WILL THE RECENT ROBUST ECONOMIC GROWTH CREATE A BURGEONING MIDDLE CLASS IN THE PHILIPPINES? Romulo A. Virola

Transcription

1 12th National Convention on Statistics (NCS) EDSA Shangri-La Hotel, Mandaluyong City October 1-2, 2013 WILL THE RECENT ROBUST ECONOMIC GROWTH CREATE A BURGEONING MIDDLE CLASS IN THE PHILIPPINES? by Romulo A. Virola, Jessamyn O. Encarnacion, Bernadette B. Balamban, Mildred B. Addawe, and Mechelle M. Viernes For additional information, please contact: Author s name Designation Romulo A. Virola Former Secretary General, National Statistical Coordination Board, & Consultant Statistically Speaking Consultancy Services (SSCS) # 2 Camia St., Vergonville, Las Piñas City, Philippines Affiliation Address Tel. no ; ravirola@yahoo.com Co-authors names Jessamyn O. Encarnacion, Bernadette B. Balamban, Mildred B. Addawe, and Mechelle M. Viernes Designation Director III, Statistical Coordination Officer (SCO) VI, SCO V, and SCO III Affiliation National Statistical Coordination Board Address 403 Sen. Gil Puyat Avenue, Makati City Tel. no addresses jo.encarnacion@nscb.gov.ph; bb.balamban@nscb.gov.ph; mb.addawe@nscb.gov.ph; mm.viernes@nscb.gov.ph; 1

2 WILL THE RECENT ROBUST ECONOMIC GROWTH CREATE A BURGEONING MIDDLE CLASS IN THE PHILIPPINES? by Romulo A. Virola, Jessamyn O. Encarnacion, Bernadette B. Balamban, Mildred B. Addawe, and Mechelle M. Viernes 1 Abstract It is now widely-accepted that the development of a nation hinges on building its middle class. With the impressive 6.8% growth of the country s Gross Domestic Product (GDP) for 2012, and four consecutive quarters of GDP growth of more than 7.0% since the third quarter of 2012, as well as the ratings upgrade to investment grade by Standard & Poor s and Fitch, and the expected similar upgrade by Moody s, the prospects for the Philippines joining the Asian tigers have surely become rosier. To achieve this goal, the Filipino middle class will have to play its role. However, aside from the fact that there is no internationally-adopted definition of the middle class, the systematic generation of data on the middle class has not been institutionalized in the Philippine Statistical System (PSS). During the 10 th National Convention on Statistics (NCS), Virola, Addawe & Querubin presented a paper that used cluster analysis and multiple regression to propose two possible definitions of the middle class, one based on income and the other based on socio-economic characteristics. The paper used the 1997, 2000 and 2003 data from the Family Income and Expenditures Surveys (FIES) and the January 2001 and 2004 Labor Force Surveys (LFS). During the 11 th NCS, Virola, et. al. updated and improved on the 10 th NCS paper, adding auxiliary variables as well as two-way interaction among independent variables in the multiple regression component, and using data from the 2000, 2003, and 2006 FIES and the January 2001, 2004 and 2007 LFS. The most worrying result that the two papers showed is a shrinking middle class in the Philippines. Convinced that the PSS should sustain the generation of statistics that can contribute to policy formulation towards the protection of the middle class, in particular, and to evidence-based decision making towards national progress in general, this paper will highlight the socio-economic and demographic characteristics as well as the province of residence of the Filipino middle class. It will also assess whether the middle class has started to expand preparing the country better for the development challenges ahead. As in the previous papers, data from the FIES and the LFS will be used. Key words and phrases: middle-income class; cluster analysis; multiple regression; socio-economic characteristics; demographic characteristics, province of residence. 1 Former Secretary General, Director III, Division Chief, Statistical Coordination Officer V, and Statistical Coordination Officer III, respectively, of the National Statistical Coordination Board. The views expressed in this paper are those of the authors and do not necessarily reflect the views of the NSCB. The authors acknowledge the assistance of Noel S. Nepomuceno and Albert A. Garcia and the cooperation of the National Statistics Office, particularly the Income and Employment Statistics Division (IESD) in the preparation of this paper. 2

3 I. Introduction It is now widely-accepted that the development of a nation hinges on building its middle class 2. In 2012, the economy recorded an impressive 6.8% growth of the country s Gross Domestic Product (GDP) and for four consecutive quarters since the third quarter of 2012, GDP has grown by more than 7.0% 3 (see [1]) Prior to the release of the 2013 Q2 GDP by the National Statistical Coordination Board (NSCB) Technical Staff (TS), the Philippines has been included in the N11 economies 4, which according to Goldman Sachs (see [2]) could follow the BRIC 5 countries Recently, the Philippines was upgraded to investment grade by Standard & Poor s and Fitch, with the understandable expectation that the third major rating agency Moody s, will follow suit. In addition, the Philippines jumped several notches up from 85 th out of 139 countries in to 65 th out of 144 in and to 59 th out of 144 countries in in the World Economic Forum Global Competitiveness Index. Surely, the prospects for the Philippines joining the Asian tigers have become rosier. To achieve this goal, the Filipino middle class will have to play its role Despite all the good news and the worldwide recognition of the robust performance of the Philippine economy, the news about the overarching goal of poverty reduction has not been that good. In the words of the NSCB TS, the poverty situation has remained practically unchanged from the first semester of 2009 with a poverty incidence among families of 22.9 % to the first semester of 2012 with a poverty incidence among families of 22.3%. (see [3]) In contrast, the British Broadcasting Company says hundreds of millions of people around the world are escaping poverty and becoming middle class. The explosion of new consumers in China, India and other economic powerhouses is changing the global balance of power. (see [4] ) Related references on the middle class in the developing world are [5] and [6]. But why has the economic growth in the Philippines not been trickling down to the poor, as claimed by many? Is it only the rich and the Senators, Congressmen and all those 2 In a speech last 31 August 2010, US President Barack Obama underscored the importance of the middle class calling them the bedrock of their prosperity, and hence, the need for them to strengthen their middle class by giving their children the education they deserve and the workers the skills they need to compete in the global economy. On the other hand, the Asian Development Bank (ADB) stresses the role of the middle class, specifically in Asia, referring to it as the main source of social activists who typically found and operate non-government organizations that demand greater government accountability. (Asian Development Bank. Key Indicators for Asia and the Pacific Special Chapter: Rise of the Asia s Middle Class) 3 The GDP grew by 7.3% in 2012 Q3, by 7.1% in 2012 Q4, by 7.7% in 2013 Q1 and by 7.5% in 2013 Q2, at constant prices. 4 Next Eleven Economies include Bangladesh, Egypt, Indonesia, Iran, Mexico, Nigeria, Pakistan, Philippines, South Korea, Turkey and Vietnam 5 The BRIC countries are Brazil, Russia, India, and China. 3

4 involved in the pork barrel scam who have been unconscionably reaping the benefits of this growth? Or are our economic managers really not smart enough to have learned the ropes of development and come up with programs and policies that will translate into better quality of lives for the marginalized sectors of our society? Could it be that our development agenda simply fails to recognize or does not consider it imperative that the middle class must expand if we want to achieve inclusive growth and meaningful development? And could this be a reason why numerous studies have been done and indicators/statistics developed with primary focus on the poor, and very little focus on the vulnerable sectors of society including the middle-income class? In fact, the Millennium Development Goals do not include an indicator on the middle class.. In order to effectively monitor the situation of the middle class, high quality and relevant statistics are needed. This paper advocates for the generation and dissemination of statistics on the middle class in the Philippine Statistical System (PSS). However, there is no internationally-adopted definition of the middle class 6. Some have defined it based on relative measures while others use absolute measures. Partly because of this, the systematic generation of data on the middle class has not been institutionalized in the PSS. During the 10 th National Convention on Statistics (NCS), Virola, Addawe & Querubin presented a paper (see[7]) that used cluster analysis and multiple regression to propose two possible definitions of the middle class, one based on income and the other based on socioeconomic characteristics. The paper used the 1997, 2000 and 2003 data from the Family Income and Expenditures Surveys (FIES) and the January 2001 and 2004 Labor Force Surveys (LFS). In June 2009, this was updated through an NSCB website Statistically Speaking article (see [8]) by Virola and Addawe. Another Statistically Speaking article (see [9]) by Virola, Encarnacion, and Viernes looked into the income and expenditure pattern of the rich in comparison with the middle class and the poor. During the 11 th NCS, Virola, et. al. updated and improved (see[10]) on the 10 th NCS paper, adding auxiliary variables as well as two-way interaction among independent variables in the multiple regression component, and using data from the 2000, 2003, and 2006 FIES and the January 2001, 2004 and 2007 LFS. The studies showed that the ranks of the rich dwindled from 2000 to The most worrying result shown by the papers is a shrinking middle class in the 6 In a study done by the ADB, they have defined the middle class as those with consumption expenditures of $2-$20 per person per day in 2005 PPP$. Ravallion (2009), on the other hand, defined the developing world s middle class as those who live above the median poverty line of developing countries but are still poor by US standards or those with income between $2 per person per day and $13 per person per day. Another study done by Birdshall, Graham and Pettinato (2000) defined the middle class as those earning between 75% and 125% of a society s median per capita income. 4

5 Philippines, losing its members to the low-income class. However, these papers did not say if the decline was statistically significant. Convinced that the PSS should sustain the generation of statistics that can contribute to policy formulation towards the protection of the middle class, in particular, and to evidence-based decision making towards national progress in general, this paper will highlight the socio-economic and demographic characteristics as well as the province of residence of the Filipino middle class. It will also assess whether the middle class has started to expand at a statistically significant pace, thereby preparing the country better for the development challenges ahead. As in the previous papers, data from the FIES and the LFS including the 2009 FIES and the January round of the 2010 LFS will be used. The paper uses Cluster Analysis to define the income bracket of the middle class. T- tests of statistical significance are used to assess the changes in the share of the middle class in the population distribution at the national level. The limitations of the methodology are described in the 11 th NCS paper ( see [10]). The next section will present an overview of the data sources and the methodology. Section 3 will show the results and the last section will give some concluding remarks and update the recommendations to advance the research agenda on the middle class of the Philippines. II. Data Sources and the Methodology Data Sources As in the other NSCB TS studies by the authors on the middle class, the data used come mainly from the FIES and the LFS conducted by the NSO. More specifically, this paper uses the 2003, 2006 and 2009 FIES merged with the January round of the 2004, 2007 and 2010 LFS, respectively. The authors had planned to use the 2012 FIES for more updated information; unfortunately, the microdata files of the 2012 FIES have not yet been released to the public. The FIES is a nationwide household survey, conducted by the NSO every three years to collect data on family income, sources of income as well as family expenditure and other related information, which can be used to determine the degree of inequality among families, provide information to update the weights used in the compilation of the CPI and in the estimation of poverty statistics in the country. However, most of the information collected by the FIES refers to the collective characteristics of the family and the household head only, hence, the need to merge with the LFS, to provide greater flexibility in 5

6 the analysis. The LFS is also a nationwide survey conducted by the NSO every quarter, designed to provide statistics on the levels and trends of employment, unemployment and underemployment 7 in the country. It contains data on the characteristics of the different household members, particularly on their education and employment. With the FIES being a rider to the LFS, these two data sets can be merged to produce a dataset containing characteristics of the family as well as its individual members, which can all be useful in the analysis of the characteristics of the middle class. Methodology A. Identifying Middle-Income Class Based on Income The methodology used in this paper to define the middle class is the approach based on income (see [10] ) and uses cluster analysis. Cluster analysis is used to divide the population into clusters. This tool is a multivariate analysis technique that classifies objects or individuals into a small number of mutually exclusive groups based on the similarities among the entities so that each object is very similar to others in the cluster with respect to some predetermined selection criterion (see [11]). In the 2013 paper, the option of five clusters was specified, as was chosen in the 2010 paper when the cluster analysis was done on 3, 4, 5, and 6 clusters. The 5-cluster option gives the most meaningful results in identifying the middle class. The cluster analysis allows us to define the per capita income boundaries of the middle class in the reference year. These boundaries are extrapolated to future years using the CPI instead of doing separate cluster analyses for each of the FIES years to maintain consistency in the standard of living defined for the middle class. Redoing the cluster analysis for each FIES could mean a changing conceptual definition of the middle class with every FIES. This is similar to the issue of whether to use a fixed or changing FE/TBE ratio in the generation of poverty statistics 8. The differences in the methodology between the 2010 NCS and the 2013 NCS papers are as follows: 1. Cluster Analysis is performed on the 2003 FIES for the 2013 paper while the 2000 FIES was used for the 2010 paper. The current master sample used by the NSO in its household surveys was first used in Available poverty estimates are based on data using this master sample and therefore using 7 Starting 2003 when a new master sample was used with region as domains, the NSO only generates employment data at the provincial level and no longer generated unemployment and underemployment data.. 8 During the workshop of the Technical Committee on Poverty Statistics on the overall review of the official poverty estimation methodology held last August 27-29, 2010, the TC PovStat recommended the use of a constant FE/TBE ratio for the indirect estimation of the non-food threshold for a period of 12 years to ensure consistency of the estimates across time. 6

7 the 2003 FIES instead of the 2000 FIES in the cluster analysis will enhance the comparability/consistency of statistics on the middle class with official poverty statistics. Thus the base year/reference year was changed from 2000 to The income variable used for the 2013 paper is per capita income; the 2010 paper used total family income. Again, this is consistent with poverty estimation and it takes into consideration the size of the family that contributes to and spends the total family income. This was also one of the comments raised during the presentation of the 2010 paper. 3. The CPI used to define the per capita income boundaries of the middle class for years other than the reference year was the 2000-based CPI for the 2010 paper and the 2006-based CPI9 for the 2013 paper. A series of runs on 3, 4, 5, and 6 clusters was undertaken but the five-group cluster analysis performed on the 2003-merged FIES-LFS provided the most meaningful results in classifying low, middle or high-income class families. For example, as shown in Annex Table 1 using 3 clusters, the maximum per capita annual income for the low-income class is too high resulting in 94.03% of the families being classified as low income. Using 4 clusters, the maximum per capita income for the low-income class may still be high while the maximum per capita income for the middle class seems to be too high [Annex Table 2]. Using 6 clusters, both the minimum per capita income and the maximum per capita income for the middle-income class seem too high [Annex Table 3]. For purposes of defining the income boundaries of the three income classes using the results of the 5-cluster run, the actual limits of the income intervals were used. Thus, the class intervals are not continuous. While recomputations using continuous intervals did not show differences in the share of the three income classes, in future work, interpolation of the class limits should nevertheless be done to come up with continuous, contiguous intervals. B. Characterizing the Middle-Income Class Based on Socio-Economic Characteristics In addition to the distribution of families by income class from 2003 to 2009, this paper presents some socio-economic characteristics of the middle-income class families identified based on their income. 9 The use of 2006 as base year for the CPI, replacing 2000, was approved by the NSCB Executive Board thru Board Resolution No.7 Series of

8 The merged FIES-LFS was used to draw the profile of the middle-income class families based on their location to examine whether the distribution of families belonging to the middle-income class varies considerably across location by region, province and by urban and rural and over time. Moreover, the characteristics of the housing units such as the type of roof materials, type of wall materials, tenure status, type of building and type of toilet facility for the low-, middle-and high-income class families were also analyzed. The household composition, which includes the educational attainment of the working age population, average percentage of school children who are currently in school and the average percentage of working age population who are employed were also examined. Furthermore, an analysis on the household head s age, sex, marital status, highest educational attainment, employment status and occupation was done. III. Results and Discussions A. Distribution By Income Classes This section presents the distribution of families into low-, middle- and high- income class based on the income cut-off defined using the 5-Cluster Analysis. 1) Per Capita Annual Income of the Low-, Middle-, and High-income classes (Table 4 and 5) As shown in Table 4, Cluster1 is the appropriate cluster to define the income limits of the low-income class, Clusters 2 and 3 for the middle-income class, and Clusters 4 and 5 for the high-income class. Thus, the per capita annual income limits of the middleincome families for 2003 are P 41,972 P 513,950. Families with per capita annual income below the lower limit (P 41,972) will be low income, and families with per capita annual income higher than the upper limit (P 513,950) will be high income. Table 5 shows that the annual per capita income of the middle income class for 2003, 2006, 2009, 2012, and 2013 applying the 2006 based CPI. For 2013, the middle income families are those with per capita income of PhP 65,787 to PhP 805,582. For different family sizes, the required income for a family to be classified as middle class is given in Table No economies of scale principle applied; computed simply as per capita income requirement multiplied by family size. 8

9 Thus, a national government employee without a dependent and with no other sources of income will be classified as middle income class, if he/she is holding a position with Salary Grade of at most 26 (about PhP 790,364/annum). A Department Undersecretary with Salary Grade (at least PhP 1,062,298/annum) and any Municipal Mayor with Salary Grade (at least PhP 769,239/annum) and even a Cabinet Secretary with nonworking, nonearning wife/husband and three children 13 will still belong to the middle income class, not to the high income class unless he/she has other sources of income, or unless he receives huge, possibly illegal allowances. The unmarried President with Salary Grade 33 14, even if he has no other sources of income, belongs to the high income class. Do the members of the Judiciary, the Senators, the Congressmen, the other politicians, the pork barrel scammers, etc. with four nonworking dependents still belong the middle class? Even BIR Commissioner Kim Henares probably does not know! 2) Trends in the Structure of the Distribution of Families by Income-Class (Tables 5, 6, and 9) At the national level, almost 24% of the total families in 2003 and 2006 were classified as middle income families. This increased to 25.2% in 2009 with the additions coming from a reduction of the share of low income families from 76% in 2003 and 2006 to 74.7% in On the other hand, the high income families represented barely 1% of the total families in the country for 2003, 2006 and (Table 7) Table 7 shows that the percentage of families belonging to the middle-income class expanded from 2003 to 2006 and 2006 to 2009, in contrast with the findings in the earlier papers of Virola, et. al. This is due to the change in the All municipal mayors have Salary Grade 27, except those in Metro Manila with Salary Grade 28. The Salary Grade indicated was sourced from 13 The CPH provides information on the average household size and not on the average family size. Thus, an adjustment factor is needed to obtain the latter using the information on the former. Based on the results of the 2010 CPH, the average household size is The adjustment factor is computed as the ratio of the 2009 FIES average family size (average of the 2 visits in July 2009 and January 2010) to the 2009 LFS average household size (average of the July 2009 and January 2010 LFS). Based on this, the estimated average family size of Filipinos for 2010 is Moreover, using the results of the 2012 FIES, the estimated average family size in 2012 is

10 methodology for the 2012 paper adopting 2003 FIES as reference year as against 2000 FIES and adopting the 2006-based CPI over 2000-CPI. The other changes in the methodology did not contribute to the reversal in the trend of the share of the middle income class from the previous papers. In 2003, 23.8 percent of the families were classified as middle-income class. The share of the middle-income class increased by 0.1 percentage points from 23.8% in 2003 to 23.9% in 2006, but the increase is not statistically significant. On the other hand, the increase in the percentage of middle income class families from 23.9% in 2006 to 25.2% in 2009 is statistically significant. Likewise, the decrease in the percentage of families belonging to the low-income class from 76.0% in 2006 to 74.7% in 2009 is statistically significant. 3) The Poor and the Low Income Class Table 8 shows the distribution the low income group into poor and non-poor families. The number of poor families is obtained from the official poverty statistics (see [12]) and is subtracted from the number of low income families to obtain the number of non-poor low income families. Many families in the low income class are not poor. In terms of the magnitude of families, Table 8 shows an increasing number of middle-income families from 2003 to In particular, between 2006 and 2009, one family per hundred was added to the middle-income families. Both in terms of absolute number and percentage share, the middle-income class of the Philippines has been expanding. On the other hand, the high income class which is already below 0.2% of the distribution is showing indications of shrinking, just like the low income class.. B. Characteristics of them middle income class 1) Characteristics of Middle-Income class families a. Location by Region Among the regions, it is only in the National Capital Region (NCR) where more than 50% of the families belong to the middle income class. The increased share of middle income class families in this region comes from the reduced 10

11 share of the low income class families, from 47.9% in 2003 to 46.9% in 2006 and 46% in (Table 9) In 15 of the 17 regions in the country, more than 70% of the families belong to the low income class. Only NCR and Region IVA have less than 70%: % for NCR and 65-67% for Region IV A. (Table 9) The highest proportion of low income families at 94-96% is in ARMM. For the years 2003, 2006 and 2009, no family from ARMM belonged to the high income class. (Table 9) The regions where the middle income class families comprise relatively the highest shares are NCR (51-54%), Region IV A (32-35%), Region III (28-29%) and CAR (26-29%). (Table 9) In terms of absolute number, the biggest concentration of middle income families in 2009 is in NCR with 28.3% share of the middle income families in the country, followed by Region IV-A with 17.5%, Region III with 12.7%, and Region VII with 5.9%. (Table 9.1) Three regions had a faster growing middle class than the national rate both from 2003 to 2006 and from 2006 to 2009: Region II, Region VIII, and Region XI. On the other hand, those with slower growing middle class than the national rate both from 2003 to 2006 and from 2006 to 2009 are Region III, Region IV- A, Region V, ARMM, and Caraga. (Table 9) b. Location by Province In the districts of the NCR except for the 3 rd District, more than 50% of the families are middle income. In the 3 rd district which includes the cities of Caloocan, Malabon, Navotas and Valenzuela, the middle income class comprises only 42-45%, with the low income class comprising 55-58%. (Table 10.1) In 2009, the middle income class comprised more than 30% of the total families in only 8 out of the 79 provinces, all of which are in Luzon: Benguet (46.7%), Cavite (43.6%), Batanes (41.7%), Bataan (39.4%), Bulacan (37.6%), Laguna (35.7%), Aurora (33.5%), and Pampanga (30.1%).(Table 10) In the Visayas region, only Cebu (24.8%), Iloilo (23.8%), and Biliran (22.6%) had more than 20% of families belonging to the middle income class in For Mindanao, the provinces with the highest relative share of the middle income families in 2009 are Davao del Sur (23.9%), Misamis Oriental (23.8%) and South Cotabato (22.6%) (Table 10) The provinces with the smallest relative sizes of the middle income class are Sulu (1.0%), Maguindanao (1.5%), Siquijor (2.0%), Tawi Tawi (4.5%) and Davao Oriental (6.0%)., four of which are in Mindanao.(Table 10) Outside of NCR, in terms of absolute number, the biggest concentration of middle income families among the provinces in 2012 are in Cavite with 324,609 or 5.8% share of the middle income families in the country, followed 11

12 by Laguna with 260,309 or 4.6%, Bulacan with 232,110 or 4.1%, and Rizal with 224,327 or 4.0%. (Table 10.1) Most impressive gains in the size of the middle class were achieved by the provinces of Aurora (from 18.0 % in 2003 to 24.4 % in 2006 to 33.5 % in 2009) and Marinduque (from 8.9 % in 2003 to 13.3 % in 2006 to 16.3 % in 2009 ).(Table 10) c. Location by Urban-Rural Area More than 35% of families in the urban area were classified as middle income class for years 2003, 2006 and On the other hand, middle income class families in the rural area range from 10-12% in the same period. It may be noted that for both areas, the share of middle income class increased from 2006 to 2009 while percentage of low income families declined from 2006 to (Table 11) The distribution of the middle income families across the major island groups highlights the disparity in development in the country. This development gap, which has persisted for years requires new and innovative approaches both in planning and implementation that must be addressed by the national and local leadership. d. Family Size of the Middle Class For years 2003, 2006, and 2009, the average family size has consistently been 5 among low income families and 4 among middle income class families. Average family size among high income class ranges from 2 to 3 family members. (Table 12) e. Housing About 9 out of 10 middle-income families have houses with roof and wall made of strong materials, i.e., either galvanized, iron, aluminum, tile, concrete, brick, stone, asbestos. (Table 20 and 21) There is an increasing number of middle income class families who live in a Single House type of housing unit: from 83.6% in 2003 to 85.6% in 2006 to 87.7% in On the other hand the number of middle income class families living in apartments/accessoria/condo/townhouses has decreased: from 10.4% in 2003 to 9.3% in 2006, to 8.6% in Is the condo bubble about to burst? (Table 22) About 7 to 8 out of 10 middle-income families own or have owner type possession of their house and lot., compared to 8 to 9 among the high income families. (Table 23) Not quite all middle-income families use water-sealed toilet facilities but the proportion has been increasing: 92.6% in 2003 to 95.5% in 2006 to 97.1% in (Table 24) 12

13 f. Household Head The most common occupation of the Household Head of middle-income families is that of officials of government, executives, managers or supervisors, (34 % in 2009) followed by farmers/fishermen/foresty (11.3 % in 2009), professionals. (10.2 % in 2009), and service workers (9.4 % in 2009). (Table 15) Heads of low-income families, consistently had the highest average percentage of employed at 88.7, 84.0 and 77.6 for the periods 2003, 2006 and 2009, respectively. On the other hand, heads of high-income families have average percentage of employed at 76.9%, 72.7% and 74.1% in the same period.(table 16) So where do the high income families with unemployed household heads get their sources of livelihood? From pork barrel? g. Presence of OFWs The proportion of families with an OFW belonging to the low income class has declined: 53.2% in 2003, 45.6% in 2006, and 44.9% in 2009 while the proportion belonging to the middle income class has increased: from 46.6% in 2003, to 54.2% in 2006, to 55% in That explains why many young women and men have joined the Filipino diaspora, seeking greener pasture abroad. (Table 13) h. Working Age Population In general, among the three income classes, the high income families (87.2% in 2009) have the highest average percentage of working age population who are employed, followed by the middle class (66.1% in 2009) and the low income class (62.2% in 2009). Indeed, jobless growth should not be allowed to happen! (Table 14) IV. Conclusions and Recommendations Based on what could possibly be internationally comparable statistics on might be a working definition of the lower middle class, the size of the Filipino lower middle class is not too small. For this purpose, we consider as lower middle class families those with per capita income/consumption of; $2 to $20. Under this definition, the middle-income class in China would be about 56% in 2007; in Indonesia, about 43% in 2009; in.india, about 38% in and in the Philippines about 54% in 2006 ([see[13]) It is the upper middle class where the Philippines probably has a lower share compared to other countries. And so it is in this area where our government might need to do infuse some radical changes in the development agenda. Regardless of the current size of the middle class in the Philippines, the generation of statistics must continue, must improve and must be institutionalized in the PSS. Currently, precious too little is being done in the PSS to generate statistics on the middle class. The NSCB TS has devoted a fair share of its very limited manpower resources to provide information on whether the middle class is expanding or shrinking, but if there are surveys/registries on the informal sector, the basic sectors in agriculture, the unemployed 13

14 and underemployed, as well as other vulnerable sectors of the Philippine society, there should be as strong a reason for a data gathering system that tries to know and understand better the socio-economic and demographic characteristics, the aspirations, the government programs for and the support systems of the middle class. Toward this end the following recommendations are made/reiterated: 1. The job of the National Statistician will be extremely super challenging but he/she is given the opportunity to provide the statistical leadership that will steer the PSS towards an even greater national statistical system. The mandate is daunting, to say the least, but he/she should take on the challenge with patriotic fervor and utmost dedication to public service. 2. The PSA-to-be should give top priority to the generation of statistics on the poor and the middle class. In the past, the FIES data file was made available to the NSCB TS 12 months after the reference period considering the significant time needed for the processing of the FIES; poverty statistics were released a month later, i.e., 13 months after the reference period. For the 2012 FIES microdata, the NSO has shortened the time lag of releasing the FIES microdata, that is, nine months and eight months after the reference period for the first semester and full year 2012 FIES microdata files, respectively. While there has been a decrease in the time lag of release of FIES data files, this can be further shortened to respond to the need to release poverty statistics earlier. 3. The PSA should put up a Microdata Center where data users will be given access to statistical information without violating the Fundamental Principles of Official Statistics. Initially, the PSA should exert best efforts to generate anonymized microdata files (such as the Public Use Files of the FIES, the LFS, the agricultural surveys etc. ) in a much more timely manner. This way, the many excellent researchers from the academe and research institutions will have the opportunity to contribute their expertise in generating information, such as on the poor and the middle class, if for some reason, this could not be given top priority by the PSA. 4. Regular generation of provincial level information on the middle class should be considered. If the PSA could not give this top priority, partnerships with the LGUs and the private sector should be established. 5. The middle income methodology should continue to be enhanced. 6. Find ways to restore the high response rate that the FIES used to have.. The FIES response rates for years 2003, 2006 and 2009 were 95.7%, 86.4% and 90.8%, respectively. 7. Finally, this paper reiterates the call for strengthening the partnership as well as the political will of the government, the NGOs and the private sector to facilitate the process for the middle class to emerge as the driver of development that it could be, that it should be. 14

15 ACRONYMS CPI FE/TBE FIES GDP LFS MDG NCR NCS NSCB NSO PSS TS Consumer Price Index Ratio of Food Expenditures to Total Basic Expenditures Family Income and Expenditures Survey Gross Domestic Product Labor Force Survey Millennium Development Goals National Capital Region National Convention on Statistics National Statistical Coordination Board National Statistics Office Philippine Statistical System Technical Staff REFERENCES [1] National Statistical Coordination Board. The National Accounts of the Philippines, various issues. [2] economies [3] [4] [5] Ravallion, Martin. Policy Research Working Paper 4816: The Developing World s Bulging (but Vulnerable) Middle Class. [6] Birdsall, Nancy, Graham, Carol and Pettinato, Stefano. Brookings Institution Center Working Paper No. 14: Stuck in Tunnel: Is Globalization Muddling the Middle? [7] Virola, Romulo A., Addawe Mildred B. and Querubin, Ma. Ivy T. Trends and Characteristics of the Middle Class in the Philippines: Is it Expanding or Shrinking? [8] National Statistical Coordination Board. Statistically Speaking: Pinoy Middle Class Before the Crisis. 03 June 2009, [9] National Statistical Coordination Board. Statistically Speaking: How Rich is Rich? 15 June 2010, [10] Virola, Romulo A., Encarnacion Jessamyn O., Balamban Bernadette, Addawe Mildred, Viernes Mechelle and Pascasio Mark.The Pinoy middle-income class is shrinking: Its impact on income and expenditure patterns [11] Hair, Anderson, Tatham, and Black (1998) Multivariate Data Analysis. Prentice-Hall, Inc. U.S.A. [12] National Statistical Coordination Board. [13] Asian Development Bank. Key Indicators for Asia and the Pacific

16 ANNEX Table 1. 3-Cluster Analysis of Annual Per Capita Income using 2003 FIES data Cluster mean median min max Percent Families Income Class 1 24,200 18,236 1,257 87, Low 2 154, ,667 87,539 1,205, Middle 3 2,249,815 1,418,590 1,383,386 8,064, High Total 32,637 19,500 1,257 8,064,012 Table 2. 4-Cluster Analysis of Annual Per Capita Income using 2003 FIES data Cluster mean median min max Percent Families Income Class 1 19,469 16,365 1,257 50, Low 2 83,423 72,300 50, , Middle 3 302, , ,850 1,205, Middle 4 2,249,815 1,418,590 1,383,386 8,064, High Total 32,637 19,500 1,257 8,064,012 Table 3. 6-Cluster Analysis of Annual Per Capita Income using 2003 FIES data Percent Income Cluster mean median min max Families Class 1 16,555 14,767 1,257 36, Low 2 58,201 53,185 36, , Middle 3 157, , , , Middle 4 457, , , , Middle 5 1,798,485 1,418,590 1,205,592 3,129, High 6 6,585,019 8,064,012 4,096,752 8,064, High Total 32,637 19,500 1,257 8,064,012 16

17 Table 4. Annual Per Capita Income of the Five Clusters from the 5-Cluster Analysis of the 2003 FIES data Cluster 2003 PER CAPITA INCOME Mean Median Minimum Maximum Percent families Income Class 1 17,738 15,463 1,257 41, Low 2 67,240 60,315 41, , Middle 3 201, , , , Middle 4 827, , ,950 1,998, High 5 4,331,790 3,129,097 3,129,097 8,064, High Table 5. Annual Per Capita Income and Size of the Middle-Income Class: 2003, 2006, 2009, 2012 and 2013 Year Low Minimum Middle Maximum High 2003 <41,972 41, ,950 >513, <49,436 49, ,359 >605, <57,396 57, ,822 >702, <64,317 64, ,572 >787, <65,787 65, ,582 >805,582 Note: CPI (2006=100): , , , , (January- July) Source: Special computations made by the NSCB Technical Staff using the 2006-based Consumer Price Index of the National Statistics Office Table 6. Annual Family Income of the Low-, Middle-, High-Income Class by family size: 2013 Family Low Middle High Size (Up To) Minimum Maximum (At Least) 1 65,708 65, , , , ,416 1,604,126 1,604, , ,124 2,406,189 2,406, , ,832 3,208,252 3,208, , ,540 4,010,315 4,010, , ,248 4,812,378 4,812, , ,956 5,614,441 5,614, , ,664 6,416,504 6,416, , ,372 7,218,567 7,218, , ,080 8,020,630 8,020,630 Source: Special computations made by the NSCB Technical Staff using the 2006-based Consumer Price Index of the National Statistics Office 17

18 Table 7. Structure of the Distribution of Families by Per Capita Income Class, 2003, 2006 and 2009 Estimate Standard Error Difference low * middle * high Note: */ statistically significant at 5% level of significance Table 8. Distribution of Families by Income Class including the Distribution of Low-Income Class into Poor and Non-Poor : 2003, 2006, and 2009 Year Income Class Low Middle High Poor Non-Poor Total Level Percent Level Percent Level Percent Level Percent Level Percent ,293, ,234, ,527, ,929, , ,670, ,559, ,230, ,152, , ,855, ,914, ,770, ,659, , Table 9. Trends in the Structure of the Distribution of Families, by Income Class and by Region, 2003, 2006 and 2009 (Share to Row Total) Region Philippines NCR CAR I II III IV-A IV-B V VI VII VIII IX X XI XII ARMM Caraga

19 Table 9.1. Trends in the Structure of the Distribution of Families, by Income Class and by Region, 2003, 2006 and 2009 (Share to Column Total) Region Philippines NCR CAR I II III IV-A IV-B V VI VII VIII IX X XI XII ARMM Caraga

20 Table 10.1 Distribution of Families by Per Capita Income Class by Province, 2003, 2006 and 2009 Share To Row Total Region Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent NCR 1st District 145, , , , , , , , nd District 230, , , , , , , , , rd District 472, , , , , , th District 249, , , , , , , , , CAR Abra 35, , , , , , Apayao 18, , , , , , Benguet 71, , , , , , Ifugao 26, , , , , , Kalinga 28, , , , , , Mt. Province 26, , , , , , Region I Ilocos Norte 88, , , , , , Ilocos Sur 99, , , , , , La Union 107, , , , , , Pangasinan 413, , , , , , Region II Batanes 1, , , , , , Cagayan 170, , , , , , Isabela 222, , , , , , , Nueva Vizcaya 62, , , , , , Quirino 26, , , , , , Region III Aurora 30, , , , , , Bataan 83, , , , , , Bulacan 341, , , , , , Nueva Ecija 301, , , , , , Pampanga 260, , , , , , Tarlac 171, , , , , , Zambales 105, , , , , , Region IV-A Batangas 292, , , , , , Cavite 273, , , , , , Laguna 276, , , , , , , , Quezon 320, , , , , , Rizal 261, , , , , , , Region IV-B Marinduque 41, , , , , , Occidental Mindoro 72, , , , , , Oriental Mindoro 123, , , , , ,

21 Region Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Level Percent Palawan 149, , , , , , Romblon 51, , , , , , Region V Albay 186, , , , , , , Camarines Norte 84, , , , , , Camarines Sur 267, , , , , , Catanduanes 35, , , , , , Masbate 131, , , , , , Sorsogon 117, , , , , , Region VI Aklan 82, , , , , , Antique 84, , , , , , Capiz 115, , , , , , , Guimaras 26, , , , , , Iloilo 319, , , , , , Negros Occidental 452, , , , , , Region VII Bohol 203, , , , , , Cebu 564, , , , , , , Negros Oriental 214, , , , , , Siquijor 16, , , , , Region VIII Biliran 24, , , , , , Eastern Samar 66, , , , , , Leyte 297, , , , , , Northern Samar 90, , , , , , Southern Leyte 66, , , , , , Western Samar 116, , , , , , Region IXb Zamboanga del Norte 159, , , , , , Zamboanga del Surc 258, , , , , , Zamboanga Sibugay 91, , , , , , Isabela City 12, , , , , , Region X Bukidnon 191, , , , , , Camiguin 11, , , , , , Lanao del Norte 133, , , , , , Misamis Occidental 92, , , , , , Misamis Oriental 196, , , , , ,