990) 45% 40% 35% 30% 39% 30% 29% Before 1990 (n=416) 1990 or later (n=97) 28% 25% 20% 15% 10% 21% 20% 13% 17% 5% 0% Electricity Natural gas Oil Bottled gas Other 1% 3% Fifty-two percent of customers whose water heaters serve only their home or apartment indicated that their water heaters are wrapped with insulation blankets. Fifty percent of the same group indicate that they have low-flow showerheads on all of the showers in their homes, while an additional 9 percent have low-flow showerheads on some of their showers (see Figure 3-11). Figure 3-11 Water Heater Energy Efficiency Measures Among Customers Whose Water Heaters Heat Only Their Home or Apartment 70% 60% 50% 40% 30% 20% 10% 0% 52% Insulation Blanket (n=489) Some Showers All Showers 59% 9% 50% Low-Flow Shower Heads (n=568) 3 15

SECTION 3 APPLIANCE SATURATION SURVEY Regionally, among Vermont customers whose water heaters serve only their homes or apartments, water heating energy-efficiency measures are slightly more prevalent in the southeastern region of the state than in the other three regions of the state (Figure 3-12). Figure 3-12 Water Heater Energy Efficiency Measures by Region Among Customers Whose Water Heaters Heat Only Their Home or Apartment 7 70% 60% 50% 40% 30% 20% 10% 0% 47% Northwest Northeast Southwest Southeast 58% 55% 58% 58% 51% 52% Blanket on One or More Water Heater(s) 64% Low-Flow Showerhead on One or More Shower(s) 3.5 LAUNDRY EQUIPMENT Ninety-two percent of the Vermont residential customers surveyed have laundry equipment in their homes for their own private use. An additional 2 percent use laundry facilities located in a common area in their building, and 6 percent do not have laundry facilities in their buildings. Ninety-two percent have clothes washers in their homes, and 87 percent have clothes dryers (Figure 3-13). 7 Blanket n: Northwest, 200; Northeast, 56; Southwest, 108; Southeast; 125; Overall n, 489. Showerhead n: Northwest, 232; Northeast, 67; Southwest, 132; Southeast, 137; Overall n, 568. 3 16

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-13 Penetration of Laundry Equipment Laundry Equipment in Home 92% 92% Clothes Washers 87% Clothes Dryers Laundry in No Laundry Equipment N = 600. Nearly 98 percent of survey respondents who live in single family homes indicated that they have laundry equipment in their homes for their own private use, compared to lower percentages in duplexes or triplexes (78 percent) or in apartments or condominiums with 2 to 4 units (53 percent). The results that follow are based on customers who have laundry equipment for their own private use (i.e., not located in a common area of a multi-family dwelling). Clothes Washers Penetration of clothes washers among the Vermont residential customers surveyed is at 92 percent. The following points apply to customers who have clothes washers for their own private use. Eighty-four percent of customers have top-loading clothes washers, while 16 percent have front-loading clothes washers (Figure 3-14). 3 17

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-14 Clothes Washer and Dryer Characteristics Among Those with Equipment for Private Use Clothes Washer (n=547) Clothes Dryer (n=518) Front Front Top Loading Loading Loading 16% 16% Natural Gas 4% Bottled Gas 10% Electric 0% 20% 40% 60% 80% 100% Four out of five of the Vermont customers surveyed have clothes washers that are less than 5 years old, while 16 percent have clothes washers that are more than 15 years old (Figure 3-15). Figure 3-15 Age of Clothes Washers Among Those with Equipment for Private Use 40% 30% 31% 26% 20% 19% 10% 9% 11% 5% 0% <1 yr 1-4 yrs 5-9 yrs 10-14 yrs 15-19 yrs 20+ yrs Appliance Age N =531. 3 18

SECTION 3 APPLIANCE SATURATION SURVEY 3.5.2 Clothes Dryers Penetration of clothes dryers among the customers surveyed is 87 percent. The following points apply to the those who have laundry equipment in their homes for their own private use: As shown in 3 19

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-15 above, 85 percent of the consumers who have clothes dryers have electric dryers, 10 percent use bottled gas to fuel their gas dryers, and 4 percent have natural gas dryers. Seventy-two percent of respondents who use electric clothes dryers use oil as their primary space heating fuel. The majority of respondents who use natural gas and bottled gas as their primary heating fuel use the same fuel for their clothes dryers. 3.6 FOOD PREPARATION 3.6.1 Penetration Ninety-nine percent of respondents indicate they have a stovetop or range. Ninety-nine percent have ovens, 57 percent dishwashers, and 92 percent microwave ovens (Table 3-6). Table 3-6 Penetration of Food Preparation Equipment Penetration Equipment Type (n = 600) Stovetop or Range 99% Oven 99% Dishwasher 57% Microwave Oven 92% 3.6.2 Fuel Shares of Cooking Equipment Ninety-nine percent of the residential customers surveyed have ovens and stovetops or ranges (Table 3-6 above). Of the survey respondents who have ovens, 65 percent have electric ovens, 22 percent bottled gas, and 13 percent natural gas ovens. Of respondents who have stovetops or ranges, 61 percent have electric stovetops or ranges, 25 percent have bottled gas, and 13 percent natural gas. Less than one percent use other fuels for their stovetops or ranges (Figure 3-16). 3 20

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-16 Fuel Shares of Food Preparation Equipment Among Customers Who Have The Equipment Electr ic Natur al Gas Bottled Gas Oven (n=598) 65% 13% 22% Stovetop or Range (n=595) 61% 13% 25% 0% 20% 40% 60% 80% 100% Seventy-five percent of homes with oil heat have electric ovens and 70 percent have electric stovetops or ranges. The majority of homes with natural gas or bottled gas for space heating also have electric ovens and/or stovetops. 3.6.3 Dishwashers Fifty-seven percent of the customers surveyed have dishwashers. Customers who live in newer homes are more likely to have dishwashers than those who live in older homes: 82 percent of customers who live in homes built during or after 1995 have dishwashers, compared with only 58 percent of those in homes built prior to 1995. Thirty-six percent of customers with dishwashers have units less than 5 years old, while 13 percent have dishwashers that are 15 years old or older (Figure 3-17). 3 21

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-17 Age of Dishwashers Among Customers Who Have the Equipment 40% 30% 28% 27% 23% 20% 10% 8% 8% 5% 0% <1 yr 1-4 yrs 5-9 yrs 10-14 yrs 15-19 yrs 20+ yrs Appliance Age N = 331. 3.7 REFRIGERATORS AND FREEZERS 3.7.1 Penetration and Saturation Refrigerator penetration among the survey respondents is 100 percent. Thirteen percent have two or more refrigerators, and saturation of refrigerators is 114 percent (Figure 3-18). Penetration of stand-alone freezers in Vermont is 42 percent. Figure 3-18 Penetration and Saturation of Refrigerators and Freezers 120% 100% 80% 60% 40% 20% 0% Penetration 114% 100% Refrigerators (n=600) Saturation 42% 44% Stand-Alone Freezers (n=600) 3 22

SECTION 3 APPLIANCE SATURATION SURVEY Fourteen percent of the Vermont customers surveyed have more than one refrigerator in their home. Forty-two percent have at least one stand-alone freezer, while 2 percent of customers have two or more freezers (Table 3-7). Nearly three in five customers do not have separate stand-alone freezers (58 percent). Table 3-7 Number of Refrigerators and Freezers Number of Refrigerators Number of Stand-Alone Freezers None 0% None 58% One 87% One 40% Two 13% Two or more 2% Three or more 1% Number of Responses 600 Number of Responses 600 Customers in the northwest are somewhat more likely to have multiple refrigerators than customers in other parts of the state, and they are also more likely to have one or more stand-alone freezers. Use of multiple refrigerators is slightly more common in newer homes: 18 percent of homes built during or after 1990 have more than one refrigerator, compared with approximately 12 percent in homes built before 1990. 3.7.2 Age of Equipment As shown in Figure 3-19, 39 percent of first-listed refrigerators are less than 5 years old and only 6 percent are 20 years old or older. Nearly one in four stand-alone freezers is less than 5 years old (24 percent); 37 percent are 15 years old or older, indicating that freezers are replaced with far less frequency than refrigerators. 3 23

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-19 Age of First Refrigerator and Freezer Among Customers Who Have The Equipment 40% Refrigerator (n=556) Stand-Alone Freezer (n=242) 30% 20% 10% 9% 5% 29% 26% 19% 21% 21% 17% 12% 8% 6% 25% 0% <1 yr 1-4 yrs 5-9 yrs 10-14 yrs 15-1 9 yrs 20+ yrs Appliance Age 3.7.3 Equipment Features Refrigerators Table 3-8 displays the sizes, styles and other features for first and second refrigerators. 8 Ninety percent of customers primary refrigerators are at least 15 cubic feet in size. Twothirds of primary refrigerators are in the 15-20 cubic feet range (66 percent), and nearly one in four respondents had refrigerators larger than 20 cubic feet (24 percent). Only ten percent of respondents have refrigerators smaller than 15 cubic feet in size. Electric heat customers and customers in older homes tend to have smaller refrigerators than other customers. Customers who live in multi-unit buildings also have smaller refrigerators than customers who live in single-family homes. Sixty-one percent of Vermont customers have top freezer / bottom refrigerator units as their primary refrigerators. Twenty-four percent of refrigerators are side-by-side units, and 8 percent are top refrigerator / bottom freezer units. Ninety percent of customers have primary refrigerators with automatic defrost, and 10 percent have door ice and water dispensers. 8 Note: Only three respondents had 3 or more refrigerators, so these results are not discussed. 3 24

SECTION 3 APPLIANCE SATURATION SURVEY Table 3-8 Refrigerator Size, Style, and Features Size First Refrigerator (n=600) Second Refrigerator (n=75) Very Small (<10 c.f.) 1% 21% Small (10-14 c.f.) 9% 21% Medium (15-20 c.f.) 66% 49% Large (>20 c.f.) 24% 10% Style Single door 8% 38% Side-by-side 24% 14% Top freezer / bottom refrigerator 61% 41% Top refrigerator / bottom freezer 8% 7% Features Automatic defrost 90% 62% Door ice & water dispenser 10% 35% Refrigerators with side-by-side doors are prevalent in newer homes (post-1995), while refrigerators with top freezers are most prevalent among older homes (pre-1995). Figure 3-20 provides additional detail. Figure 3-20 First Refrigerator Style by Year Home Built 70% 60% 61% Before 1995 (n=453) 1995 or later (n=59) 50% 46% 40% 37% 30% 23% 20% 10% 8% 7% 8% 10% 0% A single door with the freezer inside 2 side by side doors A top freezer A bottom freezer 3 25

SECTION 3 APPLIANCE SATURATION SURVEY Stand-Alone Freezers Sixty-one percent of first stand-alone freezers are at least 15 cubic feet in size. More than half of first freezers are upright models (53 percent: 28 percent frost-free, 25 percent manual defrost). Forty-eight percent of first freezers are reach-in chests (Table 3-9). Table 3-9 Stand-Alone Freezer Size, Style, and Features Among Customers Who Have Stand-Alone Freezers Size First Freezer (n=254) Second Freezer (n=12) Small (<15 c.f.) 39% 36% Medium (15-20 c.f.) 46% 36% Large (>20 c.f.) 16% 27% Style Upright, frost-free 28% 25% Upright, manual defrost 25% 25% Chest, frost-free 25% 33% Chest, manual defrost 23% 17% 3.8 HOME OFFICE EQUIPMENT Nineteen percent of Vermont customers operate a business and/or work at home. Among those who work at home, 58 percent work 30 or more hours per week, 25 percent work 11 to 30 hours per week, and 18 percent work 10 or fewer hours per week at home. Not surprisingly, penetration of personal computers and Internet access are high: approximately three in four survey respondents indicated that they had at least one personal computer and/or Internet access (see Table 3-10). Saturation of personal computers is 107 percent. 3 26

SECTION 3 APPLIANCE SATURATION SURVEY Table 3-10 Computer and Office Equipment Penetration and Saturation (n=600) Equipment Type Penetration (n = 600) Saturation (n = 600) Personal Computer 76% 107% Printer 43% 51% Internet Access (any type) 73% Internet access via traditional phone line 42% Cable Modem or DSL 28% Web TV 1% Other Internet Access < 1% Multifunction Machines 30% 36% Scanner 17% 18% Fax 11% 11% Photocopier 9% 9% Eighteen percent of customers have two or more personal computers, and 5 percent have three or more. Seventy-three percent of customers surveyed have access to the Internet. Customers who live in newer homes have higher rates of Internet access than do customers who live in older homes; 94 percent of customers who live in homes built since 1990 have Internet access compared with 85 percent in older homes (homes built prior to 1990). Forty-two percent of customers access the Internet via traditional telephone, while 28 percent use cable modems or DSL, and less than 2 percent use other means of accessing the Internet. 3 27

SECTION 3 APPLIANCE SATURATION SURVEY 3.9 ENTERTAINMENT EQUIPMENT Customers were asked about ten different types of television and entertainment accessories that they use in their homes. The penetrations and saturations are listed below in Table 3-11. Table 3-11 Entertainment Equipment Penetration and Saturation Penetration Saturation Equipment Type (n = 600) (n = 600) Standard Television (36" or less) 95% 183% VCR 87% 118% Stereo 71% 85% DVD Player (or combined DVD/VCR) 67% 86% Digital Satellite Box 38% 48% Analog or Digital Cable Box 32% 40% Video Game Console 25% 34% Video Equipment 19% 21% Large Screen Television (>36") 11% 11% Personal Video Recorder (e.g., TiVo) 5% 6% Penetration of standard-size color television sets is 95 percent and for large-screen televisions (greater than 36 inches), 11 percent. Given the increasing popularity of largescreen televisions this figure is likely to rise in the coming years as costs for larger televisions decline. Customers living in newer homes are more likely to have multiple standard-size televisions than older homes: 62 percent of homes built during or after 1990 have two or more televisions, while only 49 percent of homes built prior to 1990 have multiple televisions. Digital satellite boxes are also more common in newer homes, with 47 percent of homes built during or after 1990 having one or more digital satellite box compared to 37 percent of homes built before 1990. 3.10 MISCELLANEOUS APPLIANCES RASS respondents were asked about whether they use twenty-two types of appliances including fans, hot tubs, pumps, and other plug loads. The penetrations and saturations are listed below in Table 3-12. A number of these appliances (including cordless telephones, portable appliances or tools, kitchen range vent fans, and others) have achieved fairly high penetrations and are thus likely contributors to rising residential electricity demand in Vermont. 3 28

SECTION 3 APPLIANCE SATURATION SURVEY Table 3-12 Miscellaneous Appliance Penetration and Saturation Penetration Saturation Equipment Type (N = 600) (N = 600) Fans Kitchen Range Vent Fan 71% 72% Bathroom Fan 63% 93% Ceiling Fan 58% 114% Portable Fan 56% 119% Attic or Whole-House Fan 11% 15% Radon mitigation fans or pumps 1% 2% Pumps Electric pump for well water 44% 45% Swimming pool pump 11% * Aquarium with a pump 7% 7% Whirlpool bathtub 7% * Heaters, Hot Tubs, and Saunas Hot tub or spa 6% * Heat pump water heater 4% 4% Heated waterbed 3% 3% Swimming pool heater 2% 2% Instant hot water dispenser 2% 2% Sauna 1% * Other Plug Loads Cordless telephones 86% 134% Portable appliances or tools 63% 136% Dehumidifier 29% 30% Humidifier 25% 28% Backup portable generator 14% 15% Electronic household air cleaner 13% 15% * Data not acquired by telephone RASS (i.e., survey did not ask how many). Customers living in newer homes (those built during or after 1990) are more likely to use kitchen range, bathroom, and ceiling or paddle fans than customers in older homes (those built prior to 1990). Customers in newer homes are also more likely to use electric pumps for well water (65 percent) than customers in older homes (43 percent). 3 29

SECTION 3 APPLIANCE SATURATION SURVEY 3.11 LIGHTING 3.11.1 Lighting Fixture Inventory The on-site portion of the RASS contained an inventory of all lighting fixtures in the sample homes. We used the data generated by this lighting inventory to assess the extent of the remaining market for screw-in CFLs. Specifically, the RASS collected information on the following: Total number of lighting fixtures in the home; Number of fixtures currently fitted with incandescent, compact fluorescent, and tube fluorescent bulbs; The number of sockets that could be retrofitted with CFLs. These latter included recessed cans. The criteria applied here included the size and shape of the fixture and whether it required specialty type bulbs (in chandeliers for example). Table 3-13 summarizes these results. On average, we found that the full sample of respondent homes could accommodate an additional 24.6 CFLs. Among customers who already had one or more CFL installed, the corresponding figure was 20.3. We thus conclude that, even among current users of CFLs, there is still substantial room for further purchases. Table 3-13 Results of On-Site Surveys: Lighting Fixture & Bulb Inventory Full Sample n = 83 At Least 1 CFL n = 52 No CFLs n = 31 Number of Fixtures 36.6 35.7 38.3 Number of Bulbs 49.7 48.0 52.5 Number of Fluorescent Tubes 6.3 5.9 7.0 Number of CFL Bulbs Installed 4.7 7.5 0.0 Number of Incandescent/Other Bulbs 38.7 34.6 45.5 Number of Bulbs w/ Potential CFL Retrofit 24.6 20.3 31.8 3.12 BUILDING SHELL FEATURES 3.12.1 Windows Numbers of windows per home. The average number of windows in the 83 homes visited for the onsite survey was 22. Of these, 57 percent were of medium size (2 x4 up to 3 x6 ) and 34 percent were small (less than 2 x4 ). The remaining 18 percent were picture windows with and without dividers over 3 x 6. 3 30

SECTION 3 APPLIANCE SATURATION SURVEY Characteristics of windows installed. Table 3-14 displays the distribution of windows in the 83 homes in the on-site sample by window frame material, type of glazing, and the auditor s assessment of window condition. 9 Nearly 80 percent of all windows observed had wood frames, with 16 percent in vinyl. Sixty percent of all windows observed were double glazed. This figure is relatively high given the age distribution of the homes in the sample. Most of the remainder were single glazed. Only six percent of windows were judged to be in poor condition, with 23 percent judged to be in fair condition. Table 3-14 Observed Characteristics of Windows in Onsite Sample Window Frame Material Metal Metal w/break Wood Vinyl Fiberglass Percent of Total Windows 5% 1% 79% 16% 0% Type of Glazing Single Double Triple Other Percent of Total Windows 36% 60% 0% 3% Auditor Rating of Window Condition Very Poor Poor Fair Good Very Good Percent of Total Windows 2% 4% 23% 38% 32% 3.12.2 Insulation Auditors conducting the onsite inspections were instructed to observe and record the thickness and condition of floor and attic insulation. Unfortunately, it was not always possible to gain access to attics and crawlspaces to make those observations. Table 3-15 summarizes observations of the amount of floor and attic insulation recorded. There are only12 observations of floor insulation, but the amount observed is consistent with good practice. Interestingly, only 17 homes, one-third of the total for which observations are available, have attic insulation at or above the RBES-required level of R-30 (for sloped roofs). 9 Factors considered in condition assessment: number and size of infiltration points, condition of glass, tightness of fit, condition of frame, condition of weather stripping. 3 31

SECTION 3 APPLIANCE SATURATION SURVEY Table 3-15 Amount of Insulation R-Value Floor Attic Not Recorded 71 42 Less than R-18 0 5 R-18 to R-24 10 13 R-25 to R-29 0 6 R-30 to R-34 2 4 R-35 or Above 0 13 Total 83 83 3.13 ENERGY EFFICIENCY 3.13.1 Lighting Approximately 71 percent of the customers surveyed were aware of compact fluorescent lamps (CFLs) prior to the survey when asked about CFLs without prompting; with prompting, an additional 20 percent indicated that they had heard of CFLs prior to the survey. The phone RASS data show that fifty-seven percent reported using use at least one compact fluorescent lamp (CFL) inside or outside their homes. The 2005 onsite surveys showed that 63 percent of the participants had at least one CFL installed in their home, compared with 54 percent in 2002. 14 percent use 1 or 2 CFLs, 18 percent use between 3 and 5, 16 percent use between 6 and 10, and 7 percent use 11 or more. Customers in the northeastern part of the state are more likely to have at least one CFL installed than customers in any other part of Vermont (Figure 3-21). Customers in the southwest and southeast are the least likely to use CFLs. 3 32

SECTION 3 APPLIANCE SATURATION SURVEY Figure 3-21 CFL Use by Region 70% 60% 1 or more CFLs 67% 50% 57% 58% 53% 54% 40% 30% 20% 10% 0% All Regions (n=544) Northwest (n=218) Northeast (n=63) Southwest (n=121) Southeast (n=142) Thirty-eight percent of survey respondents indicated that they had purchased one or more CFL within the past two years; the proportion of customers reporting that they received a rebate for their purchase is 14 percent (or 41 percent of customers who reported purchasing a CFL in the past two years). Of the customers who received rebates for the CFL or CFLs they purchased within the 2 years prior to the survey, 90 percent indicated that they would have purchased the bulb(s) if no rebate had been offered. 3.13.2 ENERGY STAR Awareness Seventy-two percent of Vermont customers who participated in the survey indicated that they had heard of the Energy Star label prior to participating in the survey. Thirty-five percent of survey respondents indicated that they made recent purchases of one or more Energy Star labeled appliances. 10 Of the respondents who indicated a recently purchased appliance had the Energy Star label, 78 percent indicated that the Energy Star label was at least somewhat important in their decision to purchase the particular appliance model. 10 Note: the RASS did not limit the term ENERGY STAR appliances responses thus represent purchases of any appliances bearing the ENERGY STAR logo, whether kitchen/laundry appliances, entertainment equipment, or other equipment. 3 33

SECTION 3 APPLIANCE SATURATION SURVEY Roughly fourteen percent of survey respondents indicated that they had purchased a new natural gas or fuel oil furnace or boiler within the past 4 years. 11 Of these, 39 percent indicated that their contractor offered them the opportunity to purchase an Energy Star high efficiency unit. Of those who were offered the Energy Star unit, 83 percent indicated that they purchased the unit, indicating that contractors recommendations may have significant influence over customers purchasing decisions for new heating systems. 3.13.3 Efficiency Vermont Awareness Fifty percent of respondents indicated that they were aware of Efficiency Vermont prior to participating in the RASS. Eight percent of survey respondents indicated that they had received a rebate from Efficiency Vermont for the purchase of an energy-efficient appliance; of these, 54 percent indicated that the rebate was at least somewhat important in their decision to purchase the particular appliance model. Four percent of survey respondents indicate having used other services provided by Efficiency Vermont. When asked to name these services, several respondents mentioned that Efficiency Vermont provided a home energy audit or that they had purchased energy-efficient CFLs from Efficiency Vermont. Thirty-six percent of respondents indicated that they were aware the Efficiency Vermont offers instant store rebates on CFL bulb and fixture purchases. 3.14 CUSTOMER DEMOGRAPHICS The telephone RASS included several questions on customer demographics. Key points include: More than one in four Vermont households (27 percent) have one or more residents over the age of 65, and 15 percent have one or more resident under the age of 6. Data from the 2000 Census show that 13 percent of households have at least one resident aged 65 or older, and 6 percent of households have children under 5. The telephone RASS thus included a slightly higher proportion households with young children and elderly occupants than the general population. Telephone RASS results show that 84 percent of respondents live in owner-occupied homes, with an average of 2.5 occupants per home; this data is fairly consistent with data from the 2000 Census which show 70 percent of residents in owner-occupied homes and an average household size of 2.6 occupants. RASS results show that 78 percent of homes have 3 or fewer occupants, while 8 percent have 5 or more. 11 Note: this question was asked only if the respondent s heating system was a natural gas or fuel oil furnace or boiler and the system was between 0 and 4 years old. 3 34

SECTION 3 APPLIANCE SATURATION SURVEY Vermont residential customers are highly educated: according to telephone RASS results, 46 percent indicated at least one of their heads of household earned a 4-year college degree, compared with 29 percent of residents over age 25 in the 2000 Census. 3 35

4 REFRIGERATOR TURN-IN ASSESSMENT 4 REFRIGERATOR TURN-IN ASSESSMENT 4.1 INTRODUCTION 4.1.1 Objectives The purpose of this analysis is to estimate the net electricity savings the Energy Efficiency Utility can hope to achieve through a refrigerator and freezer turn-in initiative. This program would be designed to achieve net energy savings through the following mechanisms: Provide incentives to customers using second refrigerators in their home to turn those refrigerators in for recycling, versus continuing to use them. Provide incentives for customers discarding operable primary refrigerators to turn them in for recycling, rather than turning them over to haulers, appliance dealers, or other organizations that will likely sell or donate them for subsequent use in the Vermont market. KEMA used data collected through the 2005 Residential Appliance Saturation Survey (RASS), EVT program data, and evaluations of appliance turn-in programs in other jurisdictions to estimate the following: Likely volume and characteristics of appliances that would be turned in; Average unit annual energy consumption (UEC) for units likely to be collected by the program (a measure of gross savings per unit); A net-to-gross ratio (NTG) that reflects the net effect of a turn-in initiative on a customer s decision to recycle an appliance, versus retaining it in use or transferring it to another user; A measure effective useful life (EUL) that reflects how long the appliance would have remained in service in the absence of a turn-in initiative. KEMA anticipates the DPS and the Energy Efficiency Utility can use these findings, along with cost information from appliance recycling service providers, to assess the cost-effectiveness of pursuing an appliance turn-in initiative. 4.1.2 Overview of Data Sources and Approach The analysis presented below proceeds in the following steps. 1. Estimate the average gross savings associated with the collection of a secondary or discarded primary refrigerator. We attempted to bracket the range of likely average gross savings by taking two approaches to this task. First, we compiled information on 4 1 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT the characteristics refrigerators recently discarded by respondents to the Residential Appliance Saturation Survey (RASS) conducted for the Phase 2 Residential Program Evaluations. 1 We then estimated the average Unit Energy Consumption (UEC) of these appliances by applying the results of a study that modeled the metered UEC of refrigerators turned in to a recycling program operated by Southern California Edison (SCE), using appliance characteristics such as age, size, and defrost type as the independent variables. 2 We put forward the results of this analysis as a minimum gross savings, given that the refrigerators recovered by a recycling program will likely be older and in worse condition than the general population of discarded units. We use results from metering units that EVT collected as part of its Low-Income Single Family (LISF) program as an upper bound for the estimate of gross savings, under the assumption that refrigerators replaced through that program are likely to be less efficient than those recovered through a recycling program targeted to all customers. Ideally, we would have estimated likely savings by applying the SCE model to information on the characteristics of second refrigerators held by RASS respondents. However, there were too few such appliances in the sample to follow that procedure. 2. Estimate likely levels of participation and net-to-gross ratios. We used the results of the aforementioned SCE study, which assessed a program that sought turn-ins of both primary and second refrigerators, to estimate likely levels of participation and net-togross ratios. 3. Estimate the Effective Useful Life (EUL) of savings achieved through for refrigerators turned in through the program. Given the definition of net savings given above, the time of failure for this measure can be defined as one of two events. First, the program participant may acquire another second refrigerator. Second, the time of failure for the measure can be defined as the point at which the recycled refrigerator would otherwise fail or be recycled. Estimating this interval requires a fairly complex set of calculations to estimate a survival curve based on participant survey data. These survival curves yield estimates of the median time to failure for the measure of removing an operable refrigerator from service. KEMA has conducted such a study in California. 3 We use the results of this study to provide an estimated EUL for refrigerators recycled through a prospective program in Vermont. 1 See Section 3 for a full report of the methods and results of the RASS. 2 KEMA-XENERGY, Measurement and Evaluation Study of 2002 Statewide Residential Appliance Recycling Program, Prepared for Southern California Edison. February 13, 2004. 3 KEMA-XENERGY, Ninth Year Persistence Study of Southern California Edison s 1994 1997Appliance Recycling Program, Prepared for Southern California Edison. July 22, 2004. 4 2 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT 4.1.3 Key Findings and Recommendations Market characteristics Saturation of secondary units. Twelve percent of Vermont housing units in 1-4 family structures have two refrigerators; one percent have three. High usage of secondary units. The vast majority (roughly 90 percent) of respondents who reported having a second unit report that their second and third units are always plugged in. Secondary units tend to be older than primary units. Thirty-five percent f RASS respondents report that their primary refrigerator is more than 10 years old. However, 60 percent of respondents who indicated they have a second refrigerator currently in use report that their second refrigerator is more than 10 years old. Fourteen percent of all primary units are fifteen years or older; as are 30 percent of secondary refrigerators. Trends in Refrigerator Disposal Of the 600 RASS respondents, 128 (21 percent) reported that they had had a refrigerator and/or freezer removed from their home in the past three years. The majority of all reported discarded units (62 percent) were 15 years or older. The vast majority (98 percent) of respondents who disposed of one refrigerator in the past 3 years said that the refrigerator had been their primary refrigerator (not an extra or a back-up unit). Of the primary units discarded, 64 percent working units (just old or replaced). Twelve percent were working, but in need of repair when discarded, and nearly 24 percent said these units were not working. Gross Savings Estimates Results of applying UECs of California turn-ins to population of discarded refrigerators. When we re-weight the UECs estimated from a recent KEMA SCE Appliance Recycling Program based to reflect the characteristics of the refrigerators that Vermont RASS respondents discarded in the past three years, we estimated the UEC for those units at 1,899 kwh per year. We believe this represents a lower bound of the average consumption of refrigerators that would be turned in to a prospective program and is thus a good conservative estimate of average unit gross savings. Indeed, secondary refrigerators, which would be more likely to be turned in than primary refrigerators under most program designs, are older on average than the general population of discards. UECs of refrigerators collected by the EVT Low-Income Single Family Program. The average UEC of refrigerators replaced through EVT s LISF program was 1,624 kwh per year. It is difficult to compare this result to the average UEC yielded by the California model because the age distribution of the models collected is not known. It would be reasonable to assume that the median age of the units replaced by the LISF 4 3 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT program is lower than that of refrigerators that would likely be recovered by a recycling program. That is because the units recovered through the LISF program are not necessarily near the end of their useful lives. Participation and Net-to-Gross Ratios Participation. Given the high levels of residential program participation that EVT has achieved (relative to other program operators), one must use caution in using the results of programs in other states as a guide. Also, participation will depend on program design. For example participation will be lower if the program is restricted to collecting only operating second refrigerators as opposed to all discards of a certain age or energy consumption. The 2002 Southern California Edison Program attracted the participation of 1.1 percent of residential customers. For the lack of additional data points, we use this figure for estimating potential savings from the program. Net-to-Gross Ratios. As mentioned above, the NTG ratio reflects what participants would have done with refrigerators they recycled in the absence of the program. In the SCE study, the NTG ratio was developed using data from surveys of program participants, surveys of non-participants who discarded a refrigerator in the program timeframe, and surveys of nonparticipants who acquired a used refrigerator in the program time frame. KEMA used the results of these surveys to estimate the likelihood of alternative dispositions for the recycled equipment (i.e. retention in service somewhere in the SCE territory) and the effect of the program in preventing that alternative disposition. For the 2002 program, KEMA estimated an NTG ratio of 0.35 for refrigerators and 0.54 for freezers. This estimate has varied significantly from year to year, using the same basic methods for estimation. According to the study s authors, this variation can be attributed in part to annual fluctuations in the percentage of primary refrigerators recycled. Customers disposing of primary units report that they would have discarded the unit in the absence of the program more frequently than discarders of secondary units. These results are reviewed in greater detail below. Net Energy Savings Potential By multiplying the estimated average UEC, participation rates, number of eligible households, and NTG ratios together, we arrive at an estimate of 1,824 MWh/Year in potential net electricity savings for a refrigerator recycling program in Vermont. We are confident that the unit gross energy savings estimate provides a reliable guide to expected savings, with 5 10 percent. We have less confidence in the participation and Net-to-Gross estimates, because they have been shown to vary significantly over time in other jurisdictions and because we have little data on which to base a judgment regarding how Vermont customers and appliance dealers will respond to the program. We note that the 2002 SCE program was deemed to be cost-effective, but just barely, using the Total Resource Cost Test. 4 4 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT 4.2 MARKET CHARACTERISTICS 4.2.1 Overview The purpose of this section is two-fold: To identify specific attributes of refrigerator and freezer models currently in use in Vermont households, and To summarize the refrigerator and freezer characteristics and disposal methods of units reported by RASS respondents as recently discarded. These data will also serve as the basis for unit energy consumption (UEC) analysis. 4.2.2 Refrigerator and freezer units currently in Vermont households This section will focus on the RASS data collected regarding the number and characteristics refrigerator and freezer units currently in Vermont homes. In addition to quantifying the number of units, we will present RASS data in order to explore: Multiple units in each household, Usage frequency of second and third units, and Existing unit characteristics, such as age, size, and configuration for primary, secondary, and tertiary refrigerators and freezers currently in Vermont households. Table 4-1 shows the distribution of RASS respondents who report one or more refrigerator and freezer units currently in their household, by dwelling unit type. Holdings of multiple refrigerators. A small proportion of RASS respondents (12 percent) report that they currently have two refrigerators. An addition 1 percent own 3 refrigerators. Customers in single family homes are more likely to have multiple units than residents of multifamily buildings or mobile homes. Freezers. Forty-two percent of customers reported owning stand-alone freezers. None reported owning more than one. Again, customers in single-family homes were more likely to own freezers than customers in multifamily buildings or mobile homes. 4 5 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT Table 4-1 Number of Existing Refrigerator and Freezer Units by Household Type Number of Units Reported at Each Residence ALL (n = 600) Manufactured or mobile home (n = 59) Single family home (n = 448) Duplex or triplex (n = 32) Multifamily building with 4 or fewer units (n =57) Refrigerators One Refrigerator 87% 95% 85% 88% 91% Two Refrigerators 13% 5% 14% 13% 9% Three Refrigerators 1% 0% 1% 0% 0% Freezers No Stand-Alone freezers 58% 59% 53% 69% 88% One Stand-Alone freezer 40% 39% 45% 28% 12% Two Or More Stand-Alone freezers 2% 2% 2% 3% 0% Source: 2005 Vermont RASS Market segmentation. When a homeowner (as opposed to a renter) has more income and more space, the RASS data reflect that he or she is three times more likely to have a second refrigerator. Homeowners are more likely to have multiple refrigerators than renters. Fifteen percent of homeowners have more than one refrigerator, while only 3 percent of renters have more than one. Older homes. The existence of multiple refrigerators is more common in older homes with year-round occupancy: 72 percent of homes built before 1960 have more than one refrigerator, compared with less than 23 percent in homes built after 1990. Larger homes with two or more floors. 73 percent of respondents who reported having two or more refrigerators live in a home with two or more levels. In addition, threefourths of those respondents reporting a second or third refrigerator unit live in a household of more than 2,000 square feet. 4.2.3 Usage Factors Table 4-2 shows that the majority of RASS respondents (between 88 percent and 100 percent) report that their second and third units are always plugged in, meaning that units other than their primary refrigeration units do not tend to be in storage or unplugged. Ninety-eight percent of respondents indicated they lived in a year-round residence (as opposed to a seasonal or vacation home). All but one respondent who reported having more than one refrigerator and/or freezes currently in use resides in a year-round residence. 4 6 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT Frequency unit is plugged in Table 4-2 Frequency Current Units are Plugged-In Second Refrigerator (n = 77) Third Refrigerator (n = 3) Second Stand- Alone Freezer (n = 12) Always 88% 100% 92% Sometimes 9% -- 8% Never 3% -- -- Source: 2005 Vermont RASS 4.2.4 Characteristics of units in service Tables 4-3 and 4-4 summarize the characteristics (age, size, style, configuration) of the refrigerators and stand-alone freezers RASS respondents report are currently in their dwelling unit, revealing several trends that point to market potential for an expanded refrigerator/freezer turn-in strategy: Age Refrigerators. According to the Technical Resource Manual, refrigerators have a measure life of 17 years. Fourteen percent of primary refrigerators and 30 percent of secondary refrigerators are 15 years old or more. Freezers. Thirty-seven percent of primary freezers are 15 years or older, as are 11 percent of secondary units. Size / Configuration Among respondents who indicate they have a second refrigerator in their home, most of these units (49 percent) are medium in size (15 to 20 cubic feet). Nearly 30 percent of all second stand-alone freezers were categorized as large (more than 20 cubic feet). Two-thirds of all refrigerators currently in respondents homes are automatic/frost-free units; more than 50 percent of all stand-alone freezers are frost-free units. 4 7 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT Existing Refrigerator Characteristics Table 4-3 Existing Refrigerator Characteristics 1 st Refrigerator 2 nd Refrigerator 3 rd Refrigerator Age (n = 556) (n = 73) (n= 3) Less than 1 year old 9% 8% -- 1 to 4 years old 30% 19% -- 5 to 9 years old 26% 12% -- 10 to 14 years old 21% 30% 67% 15 to 19 years old 8% 5% -- 20 years or older 6% 25% 33% Size (n = 563) (n = 73) (n = 3) Mini: (less than 10 cubic feet) 2% 21% 33% Small: (10 to 14 cubic feet) 9% 21% -- Medium: (15 to 20 cubic feet) 66% 49% 33% Large: (more than 20 cubic feet) 24% 10% 33% Style (n= 570) (n = 71) (n= 3) Single door with the freezer inside 9% 38% 33% 2 side-by-side doors 24% 14% -- Top freezer 61% 41% 67% Bottom freezer 7% 7% -- Defrost (n= 587) (n = 77) (n= 3) Automatic / Frost-free 90% 62% 67% Manual 10% 35% 33% (none / no freezer) 1% 3% -- Ice/Water (n= 600) Through-door ice & water 17% -- -- Source: 2005 Vermont RASS Note: Number of responses varies in each category since don t know responses were removed; percentages may not sum to 100% due to rounding. Table 4-4 Existing Stand-Alone Freezer Characteristics Existing Freezer Characteristics One Stand-Alone Freezer Two or More Stand- Alone Freezers Age (n = 242) (n = 9) Less than one year old 5% -- 1 to 4 years old 19% 33% 5 to 9 years old 21% 11% 10 to 14 years old 17% 44% 15 to 19 years old 12% -- 20 years or older 25% 11% Size (n = 241) (n = 11) Small: (less than 15 cubic feet) 39% 33% Medium: (15-20 cubic feet) 46% 33% Large: (more than 20 feet) 16% 27% Style (n = 252 ) (n = 12) Upright, frost-free 28% 25% Upright, manual defrost 25% 25% Chest, frost-free 25% 33% Chest, manual defrost 23% 17% Source: 2005 Vermont RASS Note: Number of responses varies in each category since don t know responses were removed; percentages may not sum to 100% due to rounding. 4 8 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT 4.2.5 Recently Discarded Units Of the 600 RASS respondents, 128 (21 percent) reported that they had had a refrigerator and or freezer removed from their home in the past three years. This section presents RASS findings in regard to unit characteristics for discarded refrigerators and freezers. Table 4-5 shows the breakdown of the units that RASS responded they had removed from their home by disposal method. A majority of respondents reported that the unit(s) they discarded (51 percent) were removed by an appliance dealer when a new unit was purchased. The second-most reported method of discarding one or more units was to give or sell the unit(s) to another individual (27 percent). This transfer of high use refrigerators between residential dwelling units is an important aspect of the secondary market supply chain that is difficult to track. Table 4-5 Disposal Methods Reported Over the Past 3 Years Disposal Method Removed by an appliance dealer when a new unit was purchased Removed by a private hauler Removed by a refuse service Given or sold privately to another Sold to a second-hand appliance dealer Given to a recycling center/area First Unit (n = 117) Second Unit (n = 10) Three or More Units (n = 1) TOTAL (by disposal method) (n = 128) 52% 40% -- 51% 9% -- -- 9% 6% -- -- 5% 25% 50% 100% 27% 6% 10% -- 8% 2% -- -- 2% Source: 2005 Vermont RASS Of these 128 respondents, 104 were able to offer details regarding the units they discarded: 83 percent disposed of 1 refrigerator; 11 percent disposed of 1 refrigerator and 1 freezer; and 7 percent of households disposed of only 1 freezer. The characteristics of these discarded refrigerators and freezers (for example, age, size, and configuration) are reported in Table 4-6 in the context of UEC analysis of these discarded units. Several important characteristics of recently discarded units are reported in this analysis: 4 9 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT The overwhelming majority of all reported discarded units (62 percent) were older models, 15 years or older. The vast majority (98 percent) of respondents who disposed of one refrigerator in the past 3 years said that the refrigerator was their primary refrigerator (not an extra or a back-up unit). Of the respondents who reported discarding a unit in the past three years and recalled whether their unit was working or not when they discarded it, 98 percent of these respondents characterized the unit they discarded as a main (not a back-up) refrigerator unit, and all of these units were more than 15 years old. Interestingly, 64 percent of these older model discarded primary units reported were working units. Twelve percent of these respondents said these units were working, but in need of repair when discarded, and nearly 24 percent said these units were not working when they decided to discard them. 4.3 UNIT ENERGY CONSUMPTION (UEC) ANALYSIS Estimating the gross unit energy consumption (UEC) for turned-in models is key to determining an initiative s potential savings impacts. In order to consider potential savings from a future statewide Vermont refrigerator turn-in strategy, we estimate UEC in two ways: We compiled the characteristics (age, size, configuration) of the refrigerators and freezers that RASS respondents report they discarded over the past three years. By re-weighting the UEC data KEMA compiled in a previous study using RASS turn-in data, we were able to identify that the distribution of turned-in characteristics among Vermont RASS respondents and SCE s turn-in program were similar enough to offer a good estimate for gross per turned-in unit savings, which we are then able to use to suggest gross program savings. We believe the UEC estimate generated in this fashion provides a good, lower bound or conservative estimate of gross unit savings that can be achieved through the prospective initiative. We also collected data from EVT s Low-Income/Single Family (LISF) Refrigerator Replacement Program. LISF data over the past five years includes UEC of both replaced and new units, allowing us to estimate per unit gross savings reported in this data. As discussed above, we believe the average UEC of these units represents a reasonable upper bound for an estimate of the gross unit savings to be achieved through a prospective program. 4.3.1 UEC Estimates Developed by Modeling RASS Data In order to estimate Unit Energy Consumption (UEC) for Vermont residential refrigerator and freezer units, KEMA first utilized the characteristics (age, size, and configuration) of the refrigerators and freezers that RASS respondents report they discarded over the past three years. We applied the distribution of these RASS-reported discarded unit characteristics to the UEC 4 10 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT model KEMA calculated for Southern California Edison s (SCE) Residential Appliance Recycling Program (RARP) 4. We chose to re-weight the SCE UEC data using 2005 Vermont RASS-reported discarded unit characteristics, since amperage (as a basis for UEC per unit) was unknown for the discarded unit reported through the RASS. This approach also allowed us to identify whether the characteristics of the discarded units Vermont RASS respondents reported were comparable (in terms of age, size, and configuration) to units turned-in under SCE s program. The UEC data KEMA had utilized to develop program savings in the evaluation of SCE s RARP came from two sources: (1) a large database of existing lab-metered data, and (2) a new metering study of a sample of 100 units recently collected by the program. Both of these data sources used the Department of Energy (DOE) lab metering protocols to collect metered data. The metering study was conducted over a 6-month period, from late May to late November 2003. This sample of metered units was stratified by the following characteristics: Unit type (refrigerator or freezer); Size (cubit feet); Defrost type (manual, automatic, partial defrost); and Configuration SCE Model application to Vermont RASS data When designing the RASS, KEMA included questions about refrigerator unit type, size, defrost type, and configuration for both models that were currently in use and models that respondents reported they discarded in the past three years. Utilizing the same model characteristics that were stratified in the SCE analysis facilitates the comparison. However, unlike the SCE evaluation, we did not have amperage estimates for the discarded models RASS respondents report 5 Of the 600 RASS respondents, 128 (about 21 percent) reported discarding or selling a refrigerator or freezer in the past 3 years. Of these 128 respondents, 104 were able to report specific characteristics of their discarded units. The characteristics we inquired about were stratified in the same categories as the SCE model. Table 4-6 presents the distribution of SCE refrigerator/freezer model characteristics and reweights these UEC data based on the distribution of discarded units that Vermont RASS respondents reported in each characteristic category. 4 KEMA-XENERGY, Measurement and Evaluation Study of 2002 Statewide Residential Appliance Recycling Program Prepared for Southern California Edison. February 13, 2004. 5 We considered inquiring through the RASS whether the respondent knew (or could estimate) the amperage of their discarded units, but determined that these responses would most likely not approximate accuracy for the purposes of calculating savings estimates. 4 11 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT As Table 4-6 shows, despite the variations in data collection, applying the distribution of Vermont discarded units to the SCE UEC model clearly shows that there is a similar distribution of unit characteristics for those turned-in under the 2002 SCE RARP and the distribution of units that the 2005 sample of Vermont residential respondents reported discarding over the past three years. One important caveat when comparing the distribution of 2005 Vermont RASS-reported discarded units and SCE turn-ins is the effect of the program incentive structure and program information (marketing). The distribution of refrigerators (89 percent) and freezers (11 percent) turned-in under the SCE RARP closely matches the distribution of refrigerators that Vermont RASS respondents reported discarding (83 percent and 17 percent, respectively). The units turned-in under the SCE program that were 15 years or older (76 percent) is slightly higher than the older units that RASS respondents reported discarding (52 percent). Both SCE participants and Vermont RASS respondents reported that the vast majority of refrigerators they discarded were frost-free, with a significantly higher annual UEC when compared with manual or partial defrost model configurations. Both SCE participants and Vermont RASS respondents reported a higher rate of refrigerators turned in with a top-freezer configuration, though the SCE program reported a higher ratio of top freezer units compared with side-by-side turned in under the RARP program (3-to-1). Both SCE participants and Vermont RASS respondents reported discarding refrigerators with 18 or more cubic feet predominately (63 percent in the SCE analysis for larger unit turn-ins, compared with 74 percent of Vermont respondents reporting discarding refrigerators 18 cubit feet or greater). 4 12 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT Table 4-6 Vermont UEC Estimates based on RASS Turn-in Characteristics 2002 SCE Model Data* Vermont Recent Discarders Defrost Configuration Size Agebins Unit Characteristics Overall Freezer Refrigerator Freezer Refrigerator Freezer Refrigerator Freezer Refrigerator Freezer Refrigerator # of 2002 CA Participants % of Participants UEC (estimated kwh per year) % of VT RASS Respondents (n=104) Reweighted Vermont UEC (estimated kwh per year) 42945 1914.96 1897.97 4735 11% 1661.98 17% 38210 89% 1946.31 83% Frost-Free 1861 4% 2234.3 13% Manual Defrost 2584 6% 1286.14 5% Partial Defrost 290 1% 1338.05 0% Frost-Free 34559 80% 2074.41 62% Manual Defrost 2995 7% 641.6 19% Partial Defrost 656 2% 1154.58 0% Chest 1708 4% 1631.52 6% Upright 3027 7% 1679.16 12% Bottom Freezer 1088 3% 2195.66 3% Sindgle Door 2185 5% 908.35 19% Side by Side 9250 22% 2602.69 27% Top Freezer 25687 60% 1787.67 33% < 18 cu. ft 2596 6% 1451.93 6% 18 to 20 cu. ft. 1502 3% 1827.17 5% > 20 cu. ft. 637 1% 2128.47 7% < 18 cu. ft 11528 27% 1485.67 10% 18 to 20 cu. ft. 17831 42% 1962.35 36% > 20 cu. ft. 8851 21% 2513.94 38% 5yrs 34 0% 1786.7 2% 10yrs 550 1% 1609.47 4% 15yrs 708 2% 1650.67 3% 20yrs 1466 3% 1654.36 3% 25yrs 768 2% 1715.47 4% 30yrs 1209 3% 1664.23 0% 5yrs 350 1% 1595.89 10% 10yrs 5284 12% 1801.23 21% 15yrs 9631 22% 1956.15 17% 20yrs 13476 31% 2031.31 21% 25yrs 5214 12% 2027.33 13% 30yrs 4255 10% 1764.53 0% 1775.38 1828.56 2098.76 1865.75 * SCE model data is based on KEMA's "Measurement and Evaluation Study of SCE's 2002 Statewide Residential Appliance Recycling Program." (February 2004). 4 13 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT 4.3.2 UEC Estimates from Refrigerators Replaced by EVT LISF Program The Refrigerator Replacement Component of EVT s Low Income Single Family (LISF) initiative requires contractors pick-up and remove from the existing refrigerator of an eligible household and replace it with an approved ENERGY STAR model. The LISF budget absorbs 100% of the cost of the eligible ENERGY STAR replacement models, and results in no flow of old units to the secondary market, as contractors are required to render inoperable and dispose of old refrigerators and must sign a disposal agreement. In addition, this initiative specifies verification and tracking procedures, to serve as the backup for existing and new kwh assumptions in order for Efficiency Vermont to track measure savings. In previous years, Burlington Electric Department (BED) attempted to remove high use (> 1,100 kwh) refrigerator units from the secondary market by offering a $100 per unit incentive to ReCycle North. However, in 2003, ReCycle North reported that they purchased five (5) refrigerators > 1,100 kwh this program. According to a BED manager, since there is no way to track or verify the models that are bought and sold by a recycling center, the market penetration for re-sale is unclear. One approach being pursued by BED is to collaborate with ReCycle North to include a version of the Energy Guide label to let consumers know the kwh of the repaired models they are selling so consumers can make a more informed decision. Methodology As part of the Refrigerator Replacement Component of EVT s Low Income Single Family (LISF) initiative (offered since 2001), EVT verifies and tracks information relating to the unit including being replaced, including: make, model, volume (cu. ft.), age and condition of the existing unit. Existing and new kwh assumptions are tracked in order for Efficiency Vermont to track measure savings. Savings estimates (kwh per year) for this initiative are ideally developed as a result of metering. When auditors are unable to meter the existing unit, the auditor will estimate usage by looking up the model number in the Planergy Refrigerator Manual. The manual provides the manufacturer s estimated annual kwh value that may be de-rated based on age and model type. Results Table 4-7 shows the UECs for refrigerators and freezers replaced as a result of the LISF initiative from 2001 through June 2005 (1,468 refrigerators; 98 stand-alone freezers). EVT reports gross average savings of 1,199 kwh per refrigerator unit replaced (and 1,078 kwh per freezer replaced) through this targeted initiative. 6 The number of both refrigerators and freezers turned-in and recycled (removed from the supply chain) has increased each year, with the average UEC occurring in 2003 (1,168 6 2005 data is included, but is weighted by number of units reported through June. We expected the number of units replaced at the end of 2005 to be higher than what is reported in Table 10. 4 14 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT kwh savings for an average turned-in refrigerator compared with an average per unit savings of 1,1341 per year for freezers). The majority of turned-in refrigerator units were medium-sized (15-20 cubic feet), turned-in nearly twice as often as small or larger units. Table 4-7 LISF Refrigerator and Freezer Characteristics and Savings: 2001--2005* Refrigerators Size (Turned-In Units) UEC Measure Type** Annual Gross Savings (kwh) Average Annual UEC (kwh) Freezers Size (Turned-In Units) UEC Measure Type Annual Gross Savings (kwh) Average Annual UEC (kwh) Unit Characteristics Source: EVT LISF Program Data 2001 (n = 119) 2002 (n = 375) Program Year 2003 (n = 430) 2004 (n = 422 *) 2005 (n = 122) OVERALL (n = 1468) Small 31 100 96 94 31 352 Medium 72 212 280 273 67 904 Large 16 63 54 55 24 212 Metered 104 266 236 219 57 882 Estimate 15 109 194 202 65 585 Old Units (Total kwh) 211,707 611,615 690,440 654,557 189,831 2,358,150 New Units (Total kwh) 51,417 159,054 188,004 177,092 50,800 626,367 Total Gross Savings 160,290 452,561 502,436 477,465 139,031 1,731,783 Old unit average UEC 1,779.05 1,630.97 1,605.67 1,551.08 1,555.99 1,624.55 New Unit average UEC 432.08 424.14 437.22 419.65 416.39 425.90 Annual Average Per Unit Saving 1,346.97 1,206.83 1,168.46 1,131.43 1,139.60 1,198.66 2001 (n = 0) 2002 (n = 4) 2003 (n = 23) 2004 (n = 54) 2005 (n = 17) OVERALL (n = 98) Small -- 2 13 34 11 60 Large -- 2 10 20 6 38 Metered -- 4 21 49 17 91 Estimate -- 0 2 5 0 7 Old Units (Total kwh) -- 6,156 41,659 80,329 22,263 150,407 New Units (Total kwh) -- 2,019 10,818 23,100 7,382 43,319 Total Gross Savings -- 4,137 30,841 57,229 14,881 107,088 Old unit average UEC -- 1,539 1,811 1,488 1,310 1,537 New Unit average UEC -- 505 470 428 434 459 Annual Average Per Unit Saving -- 1,034 1,341 1,060 875 1,078 4.4 NET-TO-GROSS (NTG) ANALYSIS 4.4.1 Details of the California Analysis In KEMA s recent evaluation of SCE s RARP initiative, the NTG ratio was derived by using uses data from Participant and Nonparticipant Surveys to estimate the credit attributable to the program for picked up units that would otherwise have been discarded. Two components of the NTG were considered: the attribution factor, which indicates how much credit for removal should be given to the program, and the part-use factor. The attribution factor adjusts for the percentage of participants that would have disposed of the unit anyway, and gives partial credit to the program for destroying a unit that would otherwise have been transferred to another user. The part-use factor adjusts for the fraction of the time that participants would have used the unit had they had kept it. To estimate the attribution and part- 4 15 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT use factors, KEMA used survey data from program participants, nonparticipants who had recently discarded a refrigerator, and nonparticipants who had recently purchased a used refrigerator to estimate the following. 1. The fraction of units that would be kept or discarded in some way. These fractions are determined from the Participant Survey. 2. For units that would have been discarded in the absence of the program, the fractions that would have been destroyed and the fractions that would have been transferred. These fractions are determined from the supplemental survey of discarders. 3. The fraction that would have been used as a main unit and what fraction as a spare. Also, for both refrigerators and freezers, determine the proportions of alternate actions that took place because the transfer did not occur. These determinations are based on the survey of customers who acquired used units in some way. KEMA assigned attribution factors to the various possible dispositions and compute the weighted average attribution based on the proportions determined in Steps 1 through 3. We then combined the attribution factor with the part-use factor for each group to determine the NTG factor. Figure 4-1 depicts the calculation of the NTG ratio for refrigerators collected in the 2002 program year. Figure 4-1 Calculation of the NTG Ratio for Recycled Refrigerators: SCE 2002 4 16 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT In order to put these factors in context, the calculated NTG factor for refrigerators in KEMA s SCE evaluation was 0.35. However, this ratio was substantially lower than the value in the last evaluation (0.53). The attribution factor (that is, what savings can be attributed to the program specifically) was the primary source of the lower NTG ratio. There were several reasons KEMA cited for the drop in the attribution factor: The dramatic increases in the shares of participants that were using the recycled unit as a primary, rather than a spare unit (79 percent in 2002 vs. 24 percent in 1996 for refrigerators); An increase of similar magnitude in the percentage of participants that would have disposed the unit without the program (86 percent in 2002 versus 41 percent in 1996 for refrigerators); and A reduction in the share of participants that would have kept the unit in use if the program hadn t picked it up (for refrigerators, 9 percent in 2002 versus 45 percent in 1996). 4.5 EFFECTIVE USEFUL LIFE (EUL) OF SAVINGS As part of KEMA s SCE RARP evaluation, KEMA estimated the length of time that program savings estimates would continue to be realized from two net savings considerations: savings resulting from removing refrigerators/freezers from a household that the dwelling might otherwise (without the program) would have kept (direct savings); and savings resulting from the transfer of older, inefficient appliances to premises within the SCE service territory (indirect savings). Given this consideration of net savings, an appliance would not contribute to net savings is the dwelling added a refrigerator or freezer without removing/ceasing to use the existing unit, or if the existing unit would have been inoperable ultimately. Utilizing program data from SCE, the analysis concluded that the EUL (median survival time) at which half of the first-year impact evaluation savings were no longer realized was between 10 and 13 years. 7 However, because the survival curves calculated for this analysis were based on multi-year program data specific to SCE s Appliance Recycling Program, we present this EUL as illustrative of impacts that would necessarily vary under different program designs. Savings persistence would be a function of program design elements and the survival curves generated from program participation data. We recommend a more specific investigation of EUL specific to Vermont once a statewide refrigerator turn-in and/or recycling initiative is in place. 7 The EUL does not represent the average time the unit is in use, but rather, the median number of years at which half of the first year savings from the program had been realized. This variable is utilized to estimate program savings duration rather than unit life. 4 17 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT 4.6 CONCLUSION AND RECOMMENDATIONS 4.6.1 Conclusions Figure 4-2 shows the four major factors in the estimate of potential annual net electricity savings. Combining these factors together we arrive at estimated net annual savings of 1,814 MWh. By way of comparison, the planning estimate of savings for the clothes washer component of the 2004 Appliance program was 1,238 MWh/Year. Thus, it is clear that an appliance recycling program offers significant savings. As discussed in Section 4.1, we are confident that the estimate of annual unit gross savings represents a reasonable estimate of savings that can actually be achieved with the collection of each unit. We are less confident about the assumed participation rate and NTG ratio. In the case of the participation rate, we have observed in Sections 1 and 2 that Vermont customers tend to participate in energy efficiency programs at a higher level than their counterparts in California. However, any attempt to project a participation rate for Vermont on the basis of observations in other states is somewhat speculative. In the case of the NTG ratio, we note that this performance parameter has varied in unpredictable ways within single jurisdictions. Figure 4-2 Preliminary Estimate of Net Program Electricity Savings Potential Number of Households 249,450 Annual Participation Rate Annual unit gross savings (UEC) Net-to-Gross Ratio Annual Gross Savings 1.1% x 1,899 kwh/year x 0.35 x 1,824 MWh/Year 4.6.2 Recommendations The results of this analysis clearly suggest that further efforts to assess the cost-effectiveness of a prospective appliance recycling program are warranted. We recommend taking the following steps. Approach vendors such as Recycling North and Appliance Recycling Centers of America to obtain non-binding estimates of the costs of operating a program that involves recycling of 1,500 to 2,500 units. Conduct cost-effectiveness screening using the net and gross savings estimates developed for this analysis. Conduct sensitivity analysis by varying the average gross savings figure as high as 2,000 kwh per year, and the NTG ratio as high as 0.50. 4 18 VT DPS

SECTION 4 REFRIGERATOR TURN-IN ASSESSMENT Based on the results of this analysis, assess the likelihood of the program becoming cost effective. 4 19 VT DPS

5 RESIDENTIAL NEW CONSTRUCTION 5 VERRESIDENTIAL NEW CONSTRUCTION 5.1 OVERVIEW 5.1.1 Evaluation Objectives This section provides a characterization of current construction and marketing practices among Vermont homebuilders, especially in regard to energy efficiency in new homes, recognition and understanding of the Vermont Energy Star Homes program, and the perceived effects of this program by builders in Vermont. Data to support this analysis comes primarily from the builder surveys and numerous secondary sources. This section has three primary objectives: 1. Update elements of the new construction baseline developed as part of the Phase 1 Evaluation (2002); 2. Determine builder perceptions of opportunities presented by the new construction program and its effects on new construction practices; and 3. Gauge builders impressions of changes in market and the role of the Vermont ENERGY STAR Homes program in effecting those changes. This section of the report also provides an overview of Efficiency Vermont s residential new construction program (the Vermont ENERGY STAR Homes program), including new construction program activity and participation; a discussion of the data sources from which data was collected and methods used to analyze the data; key findings; and detailed results of the analyses. 5.1.2 Key Findings Key findings from Phase 2 of the residential new construction study include the following: The portion of single-family new homes that enroll in the program is very high compared to participation rates for similar programs elsewhere. In the years for which data were available (2001 through 2003), the program developed leads on 1,551 to 1,950 projects, which corresponds to 61 to 74 percent of the typical volume of single-family homes permitted each year (around 2,500). The program also does a good job converting leads to enrollments. This ratio ranged from 39 to 52 percent in the three years covered. The percentage of permitted single family homes that enrolled in the program remained stable in the range of 29 to 32 percent from 2001 to 2003. EVT has reduced the portion of projects that drop out of the program prior to certification and the pace at which projects are processed. Portion of projects completed in same year as enrollment increased from 14% in 2001 to 36% in 2003. Program participation continues to be concentrated in the Northwest to a greater extent than overall residential new construction activity. In 2003, 68 percent of the \ 5 1 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION completed units were located in the Northwest region v. 48 percent of permitted new single-family homes. This pattern may be attributed in part to the residual effects of having a strong program offered by Vermont Gas Service operating in the region which continues to partner with Efficiency Vermont. The concentration of participating units in the Northwest has decreased slightly since the inception of the program in 2000. Number of builders with projects enrolled and number of builders participating for the first time increased steadily from 2001 through 2003. As of the end of 2003, 200 builders had completed projects through the program. The number of builders completing multiple projects in one year increased to 30 in 2003 v. 12 in 2002. Repeat participation is key to the execution of EVT s strategy to transform the residential new construction market. The depth and quality of energy efficiency measures in participating homes increased significantly from 2002 through 2004. The portion of completed projects with 6+ enduses addressed increased from 38 percent in 2002 to 69 percent in 2004. The portion of completed projects meeting Energy Star rating criterion (86.0) increased from 77% in 2002 to 92% in 2004. Baseline energy efficiency practices improved between the Phase 1 evaluation (2001) and the Phase 2 study (2005), particularly for measures promoted by the EVT program. Inclusion of selected energy efficiency measures, including Energy Star appliances, hard-wired fluorescent lighting fixtures, and use of blower door tests to measure air infiltration, has increased since the Phase 1 study. These increases are likely a result of program influence. KEMA uncovered additional evidence of market transformation in interviews with builders. Key findings in this regard were as follows. Builder awareness of the non-energy benefits of energy-efficient equipment (including increased comfort and lower equipment maintenance costs) increased by statistically significant margins since 2001. Customer requests for Energy Star rated high-efficiency heating and cooling equipment as priced options have increased dramatically since the Phase 1 evaluation: builders representing 39 percent of the new construction market in Vermont indicated that customers request the equipment as compared with only 8 percent in 2001. Builder perceptions of the importance of energy efficiency to the success of their businesses have increased, another likely demonstration of the program s success in transforming the new construction market. Overview of Data Data presented in this section are from the 2005 and 2002 surveys of builders as well as a number of secondary sources. This section provides a brief overview of the data. 5 2 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Homebuilder Surveys. Using a stratified sampling approach based on establishment size and market area, KEMA conducted surveys of homebuilders in Vermont in 2002 and again in 2005. Results were analyzed using population weights or a ratio estimation procedure that yields estimates of market share in terms of total units built (as opposed to the percentage of builders adopting the practice), depending upon the appropriateness of each procedure for each survey question. Results from the 2005 survey are compared with those from the 2002 study, where appropriate, to show trends in new construction practices and energy efficiency program influence over time. KITT Database. KEMA made extensive use of data contained in the program tracking system and of reports prepared from those data by EVT. Secondary Sources. In addition to data from the builder surveys, we relied upon numerous other sources of data to provide an accurate snapshot of home construction activities in the state of Vermont. Most significant among these sources are the following: U.S. Bureau of the Census Building Permit data. The Residential Construction Branch of the U.S. Bureau of the Census provides data on local building permits collected using a mail survey. This data was analyzed to yield a count of new homes constructed in Vermont in 2004. Vermont Department of Taxes Home Sales Price data. Properties sold in the State of Vermont are subject to a Property Transfer Tax based on the selling price, which is recorded by the Vermont Department of Taxes. These data were analyzed to yield estimates of selling prices for homes in Vermont. This database does not distinguish between existing and newly-constructed homes. Vermont Housing Finance Agency (VHFA) and the University of Vermont's Center for Rural Studies Housing Affordability data. Multiple data sources published by VHFA and the Center for Rural studies were used to determine housing affordability in Vermont. Dun & Bradstreet imarket Database Establishment-Level data. The imarket database includes establishment-level data for new construction businesses in Vermont. This database has proven to be reasonably accurate source of data on construction establishments activities and number of employees. 5.1.3 Program Overview Objectives and Operations The objectives of Efficiency Vermont s Residential New Construction program, as stated in the original program plan are to: Increase market recognition of superior construction associated with ENERGY STAR qualified homes; 5 3 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Increase awareness and compliance with the Vermont Residential Building Efficiency Standard; Increase penetration of cost-effective electric and fossil-fuel measures in single family new construction; Improve occupant comfort, health and safety; Institutionalize Home Energy Ratings, and Increase the use of mortgage benefits for energy-efficient homes. The new construction program pursues these objectives by reaching out to builders and other principals in residential new construction projects and offering technical assistance and financial incentives to incorporate energy-efficient design features and equipment. New construction programs have been operating in Vermont for well over a decade. It was not until 2003, when EVT began providing new construction services in the Washington Electric Coop, that one program served all customers statewide. The key elements of new construction program operations are as follows. Marketing. EVT pursues a wide range of strategies to market the program to builders. These practices include mounting an annual conference on energy efficiency and building, placement of articles in trade and popular publications, appearance at home shows, presentation at builders and trade association meetings. Lead Development and Tracking. Leads are developed through a variety of sources, including requests for electric and gas service, Act 250 postings, outreach events, and builders already active in the new construction program. Once a principal in a new construction project is identified, EVT mails the potential participant an application packet. There are separate packets for consumers, builders, and first-time builders. Once the initial packet is sent, Efficiency Vermont follows up in an attempt to secure project enrollment: The Enrollment Process. Efficiency Vermont sends a package to identified leads that includes an Enrollment Agreement outlining the participant s responsibilities and an Energy Features Form for information on the construction project. The participant returns the Agreement and Form, and project plans to Efficiency Vermont. Staff review the materials for completeness, enroll the project in the tracking system, forward plans and/or energy features form to EVT, and contact the participant to confirm enrollment or request additional information. EVT then reviews the plan and provides technical assistance. Project Management, Inspection, and Closeout. When EVT receives the project package, staff contact the participant, review the plan, and develop an initial energy rating. EVT provides technical assistance to ensure that the home as will achieve the needed energy rating level and that the home meets ventilation and lighting criteria. EVT monitors the construction process through follow-up calls to the participant. Once the project is complete 5 4 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION EVT conducts the final inspection. Efficiency Vermont notifies the participant if changes need to be made to qualify for certification and incentives through the new construction program. Once the project qualifies, Efficiency Vermont notifies the participant, computes the rebate amount, and notifies EVT of completion. To qualify for the Vermont Star Home designation, a house had to achieve a Home Energy Rating of 86, which is equivalent to the U. S. Environmental Protection Agency s 5-star ENERGY STAR home rating. Generally, homes must contain high levels of insulation, efficient heating and hot water equipment, and high-quality air sealing measures to meet this rating. (Homes that score 86 or above in the Home Energy Rating will use approximately 20 percent less energy for heating, cooling, and hot water than those that meet the minimum requirements of Vermont s Residential Building Energy Standard.) In addition, qualifying homes need to contain at least 10 energy efficient lighting fixtures or 30 percent (whichever was lower) and efficient mechanical ventilation systems. The incentive structure has been substantially revamped in the past few years. Participating builders now receive a $100 incentive for homes that meet the basic thermal requirements needed to qualify for the 86 HERS rating plus low wattage mechanical ventilation, heating equipment that draws combustion air from the exterior, and a minimum of 4 qualified lighting fixtures. Builders also receive incentives of $15 for each surface mounted energy efficient lighting fixture and $25 for recessed cans. EVT will also procure a certificate of compliance for the Residential Building Energy Standard (RBES). Builders may earn an additional $700 in incentives for the installation of at least 10 qualifying lighting fixtures plus 3 ENERGY STAR appliances, which may include central heating, central air conditioning, refrigerators, clothes washers, or dishwashers. 5.2 NEW CONSTRUCTION PROGRAM ACTIVITY AND PARTICIPATION Findings and recommendations from the first evaluation. The first evaluation, completed in 2003, identified a number of issues that needed to be addressed in order to strengthen the performance of the new construction program. These were as follows. Conduct outreach to improve builder and customer understanding of the new construction program. The first evaluation found that most builders and customers were confused regarding program benefits and procedures. Some of this confusion may be due to frequent changes in the program name and features between 1999 and 2003. To address this situation, EVT had already begun to undertake a number of programs in 2002, including targeted mail and phone call campaigns to builders statewide, outreach to municipal officials, targeted outreach to builders of manufactured homes, and assignment of business development specialists to identify opportunities in regions outside Chittenden County, which were generally underrepresented by builders and participants. KEMA will explore the effects of these programs through interviews with builders and other principals in new construction projects. 5 5 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Improve the capability of the program tracking system. At the time of the last evaluation, EVT was in the process of implementing its current program tracking system. Program contractors and EVT had experienced difficulties producing useful reports of program activities, particularly in regard to the progress of projects from one stage to another. It thus appeared that there was a large amount of attrition once initial contacts were made, but little information available to understand such patterns or to manage the process effectively. Also, the final disposition of some projects was difficult to ascertain. Based on review of the current tracking system, we believe these issues have, to a large extent, been addressed successfully. We have made extensive use of information from the tracking system and the Beehive milestone tracking application in compiling the quantitative information discussed below. In the paragraphs that follow we use analysis of the program database to explore how the program has addressed some of the issues brought up in the last evaluation. Patterns of Recent Program Activity: Success in Enrollment. One way to assess the operation of the program is to estimate the percentage of new single-family homes permitted each year that reach the various stages of project development. Table 5-1 shows the number of 1-2 family homes permitted each year, per the U. S. Census Survey of Construction Permits, as well as the number of leads qualified and projects enrolled. We focus on the years beginning in 2001, the year the program began to take its current statewide shape. Since many factors can delay project completion for months, if not years, it is not appropriate to compare completions to permits issued in a year. It is more appropriate to look at the change from year to year in the elapsed time required to move projects from lead identification to completion. We examine that in another table. Table 5-1 Trends in Lead Identification and Enrollment Characteristic 2003 2002 2001 Permits Issued for 1 & 2 unit homes 2,544 2,599 2,431 Leads Qualified as Eligible 1,553 1,950 1,551 Single Family Participants Enrolled 803 760 699 VESH or VT-Star Homes Completed 317 266 194 Drop Outs (Terminations) 99 260 190 Leads as % of Permits 61% 75% 64% Enrollments as % of Leads 52% 39% 45% Enrollments as % of Permits 32% 29% 29% As Table 5-1 shows, Efficiency Vermont has done a good job of identifying and reaching principals in single-family new construction projects. In the years covered, the program has reached principals of 1,551 to 1,950 projects, which corresponds to 61 to 74 percent of the typical volume of single-family homes permitted each year (around 2,500). This is a very high 5 6 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION proportion compared to other states. Vermont s small size and the long tenure of the Efficiency Vermont staff in their positions contribute to this record. The program also does a good job converting leads to enrollments. This ratio ranged from 39 to 52 percent in the three years covered. Generally speaking, the performance of the program in identifying leads and enrolling customers is good, but has not changed significantly since the program took its current form. Finally, attrition from the program decreased by 63 percent between 2002 and 2003. However, it is too soon to determine whether this is a trend. Time to Completion. Table 5-2 shows the distribution of completed units by year of enrollment for the units completed in 2001 through 2003. The percentage of units completed in the same year as they enroll increased from 14 to 36 percent. However, the percentage of units completed in the two years prior to completion varied from 60 percent in 2002 to 92 percent in 2001. These results reflect the fact that many factors beyond the control of EVT and, at times, builders affect the completion time for projects enrolled in the program. Table 5-2 Year of Completion v. Year of Enrollment Year of Enrollment v. Completion 2003 2002 2001 Enrolled Same Year as Completion 36% 31% 14% Y 1 41% 29% 78% Y 2 4% 38% 8% Y 3 8% 1% 1% Y 4 1% 2% - Y 5 10% - - Total 100% 100% 100% n = 316 238 160 Regional Distribution of Enrollments. The previous evaluation found that the RNC program had had more success in reaching builders and owners in the Northwest region of the state than elsewhere. 1 This was due in part to historical patterns of program activity and the concentration of larger builders and developments in the Northwest. This pattern has continued through 2003. Sixty-eight percent of program enrollments came from the Northwest, versus the 48 percent of total units built, per analysis of state tax records. However, the program has done a better job over the years in reaching builders and owners outside the Northwest. See Table 5-3. 1 See Table 5-34 on page 5 41 for complete definitions of market areas. 5 7 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-3 Enrollments by Region v. Permits Issued Enrollments Region 2003 2002 2001 2000 Units Built 2003 Northwest 68% 71% 74% 79% 48% Southwest 10% 7% 8% 4% 17% Northeast 4% 3% 2% - 14% Southeast 13% 19% 11% 15% 21% Unknown 5% - 6% 1% Breadth of Builder Participation. Table 5-4 shows the total number of builders that enrolled projects for each program year and the number of those builders who enrolled projects for the first time. Over the five program years, a total of 179 unique builders enrolled projects in the program. These establishments enrolled units in the program in 308 separate instances. The number of first-time participants and total participants increased steadily over the five years. The total number of unique builders enrolled in the program accounts for 25 percent of all establishments that list single-family new construction as their primary business with Dun & Bradstreet. There are many additional establishments and individuals that build homes in Vermont. Efficiency Vermont s mailing lists include as many as 1,500 builders. These results suggest that a relatively small segment of Vermont builders participate in the program. The ideal measure of strength of participation would be the share of total annual new units built by program participants. Unfortunately, the kind of sampling required to make this estimate could be accommodated in the research plan. Table 5-4 Total Builders and First-Time Builders Enrolled by Program Year Program Year Number of Builders w/ Enrolled Projects Number of First-Time Participants 1999 37 32 2000 68 47 2001 64 33 2002 64 33 2003 75 34 Total 308 179 Repeat Participation. Much of EVT s approach to market transformation revolves around engaging market actors on the supply side, such as builders and contractors, in multiple projects. In this manner, EVT has an opportunity to demonstrate and reinforce the techniques and business benefits of promoting energy-efficient designs and products. Obtaining repeat participation by 5 8 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION builders is difficult in Vermont due to the highly fragmented nature of the residential construction industry. It is characterized by over 1,000 builders and remodelers, the vast majority of whom build no more than one or two units per year. Table 5-5 arrays participating builders by the number of units completed by year. The data show that both the number and proportion of builders completing multiple units in a year has increased steadily over time. Further analysis is required to explore the pattern of enrollments and of repeat builder participation across program years. Table 5-5 Builders Participating in RNC by Number of Units Completed by Year* Units Enrolled 2003 2002 2001 2000 1999 1 43 37 11 4 1 2-4 19 6 2 1-5-9 7 4-1 1 10+ 4 2-1 - * Note: due to conversion of data systems, the numbers for the first few years represented in this table are not likely to be accurate. Comprehensiveness of measure installations. The degree to which builders incorporate the full range of energy efficiency measures into their projects constitutes another dimension of program effects and market transformation. KEMA analyzed project-level data from RNC to assess trends in the number and types of measures installed in participating homes. We also examined trends in Home Energy Rating scores. It should be kept in mind that the program was substantially redesigned beginning January 2002 to provide incentives only to homes that received Home Energy Rating inspections and qualified for the Vermont Energy Star Homes designation. Prior to that, homes could participate in a non-inspection track and receive rebates for individual measures. Table 5-6 through Table 5-8 summarize the results of these analyses. Table 5-6 shows the distribution of completed RNC projects by number of end-uses addressed through program-supported measures. The number of end uses addressed per project has clearly increased over time. In 2000, the modal number of end-uses addressed was four. In the transition year of 2002, one-third of completed projects addressed six end-uses. In 2003 and the first 20 months of 2004, seventy percent of projects addressed six or seven end-uses. 5 9 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-6 Percent of Completed RNC Projects by Number of End-Uses Addressed Program Year Number of End-Uses Treated 2000 2001 2002 2003 2004 1 9% 10% 3% 1% 1% 2 21% 11% 9% 2% 2% 3 11% 9% 8% 8% 5% 4 41% 40% 20% 6% 8% 5 17% 15% 22% 13% 15% 6 1% 14% 32% 65% 62% 7-1% 6% 5% 7% Table 5-7 displays the percentage of projects with specific end-uses addressed by year. For almost all end-uses, the share of projects in which measures are installed increased steadily over the five program years. The increase in the number of completions with envelope, space heating, and water heating measures reflects the elimination of the Vermont Advantage home track from the program. The increase in completions with laundry and refrigeration measures reflects the integration of procedures to promote ENERGY STAR appliances into the RNC program. Table 5-7 Percent of Completed RNC Projects by End Uses Treated Program Year End Uses Treated 2000 2001 2002 2003 2004 Air Conditioning 7% 11% 17% 11% 10% Cooking/Laundry - 18% 60% 91% 95% Hot Water Efficiency (fossil fuels) 68% 76% 83% 84% 80% Hot Water Efficiency (electricity) 1% - - - - Lighting 94% 92% 94% 98% 98% Light Fixtures 92% 87% 90% 97% 95% Light Bulbs 8% 14% 34% 58% 77% Lighting Controls 18% 21% 15% - - Other fuel switch - - 1% - - Refrigeration 23% 32% 49% 77% 84% Envelope Efficiency 67% 76% 83% 84% 80% Space Heating Efficiency - 42% 75% 89% 88% Ventilation 67% 76% 83% 84% 80% 5 10 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-8 shows the distribution of completed projects (with inspections) by Home Energy Rating score. The table shows a clear upward trend in the percentage of homes that meet the Vermont Energy Star Homes standard (86.0). In 2004, 92 percent of the homes inspected met that standard, up from 53 percent in 2000. Moreover, 15 percent of the homes completed in 2002 attained a score of 88, up from 1 percent in 2001. Again, these increases clearly reflect the effects of changes in program design. However, they may also reflect some degree of market transformation or learning, as builders become more familiar with the requirements of the program and with the technical aspects of energy-efficient construction. We will explore these issues in the evaluation. Table 5-8 Distribution of Completed RNC Projects by Home Energy Rating Score HERS Score 2000 2001 2002 2003 2004 < 82.0 2% 1% 1% - 1% 82.0-85.9 46% 33% 22% 12% 8% 86.0-87.9 52% 65% 71% 76% 77% >= 88.0 1% 1% 6% 11% 15% Units Completed 322 435 414 349 348 The RNC project has clearly made a lot of progress since the end of the previous evaluation. Perhaps most importantly, the new construction program now operates statewide with a unified approach. This provides the basis from which to develop increasingly effective program marketing and outreach activities. It also contributes to containment of the costs that builders experience in participation. The RNC has also significantly improved the quality of record keeping and was able to provide KEMA with a detailed quantitative picture of its operations going back to the beginning of EVT administration. 5.3 MARKET DESCRIPTION This section uses a variety of sources to estimate the size and describe the segmentation of the residential new construction market in Vermont. On the demand side, we compile information regarding the number, location, and affordability of homes built in Vermont. On the supply side, we characterize the population of builders and develop information on segmentation by firm size, market area, and range of activities. 5.3.1 Demand Side Overview Our discussion of new construction market characteristics in Vermont is based primarily on the following two sources: 1. U.S. Bureau of the Census. The Residential Construction Branch of the U.S. Bureau of the Census provides data on local building permits. Data is collected through a mail 5 11 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION survey using Form C-404, "Report of Building or Zoning Permits Issued and Local Public Construction." Where reports are not received, data are either obtained from the Survey of Use of Permits (SUP), which collects data on housing starts, or are imputed by the Bureau. For places not in the SUP, imputations are used. 2 Based on the description of the sampling method published by the Census, it is unclear how many places in Vermont are included in the SUP. 2. Vermont Department of Taxes. Properties sold in the State of Vermont are subject to a Property Transfer Tax based on the selling price. The Vermont Department of Taxes compiles data on the average selling price of primary residences in Vermont as part of its database of property transfer tax statistics. A database of these transactions is maintained for primary residences in which the seller has full interest in the property within the state of Vermont. This database does not distinguish between existing and newly-constructed homes. Number of New Homes Built Table 5-9 shows the Census estimates of new privately-owned housing units constructed for the years 2000 through 2004 by building type based on permit data. There is no discernable annual trend in the number of units constructed during this time period across all building types; however, for single-family homes, the number of new units per year continues to rise, increasing by approximately 22 percent since 2000, and by 11 percent between 2003 and 2004. Singlefamily homes continue to dominate the proportion of new units constructed across all housing types. In structures with 5 or more units, 33 buildings accounted for the 550 new units in 2004, averaging approximately 17 units per building of this type. According to 2003 data from the National Association of Home Builders (NAHB) 3, 100 new single-family homes create 250 jobs and $11.6 million in economic activity wages and income to local businesses in the year of construction alone. Construction of 100 multi-family units creates 112 jobs and $5.3 million in economic activity in the first year of construction. Increases in home construction volume translate to higher income for homebuilders. 2 A complete description of imputation procedures is available in the Documentation files for County Level Residential Building Permit Statistics at http://www.census.gov/const/c40/sample/cntyasc.pdf (Manufacturing and Construction Division, U.S. Census Bureau, Washington, D.C.). 3 National Association of Home Builders, Economics Group, 2003, as cited in the Vermont Housing Council and the Vermont Housing Awareness Campaign, 2004. Between a Rock and a Hard Place: Housing and Wages in Vermont, online at http://www.housingawareness.org/publications/housing-wages-2004.pdf. 5 12 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-9 Annual Estimates of New Privately Owned Housing Units in Vermont by Building Type, 2000-2004 Building Type % Change Year Single- Family Two Units 3-4 Units 5 or More Units Total Units from Prev. Year 2000 2,212 68 39 187 2,506-4% 2001 2,349 82 49 267 2,747 10% 2002 2,451 148 50 423 3,072 12% 2003 2,430 162 79 172 2,843-7% 2004 2,688 234 99 550 3,571 26% Source: Residential Construction Branch, U. S. Bureau of the Census. 19,000 Place Data Series, Table 2au. New Privately Owned Housing Units Authorized (Unadjusted Units) for Regions, Divisions, and States. Online at http://www.census.gov/const/www/c40/table2.html#annual. The Residential Construction Branch of the Census Bureau also gathers annual data on the number of building permits by county within Vermont. As shown in Table 5-10, new units in Chittenden County represent nearly a quarter of all units permitted in the state in 2004; 85 percent of these units are in buildings with 5 or more units. Approximately 36 percent of all new single-family homes permitted in Vermont in 2004 were in Chittenden County. The county with the second highest proportion of new housing units in Vermont in 2004 was Windsor, representing 11 percent of the state total across all housing types but only 4 percent of new single-family homes in the state. In Windham County, which represents only 7 percent of the total new units permitted in 2004, 58 single-family units were permitted in 2004, representing 25 percent of the total new single-family units in the state. 5 13 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-10 Estimates of New Privately Owned Housing Units in Vermont by County and Building Type, 2004 4 Building Type County Single- Family Two Units 3-4 Units 5 or More Units Total Units % of State Total Addison 187 - - - 187 5% Bennington 185 34 21 56 296 8% Caledonia 183-3 - 186 5% Chittenden 397 84 37 336 854 24% Essex 36 - - - 36 1% Franklin 276 14 6 40 336 9% Grand Isle 37 - - - 37 1% Lamoille 161 18 - - 179 5% Orange 100 - - - 100 3% Orleans 177 2 4 82 265 7% Rutland 168 2-12 182 5% Washington 259 12 8-279 8% Windham 205 58 3-266 7% Windsor 332 10 17 43 402 11% Total/Overall 2,703 234 99 569 3,605 100% Source: Residential Construction Branch, U. S. Bureau of the Census: data file CO2004A.txt. Obtained by personal communication (via email) on June 23, 2005. When the data in Table 5-10 is compiled by market areas as designated by the study, the northwest region dominates the state in permits issued for new construction in 2004 (Table 5-11). More than half of the state s multi-family homes were constructed in the northwest as well as approximately 42 percent of the state s single-family homes. The northeast region of the state represents the smallest proportion of new construction permits issued in 2004 (14 percent). 4 Although sources for Table 5-9 and Table 5-10 are the same, the total number of housing units in 2004 differs by 34 between the two tables. The U.S. Bureau of the Census postulates that the reason for this discrepancy may be a result of compiling data for the two different tables at different times (before and after changes or corrections had been made). 5 14 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-11 Estimates of New Privately Owned Housing Units in Vermont by Market Area and Building Type, 2004 Building Type Market Area Single- Family Two Units 3-4 Units 5 or More Units Total Units % of Total Northwest 1,130 128 51 376 1,685 47% Northeast 396 2 7 82 487 14% S. West / S. Central 540 36 21 68 665 18% Southeast 637 68 20 43 768 21% Total 2,703 234 99 569 3,605 100% Source: Residential Construction Branch, U. S. Bureau of the Census: data file CO2004A.txt. Obtained by personal communication (via email) on June 23, 2005. Trends in New Home Prices Average Price for Primary Residence. According to the Vermont Housing Council and the Vermont Housing Awareness Campaign, while the median 2004 purchase price of a home in Vermont was $165,000 (up 67 percent from 1996), the median price of a newly constructed home was approximately $294,000. 5 Because there is no non-proprietary source of detailed data on new housing prices readily available at a more granular level, we relied upon data from the Vermont Department of Taxes that do not distinguish between new and existing homes. Table 5-12 shows annual estimates of average home prices for primary residences in Vermont based on Property Transfer Tax statistics. The data show that housing prices in Vermont have been climbing rapidly after some price fluctuation in the late 1980s through mid-1990s (Figure 5-1); rising housing prices result in increased income for homebuilders. The largest increase in per-unit prices between 2003 and 2004 was for condominiums (12 percent), while single-family home prices increased by approximately 10 percent over the previous year. While single-family homes continue to be the most expensive across housing types, the price gap between singlefamily homes and condominiums has fallen from more than $34,000 in 2000 to less than $27,000 in 2004. 5 Vermont Housing Facts, http://www.housingawareness.org/facts.htm. 5 15 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-12 Annual Estimates of Average Price for Primary Residences (New and Existing) in Vermont by Housing Type, 2000-2004 6 Housing Type Year Single-Family Condominiums Mobile Homes 2000 $138,493 $104,194 $38,508 2001 $150,859 $119,686 $38,990 2002 $166,628 $143,208 $45,437 2003 $180,310 $152,896 $52,006 2004 $198,712 $171,914 $56,582 Source: Vermont Department of Taxes, Statistics: Property Transfer Tax, as compiled by the Vermont Housing Finance Agency (VHFA) and the University of Vermont's Center for Rural Studies at Housingdata.org. Online at http://www.housingdata.org/profile/profilemainresult.php?submitted=stateprofile. $200,000 $175,000 $150,000 $125,000 $100,000 $75,000 $50,000 $25,000 Figure 5-1 Trends in Average Price for Primary Residences (New and Existing) in Vermont by Housing Type, 1988-2004 Single Family Condominiums Mobile Homes $0 1988 1990 1992 1994 1996 1998 2000 2002 2004 Source: Vermont Department of Taxes, Statistics: Property Transfer Tax, as compiled by the Vermont Housing Finance Agency (VHFA) and the University of Vermont's Center for Rural Studies at Housingdata.org. Online at http://www.housingdata.org/profile/profilemainresult.php?submitted=stateprofile. Single-Family Housing Affordability. As shown in Table 5-12, average prices for singlefamily homes have increased by an average of 9 percent annually since 2000 (8 to 10 percent per year), while median income for a family of four has increased by approximately 8 percent annually in the same timeframe (Table 5-13; 3 to 16 percent per year). According to an online calculator published by the Vermont Housing Finance Agency (VHFA) and the University of 6 Note: In July 2004, the minimum transaction amount for inclusion of condominiums and single-family homes in the Vermont Department of Taxes Property Transfer Database was set at $30,000. 5 16 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Vermont's Center for Rural Studies 7, median household income is just high enough to render houses affordable for the average family of four in Vermont, with 2004 income for a family of four only 6 percent higher than required to afford a single-family home at the state average price of $198,712. Table 5-13 Single-Family Housing Affordability for a Family of Four in Vermont, 2000-2004 Year Average Price for Single-Family Home (a) Income Required to Afford an Average- Priced Single- Family Home (with 20% down) (b) Median Income for a Family of Four (c) 2000 $138,493 $38,487 $43,000 2001 $150,859 $41,924 $45,500 2002 $166,628 $46,306 $46,800 2003 $180,310 $50,108 $55,700 2004 $198,712 $55,222 $58,600 Sources: (a) Vermont Department of Taxes, Statistics: Property Transfer Tax, as compiled by the Vermont Housing Finance Agency (VHFA) and the University of Vermont's Center for Rural Studies at Housingdata.org. Online at http://www.housingdata.org/profile/profilemainresult.php?submitted=stateprofile; (b) Vermont Housing Finance Agency (VHFA) and the University of Vermont's Center for Rural Studies, Home mortgage calculator: How much income do you need to afford a house? Online at http://www.housingdata.org/calculator/priceform.php; and (c) U.S. Department of Housing and Urban Development (HUD) 8, 2005, as compiled by the Vermont Housing Finance Agency (VHFA) and the University of Vermont's Center for Rural Studies at Housingdata.org. Online at http://www.housingdata.org/profile/profilemainresult.php?submitted=stateprofile. 5.3.2 Supply Side Overview Our discussion of the firmographic characteristics of construction establishments in Vermont is based primarily on the following two sources: 1. Dun & Bradstreet imarket Database. To develop a preliminary profile of the population of Vermont builders, KEMA analyzed establishment data from Dun & Bradstreet contained in the imarket database. In this and a number of other recent studies, we have found that builder lists developed from Dun & Bradstreet have proven to be generally accurate in terms of reported construction activities and classification by number of employees. 9 2. Builder Surveys. As part of this evaluation, KEMA conducted surveys of 61 establishments that listed residential new construction or remodeling as their primary or 7 Based on home price, 20% downpayment, 6%mortgage interest rate, 30-year length of mortgage, and other customary monthly housing expenses including taxes and insurance. The additional monthly expenses are added to the monthly mortgage payment, which is annualized and based on the assumption that these housing costs will consume 30 percent of a household's income. 8 For more information, see www.huduser.org/datasets/il.html. 9 Due to the intensive use of subcontractors, some builders in the medium category (5 24 employees) are likely responsible for large volumes of construction. 5 17 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION secondary SIC. The surveys were designed to yield information on a number of key issues, including business characteristics of the targeted establishments, current construction and marketing practices in regard to energy efficiency, and knowledge of and response to the Vermont Energy Star Homes program. Results from the builder survey are presented in Section 5.3.3. Number and Distribution of Builders The imarket data analysis yielded a count of 1,083 businesses in Vermont that list new construction or remodeling for single- or multi-family as their primary or secondary SICs. Counts of these businesses were obtained by market area and business size category and are shown in Table 5-14. Comparisons between the number of builders by region in the 2002 and 2005 studies are shown in Table 5-15. Market Area Table 5-14 Distribution of Residential Construction and Remodeling Establishments by Size of Business and Market Area, 2004 Small (1-4 emps) Size of Business Medium (5-24 emps) Large (25+ emps) Total Businesses Percent of Total Businesses Northwest 395 86 10 491 45% Northeast 69 22-91 8% S. West /S. Central 200 38 8 246 23% Southeast 208 45 2 255 24% Total Businesses 872 191 20 1,083 100% Percent of Total Businesses 81% 18% 2% 100% Source: Dun & Bradstreet s imarket Database, First Quarter (Jan-Mar), 2005. The D&B Corporation, Short Hills, NJ. Table 5-15 Distribution of Residential Construction and Remodeling Establishments by Market Area, 2001 and 2004 2001 2004 Market Area N % N % Northwest 428 47% 491 45% Northeast 80 9% 91 8% S. West /S. Central 195 21% 246 23% Southeast 213 23% 255 24% Total 916 100% 1,083 100% Sources: Dun & Bradstreet s imarket Database, First Quarter (Jan-Mar), 2005. The D&B Corporation, Short Hills, NJ; and 2002 study. 5 18 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Key findings from the analysis of the imarket data include the following: Number of establishments. The number of establishments increased by 16 percent between 2002 and 2005, potentially due to the increase in construction volume and prices, while the distribution of establishments by market area remained largely unchanged. Size distribution of establishments. These establishments are generally very small: 81 percent of all builders employ fewer than 5 persons (compared with 76 percent in the 2002 study). Overall, builders averaged 7 employees per establishment, with an average of 4 employees for small builders, 15 for medium-sized, and 45 for large builders. Geographic distribution. The geographic distribution of the listed builders by market area mirrors the regional distribution of new home construction, with the majority of businesses and construction activity located in the northwest (see Table 5-11 on page 5 15). This finding implies that home building is very much a local activity in Vermont, a suggestion supported by survey data showing that 68 percent of builders conduct business only within the state of Vermont. 5.3.3 Detailed Builder Characteristics Detailed builder characteristics were obtained primarily through the builder survey. This section of the report discusses the survey methods and discusses findings related to new construction establishments in Vermont and compares these with findings from the 2002 study. See Section 5-8 for a description of sampling and survey methods. Sources of Revenue/Involvement in Remodeling. Remodeling accounts for a substantial portion of business revenues among all businesses involved in new construction. Table 5-16 displays information from the builder survey on the portion of sample firms involved in various kinds of construction activities, and the average percentage of total revenue derived from those activities. On average residential remodeling provided 42 percent of total revenues for the sample builders, with a range of approximately 26 percent for large firms up to 47 percent for medium-sized firms. Twenty-six percent of all builders sampled engage in commercial new construction, 78 percent are involved in residential remodeling, and 23 percent pursue commercial remodeling. The percentage of establishments involved in activities other than residential construction is highest among larger firms. Similarly, larger firms derive a lesser portion of their total revenues (41 percent) from residential new construction than small and medium-sized firms. 5 19 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-16 Involvement in and Revenue From Various Construction Activities, 2004 and 2001 (Weighted by Population of Builders) Construction Practice Small (n = 33) Size of Business Medium (n =21) Large (n = 7) 2004 (n = 61) All Builders 2001 (n = 54) Percent of Establishments General Contracting: Residential NC 100% 94% 100% 99% 100% General Contracting: Commercial NC 22% 36% 76% 26% 28% Residential Remodeling 77% 77% 87% 78% 70% Commercial Remodeling 21% 32% 47% 23% 32% Other 2% - - 2% 3% Mean Percent of Total Revenues * General Contracting: Residential NC 64% 50% 41% 61% 75% General Contracting: Commercial NC 11% 9% 24% 11% 3% Residential Remodeling 40% 47% 26% 42% 17% Commercial Remodeling 6% 10% 26% 9% 3% Other 10% - - 10% 2% * 2005 survey did not force the 5 construction practices to add to 100% as in the 2002 study; overall results are thus incomparable. Extent of construction activity in Vermont. Builders representing approximately 68 percent of the volume of new homes built in 2004 indicated that Vermont is the only state in which they provide their services. As shown in Table 5-17, this distribution has not changed since the 2002 study. Among contractors who build in other states, respondents indicated that 71 percent of their company s total revenues came from the state of Vermont. Table 5-17 Geographic Distribution of Work Among Builders by State, 2004 and 2001 (Weighted by Volume of New Homes Built) Location Percent of Total Market 2004 2001 (n = 61) (n = 54) Build only in VT 68% 69% Build in Other States * 32% 31% NH 62% 45% NY 37% 55% MA 46% 30% ME 4% <1% CT / Other 2% 14% * Respondents allowed to indicate more than one other state. 5 20 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Builders representing 69 percent of the new construction market in the southeast market area reported activity in multiple states, compared with builders representing only 7 to 22 percent of the market share in other areas. Builders representing 20 percent or more of the new construction market reported that their business was concentrated in Windham (24 percent), Washington (22 percent), and Windsor Counties (20 percent). None of the builders activities were concentrated in Essex County. 5.4 BASELINE CONSTRUCTION PRACTICES 5.4.1 Characteristics of Homes Built Volume of new construction and market share of size segments. Table 5-18 shows the estimated total number of units built by all Vermont builders by size category, along with the percentage of total units accounted for by establishments in the size category and the average number of units built. Small builders (those with 4 or fewer employees) accounted for the largest share of total units built (73 percent), although each establishment completed, on average, approximately 3 houses per year. Small firms have increased their dominance in the new construction market since 2001; the number of small firms has increased by approximately 60 percent and the number of units constructed by small firms has more than doubled. Upward trends in housing prices have likely made homebuilding a more attractive business. Medium sized firms accounted for 24 percent of total construction, and the 7 largest firms in the state accounted for an estimated 87 units, or 2 percent of total units constructed. Clearly, residential new construction activity in Vermont is highly fragmented, especially when one takes into account the 15 to 20 percent of homes that are owner-built. 5 21 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-18 Volume of New Construction and Market Share by Size Segment, 2004 and 2001 10 (Weighted by Population of Builders) Size of Business Characteristic Small (n = 33) Medium (n =21) Large (n = 7) Number of Establishments in Population 2004 (n = 61) 872 191 20 2001 (n = 54) 544 125 12 Estimated Total Units Built 2004 2,780 917 87 2001 1,301 1,076 229 Share of Total Units 2004 73% 24% 2% 2001 50% 41% 9% Average Units Built per Establishment 1 2004 3.2 9.2 18.6 2001 2.3 8.6 19.1 1 Unweighted Home Characteristics. Each survey respondent was asked to indicate the number of housing units they built in 2004 by type of housing and other characteristics. These responses were averaged to yield information about the volume of new construction in 2004; Table 5-19 shows the results of this analysis. Small builders were more active in the multi-family market than large builders, while medium-sized builders were most active in the single-family market. The majority of homes built by all builders were custom homes built for year-round occupancy. 10 Readers will notice that the total number of units built estimated from the builder surveys exceeds the number of permits isused: 3,784 v. 3,036. We believe much of this discrepancy is due to the effects of the sampling and weighting system used for the builder survey. In order to get information about building practices for a significant number of units, we needed to oversample large builders, and to interview relatively few small builders. In this round of the survey, we randomly interviewed a number of small builders who had built a relatively large number of units in the past year. See the comparison of 3.2 units v. 2.3 units built in Table 5-18. However, because these few sample builders represent such a large population, the impact of their response on the estimate is enormous: 2,708 homes built in 2004 by small builders v. 1,301 in 2001. 5 22 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Housing Type/Characteristic Table 5-19 Mean Number of New Housing Units Per Builder by Size of Business and Housing Type/Characteristic, 2004 * (Weighted by Volume of New Homes Built) Small (n = 33) Size of Business Medium (n =21) Large (n = 7) All Builders 2004 (n = 61) Single-Family Homes 4 23 5 8 Two-Family Homes < 1 < 1 3 < 1 Three-Four Family Homes 1 3 <1 1 Homes with 5 or More Units 5 <1 <1 4 Production Homes 1 <1 <1 1 Built for Year-Round Occupancy 4 25 3 9 Custom-Built Homes 4 25 5 9 * Phase 1 results not included: 2001 survey asked for proportion of total business represented by each housing type rather than the number of units. Sales Prices. The sample builders were asked to indicate price ranges for the custom and production homes they built in Vermont and sold in 2004. Builders representing 44 percent of the new construction market indicated that the typical selling price for custom-built units was between $300,000 and $400,000. For production homes, builders representing a similar proportion of the new construction market (45 percent) reported the $200,000 to $300,000 range as the typical selling for production homes, with a similar proportion selling for less than $200,000 (41 percent). Further detail is reported in Table 5-20. These results are consistent with data from the Vermont Housing Council discussed above; one-quarter of the new homes built in 2004 sold for more than $400,000. Table 5-20 Typical Selling Price Range for Custom and Production Homes, 2004 * Custom Homes Production Homes Price Range (n = 59) (n = 3) Less Than $200,000 5% 41% $200,000 - $300,000 26% 45% $300,000 - $400,000 44% - $400,000 - $500,000 8% 14% More than $500,000 17% - Total\Overall 100% 100% * Phase 1 results not included: 2001 survey asked mean price for custom and production homes rather than percentage within price range. 5 23 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION 5.5 ENERGY EFFICIENCY IN MARKETING AND CONSTRUCTION PRACTICES This section of the report characterizes builders construction and marketing practices regarding energy efficiency in new homes, their recognition and understanding the Vermont Energy Star Homes Program, and the program effects perceived by builders. Data to support this analysis comes primarily from the builder surveys (n = 61) as compared (where possible) with data from the Phase 1 evaluation (2002; n = 54). 5.5.1 Energy efficiency in the home sales and planning process Sample builders were asked whether they discussed the benefits of energy efficiency with buyers in the course developing plans for a new home. They were also asked to name the benefits of energy efficiency that they identified to customers. The most striking change between 2002 and 2005 is that builders representing a large proportion of 2004 market share indicated that they emphasize comfort and reduced maintenance costs with homeowners (see Table 5-21 for details). Additional findings include: Builders representing approximately 99 percent of the 2004 new construction volume in Vermont indicated that they discuss energy efficiency considerations in all, most, or some cases when developing plans for custom-built homes (as compared with 91 percent in the 2002 study). Among builders who discuss energy-efficiency considerations with their customers, builders accounting for 92 percent of new construction volume indicated that they discuss reduced energy costs with homebuyers (no change from the 2002 study), and more than 60 percent mentioned the additional benefits of greater comfort and lower maintenance costs, both statistically significant increases since the 2002 study. The only benefits discussed by builders representing a larger segment of the overall market share in the 2002 study than in the 2005 study were environmental benefits. Table 5-21 Energy-Efficiency Benefits Discussed with Homeowners When Developing Plans for New Homes, 2004 and 2001 (Weighted by Volume of New Homes Built) Frequency 2004 (n =59) All Builders 2001 (n = 33) Reduced energy costs 92% 91% Greater comfort * 65% 24% Lower maintenance costs * 60% 12% Environmental benefits 20% 27% Higher resale value 19% 21% Longer component life 10% 9% * Statistically significant change between 2001 and 2004 at 95% CI. 5 24 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION 5.5.2 Energy-Efficient Features Inclusion of Features. Builders in the sample were asked to indicate whether they include specific energy-efficient features in all, most, some, or none of the homes they built in 2004. Table 5-22 shows the proportion of market share represented by builders who indicated that the features were included in all of the homes they built in 2004. Key findings include: Although large and medium-sized establishments were more likely to include some energy-efficient features than smaller firms, this was inconsistent; in general, there was little variation by business size. Marked variation among builders was apparent at the market area level: the largest proportion of homes constructed in 2004 that included many of these features were in the northwest region. Feature Shell Features Table 5-22 Inclusion of Energy-Efficient Features in All New Homes Built by Market Area, 2004 (Weighted by Volume of New Homes Built) NE (n = 5) Market Area NW (n = 21) SE (n = 15) SW (n = 12) Low-e windows 100% 98% 74% 100% Basement insulation above R-10 33% 84% 66% 48% Wall insulation above R-19 61% 64% 75% 45% Attic insulation above R-38 100% 72% 43% 39% Reduced air infiltration measured w/blower door test 22% 53% 50% 6% Argon-filled windows 100% 45% 25% 47% Floor insulation greater than R-10 33% 58% 30% 18% HVAC Energy Star high-efficiency heating equipment 50% 95% 62% 30% Programmable thermostats 42% 85% 35% 32% Energy Star high-efficiency cooling equipment < 1% 46% 1% 17% Advanced controls (timers and sensors) 32% 10% 43% 9% Energy-efficient ventilation systems 32% 78% 50% 16% Duct sealing and leakage testing < 1% 44% 28% 3% Lighting and Appliances Energy Star high-efficiency appliances 56% 67% 38% 9% Energy Star hardwired CFL fixtures 25% 57% 22% 64% Changes since 2002. While builders in the 2005 study were asked to indicate whether they include specific energy-efficient features in all, most, some, or none of the homes they built in 5 25 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION 2004, builders in the 2002 study were asked to indicate whether the features were standard on all homes, offered as options to customers, or generally not offered. Table 5-23 compares the 2005 respondents who indicated they included the feature in all or most homes with the 2002 respondents who indicated that the feature was standard. 11 Key findings include: Builders representing 94 percent of the 2004 new construction market indicated that they include low-e windows in all or most of the new homes they build, followed by basement insulation above R-10 (86 percent) and Energy Star rated high-efficiency heating equipment (82 percent). Energy Star high-efficiency appliances are the feature for which the largest increase in market share occurred between 2001 and 2004, increasing from 31 percent to 78 percent (a statistically significant change). Energy Star hardwired compact fluorescent lighting fixtures increased from 20 to 53 percent, also statistically significant. This change seems clearly related to EVT s efforts with both the lighting/appliance bonus within the new construction program and the retail promotion of these products. Argon-filled windows exhibited the largest decline in market share, from 76 to 51 percent. 5 26 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-23 Inclusion of Energy-Efficient Features in All or Most New Homes Built by Market Area, 2004 and 2001 * (Weighted by Volume of New Homes Built) Feature Shell Features All Builders 2004 (n = 61) 2001 (n = 54) Low-e windows 90% 94% Basement insulation above R-10 86% 72% Wall insulation above R-19 75% 56% Attic insulation above R-38 59% 55% Reduced air infiltration measured w/blower door test 49% 21% Argon-filled windows 51% 76% Floor insulation greater than R-10 42% 31% HVAC Energy Star high-efficiency heating equipment 82% 65% 1 Programmable thermostats 74% 51% 2 Energy Star high-efficiency cooling equipment 23% 65% 1 Advanced controls (timers and sensors) 29% 51% 2 Energy-efficient ventilation systems 65% -- 3 Duct sealing and leakage testing 31% 47% Lighting and Appliances Energy Star high-efficiency appliances 78% 31% Energy Star hardwired CFL fixtures 53% 20% * 2005 data represents market share of builders who indicated inclusion of feature in all homes, while 2002 data represents market share of builders who indicated inclusion of features was standard practice. 1 Energy Star heating and cooling systems were combined in the 2002 study. 2 Thermostats and other controls were combined in the 2002 study. 3 Energy-efficient ventilation systems were not included in the 2002 study. Statistically significant change between 2001 and 2004 at 95% CI. 5.5.3 Energy-Efficient Options Sample builders were asked to indicate whether they offer any energy-efficiency features as priced options for customers in new homes rather than as standard features. Builders who responded in the affirmative were asked to describe the features most often requested by customers as well as any objections to energy-efficient features that customers typically raise. Priced Options. Overall, builders representing nearly three-quarters of the new construction market share in Vermont (72 percent) indicated that they offer energy-efficient features as priced options in new homes. Builders representing 79 percent of the market share for small firms indicated that they offer priced energy-efficiency options, compared with builders representing 5 27 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION 56 percent of the market in both the medium and large business size categories. Regionally, builders representing the majority of market share in the northwest (91 percent) and southwest (88 percent) indicated they offer priced energy-efficiency options to their customers as compared with lower proportions in the northeast (52 percent) and southeast (46 percent). Table 5-24 Energy-Efficient Features Most Frequently Requested As Priced Options by CustomersIn New Homes, 2004 (Weighted by Volume of New Homes Built) Feature Shell Features All Builders 2004 (n =61) Wall insulation above R-19 7% Attic insulation above R-38 6% Low-e windows 5% Argon-filled windows 1% Basement insulation above R-10 1% Floor insulation greater than R-10 < 1% Reduced air infiltration measured w/blower door test - HVAC Energy Star high-efficiency heating and cooling equipment 39% Duct sealing and leakage testing 1% Programmable thermostats - Advanced controls (timers and sensors) - Lighting and Appliances Energy Star high-efficiency appliances 11% Energy Star hardwired CFL fixtures - Offer No Priced Options to Customers 28% Builders representing 39 percent of the new construction market in Vermont indicated that customers request Energy Star rated high-efficiency heating and cooling equipment as priced options, compared with only 8 percent in 2001, a statistically significant increase. 12 Builders report that customers request other types of energy-efficient equipment to a far lesser extent, which may reflect effective marketing of Energy Star rated high-efficiency HVAC equipment in Vermont (Table 5-24). Other noteworthy findings include: Builders representing only 3 percent of the market share for large firms indicated that customers request Energy Star rated high-efficiency heating and cooling equipment as a priced option, compared with 39 percent of medium and 40 percent of small builders. 12 Statistically significant difference between 2001 and 2004 results at 95% CI. 5 28 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Builders representing 31 percent of the market share for large businesses indicated that customers most frequently request basement insulation greater than R10 as a priced option, compared with builders representing less than 1 percent of the market share in other business size categories. Regionally, builders in the northeast who indicated that customers most frequently request attic insulation greater than R38 as a priced option represented a higher market share than in any other region (32 percent as compared with 12 percent or less in other market areas). Customer Objections to Efficient Equipment. Builders representing approximately 28 percent of the 2004 new construction volume indicated that they offer no priced options to their customers. Builders representing 59 percent of Vermont s new construction market share indicated that customers object to the initial cost of the feature when offered as priced options (Table 5-25). Builders representing approximately 38 percent of Vermont s new construction volume in 2004 indicated that customers objections relate to uncertainty about equipment performance. Table 5-25 Builder Observations of Customer Objections to Efficient Equipment Offered as Priced Options by Size of Business, 2004 and 2001* (Weighted by Volume of New Homes Built) Size of Business All Builders Feature Small (n = 33) Medium (n =21) Large (n = 7) 2004 (n = 61) 2001 (n =33) No priced options offered 21% 45% 51% 28% 32% Equipment cost 62% 52% 49% 59% 36% Uncertainty about equipment performance 51% 2% < 1% 38% Equipment quality 11% < 1% 49% 9% Equipment manufacturer 9% < 1% < 1% 7% Other Reason 1% 2% < 1% 1% * Builders allowed to indicate more than one customer objection ; total may be greater than 100 percent. 2001 responses not comparable (question phrasing was different in 2002 study). 5.5.4 Energy Efficiency as a Business Proposition Builders were asked to rate the importance of marketing and delivering energy efficient homes to the overall success of their business. Table 5-26 shows their responses. Builders representing approximately half of the 2004 new home volume in Vermont gave a rating of 5 ( Very Important ) and builders representing an additional 36 percent gave a rating of 4. While data from the 2002 study is not directly comparable because the question was asked only of builders who were aware of at least one energy efficiency program, the 2002 data should logically show that builders indicated higher importance for energy efficiency as a result of this awareness; the 5 29 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION data, however, show the opposite. Builders representing approximately 85 percent of the new construction market in 2004 indicated that energy efficient homes are at least somewhat important to the success of their businesses, compared with only 70 percent in 2001: this may reflect some measure of program success in increasing builders recognition of the importance of energy efficiency in new construction. Importance Table 5-26 Importance of Energy Efficient Homes to Builders Business Success by Size of Business, 2004 * (Weighted by Volume of New Homes Built) Small (n = 33) Size of Business Medium (n =21) Large (n = 7) 2004 (n = 61) All Builders 2001 (n = ) 1 Not at all important 1% 37% <1% 10% 24% 2 Somewhat unimportant < 1% 1% 2% < 1% 1% 3 Neither important nor unimportant 5% 2% <1% 5% 5% 4 Somewhat important 32% 43% 90% 36% 40% 5 Very important 61% 17% 80% 49% 30% * Note: results not comparable to 2002 study, as this question was asked only of builders who reported awareness of at least one energy efficiency program in Vermont in 2001. 5.6 AWARENESS OF ENERGY RATINGS AND STANDARDS Residential Building Energy Standards (RBES). Approximately 69 percent of newlyconstructed homes in Vermont in 2004 were built by builders who indicated that they were aware of the RBES, compared with 85 percent in 2001. Of the builders who reported they were aware of the RBES, those who reported posting certificates of compliance (as required by the RBES) represented only 23 percent of new construction volume in the state (compared with 37 percent in 2001). Despite relatively high awareness of the standard, compliance continues to be low priority for builders in Vermont. Builders who indicated awareness of the RBES were asked to identify, without prompting, home features that were required by the RBES. Table 5-27 shows the percentage of new construction volume represented by builders who were able to name particular features necessary for compliance, by feature and size of business. Builders representing less than 70 percent of new homes built in 2004 were familiar with attic and wall insulation levels required to meet RBES standards, compared with more than three-quarters in 2001. We should note that builders who use VESH to check compliance or who use software packages may be unaware of individual building components considered in assessing compliance. 5 30 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Feature Table 5-27 Unaided Recall of Features Necessary for RBES Compliance by Size of Business and Feature, 2004 and 2001 (Weighted by Volume of New Homes Built) Small (n = 24) Size of Business Medium (n =17) Large (n = 7) 2004 (n = 48) All Builders 2001 (n = 27) Low-e windows 68% 86% 69% 72% 40% Attic insulation at least R-38 78% 36% 69% 69% 77% Wall insulation at least R-19 76% 40% 69% 68% 77% Basement insulation at least R-10 68% 25% 69% 59% 44% Floor insulation at least R-10 65% 16% 62% 55% 35% High efficiency heating and cooling equipment 45% 83% 85% 55% 28% Mechanical ventilation systems 40% 21% 64% 36% - Duct sealing and leakage testing 34% 2% < 1% 26% - Reduced air infiltration 13% 65% 34% 24% 24% Other 13% < 1% < 1% 10% - Home Energy Ratings. Builders representing one-third of the new homes built in Vermont during 2004 indicated that they purchase home energy ratings from third party agencies for any of the homes they build, up from 25 percent in 2001. The most frequently cited reason for not including the ratings is that customers did not request the service. Builders representing approximately 31 percent of 2004 new construction volume indicated that they recommend that their customers obtain home energy ratings, up from 23 percent in 2001. While changes between the 2002 and 2005 studies are significant, they suggest that despite the expense, a fairly sizeable proportion of builders continue to purchase and/or recommend Home Energy Ratings to their customers. We should note that the VESH program staff believe that the respondents significantly overreported their purchase of Home Energy Ratings. 5.7 VERMONT ENERGY STAR HOMES PROGRAM 5.7.1 Reported Program Awareness and Participation Builders representing approximately 92 percent of the state s new construction in 2004 indicated awareness of the Vermont Energy Star Homes program, showing no change since 2001; the program continues to retain a high level of awareness among builders (Table 5-28). However, of builders who reported awareness of the program, builders representing a larger proportion of market share in 2004 reported participating in the program than in 2001 (43 and 24 percent, respectively). Reported participation in 2004 was lowest among large builders. 5 31 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-28 Builder Recognition of and Participation in Vermont [Energy] Star Homes Program by Size of Business, 2004 and 2001 13 (Weighted by Volume of New Homes Built) Size of Business All Builders Characteristic Small Medium Large 2004 2001 (n = 54) Awareness 89% 99% 100% 92% 95% Participation (of those aware) 41% 52% 31% 43% 24% Table 5-29 presents the same data as Table 5-28 broken out by region rather than size. Awareness of the Vermont Energy Star Homes program was fairly high in all regions; participation, however, was negligible in the northeast and southwest regions of the state. Participation in these regions is a persistent problem; data from 2001 show similarly low levels of participation in the northeast and southwest, indicating that recruitment in these areas requires continued persistence. Table 5-29 Builder Recognition of and Participation in Vermont [Energy] Star Homes Program by Market Area, 2004 and 2001 14 (Weighted by Volume of New Homes Built) Market Area All Builders Feature NE NW SE SW 2004 2001 (n = 54) Awareness 81% 96% 84% 99% 92% 95% Participation (of those aware) < 1% 49% 61% 7% 43% 24% 5.7.2 Participating Builder Response to Vermont ENERGY STAR Homes Builders representing 43 percent of the new construction market share indicated that they had received certification and financial assistance from Vermont Energy Star Homes program for homes they built in 2004. For the purposes of this study, builders who received certification and financial assistance from VESH are considered participants. It should be noted, however, that this reflects somewhat of a narrow view of the Energy Star Homes program s influence on new construction practices. Non-participants (as defined by this study) may still incorporate energyefficient components into their new homes as a result of advice from Efficiency Vermont without receiving certification or financial assistance through a formal new construction program. In 13 Awareness N: Small = 33; Medium = 21; Large = 7; Overall = 61 Participation N: Small = 28; Medium = 20; Large = 7; Overall = 55 14 Awareness N: NE = 5; NW = 29; SE = 15; SW = 12; Overall = 61 Participation N: NE = 4; NW = 27; SE = 14; SW = 10; Overall = 55 5 32 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION other words, the influence of EVT s new construction programs is likely broader than indicated by the data on formal participation. The following paragraphs summarize findings from the builder survey about the basic characteristics of participating builders, their motivations for participation, and their experiences with the new construction programs. The 2002 study queried builders on their experiences with the Vermont Star Homes program, and results from questions in the 2005 survey on Energy Star Homes are compared with these results where applicable. Volume of construction and share through the program. The fourteen sample builders who reported participating in the Vermont Energy Star Homes Program in 2004 completed a total of 95 homes in 2004. They reported receiving program certification for 75 of these homes, or 79 percent of the total number of homes they built. In the 2002 study, the 12 builders in the sample who participated in the Vermont Star Homes program reported receiving certification for approximately 57 percent of the homes they built. Although the sample sizes are small, the increase in the proportion of program-certified homes constructed by participating builders is noteworthy. Influences on Participation. Builders who reported participation in the Vermont Energy Star Homes program in 2004 were asked to identify the sources through which they had heard of the program and the one source that had the most influence on their participation in the program. Most respondents identified only one source of information. In 2001, one-third of the participating builders indicated that Homebuilders Associations were the most important source of information, while a larger proportion of 2004 participating builders indicated that the Vermont Energy Star Homes program staff were the most important source of information in their decision to participate in the program: in fact, 7 out of 10 of the 2004 respondents indicated that their most influential source of information was direct contact with program staff (Table 5-30). This change reflects markedly improved program outreach since 2001. 5 33 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-30 Most Influential Sources of Information on the Vermont [Energy] Star Homes Program Among Participating Builders, 2004 and 2001 Number of Participating Builders Source of Information 2004 2001 Vermont [Energy] Star Homes program staff 5 1 Vermont [Energy] Star Homes direct mail, other materials 3 - Efficiency Vermont program staff 2 1 Efficiency Vermont direct mail, other materials 1 - Home Builders Associations - 4 Owner 2 Efficiency Vermont direct mail, other materials - 1 Utility - 1 Other trade or professional organizations - 1 Potential homebuyers - 1 Don t Know 3 Total 14 12 Reasons for Participation. Five of the builders who participated in the program in 2004 reported that their main reason for participating was to get marketing support and extra publicity for their company, demonstrating the effectiveness of program outreach emphasizing market differentiation for participating builders. Three builders reported that their main reason for participation was to learn more about efficient building techniques, and two reported participating because they generally think that environmental issues are important. The remaining four builders each had a different reason for participating: to get rebates, because they could charge more for efficient homes, to help market the houses that get labeled, and as a result of questions on environmental aspects of homes from customers. In 2001, the main driver for participation reported by builders was that principals requested their participation (3 builders), while two builders reported that their main reason for participating was to get the rebates. These changes demonstrate a major program impact on the new construction market. Knowledge of Required Features. The participant portion of the builder survey contained a question sequence in which respondents were asked to name, unaided, the construction features and equipment required for certification as a Vermont Energy Star home. For each feature they mentioned, we also asked whether the respondent been aware of the feature prior to participating in the program. Table 5-31 shows the results of this sequence for ten key technical requirements of Vermont Star Homes. The 2004 sample program participants mentioned efficient heating equipment most frequently of all required program features (11 of 14); in 2001, 9 of 12 builders recalled efficient 5 34 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION heating and cooling equipment as a requirement. 15 Of those that mentioned high-efficiency heating equipment, 9 claimed to have been aware of the measure prior to participating. All of the sample participants were able to name at least two construction feature required by the program, while in 2001, all builders could name at least one. More than half of the builders surveyed in 2004 named at least 4 features; these changes demonstrate increased familiarity with the program among builders in the 2005 study as compared with those in the 2002 study, likely a result of effective program outreach. Awareness. In 2001, seven of the twelve HVAC contractors reported that they were aware of the Vermont Energy Star Program, with only one claiming to be even somewhat familiar with the provisions of the program. In 2004, a higher proportion of HVAC contractors (eight of nine) said they were aware of Vermont Energy Star Homes (VESH) Program, however, only two of these respondents could name one element of the program each. Table 5-31 Participant Builder Awareness and Adoption of Program Features, 2004 and 2001 Unprompted Awareness Awareness Prior to Program Participation Vermont [Energy] Star Homes Program Requirement 2004 (n = 14) 2001 (n = 12) 2004 (n = 14) 2001 (n = 12) Efficient heating equipment 1 11 9 11 8 Efficient ventilation 10 3 8 3 Efficient light fixtures 9 5 9 3 Efficient cooling equipment 1 9 9 8 8 High levels of insulation 9 6 7 6 Efficient water heating equipment 5 * 5 * Energy Star 5-star rating 5 * 5 * Air sealing 3 6 3 4 Home Energy Rating of 86 or higher 2 * 2 * 1 Efficient heating and cooling systems were combined in the 2002 study. * Feature not included in 2002 study. Marketing and Selling Vermont Energy Star Homes. Builders were asked a series of questions regarding the program s impact on marketing and sales. Key findings from the results of these questions are as follows: Effects of program requirements on construction costs. Of the fourteen participating builders interviewed, 9 indicated that installing features required to gain certification 15 Note: in the Phase 1 evaluation, efficient heating and cooling equipment were not broken out into separate features as in the 2005 study. 5 35 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION from the program resulted in added construction costs compared to homes without those features (two were not sure). Four of these builders were able to estimate the added construction costs; these costs averaged $6,375 (median = $2,500) and ranged from $500 to $20,000. The 2004 mean and range are similar to those in the 2002 study (mean: $6,766; range: $1,000 to $20,000). We note that builders may have a difficult time estimating added construction costs because these costs depend largely on general home characteristics, particularly size; for example larger homes would generally incur higher additional construction costs to include certain features necessary for certification than would a smaller home. The information above should be treated with some caution given the small number of builders who responded to the question. Effects of new construction program certification on salability. Five of the fourteen builders interviewed reported that they were able to sell certified homes more easily than uncertified homes built during the same period, while 4 were not sure whether it was more or less easy to sell certified homes. In 2001, 8 of the participating 12 builders reported that they were able to sell certified homes more easily. Effects of program certification on sales prices. Seven of the 14 participating builders interviewed reported that they were able to obtain a higher selling price for homes certified through the program (2 were not sure). Only 4 builders were to indicate the average increase in selling price for certified homes, likely because the price increase may depend on the general desirability of the home (size, location, etc) more than efficient construction or features. The four estimates of incremental home prices that respondents offered ranged from $1,000 to $40,000 with an average of $17,750. 5.7.3 Nonparticipating Builder Response to Vermont ENERGY STAR Homes Familiarity with Program. Forty-one of the builders who indicated that they d heard of the Vermont Star Homes program answered a series of questions about the construction and equipment features required for the program. Of these, builders representing approximately 19 percent of the 2004 new construction volume outside the program were unable to list any features required for homes to be certified through the program compared with 17 percent in 2001. Among the builders who were able to name at least one required feature, builders representing 70 percent of the non-participant new construction volume mentioned efficient light fixtures in 2004, a statistically significant increase over the proportion who recalled this feature in 2001 (Table 5-23). 5 36 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Table 5-32 Unaided Recall of Features Required for Vermont [Energy] Star Homes Program Among Non-Participating Builders by Size of Business, 2004 and 2001 (Weighted by Non-participant Volume of New Homes Built) Percent of Nonparticipating Builders Vermont [Energy] Star Homes Program 2004 2001 Requirement (n = 41) (n = 42) Efficient heating equipment 1 36% 27% Efficient ventilation 40% * Efficient light fixtures 70% 30% Efficient cooling equipment 1 31% 27% High levels of insulation 45% 42% Efficient water heating equipment 30% * Energy Star 5-star rating 17% * Air sealing 3% 40% Home Energy Rating of 86 or higher 18% * 1 Efficient heating and cooling systems were combined in the 2002 study. Statistically significant difference between 2001 and 2004 at 95% CI. * Feature not included in 2002 study. Familiarity with Services and Marketing Support. Non-participating builders representing approximately 45 percent of new construction volume outside the program indicated that they were familiar with the services and marketing support offered by the Vermont Star Homes Program compared with 74 percent of builders in the 2002 study who reported awareness of the services offered by the Vermont Star Homes program; because of the small sample sizes, however, this change is not statistically significant. More than half of the nonparticipating builders were unable to name any of these services, but the remaining were able to recall at least one, including builder certification, marketing support, technical assistance, and assistance with code compliance. Value of Services. Builders who indicated that they were aware of the Vermont Star Homes Program but did not participate in the program were read a list of services offered to builders by the Program. These services included the following: Review of plans to identify energy saving opportunities; Training in energy efficient construction practices; Technical assistance during construction; Free home energy rating; Financial incentives for meeting thermal requirements and for selected equipment; and Marketing assistance, including advertising and payment for certain marketing materials. 5 37 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Builders were then asked to indicate whether or not they felt the services would be useful in marketing the homes they built: Builders representing two-thirds (67 percent) of the 2004 new construction volume indicated that they felt such services would be useful, compared with builders representing only 36 percent of the market share in 2001. Of the builders who felt the services would be useful, those representing 65 percent of the new construction volume indicated that review of plans to identifying energy saving opportunities would among the most useful services offered. Builders who represented one-third (33 percent) of the state s non-program new construction volume thought the program would not be useful because they were not responsible for marketing the homes they build: Builders representing approximately 11 percent of non-program new construction volume stated that the program involves too much paperwork and is thus not beneficial. In the 2002 study, builders representing 18 percent of the non-participant new construction volume indicated that the program would not be beneficial to them because other entities (such as architects) market their homes. Reasons for Nonparticipation. Builders representing approximately 22 percent of the state s non-program new construction volume indicated that their primary reason for non-participation was that they had no trouble selling their homes without the Program s assistance; builders representing 28 percent of non-program market share in the 2002 study cited the same reason. Other reasons mentioned include the following: The belief that customers have no interest in energy efficiency; No time to become informed about the program; No time to learn about the construction techniques required by the program; and General dislike of the organized programs. 5.8 BUILDER SURVEY METHODS As part of this evaluation, KEMA conducted surveys of 61 establishments in the residential new construction and remodeling industry in Vermont. Businesses listed in the imarket database must designate primary and secondary SICs; because of the relatively low overall number of records where the primary SIC is specifically New Construction, Single-Family House, the definition of builders for the initial frame was expanded to include all records where the primary or secondary SIC is new construction or remodeling for single- or multi-family homes (see Table 5-33). 5 38 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION 8-Digit SIC Table 5-33 SICs Included in the Builder Survey, 2004 SIC Definition 1521-0000 Single-family housing construction 1521-9901 New construction, single-family houses 1521-0100 Single-family home remodeling, additions, and repairs 1521-0101 General remodeling, single-family houses 1521-0102 Mobile home repair, on site 1521-0103 Patio and deck construction and repair 1521-0104 Repairing fire damage, single-family houses 1521-9902 Prefabricated single-family house erection 1522-0000 Residential construction, nec 1522-0100 Hotel/motel & multi-family home construction 1522-0101 Apartment building construction 1522-0106 Multi-family dwelling construction, nec 1522-0107 Multi-family dwellings, new construction 1522-0200 Hotel/motel & multi-family home renovation &remodeling 1522-0201 Remodeling, multi-family dwellings The surveys were designed to yield information on a number of key issues, including business characteristics of the targeted establishments, current construction and marketing practices in regard to energy efficiency, and knowledge of and response to the RNC programs. The following paragraphs summarize key methodological aspects of the survey. Sample Design. KEMA employed a stratified sampling approach. The state was divided into the four market areas shown in Table 5-34. These market areas were established for the 2002 study based on consultation with Vermont realtors, builders, and other market observers to reflect distinctions in market conditions among the market areas. Within these areas, establishments were divided into three size categories based on the number of persons they employed. Estimates of employment from Dun & Bradstreet s imarket database were used to determine the builders business size categories, with firms employing 1 to 4 persons designated as small, firms with 5 to 24 employees designated as medium, and large firms representing those with 25 or more employees. The targeted number of completed surveys was allocated to the six regional/size strata according to the proportion of all employees in all establishments in the sample frame accounted for by each stratum as shown in Table 5-35. The version of the imarket database used to assemble this data included no large firms in the northeast area of the state. 5 39 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Market Area Table 5-34 Vermont Counties by Market Area Counties Northwest Chittenden, Franklin, Grand Isle, Lamoille, Washington Northeast Caledonia, Essex, Orleans S. West / S. Central Addison, Bennington, Rutland Southeast Orange, Windham, Windsor Table 5-35 Targeted Number of Completed Builder Surveys by Size of Business and Market Area, 2004 Size of Business Market Area Small (2-4 emps) Medium (5-24 emps) Large (25+ emps) Total Builders Northwest 14 3 10 27 Northeast 4 1 0 5 S. West / S. Central 5 1 8 14 Southeast 10 2 2 14 Total 33 7 20 60 Sample Selection. The total population of builders and remodelers in the SICs shown in Table 5-33 (1,083 establishments) was obtained from the imarket database. KEMA pulled a simple random sample of 200 establishments from the list and designated these as sample for the remodeler surveys (see Section 6), resulting in a remainder of 883 establishments for the builder surveys. Interviews Completed. Ultimately, we were able to complete interviews with 61 homebuilders in the size categories specified by the sample design. Table 5-36 shows the allocation of completed surveys among the sample strata defined by region and employment. When sample in the large size category was exhausted within any market area, additional completions were sought in the medium size category within the same market area. The highlighted cells in Table 5-36 show where completed surveys differed from targeted completions. Table 5-36 Completed Builder Surveys by Size of Business and Market Area, 2004 Size of Business Small Medium Large Total Market Area (2-4 emps) (5-24 emps) (25+ emps) Builders Northwest 14 12 3 29 5 40 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION Northeast 4 1-5 S. West / S. Central 5 4 3 12 Southeast 10 4 1 15 Total 33 21 7 61 Weighting and Analysis Procedures. Most of the items in the survey were analyzed using a ratio estimation procedure that yields an estimate of market share for practices of interest in terms of the portion of units built as opposed to the percentage of builders adopting the practice. Similarly, average values, such as insulation levels, are computed to reflect the population of houses reportedly built by the respondents. Thus, in reporting results, we generally use the formulation builders representing xx percent of the market or builders representing xx percent of all units built. Weighting and computation of values. Builder survey responses to were weighted to reflect to the number of homes reportedly constructed by the sample builder as well as the population weight of the size stratum from which the firm was drawn. Where the questionnaire sought responses in the form of a number or percentage say, the portion of homes built with energy efficient features, the survey responses were calculated using the combined ratio estimator Rˆ : N h Bh x i i h nh i Rˆ c =, N h xi n h h i c where i = sample builder, N h = number of builders in the population in sample stratum h, n h = number of builders in the sample in stratum h, B h i = builder i s response (expressed as a number or percentage), and x i = number of new homes builder i built in 2004. If the question elicited a categorical response (e.g., yes/no), a Bh was created for each i possible response. For the selected response, B = 1. For the response/s not selected, B h i = 0. Precision of estimates. The use of the combined ratio estimator supported the estimate of a standard deviation and standard error for each variable. The standard error for each estimate is shown in a table in the Appendices located directly below the results table on each page. We used the standard errors to calculate appropriate measures of precision for various kinds of hi 5 41 VT DPS

SECTION 5 RESIDENTIAL NEW CONSTRUCTION results. For estimates of totals and ratios (such as average percentage of homes built to Vermont Star standards), and proportions (such as the percentage of builders who had adopted a given practice), we generally used 90 percent confidence intervals. In some cases, the results were more informative and appropriate when presented as population-weighted estimates; these cases are noted within the report. 5 42 VT DPS

6 TRADE ALLY IN-DEPTH INTERVIEWS 6 TRADE ALLY IN-DEPTH INTERVIEWS As part of this evaluation, KEMA conducted interviews with small samples of trade allies who are in key positions to influence homebuyer and builder decisions in regard to the energy efficiency features of new and existing homes. We present analysis on trade ally knowledge of and interest in various EVT offerings, and trade ally interest in energy efficiency as it relates to their own business and markets. The trade ally groups include: Insulation Contractors. Insulation contractors have many opportunities to incorporate energy efficient options into residential projects. We present findings regarding these opportunities insulation contractors report, along with their knowledge of existing EVT initiatives and participation interest. KEMA s completed in-depth interviews with six insulation contractors. HVAC Contractors. Because builders and homeowners typically rely on HVAC contractors to specify and size central heating and cooling equipment, HVAC contractors typically exercise a fair amount of influence on these key energy-related home features. KEMA interviewed 9 HVAC contractors in depth, focusing on their perceptions of the factors that influence the efficiency of systems installed in new and existing homes in Vermont. We also compare the feedback we received from HVAC Contractors as part of the 2002 Residential New Construction analysis KEMA (previously XENERGY) conducted to identify any trends or changes in program perspectives over time. Remodelers. Remodelers are in a unique position to work with both homeowners and contractors on home remodeling and improvement projects. KEMA interviewed 16 remodelers, identifying the types of projects they were involved with in 2004, their perceived level of influence on energy efficiency recommendations, and their knowledge and assessment of state standards and initiatives available to them. 6.1 INSULATION CONTRACTORS 6.1.1 Survey Objectives and Methods Objectives. The primary objectives of the survey were to: gauge the market for insulation projects completed among sample respondents in new and existing homes in 2004; examine insulation contractor practices as they relate to energy efficiency; and solicit comments and suggestions regarding the promotion of energy efficiency in the residential construction market. 6 1

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Sample. Vermont insulation contractors who had completed at least four residential insulation projects in Vermont in 2004 were eligible for participation in the survey. We utilized Dun & Bradstreet s imarket Database to develop the sample frame. 6.1.2 Sample Characteristics Collectively, the six insulation contractors we interviewed had completed 975 residential insulation projects in 2004. The average number of insulation projects completed by respondents was 162 in 2004, ranging from a 20 to 500 residential insulation projects. The insulation projects respondents completed in new vs. existing homes vary across the six contractors interviewed: one insulation contractor reported completing the same number of projects in new vs. existing homes in 2004 (150 respectively). However, three insulation contractors did most of their work in new homes, and two predominately installed insulation in existing homes, in 2004. Table 6.1 Insulation Contractor Sample Distribution by Region Region Insulation Contractor Survey Completes Insulation Contractor Regional Allocation Northeast (NE) 1 17% Northwest (NW) 3 50% S. West /S. Central (SW/SC) 1 17% Southeast (SE) 1 17% TOTAL 6 Table 6.2 displays the 2004 establishment features and insulation project types reported by the six insulation contractors interviewed. Respondents tended to have small full-time equivalent staff, with the largest staff among respondents (10) completing 400 insulation jobs (in both new and existing homes) in 2004. 6 2

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Table 6.2 Insulation Contractor Establishment Characteristics and Project Type, 2004 Region # Employees Service Territory # Insulation Projects in # Insulation Projects in New Homes (2004) Existing Homes (2004) NE 2 VT, NH, NY, MA 150 150 NW 2 VT 78 2 SW/SC 5 VT 10 15 NW 5 VT 42 8 SE 8 VT, NH, MA, CT 250 150 VT + consults 10 NW internationally 0 20 Total 32 530 345 6.1.3 Awareness of Statewide Energy Standards Although all six insulation contractors reported that they had heard of the Residential Building Energy Standards (RBES), there was a high degree of variability when respondents attempted to summarize RBES features. Table 6.3 Awareness of RBES Standards (unprompted) # of Mentions Feature 1 Attic insulation at least R-38 1 Wall insulation at least R-19 2 Basement insulation at least R-10 0 Floor insulation at least R-10 1 Low-e windows 0 Argon-filled windows 0 Reduced air infiltration 0 High efficiency heating and cooling equipment 1 Mechanical ventilation 3 Don t know 6.1.4 Awareness and Knowledge of EVT Programs Vermont Energy Star Homes. While all six insulation contractors had heard of Vermont Energy Star Homes, none had participated in the program or received incentives from the program for homes insulated in 2004. 6 3

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Respondents reported a wide range of sources through which they learned about the Vermont Energy Star Homes program: two reported learning about the program through Vermont Star Home direct mail, one learned about the program through the Homebuilders Association, one reported learning about the program through other trade organization, and a final insulation contractor learned of the program through word of mouth. Home Performance with Energy Star Program. Only one of the six insulation contractors reported having heard of the Home Performance with Energy Star Program, though that contractor had very definitive views that the program was nonsensical, contending that the cost for higher rated insulation values exceed the benefits customers see in terms of building performance or energy savings. Only one of the other five insulation contractors indicated he would be interested in participating in a program like Home Performance with Energy Star once he was offered an overview of the program, specially citing the training and certification in building diagnostics as an attractive program offering. The remaining four respondents were split as to whether they did not want additional information about this program, or whether they were not certain if the program would offer them benefits. Among the insulation contractors who were skeptical about the program s potential effect, selected reasons for holding that opinion were as follows: It is time consuming trying to seal up older homes - crazy to displace new work to work on older houses - we don't have the time. It would be ok to pursue this program if I wanted to hire more people to keep up with the demand this created, but I am happy with my current workload. The $100 incentive does not seem like enough to justify the hassle. We don t have much control over these things they are already determined through an engineer or architect by the time we begin an insulation project. I would like to have more info to know exactly what the program offers. 6.1.5 Marketing Energy Efficiency Features Two-thirds of respondents said they discuss options for different levels of insulation in all or most cases when discussing a job with a customer; the remaining third said they discuss different insulation levels with customers in some cases (driven by specific customer preferences, rather than by contractor initiatives). Table 6.4 shows the energy-related insulation features respondents installed in their 2004 projects. Among respondents who installed a measure in less than 25% of projects, the interviewer probed as to why a measure was installed at a lower rate. 6 4

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Two respondents said that floor insulation is really only intended to soundproof, rather than serve an energy conservation purpose. One respondent said he never installs wall insulation above R-19, because it makes windows fit funny. Table 6.4 Energy Efficiency Measure Installation, 2004 Respondent Projects Feature % of respondents who installed (2004) Average % of 2004 Respondent projects where measure installed Attic insulation > R-38 83% 44% Wall insulation > R-19 67% 31% Basement insulation >R-10 67% 25% Floor insulation > R-10 17% 39% Reduced air infiltration as measured using blower door equipment 17% 17% Duct sealing and leakage Testing 33% 8% 6.1.6 Energy Efficiency Market Opportunities Four of the respondents completed an average of 86% of their insulation projects on residential new construction jobs over the past two years through their existing relations with builders (no bidding). One respondent said that about 50% of his new construction insulation projects were as a result of a written bid, with the remaining projects obtained through existing relations with builders (30%) and word of mouth (20%). Respondents vary regarding whether they submit energy efficient insulation options in bids. When submitting bids, however, two-thirds of insulation contractors interviewed said they only included pricing options for energy efficient features sometimes, with one respondent indicating he does not include this pricing detail in bids. Respondents ranged fully from 0% to 100% of the new construction insulation jobs they submit energy efficiency pricing options in three categories: insulation levels higher than code, insulation to basement walls, and air sealing, respectively. Half of respondents specify the type and level of insulation installed in most projects. Three respondents said that in 80% or more new construction projects, they (and not the builder) specified the type and level pf insulation installed. Five of six insulation contractors have been asked by builders to install insulation that exceed current minimum efficiency standards. Of these respondents, two said this request has 6 5

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS been made by builders more than 75% of the time. The most-cited reason respondents attributed to builders motivation was to improve his or her reputation as completing a higher-than-required quality job. Respondents do not have a desire to expand the insulation services they market to homebuilders. Two respondents offer air-sealing services, though they have not attempted to sell air sealing or air leakage testing to homebuilders. Respondents cited several reasons why they felt there are no business advantages to marketing high levels of insulation and air sealing to builders and general contractors, including: (air sealing and leakage tests) are a waste of time and materials. I am not out there trying to market myself - current contractors have me doing high levels of insulation; I have enough work to keep me back-logged without marketing more services. Most people we deal with already ask for high levels of insulation we don t have to market it. In general, people don't always think they need these higher-level services or that they make a difference. 6.2 HVAC CONTRACTORS 6.2.1 Survey Objectives and Methods The primary objectives of this survey were to: Assess HVAC contractors role and influence on the selection and sizing of equipment installed in new and existing homes; Assess HVAC contractors energy efficient equipment stocking and installation procedures; Gather HVAC contractors views on the factors that affect builder and owner decisions regarding selection of HVAC equipment; Assess HVAC contractors awareness and understanding of EVT programs; and The sample was initially selected at random from the imarket Database, and then supplemented by commercial listings (www.switchboard.com). Our sample targeted HVAC contractors who completed at least ten residential projects in 2004, any number of C&I projects, or some combination of both. Therefore, the survey instrument was designed to collect information about both elements of an HVAC contractor s business, if they serve both the residential and the commercial or industrial markets. Unlike the 2001 HVAC contractor sample plan, the 2004 targeted HVAC contractors based on a weighted population distribution in the four state regions rather than by firm size. KEMA 6 6

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS conducted the interviews. Our original target was 25 completes, however, a significant portion of the sample was exhausted with four or more call attempts without reaching the contact or being able to schedule an interview (or an interview was scheduled and the contractor cancelled). We attribute this trend in large part to the interviews occurring during May and June, a particularly high volume time for this segment. In addition, we found that the vast majority of individuals in the sample did not have an office or administrative staff for which to take calls or messages. Rather, most individuals in the sample represented a small operation generally with a home phone, often without voice messaging. Our call attempts were premised on reaching 25 completes. KEMA added sample from conventional directories and was able to complete 9 indepth interviews due to the call challenges noted above. 6.2.2 Respondent Characteristics Table 6.5 summarizes the business characteristics of the respondents to the HVAC contractor survey. In the 2001 evaluation, KEMA (then XENERGY) found that smaller companies generally consist of one proprietor/sales technician with one to three assistants. However, the average number of employees for respondents in 2004 was one-third what 2001 respondents reported. Table 6.5 Characteristics of Sample HVAC Contractors, 2001 and 2004 Characteristic 2001 2004 n = 12 9 Average # of Vermont heating or cooling equipment installations Average number of new Vermont homes in which heating or cooling equipment was installed Average number of employees at location Average number of technicians/installers at location Number that work outside local metropolitan area 46 43 20 34 13 4* 9 3 6 3 The range is 1 through 120 and the actual average is 17. Removing the high outlier, the average number of employees is more accurately 4. On average, HVAC contracting represents an average of nearly 60% of contractors 2004 business. Respondents predominately indicate that the remainder of their business is related to plumbing, and occasionally water heating. Of the nine HVAC Contractors interviewed in 2004, six exclusively did residential projects; one respondent only did C&I; whereas one HVAC 6 7

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS contractor reported doing work for both customer segments. We present findings on the specific types of residential and commercial and industrial installations in the section that follows. 6.1.3 2004 HVAC Installations HVAC contractors interviewed reported residential and commercial activities in Vermont in 2004. Table 6.6 shows the average number of projects and average percentage of projects that respondents report by customer type. As the table shows, about one-third of residential HVAC projects, and nearly half of C&I HVAC installations in 2004, occurred in new (as opposed to existing) homes or buildings. Table 6.6 2004 HVAC Installation Overview by Customer Type HVAC Installation 2004 Projects Residential C&I Average # 2004 HVAC installations 43 236 Average % 2004 HVAC installations in new homes/buildings 34% 44% Table 6.7 and Table 6.8 show the residential and C&I measures the HVAC contractors we interviewed installed in 2004. Respondents who installed residential HVAC measures said that almost 90% of their projects involved installing thermostats, while three-fourths of all HVAC residential installation projects involved oil-fired boilers. Respondents who installed C&I HVAC measures reported installing large furnaces (>100,000 BTUh) in about 44% of their projects, while packaged systems were installed in 56% of their C&I projects. 6 8

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Table 6.7 % of Residential HVAC Installations on 2004 Projects HVAC Residential Installation Average % of 2004 projects n = 8 Thermostats 88% Oil-Fired Boilers 72% Water heaters 26% Oil-Fired Furnaces 21% Duct work 20% Natural Gas Boilers 15% Hydronic distribution systems 14% Natural Gas Furnaces 12% Central air conditioners 10% Other Types of Heating Equipment (Propane/LP) 5% Heat pumps -- Table 6.8 % of C&I HVAC Installations on 2004 Projects HVAC C&I Installation Average % of 2004 projects n = 3 Furnaces (>100,000 BTUh) 44% Packaged systems (Unitary equipment) 11-20 Tons 33% Boilers (>1 Million BTUh) 25% Packaged systems (Unitary equipment) >20 Tons 23% Boilers (<1 Million BTUh) 14% Chillers (Built up systems) 13% Packaged systems (Unitary equipment) <11 Tons 10% Furnaces (<100,000 BTUh) < 1% 6.2.4 Awareness and Installation of ENERGY STAR-qualifying Equipment Six of the nine respondents said that they were familiar with the ENERGY STAR specifications for HVAC equipment for some types sold, whereas only one respondent indicated he is familiar with all ENERGY STAR specs for all types of HVAC systems sold. More importantly, two of the nine HVAC contractors we spoke with indicated they were not at all familiar with ENERGY STAR specifications for HVAC equipment. 6 9

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS After reviewing ENERGY STAR specifications, the interviewer asked the sample HVAC contractors to estimate what percentage of different kinds of equipment they installed in both new and existing homes met those efficiency specifications. Table 6.9 summarizes the results of this series of questions for projects reported in both 2001 and 2004. Table 6.9 Average Percent of ENERGY STAR-qualified Equipment Installed: 2001 and 2004 New Construction and Existing Homes Equipment Category 2001 2004 New Existing New Existing 1. Gas/oil furnaces 61% 81% 45% 38% 2. Gas/oil boilers 73% 72% 74% 24% 3. Central air conditioners 38% 58% 8% 8% 4. Air source heat pumps 100% 100% -- -- 5. Programmable t-stats 51% 33% 60% 40% * This figure does not manufactured housing units. ** Data were grouped exactly to reflect the ENERGY STAR specifications. Based on Table 6.9, we note the following: HVAC contractors interviewed for the prior evaluation reported that ENERGY STARrated gas/oil furnaces and boilers were installed in the highest in existing homes in 2001. Respondents reported that half as many furnace and boiler installations in 2004 new homes were ENERGY STAR qualified furnaces and boilers compared with the responses in 2001. Respondents in 2002 and 2005 consistently installed ENERGY STAR-rated programmable thermostat on heating and cooling equipment more than 50% of the time in new homes in both 2001 and 2004, and between 30 and 40% of the time in existing homes. Seven of the nine respondents in 2004 said that offering energy efficient HVAC options is somewhat or very important to their overall business. These respondents report discussing energy efficiency with their customers in all or most sales situations. Some of the messaging respondents said they emphasize with their customers includes: Energy savings = dollar savings You can save in the long run with energy efficiency equipment, and payback time is short Refer to cost savings from high oil costs Energy and cost savings, increased comfort 6 10

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS 6.2.5 Sizing and Installation Methods Because proper sizing and installation of HVAC equipment can yield energy savings compared with improperly sized or installed equipment, we explored the procedures respondents use to size and install HVAC equipment. Table 6.10 summarizes the range of responses HVAC contractors offered, revealing that there was no consistent use of any of the practices on which information was sought. Table 6.10 Respondents HVAC Equipment Sizing Methods # of Mentions Method to Size replace HVAC Equipment 2 Rule of Thumb (BTUh/Square foot) 2 Heat loss calculation 2 Defer to dealership to calculate Depends on situation - somestimes we replace 1 equipment with the same size; other times will do a heat loss calculation If the existing unit loosk like it was sized right, we match new unit to existing, but if we have any question that the 1 existing unit is sized properly, we have an engineer do calculations to size correctly Unlike HVAC respondents in the previous evaluation, none of the nine HVAC contractors interviewed were able to describe their installation procedures for HVAC equipment even on a general level. We asked respondents what percentage of existing HVAC equipment they find has been properly sized when they have done replacement installations. Table 6.11 shows that respondents feel that nearly three-fourths of existing HVAC installations they have observed have been properly sized. In addition, respondents reveal that C&I HVAC equipment is more than twice as likely to be undersized than residential equipment. Table 6.11 Respondents Perception of Existing HVAC Equipment Sizing Existing HVAC Installations Residential C&I Properly Sized 74% 70% Oversized 19% 15% Undersized 6% 15% 6 11

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS 6.2.6 Awareness and Assessment of EVT Programs Vermont Energy Star Homes Program Awareness. In 2001, seven of the twelve HVAC contractors reported that they were aware of the Vermont Energy Star Program, with only one claiming to be even somewhat familiar with the provisions of the program. In 2004, a higher proportion of HVAC contractors (eight of nine) said they were aware of Vermont Energy Star Homes (VESH) Program, however, only two of these respondents could name one element of the program each (incentives, and energy efficiency new construction guidelines). Assessment. After describing the program in some detail, the interviewer asked the respondents whether they believed the program would help them convince builders to install high efficiency HVAC equipment in residential new construction. Six of nine 2004 respondents said the thought it would, which compares favorably to HVAC contractor responses in 2001. Contractors cited rationale as to why respondents believed VESH Program would be somewhat or very effective to spur builders to include energy efficient HVAC equipment in their houses: People are very interested because of high oil prices I am seeing more and more people starting to ask for ENERGY STAR-rated equipment Among the contractors who were not convinced that the VESH Program was a magic bullet to spur energy efficiency, feedback includes: Half of the customers I encounter care about, energy efficiency, and the other half does not seem to put it at the top of their list. The decisions are all driven by what the customers want. Until the state of Vermont does a better job enforcing licenses, VESH will not go anywhere. We need to get banks/insurers involved for customer protection. HVAC contractors were also asked whether the VESH Program would be effective in encouraging proper sizing and installation of HVAC equipment in new homes. Two-thirds of the 2004 respondents said they believed VESH Program would be neither effective nor ineffective as it relates to proper sizing and installation of HVAC equipment in new homes. In fact, two customers classified this program as somewhat or very ineffective, citing their experience that customers want a less expensive product. Efficient HVAC Equipment Rebate Program Awareness. One in nine HVAC contractors indicated he was familiar with the Efficient HVAC Equipment Rebate Program and had participated in the program. This respondent was the only one interviewed in 2004 able to cite the three types of incentives the program offered (customer air conditioning, customer hearing, and dealer incentives). 6 12

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Assessment. After describing the program in some detail, the interviewer asked the respondents whether they believed the program would encourage HVAC contractors to sell and install more HVAC equipment to residential customers. Interestingly, half of respondents s thought Efficient HVAC Equipment Rebate Program would be somewhat or very effective, while half thought the program would either have no impact or be ineffective. The contractors who reached different conclusions about the impact of the program cite the effectiveness of incentives. Respondents who believe the program incentives might cause some contractors to consider selling and installing energy efficient equipment thought the program could be effective. However, the other half of HVAC contractors said that the incentives offered under this program were not enough to give them the incentive to sell and install HVAC equipment at a higher rate than currently. The Cool Choice Program Awareness. The level of respondent awareness of the Cool Choice Program is similar to that of the HVAC Equipment Rebate Program. One of the nine respondents participated in the Cool Choice Program, yet was able to explain the program s offerings or note that the initiative is targeted at Commercial and Industrial customers. However, respondents were more favorable in their assessments of the Cool Choice Program after hearing more details. Assessment. After describing the program in some detail, the interviewer asked the respondents whether they believed the program would encourage HVAC contractors to sell and install more HVAC equipment to residential customers. Seven of nine respondents thought the Cool Choice Program would be somewhat or very effective to achieve that goal. Examples of respondents positive reaction about the Cool Choice Program s incentive structure include: This program would be much more effective on the C&I end - businesses tend to be more conscious of saving money. This approach might offset the bottom line cost of projects and steer more people to push energy efficiency that might not have otherwise. 6.3 REMODELERS 6.3.1 Survey Objectives and Methods Objectives. The primary objectives of the remodeler survey was to: identify the 2004 remodeling project activity among interview respondents; determine remodeler perceptions of opportunities presented by the new construction initiative and its effects on new construction practices; 6 13

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS examine remodeler practices as they relate to energy efficiency; and solicit comments and suggestions regarding the promotion of energy efficiency in the residential construction market. Sample. We utilized Dun & Bradstreet s imarket Database to develop the sample frame, where the primary or secondary SIC is New Construction or Remodeling, Single or Multi-family. Vermont remodelers who had completed at least four remodeling projects in Vermont in 2004 were eligible for participation in the survey. Table 6-12 shows the regional distribution of respondents. Table 6.12 Remodeler Sample Distribution by Region Region Remodeler Survey Completes Remodeler Regional Allocation Northeast (NE) 0 0% Northwest (NW) 10 63% S. West /S. Central (SW/SC) 2 13% Southeast (SE) 4 25% TOTAL 16 6.1.2 Sample Characteristics Respondent Territory. Eight of the remodelers interviewed did work exclusively in Vermont in 2004; the remaining half of the remodelers interviewed did business primarily in New Hampshire, though two respondents cited additional states (Massachusetts, Florida and Ohio). Respondents reported an average of 6 full time equivalent employees per company, but several cited their significant reliance on subcontractors. 2004 Remodeling Projects. Collectively, the sixteen remodelers we interviewed had completed 112 residential projects in 2004. The average number of remodeling projects completed by respondents was 7 in 2004, ranging from a 4 to 20 residential remodeling projects. Project Specialization. Sixty-eight percent of the total remodeling projects respondents reported in 2004 involved breaching the envelope of a room to add additional space. The majority (88% of respondents) indicated that they did not specialize in the renovation of a particular room or space. 6 14

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Table 6-13 shows the average percentage of projects that involved specific activities, as specified by the sixteen remodelers interviewed. As this table reflects, most remodeling projects completed by respondents in 2004 involved installing new hard-wired fixtures (87%) or replacing windows (68%), with slightly fewer average projects (44%) involving installing new heating or cooling equipment. Purchasing appliances was nearly never a part of the remodeling projects respondents reported last year. Table 6.13 Remodeler Project Activity - 2004 Average % of Remodeling Project Activity Remodeling Projects (2004) Installation of new hard-wired lighting fixtures 87% Installation or replacement of windows 68% Installation of new heating or cooling equipment 44% Installation of new kitchen appliances 19% Installation of new clothes dryer 4% Installation of new clothes washer 2% 6.1.3 Awareness of Statewide Energy Standards Nearly all (fifteen of the sixteen respondents) indicated that they were aware of the Residential Building Energy Standards (RBES) in Vermont. However, as Table 6.14 reflects, when asked to describe the features of the RBES (unprompted), respondents familiarity was concentrated on insulation and window requirements, with less broad knowledge of the RBES. This distribution of unprompted responses among the remodelers interviewed, however, is consistent with the types of remodeling projects (primarily dealing with re-wiring and windows) they reported completing in 2004. 6 15

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Table 6.14 Awareness of RBES Standards (unprompted) # of Mentions Feature 8 Attic insulation at least R-38 8 Wall insulation at least R-19 7 Other (ranging from Energy Star requirements, to fresh air intake requirements, to it is entirely up to the customer ) 5 Low-e windows 3 Basement insulation at least R-10 3 Argon-filled windows 2 Floor insulation at least R-10 1 High efficiency heating and cooling equipment 1 Mechanical ventilation 0 Reduced air infiltration 6.1.4 Awareness and Knowledge of EVT Programs Vermont Energy Star Homes Program Awareness and Participation. The vast majority (14 of 16) remodelers interviewed reported that they were aware of the Vermont Energy Star Homes (VESH) Program. Of these fourteen, two respondents reported having received incentives from VESH on homes remodeled in 2004 (two and four homes, respectively). The two respondents who participated in the VESH Program in 2004 cited VESH direct mail pieces and EVT staff, respectively, as the most important factor as to why they participated. Other reasons cited included: discussions with other remodelers and information from professional organizations to which they belong. Both respondents who participated in the VESH Program in 2004 ranked the program as being very important in their decisions to include energy efficient features in the homes they remodeled in 2004. Both remodelers cited training and technical assistance as the most useful elements of the VESH Program as it related to marketing and selling energy efficient features. Additional feedback included: The financial incentives were good. Although it might not make or break the project, homeowners really like the idea of saving money during construction. Review and identification of energy savings opportunities were a huge selling point to customers. 6 16

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS However, one respondent suggested that the program staff consider more internal consistency, noting: I worked with three different technicians from the program and found that their responsiveness and general knowledge varied significantly. Non-participant Assessment. After describing VESH in some detail, the interviewer asked the respondents who had not participated (and who also indicated they were unfamiliar) with the VESH offerings about their perspectives on the usefulness and feedback about this program. The majority of respondents cited the review of plans and the assistance of energy efficient savings as a valuable aspect they perceive about the services VESH offers. Remodelers cited rationale as to why respondents believed VESH Program would be somewhat or very effective to spur builders to include energy efficient HVAC equipment in their houses: "The support offered to builders to review plans would help familiarize builders with energy efficient techniques and programs. The builder expects the contractor/remodeler to be knowledgeable on these services. When they are not up to speed, it requires additional time to research and makes the client unhappy. The key is to make follow-up with customer less time intensive and easier (e.g. if the program sent out catalogues or had a website where the remodeler could look up information, since I am seeing more and more people starting to ask for ENERGY STAR-rated equipment All aspects would be great, but respondent urged the program to market itself better to the public. Current materials are so vague. Home Performance with Energy Star Program Awareness and Participation. Only one of the 16 remodelers interviewed reported being familiar with the Home Performance with Energy Star Program. This respondent indicated that neither he nor his staff had participated in the program (either via a training or certification course). Non-participant Assessment. After describing the Home Performance initiative in some detail, the interviewer asked the respondents who had not participated (and who also indicated they were unfamiliar) about their perspectives on the usefulness and feedback about this program. The respondents were divided on the value they perceived from this program. Among the positive responses, remodelers indicated: We would like to associate our company with energy efficiency and savings to target younger, more educated clientele. This target group is more likely to recognize life cycle cost savings rather than fixate on higher initial outlay and have more insight on what is going on in their market. 6 17

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS I am always interested in receiving new information. The training and certification support would be especially helpful in marketing energy efficient technologies to the customer. It would be helpful to have human contact within the program to "back up the contractor" when they are discussing new energy efficient options. The entire menu sounds good. Specifically, the review of products would be helpful. I would like to market our business/company as being concerned about the environment and market energy efficiency as being an important aspect of his building techniques. Being better informed means a better product for our clients. It would be most helpful to know about new technology so we could think twice about potential options before remodeling a home. However, about half of the non-participants were not convinced that the Home Performance with Energy Star Program would be something they would participate in going forward: It appears that this program must target less expensive projects. Our company does high end, expensive, complicated projects and does not want to target less expensive projects. The amount of time to participate would be too much; $100 does not warrant that. Even if customers are interested, the reality is that the[energy efficiency] installation costs more even with incentives. Respondent deals mostly with middle class customers who are more likely to spend any extra money on kitchen appliances. Fifteen of the sixteen respondents were not able to indicate what information, if any, they would need to increase their level of interest in the program. One remodeler suggested offering seminars and more reading materials that would describe the program before determining his level of interest in the services. 6.1.5 Marketing Energy Efficiency Features On average, the sixteen respondents said that the issue of energy-efficiency is discussed with the owner of a property about 66% of the time when the remodeler is developing plans for a renovation or addition. Three of the sixteen remodelers, however, said they discuss energy efficiency with a homeowner less than 20% of the time. Table 6-15 shows the average % of projects across remodeler respondents who reported installing specific energy efficient features in their 2004 remodeling projects. 6 18

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Table 6.15 Remodeler Energy Efficient Project Activity - 2004 Average % of Feature 2004 Projects Low-e windows 94% Argon-filled windows 69% Advanced controls such as fan timers, occupancy sensors, programmable t-stats 68% Wall insulation above R-19 63% Energy Star high-efficiency heating and cooling equipment 46% Attic insulation above R 38 43% Basement insulation above R-10 36% Energy Star hard-wired compact fluorescent lighting fixtures 28% Floor insulation greater than R-10 27% Hi-efficiency mechanical ventilation 16% Duct sealing and leakage testing 15% Energy Star high-efficiency appliances 13% Reduced air infiltration as measured using blower door equipment 11% As the Table 6-15 reflects, window installation was key to energy efficient installations among the remodelers interviewed. Remodelers overwhelmingly indicate they the energy efficiency ball is always in the owners court. However, for features which were pursued least frequently, Remodelers point out that customers almost always select their own appliances. In addition, the remodelers we interviewed indicated they do not believe blower door tests are necessary, and will only pursue additional duct sealing and leakage testing if the homeowner requests it. Remodelers acknowledge cost is king to homeowners. All respondents indicated that initial cost is the most frequently cited objection that homeowners raise if the remodeler proposes an energy efficient feature during the planning phase of a project. However, all respondents said that reduced energy costs were the main benefit to homeowners if they chose an energy efficient option (generally). Other benefits remodelers cited include: greater comfort, lower long-term maintenance costs, and seven of the sixteen respondents cited energy conservation/benefiting the environment. No consistent approach to demonstrating project energy costs/savings. Of the remodelers interviewed (30%) who almost always calculate estimates of the renovation project s impact on a home s energy use and energy costs, each respondent had a different method of calculating costs and savings, ranging from: using a sub contractor s estimates; referencing the "Underwriters 6 19

SECTION 6 TRADE ALLY IN-DEPTH INTERVIEWS Laboratory Guide ; utilizing the manufacturer specifications; and one respondent who relied on EVT program staff to help with calculations. Energy efficiency advocacy is not core to remodeler planning activities. The upfront cost vs. long-term savings discussion is one that 50% of the remodelers interviewed said they will have initially with a homeowner, but respondents overwhelmingly indicate that whatever the customer wants is what they include in their plans, regardless of the efficiency levels, even if the remodeler him/herself acknowledges the many benefits of energy efficient options. While five respondents (about 30%) said the almost always calculate estimates of the renovation project s impact on a home s energy use and energy costs, nine respondents (56%) admit they calculate costs and savings rarely or never. The remaining respondents (14%) said they calculate calculate energy costs and savings as a result of their remodeling project only in some cases when a customer requests. 6 20

7 PROCESS EVALUATION 7 PROCESS EVALUATION This section presents analysis of selected issues concerning EVT s internal organization, marketing effectiveness, and performance relative to other organizations that operate rate payerfunded energy efficiency programs. 7.1 INTERNAL ORGANIZATION AND ADMINISTRATION In the course of conducting this evaluation, KEMA interviewed over a dozen EVT staff members from the including the Executive Director, Business Development Manager, Information Technology Manager, Residential Program Manager, managers of the Products, Construction, and Existing Homes programs, and a number of business development, planning, and technical staff. In addition, we reviewed the data structure of the KITT program tracking databases and other specialized program databases. We also made extensive use in the analyses contained in Sections 1 6 of raw data extracted from the KITT database by EVT IT staff and of standard and ad hoc reports prepared from those data by program and planning staff. This work enabled us to assess administrative systems and data quality. Generally, we found EVT to be a very well managed organization. To support this conclusion, we provide the following findings. Organization and Management Consistent understanding of mission, brand, and strategy. All staff we interviewed reported a uniform understanding of the Energy Efficiency Utility s mission to achieve energy savings and to change or transform the relevant markets so that efficiency gains become sustainable. Moreover, the organization had recently gone through a branding exercise that stressed the importance of external focus and excellent customer service. All of the staff we spoke with had understood this message and were attempting to adapt it to their daily responsibilities. Finally, most of the staff we spoke with were able to articulate the connection between mission and brand on one hand and their program strategies on the other. KEMA believes that the high level of public awareness for Efficiency Vermont and generally high consumer ratings for program quality reflect, in part, these organizational development efforts. Knowledgeable, capable staff. KEMA found all of the program, marketing, and technical staff we interviewed to be very knowledgeable about their markets and about developments in the technologies they were promoting. Administrative Systems Excellent design of administrative and information technology systems. The relational structure of the KITT system and smaller systems that support the New 7 1 VT DPS

SECTION 7 PROCESS EVALUATION Construction program are well-designed to provide flexible support for planning and administrative tasks. Excellent implementation of tracking systems. KEMA found the data sets used in this evaluation to be completely and accurately populated. For example, we found few participant customer or vendor records with missing or false information. 7.2 MARKETING: CUSTOMER AWARENESS AND PERCEPTIONS EVT has achieved high levels of awareness for its programs among residential customers, firms in the single-family construction and renovation industry and appliance and lighting retailers. Participation among appliance retailers and outlets that carry compact fluorescent light bulbs is nearly universal. Participation among large builders and mid-sized builders is strong and has improved significantly since the Phase 1 Evaluation. The following paragraphs provide significant details on the efficacy of EVT marketing. Customer Awareness and Participation High level of customer awareness. Forty-eight percent of RASS respondents reported that they knew of Efficiency Vermont. Regionally, the percentage of aware customers ranged from 38 in the Southeast to 55 in Northwest. In response to an open-ended question concerning what EVT does, most were able to provide accurate information to the effect that EVT administers various kinds of energy efficiency programs. Awareness and Participation among Building Trades Capture of single-family new construction projects. EVT has done a good job of identifying and reaching principals in single-family new construction projects. In the years covered, the program has reached principals of 1,551 to 1,950 projects, which corresponds to 61 to 74 percent of the typical volume of single-family homes permitted each year (around 2,500). This is a very high proportion compared to other states. Awareness and participation among builders. Builders representing approximately 92 percent of the state s new construction in 2004 indicated awareness of the Vermont Energy Star Homes program. Of builders who reported awareness of the program, builders representing a larger proportion of market share in 2004 reported participating in the program than in 2001 (43 and 24 percent, respectively). Awareness and participation among remodelers. The vast majority (14 of 16) remodelers interviewed reported that they were aware of the Vermont Energy Star Homes (VESH) Program. Awareness and Participation among Retailers Since 2001, virtually all appliance retailers in Vermont have participated in the Appliance component of the Efficient Products Program. Participation by lighting retailers. Retailers who account for 98 percent of all CFL sales participate in the lighting incentive program. 7 2 VT DPS

SECTION 7 PROCESS EVALUATION 7.3 EVT PERFORMANCE METRICS COMPARED TO OTHER STATES One of the most useful ways to assess the overall performance of EVT is to compare it to that of peer programs using a consistent set of indicators. In this case, peer programs would consist of long-established public benefit-funded energy efficiency programs that address all or most of the customers in a given state. Unfortunately, regulators in the individual states do not collect program performance information in uniform ways. Thus direct comparisons between programs need to be treated as very general in nature. Table 7-1 displays information on the budgets, first-year electric savings, first-year savings as a percent of total electric sales in the relevant market, and the average cost per kwh of first-year savings for eight states with peer programs. Overall, Vermont tied with three other programs for the highest level of electric savings as a portion of sales with 0.8 percent EVT s residential programs captured savings equal to 0.9 percent of total residential electric sales in Vermont. The cost per first year savings provides a very rough measure of cost-effectiveness for a portfolio of programs. It would be much more appropriate to consider the combined results of the portfolio using the Total Resource Cost Test, or similar measures that take measure lifetime savings, customer costs, and environmental benefits into account. However, such an analysis would require a great deal more data than is readily available from other programs. With this caveat in mind, we see that EVT ranked third among states with peer programs in terms of cost per first year kwh savings. For all of EVT s programs combined, this figure was $0.281/kWh. The range for this indicator ran from $0.231/kWh in Wisconsin to $0.580/kWh in New Jersey. State Table 7-1 Comparative High-Level Performance Statistics State Public Benefit-Funded Energy Efficiency Programs Year Budget $mm/yr % Elec. Rev s Savings New MWH/Year % Electric Sales $/kwh New 1 st Year Savings CA 2003 240.0 1.5% 933,365 0.8% 0.257 CT 2002 87.1 3.1% 246,000 0.8% 0.354 MA 2002 128.0 3.0% 241,000 0.7% 0.531 NJ 2002 99.6 1.5% 171,692 0.2% 0.580 NY 2002 129.0 1.3% 290,000 0.3% 0.445 RI 2002 16.4 2.7% 50,568 0.8% 0.324 WI 2003 49.7 1.4% 214,800 0.4% 0.231 VT 2003 13.7 3.3% 48,763 0.8% 0.281 VT Res 2003 5.7 18,439 0.9% 0.311 7 3 VT DPS