Consumers' willingness to pay for agricultural products certified to ensure fair working conditions

Size: px
Start display at page:

Download "Consumers' willingness to pay for agricultural products certified to ensure fair working conditions"

Transcription

1 RESEARCH PROJECTS 2014 Consumers' willingness to pay for agricultural products certified to ensure fair working conditions Andreas C. Drichoutis, Lecturer, Dept. of Agricultural Economics & Rural Development, Agricultural University of Athens, Greece Achilleas Vassilopoulos, PhD candidate, Dept. of Agricultural Economics & Rural Development, Agricultural University of Athens, Greece Jayson L. Lusk, Regents Professor & Willard Sparks Endowed Chair, Dept. of Agricultural Economics, Oklahoma State University, USA December 2014

2 Consumers willingness to pay for agricultural products certified to ensure fair working conditions Final report submitted to the John S. Latsis Public Benefit Foundation Andreas C. Drichoutis 1, Achilleas Vassilopoulos 1, and Jayson L. Lusk 2 1 Agricultural University of Athens 2 Oklahoma State University December, 2014 Lecturer, Department of Agricultural Economics, Agricultural University of Athens, Iera Odos 75, 11855, Athens, Greece, PhD candidate, Department of Agricultural Economics, Agricultural University of Athens, Iera Odos 75, 11855, Athens, Greece, Regents Professor & Willard Sparks Endowed Chair, Department of Agricultural Economics, Oklahoma State University, Stillwater, OK 74078,

3 Abstract: This project contributes in the discussion about the trade-offs between fair working conditions and the competitiveness of local agricultural products. We use non-market valuation techniques designed to uncover the underlying preferences of Greek consumers towards a fair labor certification system that certifies fair working conditions for the workers employed at all production stages of agricultural products. The first part of the study uses the Choice Experiment methodology to capture the determinants of individual well-being and behavior by asking consumers to choose between alternative states of the world that vary attributes relevant to the proposed label such as inflation of food prices, income of farm laborers, percentage of food imports and unemployment rate. This allows us to estimate how consumers well being differs with different levels of farm laborers income given all other trade offs that consumers might face with the introduction of stricter policies regarding farm labor. The second part employs the Contingent valuation method as well as the Inferred valuation method to elicit consumers willingness to pay for a fair labor label in order to uncover consumers preferences for a fair labor certification system. We find that consumers are willing to pay an average premium of 72 cents/kg of strawberries with fair labor certification, which is equivalent to 49% of current market prices. Keywords: fair labor label; contingent valuation; inferred valuation; choice experiment; consequentiality script; cheap talk; uncertainty scale; equivalent loss; willingness to pay. Acknowledgments: Drichoutis, Vassilopoulos and Lusk acknowledge funding for this project by the John S. Latsis Public Benefit Foundation. The sole responsibility for its content lies with its authors. We also acknowledge the help of several people: Sotiria Alexandri, Nikos Bochoridis, Christina Chliaoutaki, Froso Gkika, Almarina Gkramozi, Eirini Kostoula, Maria Michalopoulou, Alexandra Papadopoulou, Eleftheria Papamichali, Sofia Tsarsitalidou and Eleni Vasilaki for data collection; Foteini Tragou and Nikos Petrou for data coding; Panagiotis Lazaridis for additional financial help; Michalis Chletsos and Apostolis Tsiouris for supporting data collection in the city of Ioannina. 1

4 PerÐlhyh: H paroôsa èreuna suneisfèrei sthn suz thsh gôrw apì thn allhloantistˆjmish metaxô twn dðkaiwn sunjhk n ergasðac kai thc antagwnistikìthtac topik n gewrgik n proðìntwn. QrhsimopoioÔme teqnikèc apotðmhshc mh emporeôsimwn agaj n gia na ekmaieôsoume tic protim seic Ell nwn katanalwt n proc èna sôsthma diapðsteushc dðkaiwn sunjhk n ergasðac pou pistopoieð tic sunj kec ergasðac ergat n pou ergˆzontai se ìla ta stˆdia paragwg c gewrgik n proðìntwn. To pr to mèroc thc èreunac qrhsimopoieð thn mejodologða twn Peiramˆtwn Epilog c gia thn tautopoðhsh twn prosdioristik n paragìntwn thc euhmerðac kai sumperiforˆc twn atìmwn. Me thn mèjodo aut zhteðtai apì ta ˆtoma na epilèxoun metaxô diaforetik n katastˆsewn oikonomik n sunjhk n, stic opoðec metabˆllontai qarakthristikˆ sqetikˆ me thn proteinìmenh s mansh ìpwc o plhjwrismìc stic timèc twn trofðmwn, to eisìdhma twn agrergat n, to posostì twn eisagwg n trofðmwn kai to posostì thc anergðac. Autì mac epitrèpei na ektim soume pwc h euhmerða twn atìmwn diaforopoieðtai gia diaforetikˆ epðpeda tou eisod matoc twn agrergat n, dedomènou ìlwn twn ˆllwn allhloantistajmðsewn pou mporeð na antimetwpðsoun oi katanalwtèc me thn efarmog austhrìterwn politik n sqetikˆ me thn agrotik ergasða. To deôtero mèroc thc èreunac qrhsimopoieð thn mèjodo thc Endeqìmenhc apotðmhshc allˆ kai thn mèjodo thc 'Emmeshc apotðmhshc gia na ekmaieôsei thn projumða plhrwm c twn katanalwt n gia s mansh dðkaiwn sunjhk n ergasðac se gewrgikˆ proðìnta. Ta apotelèsmata deðqnoun ìti oi katanalwtèc eðnai diatejeimènoi na plhr soun èna mèso epiplèon posì thc tˆxewc twn 72 lept n/kilì fraoul n, to opoðo isodunameð me to 49% thc lianik c tim c. 2

5 Contents 1 Project overview Data collection methods The choice experiment The model Experimental Design Results The valuation experiment Experimental design and questionnaire development Randomization to treatment Descriptive data analysis Econometric analysis Conclusions and discussion 37 Appendix A Sample Questionnaire 46 Appendix B Sample Refusal Form 54 Appendix C Additional tables 56 3

6 List of Tables 1.1 Refusals and agreements to participate in the survey-experiment by location site Comparison of refusals and co-operators by gender, age group and geographical location site (percentages) Comparison of gender and age groups between survey respondents, their household members and the 2001 census (percentages) Variable names and description Attributes and attribute levels of CE Results of the Conditional Logit model from the Pilot Study Results of the Random Parameter Logit model Choice Probabilities of Price-Wage tradeoff scenarios Experimental design Descriptives statistics of subjects observable characteristics Interval regression estimates C.1 Comparison of gender and age groups between survey respondents, their household members and the 2001 census per survey location site (percentages) 57 C.2 Experimetal design of pilot study C.3 Experimetal design of final Choice Experiment C.4 Interval regression estimates (with demographics)

7 List of Figures 2.1 CDF of Marginal Utilities in Choice Scenario CDF of Marginal Utilities in Choice Scenario CDF of Marginal Utilities in Choice Scenario CDF of Marginal Utilities in Choice Scenario CDF Shifts with Daily Wages Percentage of Yes/No responses in the CVM per treatment Percentage of Yes/No responses in the IVM per treatment Percentage of Yes/No responses in the CVM per value measure Percentage of Yes/No responses in the IVM per value measure Aggregate demand curves for the CV and IV elicitation methods

8 Chapter 1 Project overview Recently there has been an upsurge of incidents of labor exploitation in the Greek agricultural farm sector. The most recent incident in April 2013, involved 33 Bangladeshi workers being shot and injured by their supervisors at a strawberry farm because they protested for being unpaid for the last seven months. This incident brought in the spotlight of mass media attention the long-term issue of labor exploitation as practiced by business in the agricultural farming sector. Subsequently, cases of mistreatment and unfair working conditions in the farm business kept coming to the spotlight one after another, causing the anger of consumers and distribution channels both within as well as outside the country. This anger was manifested by a strong (albeit temporary) decline in the demand for strawberries. On the other hand, farmers claim that in order to keep market prices at levels consumers are willing to pay and be competitive relative to imported agricultural products, they can t afford the cost of providing fringe benefits to the employees (e.g., minimum wage, fair working conditions, decent accommodation). Thus, as they claim, their only option is to settle with illegal employment practices. This project seeks to contribute in the controversy about the trade-offs between fair working conditions and the competitiveness of local agricultural products. We use nonmarket valuation techniques designed to uncover the underlying preferences of Greek consumers towards a food labeling system that certifies fair working conditions for the workers employed at all production stages of agricultural products. Our aim is to understand whether consumers alleged disapproval of unfair working practices is reflected in their willingness-to-pay (WTP) of a premium above regular prices of conventional agricultural products. Most relevant to our work is a labeling system that is already in place in the global agro-food system known as Fair Trade labels (FT). FT labels have entered the mainstream marketplace to address consumers concerns about intrinsic and/or extrinsic characteristics of agricultural products. FT labels (which can be found on numerous products, such as tea, honey, bananas, flowers etc.) allow firms to signal the presence of specific desirable attributes and create the potential for premiums (McCluskey and 6

9 Loureiro, 2003). 1 Claims related to other ethical criteria such as social justice towards employees have not been adequately picked up by the literature and the food industry so far. However, recent works by Howard and Allen (2006, 2010) and Hustvedt and Bernard (2010) provide accumulated evidence in favor of labor-related information on a variety of products. Using both hypothetical and non-hypothetical methods the above studies found that consumers are WTP a premium for products, the production of which mandates subsistence wages to all employees in the production chain. Our aim in this project is to take this line of research a step forward by introducing labels ensuring subsistence wages but also reasonable working hours, access to descent housing and personal hygiene facilities as well as health care services. Given that employment in Greek farms is mostly seasonal and based on labor exploitation of non-legal immigrants, these additions seem necessary to capture the notion of humane and social just in the Greek agro-food labor market. To better understand whether fair labor in the agro-food sector is indeed an important concept in consumers perception and also be able to quantify the magnitude of its importance relative to other potential changes that may accompany a fair labor policy, the first part of the study uses the Choice Experiment (CE) methodology. Our design, similar to Johansson-Stenman et al. s (2002) and Alpizar et al. s (2005), allows us to capture the determinants of individual well-being and behavior by asking consumers to choose between alternative states of the world. The states of the world presented to consumers vary attributes relevant to the proposed label such as inflation of food prices, income of farm laborers, percentage of food imports and unemployment rate. This allows us to estimate how consumers well being differs with different levels of farm laborers income given all other trade offs that consumers might face with the introduction of stricter policies regarding farm labor. The second part of the study employs the Contingent Valuation method (CVM) to elicit consumers WTP for a fair labor label in order to uncover consumers preferences for a fair labor certification system. The CVM involves creating a hypothetical valuation scenario in which consumers are asked to state their WTP for the product under valuation. Empirical findings from several disciplines suggest that results from the CV method can be susceptible to social desirability bias. If this is the case, the credibility of the results obtained from the CVM is questionable. Lusk and Norwood (2009b) used a modification of Levitt and List s (2007) additive utility model to illustrate how social desirability may carry over to non-market valuation and produce inflated bids that misrepresent respondents underlying preferences. Their new method, called Inferred valuation, addresses the so-called social desirability bias by asking respondents to state their beliefs about the average consumer s valuation for a good. To answer the aims of our research agenda we conducted a wide scale questionnaire 1 For a recent critical overview of the economic theory behind Fair Trade, see Dragusanu et al. (2014) 7

10 based survey-experiment in two cities of Greece, Athens and Ioannina. We collected responses from more than 3,800 subjects which allows us to make robust inferences. This report proceeds as follows. Next section describes the data collection methods. Subsequently we report the experimental design and results for the choice experiment. We then report the experimental design and results for the valuation experiment and conclude in the last section. 1.1 Data collection methods A pilot questionnaire was pre-tested in February-March 2014 in the city of Athens with 160 subjects and several adjustments were made. To economize on resources both research aims (i.e., the valuation experiment and the choice experiment) where combined in a single survey. The first 100 subjects responded to both the CE and the CV survey and the rest only to the CV survey. The full scale survey was lunched in April, 1st 2014 and questionnaires were filled in until June 11, The period of data collection was mainly dictated by the fact that strawberries, which were the chosen product for our valuation experiment (discussed momentarily), are mainly traded during April, May and early June. In all, eleven interviewers worked for this project (six in Athens). All interviewers were briefed in and trained by one of the authors. Interviewers were current or former students of University of Ioannina and Agricultural University of Athens. Both the pilot and the main questionnaires were first constructed in English and were translated and back-translated by two professional applied linguists; several discrepancies were corrected accordingly. Consumers were randomly intercepted in front of the main entrance of various supermarkets. In all, 11,510 subjects were intercepted and 3,825 agreed to take part in the survey resulting in a cooperation rate of 33.23%. Of course, several subjects walked out during an interview or opted not to respond to certain questions, which further reduces the available number of subjects for statistical analysis. For the valuation experiment when we consider subjects that at minimum answered all relevant valuation questions, we have 3,772 subjects. For the Choice Experiment we have valid responses from 3,580 subjects. Table 1.1 shows number of refusals and agreements to participate in the survey-experiment by location site. To mitigate concerns of self-selection, we systematically recorded gender and age group of persons that refused to participate in the survey-experiment (Singh, 2007, p. 84). Interviewers were instructed to record the age group of the persons that refused to participate judging from the person s overall appearance. Table 1.2 compares gender and age group of co-operators and subjects that refused to participate (non-responders). First note that with respect to geographical location there is no difference between location sites when it comes to gender. On the other hand, there appears to be proportionally more subjects in 8

11 Table 1.1: Refusals and agreements to participate in the survey-experiment by location site Refusals Agreed to Total participate Athens Ioannina Total (Athens & Ioannina) the younger age group (18-25 years old) in the city of Ioannina. This is to be expected given that the university plays an important role in the life of the city of Ioannina and university students represent a big part of the city s non-permanent population. Overall, there are more female non-responders than female co-operators ( = 7.23%), and vice versa for males. With respect to age, while the middle age groups (between 26 and 60 years old) are comparable between non-responders and co-operators, the extreme age groups (18-25 and 61 years old) differ. For example, there are more co-operators in the young age group of years old ( =11.67%) and more non-responders in the oldest age group ( =8.35%). 2 Table 1.2: Comparison of refusals and co-operators by gender, age group and geographical location site (percentages) Gender Age group Female Male Ioannina Refusals Athens Total (Athens & Ioannina) Ioannina Co-operators Athens Total (Athens & Ioannina) Regarding the demographic profile of our sample, Table 1.3 shows that the vast majority of respondents were females (66.36%). This is not as problematic as it may seem on first glance, given that primary shoppers are mainly females. For example, one study estimates that 75% of principal household shoppers in the US are females (Mediamark Research and Intelligence, 2009). Therefore, the gender composition of our sample is not representative of the population of the two cities but it might better represent the grocery shopping population. Since we also asked respondents to report on the age and gender composition of their household, we can also compare the demographic profile of respondents households with that of the 2001 census (which is the latest available census for which basic demo- 2 Having less older people in our sample of co-operators could be due to the nature of the survey which required respondents standing for the duration of the survey. 9

12 graphic information are available). The comparison shows that discrepancies with the 2001 census are rather small. Table C.1 in Appendix C compares the demographic profile of respondents and respondents households with the 2001 census per survey location site. Table 1.3: Comparison of gender and age groups between survey respondents, their household members and the 2001 census (percentages) Males Respondents Households Census

13 Chapter 2 The choice experiment A labeling scheme about fair working conditions is expected to affect the labor market and unemployment rates since, if successful, it would attract part of the workforce from other sectors of the economy or from the pool of unemployed workers whose reservation incomes are higher than current farm wages. On the other hand, given that the production of several commodities is labor intensive (e.g., strawberries) and that workers play a major role in several stages of the production chain, it may drive some farms failing to cover the increased costs to cease production, and thus lower the demand for labor. Such labeling policies are also very likely to create distortions to the domestic supply and demand as well as to the international trade of some commodities and eventually, their prices. Depending on the potential market shares and the profit margins of the labeled products, domestic producers currently exporting their yield may focus on the domestic market while others may turn to exports in order to avoid competing under the new rules of differentiation and as such, incentivize imports of unlabeled products over domestic production. Given that we do not expect consumption patterns to change dramatically, food prices are expected to fall or rise depending on the substitutability and price elasticity of labeled, unlabeled, domestic and imported food products. Although labeling policies have been extensively examined in the literature, the above presented macroeconomic interdependencies are not easy to capture using conventional non-market valuation techniques involving trade-offs between unlabeled/labeled products and prices. However, consumers alleged preferences over a fair labor certification system may fade out or weaken in face of changes brought about the policies necessary to ensure them. To circumvent this difficulty, we use an alternative choice experiment where consumers choose between states of the world which differ with respect to food prices (current level, ±5%, ±10%), unemployment rates (current level, ±2%, ±5%), food imports (current level, ±10%, ±20%) and daily wages of farm laborers (e20, e23, e26, e30, e35). Our design is similar to Johansson-Stenman et al. (2002) (albeit the scope of the two studies is different) who presented consumers with societies described by their income distribution or by the own and average income and asked them to decide in which society their grandchild 11

14 would be most content. In a similar fashion, Alpizar et al. (2005) studied choices between societies described by the own and average consumption of goods such as cars, days of vacation, insurance plans and housing. Because we use a Bayesian efficient design which uses the element of the final model, in a somehow unusual order, we present our utility and econometric model first and then the experimental design and the results. 2.1 The model In a random utility framework, the CE methodology assumes utility functions with a linear-in-attributes deterministic component (V ) and a random idiosyncratic component (ɛ) reflecting the unobserved influences. As a result, the utility from the j th alternative is given by: U j = V j + ɛ j, with V j = k β k X kj (2.1) with X kj, the value of the k th attribute for this alternative. In our application, assuming that the marginal utility of the four attributes is non-linear in their value and normalizing the utility of the status-quo to zero, the utility model specification is: U j =β 1 1 F ood P rices% + β 2 2 F ood P rices% + β 3 3 F ood P rices% + β 4 4 F ood P rices% + β 5 1 Unemployment% + β 6 2 Unemployment% + β 7 3 Unemployment% + β 8 4 Unemployment% + β 9 1 F ood Imports% + β 10 2 F ood Imports% + β 11 3 F ood Imports% + β 12 4 F ood Imports% + β 13 F arm W age 1 + β 14 F arm W age 2 + β 15 F arm W age 3 + β 16 F arm W age 4 + ɛ j (2.2) The variables along with their description are given in Table 2.1 below. McFadden (1974) shows how the unknown parameters of the above utility model can be consistently estimated from stated choice outcomes between different alternatives using the Conditional Logit model. When ɛ j s are Independently and Identically Distributed (IID), random components can be integrated out and the choice probabilities have a closed form solution. To relax the IID assumption, several models have been proposed with the most popular being the Random Parameters Logit (RPL). The RPL allows heterogeneity of parameters across individuals, correlation of random parameters and non-independence between choice observations (e.g., panel data) and as such it is the most used generalization of the CL. Non-independence between choice observations is a very important characteristic of the RPL model, since in most studies using the CE methodology (including the present one) subjects face more than one choice situations. In the RPL model, the β j s 12

15 Table 2.1: Variable names and description Variables Description 1 F ood P rices% Food Prices are 10% lower than current level 2 F ood P rices% Food Prices are 5% lower than current level 3 F ood P rices% Food Prices are 5% higher than current level 4 F ood P rices% Food Prices are 10% higher than current level 5 F ood P rices% Food Prices are at the current level 1 Unemployment% Unemployment Rate is 5% lower than current level 2 Unemployment% Unemployment Rate is 2% lower than current level 3 Unemployment% Unemployment Rate is 2% higher than current level 4 Unemployment% Unemployment Rate is 5% higher than current level 5 Unemployment% Unemployment Rate is at the current level 1 F ood Imports% Food Imports are 20% lower than current level 2 F ood Imports% Food Imports are 10% lower than current level 3 F ood Imports% Food Imports are 10% higher than current level 4 F ood Imports% Food Imports are 20% higher than current level 5 F ood Imports% Food Imports are at the current level F arm W age 1 Daily Income of Farm Laborers is 20 e F arm W age 2 Daily Income of Farm Laborers is 23 e F arm W age 3 Daily Income of Farm Laborers is 30 e F arm W age 4 Daily Income of Farm Laborers is 35 e F arm W age 5 Daily Income of Farm Laborers is 26 e(current level) Excluded from estimation to avoid multicollinearity in equation (2.2) are assumed to be randomly distributed across respondents according to some known distribution F, namely β j F (µ betaj, σ betaj ). Because the choice probabilities are conditional on the specific assumptions made about the joint distribution of the parameters that is a priori unknown, the parameters of F are estimated using simulation (e.g., Train, 2003). 2.2 Experimental Design Given that the choice sets had 2 alternative states with each state consisting of 4 attributes with 5 levels (see table 2.2), we end up with a full factorial of 5 8 choices. Clearly, Table 2.2: Attributes and attribute levels of CE Attributes Attribute levels % Change in Food Prices -10, -5, 0, 5, 10 % Change in Unemployment Rate -5, -2, 0, 2, 5 % Change in Food Imports -20, -10, 0, 10, 20 Income of Farm Laborers (e) 20, 23, 26, 30, 35 facing so many choice situations would have been a huge cognitive burden for respondents, so we had to reduce the size of the design. The option of randomly selecting a subset of the 13

16 full factorial for each respondent was discarded because it may lead to biased estimates due to attribute level imbalance. For this reason, orthogonal designs with or without blocking have been used in experimental design for a long time. Orthogonal designs satisfy attribute level balance and are able to estimate each parameter independently which, however, comes at a cost of design matrices which are larger than necessary and estimates which are not efficient for non-linear models. Street et al. (2005) proposed an alternative way to reduce the design matrix and at the same time retain orthogonality using a D-optimal design that maximizes attribute level differences and the determinant of the information matrix. However, such designs are problematic in the presence of one or more salient attributes while in general they are not efficient. 3 Given the above, we have decided to employ a Bayesian D-efficient design suggested by Rose and Bliemer (2009) aiming to minimize the elements of the Asymptotic Variance-Covariance (AVC) matrix. This option of course, does not come without caveats. Since, apart from the design, the AVC matrix depends on the parameter estimates and the specific econometric model to be used, the generation of efficient designs requires some knowledge of these elements. Such prior information can only be obtained by pilot studies and this is the first step we took in constructing our final design. Although, as explained above, our aim is to estimate a panel RPL model, the pilot (final) design was (Bayesian) D-efficient based on the CL model. According to Bliemer and Rose (2010) there are a few reasons one would want to base their design on a CL model even though a panel RPL is to be estimated. For one, designs for the CL model (especially using Bayesian priors) perform very well when the final model is a panel RPL. Another reason is that designs based on the CL model, are much easier to generate than the ones optimized for the RPL model which are very difficult or even infeasible and can take a considerable amount of time due to many required repetitions. In addition, the number of degrees of freedom necessary for the estimation of the full RPL model is 32 which, given that choices were binary, would correspond to 32 or 35 choice situations (rows in the design matrix) in the pilot study 4. With the limited number of subjects participating in the pilot study (100 respondents), it would be infeasible to obtain reliable parameters estimates. Optimizing to the CL model instead, reduced the required size in half (i.e., 20 rows) and allowed the estimation of more reliable priors to be used for the final design. In all, the pilot CE design (see Table C.2 in the appendix) was a CL D-efficient (see Table 2.2) with D-error of 0.64 and A-error of 0.92 and with all priors set to zero. We have also used 4 blocks so that each respondent faced only 5 out of the 20 choice situations. 5 From 100 subjects that were asked to answer the CE pilot questionnaire, we have obtained usable data on 351 choices. The results of the CL model from the pilot are given 3 An exemption is when all the parameters of the model are zero. 4 This is to achieve attribute level balance since the attributes had 5 levels. 5 All designs were constructed in Ngene ver

17 in Table 2.3. Table 2.3: Results of the Conditional Logit model from the Pilot Study Variables Coeff. Std.Error Z P-value 95% Conf. Interval 1 F ood P rices% F ood P rices% F ood P rices% F ood P rices% Unemployment% 1.08*** Unemployment% 0.80*** Unemployment% Unemployment% -0.59* F ood Imports% F ood Imports% 0.76** F ood Imports% -0.59* F ood Imports% -0.88*** F arm W age ** F arm W age * F arm W age * F arm W age ** ***, **, * Significance at 1%, 5%, 10% level. For the final design, all parameters are assumed to be normally distributed with means equal to their above estimates and standard deviations equal to their associated standard errors. The final design (Table C.3) is a Bayesian D-efficient design, optimized for CL with 40 rows and 8 blocks 6. The mean D-error of the final design is 0.45 (SD=0.034, [Min,Max]=[0.39,0.56]) and the A-error is 0.75 (SD=0.06, [Min,Max]=[0.64,0.97]). 2.3 Results Estimates from the RPL model are shown in Table 2.4. To study consumer s preferences towards fair labor labels, we need to investigate the trade-offs between their introduction to a market and the change of other attributes that are likely to be affected by the labeling policy. Usually, such trade-offs are examined using the marginal WTP which is derived, assuming fixed and linear cost parameter, by the division of the marginal utility of the label by the (negative) partworth of price. When the cost parameter is random and its density is positive around zero, other methods have been developed for the estimation of WTP values (see Train and Weeks, 2005; Scarpa et al., 2008). However, when partworths are non-linear and random, one should study the joint distribution of the parameters that 6 Although the number of rows needed to satisfy the degrees of freedom and attribute level balance is 35, we have added 5 additional rows for better organization of the questionnaires in combination with the other treatments of the survey (see Section 3) 15

18 correspond to the attributes affected by the introduction of the labels, in order to conclude upon the underlying preferences of consumers for such a policy. Table 2.4: Results of the Random Parameter Logit model Variables Coeff. Std.Error Z P-value 95% Conf. Interval Mean 1 F ood P rices% 0.57*** F ood P rices% 0.29*** F ood P rices% -0.19*** F ood P rices% -0.50*** Unemployment% 0.50*** Unemployment% 0.35*** Unemployment% -0.29*** Unemployment% -0.82*** F ood Imports% 1.23*** F ood Imports% 0.65*** F ood Imports% -0.85*** F ood Imports% -1.67*** F arm W age *** F arm W age *** F arm W age *** F arm W age *** Standard Deviation 1 F ood P rices% 0.53*** F ood P rices% 0.62*** F ood P rices% 0.42*** F ood P rices% 0.79*** Unemployment% 0.61*** Unemployment% Unemployment% Unemployment% 0.58*** F ood Imports% 1.25*** F ood Imports% 0.72*** F ood Imports% 0.30* F ood Imports% 1.27*** F arm W age *** F arm W age *** F arm W age F arm W age *** ***, **, * Significance at 1%, 5%, 10% level. As explained in Section 2.1, all parameters j J are assumed to be normally distributed across consumers with means (µ betaj ) and standard deviations (σ betaj ) as given in Table 2.4. Due to independence, the sum of the marginal utilities associated with the combination of any attributes is also normally distributed with mean µ betaj and standard deviation of j 16

19 j σ 2 beta j. Using this fact, in Table 2.5 we estimate the fraction of consumers that would be in favor of a state of the world where the daily income of farm laborers is 4e (3e) and 9e (6e) higher (lower) or a state of the world that food prices are 5% or 10% higher (lower) 7. All estimated shares in the table are derived as 1 F (0) where F is the CDF of the sum of the corresponding partworths. Table 2.5: Choice Probabilities of Price-Wage tradeoff scenarios Choice Daily Income Choice Food Prices Scenario of Farm Laborers Probability 1 +5% (-5%) 30e (23e) 56% (42%) 2 +5% (-5%) 35e (20e) 68% (33%) 3 +10% (-10%) 30e (23e) 37% (56%) 4 +10% (-10%) 35e (20e) 56% (40%) As shown in Table 2.5, 56% of the respondents would prefer a mild increase (5%) in food prices in return to an increase in daily wages of crop laborers of 4e (i.e., from a base level of 26e to 30e). However, only 42% would rather experience an analogous decrease in food prices if it was associated with an even lower decrease (3e) in daily wages (i.e., from a base level of 26e to 23e). This difference is depicted in Figure 2.1, which shows how the CDF of the marginal utility below zero associated with the higher daily wage stochastically dominates the choice linked to the lower wage. 8 Figures 2.2 and 2.4 reveal that a similar Figure 2.1: CDF of Marginal Utilities in Choice Scenario 1 7 Note that this representation of preferences is irrespective of attributes that are held constant so the results are indicative for any food imports-unemployment combination. 8 According to our model specification, zero represents the utility associated with the status-quo. 17

20 pattern is also observed for higher increases (decreases) in prices and daily wages. However, Figure 2.3 shows that when a 10% increase (decrease) in food prices is combined with a 4e (3e) increase (decrease) in the daily income of crop laborers, respondents are more responsive to the price change. Only 37% prefer the combination of 10% increase in food prices and 4e increase in the daily income of crop laborers while 56% prefer a 10% decrease in food prices and 3e increase in the daily income of crop laborers. Figure 2.2: CDF of Marginal Utilities in Choice Scenario 2 Figure 2.3: CDF of Marginal Utilities in Choice Scenario 3 18

21 Figure 2.4: CDF of Marginal Utilities in Choice Scenario 4 In addition, Figure 2.5 shows that holding food price changes constant, there is a significant increase (decrease) in choice probabilities of 9% to 19% for incremental changes in farm laborers income. For example, the percentage of respondents who prefer an increase in food prices of 5% is 12% higher when the associated daily wage change is +9e than when it is +4e (68% vs 56%, see Table 2.5). Figure 2.5: CDF Shifts with Daily Wages All in all, we find that Greek consumers have a high interest towards fair working conditions in agriculture which is reflected by the fact that choice probabilities are quite 19

22 responsive to changes in the level of daily wages. 20

23 Chapter 3 The valuation experiment To uncover consumers preferences for a fair labor certification system, this project uses popular stated preferences methods. This part of the study employs the Contingent Valuation method to elicit consumers WTP for a fair labor label. The CVM has become the most popular method to measure WTP values which are used for valuing the benefits of new goods, services, or amenities. Although the CVM was principally developed in environmental and transport economics, it has made considerable headway in the valuation of food products over the last decades (e.g., Buzby et al., 1998; Corsi, 2007). Most, if not all, CV studies are conducted in hypothetical contexts, particularly in environmental valuation studies where a real market with salient payments is difficult to establish and some number is considered better than no number at all (Carson, 2012; Haab et al., 2013; Kling et al., 2012). The CVM involves creating a hypothetical valuation scenario in which consumers are asked to state their WTP for the product under valuation. Empirical findings from several disciplines suggest that results from the CV method can be susceptible to social desirability bias. There are now several studies that show that when people are asked to predict other people s value, as in the IV method (discussed momentarily), they state a different value than their own (e.g., Frederick, 2012; Kurt and Inman, 2013; Loewenstein and Adler, 1995; van Boven et al., 2000, 2003). Lusk and Norwood (2009b) used a modification of Levitt and List s (2007) additive utility model to illustrate how social desirability may carry over to non-market valuation and produce inflated bids that misrepresent respondents underlying preferences. Their so called Inferred Valuation method, addresses social desirability bias by asking respondents to state their beliefs about the average consumer s valuation for a good. Lusk and Norwood (2009a,b) argued that the IV method generates valuations that are less likely to suffer from normative or moral response biases (such as social desirability bias), and they found that responses to the IV method better predicted actual shopping behavior than did those from a CV method. They also found that the IV method produced less hypothetical bias when social desirability was present. The authors showed that goods 21

24 with normative dimensions are more prone to social desirability bias and thus the IV method is more effective in bridging the gap between the laboratory and field valuations. In the spirit of Lusk and Norwood (2009a,b), Pronin (2007) argues that people tend to recognize biases in human judgment except when that bias is their own which implies that predictions over other peoples preferences should mitigate biases. In the CVM literature the favored elicitation format has been the dichotomous choice (DC) format, ever since the National Oceanographic and Atmospheric Administration (Arrow et al., 1993) criticized the open-ended format as providing erratic and biased responses. Although other alternatives that were not considered by the NOAA panel were later developed, Carson and Groves (2007) offer a typology of elicitation formats that shows that all formats can be seen as generalizations of the DC format. The DC format has been favored due to its well known property of incentive compatibility. This is due to the Gibbard-Satterthwaite theorem (Gibbard, 1973; Satterthwaite, 1975) which states that for the case of more than two alternatives (i.e., non-dc formats) no nondictatorial strategy-proof voting procedure exists. 9 This is just to say that any response format with at least three possible outcomes is subject to individual manipulation (i.e., it s not incentive compatible). This does not imply that any binary DC format is incentive compatible but that, by elimination, only a DC format could be incentive compatible, assuming subjects believe their response is consequential (meaning there is some probability the respondent s answer will actual influence the provision of the good). 10 As we discuss in the next paragraph, the consequentiality of the survey is a very important issue that only lately has been given prompt attention. Much of the early literature has evolved around the presumption that hypothetical bias prevails in CV estimates. As such, a number of approaches have aimed in calibrating WTP estimates in a way that the CVM will more closely approach criterion validity. The cheap talk method in which participants are reminded of the tendency among people to inflate their (hypothetical) valuations has been labeled as moderately effective (Kling et al., 2012). The issue of consequentiality has only been given attention in the last few years. Carson and Groves (2007) argue that for producing useful information about an agent s preferences, survey s results must be seen as potentially consequential. Under consequentiality, survey respondents are explicitly told that their responses to preference questions will influence provision of the good under valuation. As a result the literature examining the merits of consequential surveys has been accumulating fast (Carson et al., 9 The theorem was formalized by Gibbard (1973) and Satterthwaite (1975) and noted in passing by Dummett and Farquharson (1961). See also Svensson and Reffgen (2014). 10 While proponents of the DC format take this result about the incentive compatibility of the DC format as granted for any type of good, Carson et al. (1997) show that the DC format is not incentive compatible in the case of provision of a new private or quasi-public good. The incentive compatibility of the DC format can be restored for quasi-public or private goods only if the binary choice is between two different forms of the good, so that the valuation question represents a change in the good (Carson et al., 2001; Carson and Groves, 2007). 22

25 2014; Herriges et al., 2010; List and Price, 2013; Mitani and Flores, 2013; Poe and Vossler, 2011; Vossler et al., 2012; Vossler and Evans, 2009; Vossler and Watson, 2013). An additional topic we tackle with this project is the issue of reference-dependent preferences. In the conventional neoclassical theory of consumer choice (often referred as Hicksian theory ) ordering of consumption bundles is independent of individual s endowment. In Tversky and Kahneman s (1992) reference-dependent theory, preferences are defined in relation to reference states. In this theory loss aversion plays a central role and offers an explanation of the frequently observed gap of two of the most widely used valuation measures in the literature, namely willingness-to-accept (WTA) and willingness-to-pay valuations. Most studies elicit WTP values (e.g., Bateman et al., 2000) which is defined as the maximum amount a consumer would be willing to pay in return for an increase (or upgrade) in his consumption for a good. Under Hicksian preferences, the Equivalent loss (EL), defined as the maximum amount a consumer would be willing to pay in place of a reduction (or downgrade) in her consumption of a good, should be equal to WTP (EL=WTP). In two competing hypothesis regarding reference-dependent preferences, Munro and Sugden (2003), show how the WTA/WTP gap is caused both by loss aversion in the good and by loss aversion in money. This hypothesis is also verified by the results of Bateman et al. (1997) using experimental markets and Bateman et al. (2000) using the CV method. On the other hand, Kahneman et al. (1990) provide evidence of Tversky and Kahneman s (1991) original no loss in buying (NLIB) hypothesis which states that buyers do not value the money they give up in a transaction as a loss; that is money spent is not coded as a loss. This is also confirmed in Novemsky and Kahneman (2005) who firmed up the NLIB hypothesis by proposing a theory of the conditions under which the gains and losses associated with a transaction are mentally integrated prior to evaluation, rather than being evaluated separately. They propose that when a loss and an equal and opposite gain are integrated in this way, painful perceptions of loss do not arise. Because consumers normally have budget reserves, that is, reserves of money that are available for unanticipated spending, when faced an unanticipated opportunity to buy a good and are able to finance this spending from their budget reserve, gains and losses are integrated. As a result, the money that has to be spent to buy the good is already seen as a token for unspecified goods and thus money outlays are not perceived as losses. In contrast, if the individual faces an unanticipated buying opportunity which she can finance only by forgoing some specific consumption plan, the act of buying involves a definite loss, separable from the gain; and so the money payment is perceived as a loss. A subsequent adversarial collaboration paper of Bateman et al. (2005) however, provides evidence that the cost of buying is indeed subject to loss aversion. Normalizing the utility of gains to zero, if preferences are reference dependent in the form of Munro and Sugden (2003), then we should observe EL WTP (for example, see 23

26 Bateman et al., 2000, for a proof), irrespective of whether downgrading in EL is perceived as a loss or not. 11 On the other hand, if the NLIB hypothesis is true, then there are three different competing scenarios about the relative magnitude of WTP and EL. In the first scenario, which holds when money spent to avoid the downgrade is covered by budget reserves (as in WTP) and thus it is not perceived as a loss, yields an EL that is greater or equal to WTP (EL WTP) notwithstanding the loss perceptions in EL treatments. According to the other two scenarios, the cost of avoiding the downgrade as well as the downgrade per se are perceived as losses in EL and as such we expect EL ( )WTP when losses in the money (good) domain are weighted more heavily than equi-util losses in the other dimension. In conclusion, while EL>WTP is indicative of reference-dependent preferences with or without NLIB, EL<WTP is supportive of the NLIB hypothesis. We test for these competing hypotheses in our survey by framing the valuation questions either as a WTP or as an EL. Next section describes the experimental design we adopt to test the hypotheses described above. To elicit valuations for the fair labor certification system, we chose a pack of 500gr of strawberries. This was our chosen product for a number of reasons. First, we needed a product which is widely consumed and would appeal to most consumers. Second, we wanted our product to be able to be sold in packages that could carry a labeling certification system; strawberries can and are often sold in a packaged form. Third, the production process of strawberries is a (manual) labor intensive process (labor contributes to 45-50% of total cost of strawberries (Poinssot, 2013)). To verify this we looked at the Agricultural Production Indexes issued from several prefectures of Greece for various years. These indexes include detailed information about manual and mechanical labor inputs for crops and fruits produced in the respective prefecture. For example, in the prefecture of Central Macedonia, strawberries require between hours/stremma of manual labor. 12 next most labor intensive fruit, pears, requires between hours/stremma of manual labor. 13 All subjects were first informed about what a fair labor certification system actually is. Since there was the chance that subjects were familiar with a fair trade label that could confound their perception of a fair labor label, we wanted to make sure that all subjects would be actually valuing a fair labor label by providing them the script below (the original version in Greek can be found in Appendix A): Normalizing the utility of gains to zero assumes that there is no extra kick from gaining a good, other than its consumption utility. This is the original and the most commonly used formulation of prospect theory for riskless choices. 12 The stremma is a Greek unit of land area, equal to 1,000 m 2 13 Although blackberries are labor intensive as well (i.e., between hours/stremma), they do not appeal to the average consumer as much as strawberries. In addition, some tomatoes and beans which are labor intensive as well, are mostly traded in bulk quantities (not packaged). 14 Information were compiled from three relevant to the fair labor label websites: an auditing non-profit organization that formally measures and accredits best employment practices (http://www. fairworkingconditions.ie), the fair labor association (http://www.fairlabor.org) and the Food Al- 24 The

27 The Fair Labor label [show picture 1] 15 can be certified by various organizations like the Fair Working Conditions.ie which is a non-profit international organization that aims in recognizing and improving employment conditions. Such a label ensures that the product is produced in a business farm that strictly follows the standards set by the International Labour Organisation (ILO). These standards concern the maximum number of working hours per week, the legal compensation and labor benefits required by state law for each activity sector as well as workers hygiene conditions in their working place. In addition, the standards prohibit child labor and bind the employer for no discrimination on the basis of race, nationality, or any other criteria. 3.1 Experimental design and questionnaire development To answer the methodological issues we raised above, we adopt a design with elements of a within- as well as a between-subjects design (Charness et al., 2012). For the betweensubjects design we adopt a 4 2 design where we vary the scripts accompanying the valuation questions (control (no scripts) vs. cheap talk vs. consequentiality vs. cheap talk & consequentiality) as well as the elicited value measure (equivalent loss vs. willingness to pay). We use the Discrete Choice format as the elicitation format of our survey where we vary the bid amount at five levels (20 cents vs. 40 cents vs. 70 cents vs. 100 cents vs. 120 cents). The bid amounts were selected and adjusted based on projected historical prices of strawberries (discussed momentarily) as well as feedback we received from the pilot survey. The bid amounts were varied on a between subject basis so that each subject was asked for his/her valuation at only one bid amount. Table 3.1 shows our experimental design and the per treatment number of subjects. Valuations were elicited using the CV method as well as the IV method on a within subjects basis. The order was counterbalanced. The cheap talk script (the original version in Greek can be found in Appendix A) was compiled from several sources as well as our own previous work (e.g., Bulte et al., 2005; Lusk, 2003) and reads as follows: In a minute you ll be asked whether you are willing to pay a certain amount for strawberries. This question will be hypothetical, that is, you will not actually have to pay. In general, people experience difficulties in answering hypothetical questions. They often state they are willing to pay an amount larger than the amount they are willing to pay in reality. liance certification programme (http://foodalliance.org). 15 The picture can be found in Appendix A. 25

28 IV then CV CV then IV Table 3.1: Experimental design Equivalent Loss Willingness to pay Bid amount Total Control Consequentiality Cheap talk Cheap talk & Consequentiality Control Consequentiality Cheap talk Cheap talk & Consequentiality Total Notes: CV stands for Contingent valuation; IV stands for Inferred Valuation. One reason why this happens is because when the time comes to actually make the payment, they also consider that these money won t be available for other purchases. Therefore, when the question is hypothetical, it is easier to exaggerate their response. Before answering the willingness to pay question, try to think whether you are really willing to pay this amount for strawberries and that this amount will no longer be available for other purchases. The consequentiality script was adopted from Vossler and Watson (2013) and Vossler and Evans (2009) and reads as follows (the original version in Greek can be found in Appendix A): We would like to inform you that the survey results will become available to producers, traders and retailers of agricultural products as well as to the wider general public of consumers. This means that this survey could affect the decision of producers, traders and retailers of adopting a Fair Labor certification system for strawberries as well as the average price of strawberries. After the script(s) was(were) read, the valuation questions followed. In the control (no script) treatment, the valuation treatments followed right after information about the fair labor label was given. The valuation measure of willingness to pay was framed as: Assume you are given a pack of half a kilo of conventional strawberries [show picture 2]. Would you be willing to pay XX cents so that you can exchange it with a similar pack of strawberries certified with a fair labor label [show picture 3]? The pictures can be found in Appendix A. 26

29 Conversely, the equivalent loss measure was framed as: Assume you are given a pack of half a kilo of strawberries certified with a fair labor label [show picture 3]. Would you be willing to pay XX cents so that you can avoid exchanging it with a pack of conventional strawberries without any certification? [show picture 2] Following the literature on certainty scales (Champ et al., 1997; Morrison and Brown, 2009), every CV discrete choice question was followed by a question asking respondents to state how certain they were about their response on a tenpoint scale anchored with the labels Not certain at all and Very certain. Similar to the CV questions, the IV questions were formatted to elicit the valuation measure of willingness to pay: Assume that an average consumer is given a pack of half a kilo of conventional strawberries [show picture 2]. Do you think s/he would be willing to pay XX cents so that s/he can exchange it with a similar pack of strawberries certified with a fair labor label [show picture 3]? or the equivalent loss valuation measure: Assume that an average consumer is given a pack of half a kilo of strawberries certified with a fair labor label [show picture 3]. Do you think s/he would be willing to pay XX cents so that s/he can avoid exchanging it with a pack of conventional strawberries without any certification? [show picture 2] A consequentiality question was also included to allow us to test for differences between respondents with different consequentiality perceptions of the survey. The question was adopted from Vossler and Watson (2013) and Vossler et al. (2012). Participants had to indicate of the indirect consequences of the survey on a five point likert scale anchored by not at all and very much. The question was framed as: To what extent do you believe that your answers in this survey will be taken into account by producers, traders and retailers? To select the bid amounts for the discrete choice valuation question, we first examined historical data for strawberries from the Athens Central Market Organization S.A. which is one of the major markets for vegetable and meat wholesalers in Greece. 17 The historical data show that the bulk of strawberry sales takes place between April and May (smaller quantities of strawberries are also traded in February, March and June) which also determined the data collection period of our survey-experiment. A simple linear projection of 17 The Athens Central Market Organization S.A. (http://www.okaa.gr/) is a public organization supervised by the Ministry of Rural Development and Food. It serves as an organized host of about 600 vegetable and meat wholesalers and operates in an area of about 26 hectares. 27

30 strawberry prices with data from April 2008 to May 2013, projected an average mode price for April-May 2014 at e1.458/kg. This is remarkably close to the realized average mode price for April-May 2014 at e1.465/kg. Given that our valuation product was selected to be a pack of 500gr of strawberries (corresponding projected price is e0.73/500 gr) and that the valuation question concerned an upgrade (or downgrade, depending on the elicited valuation measure), we decided that our price range would center around e0.70 taking also into account feedback from our pilot survey. Therefore, the full price range was selected as: 20 cents, 40 cents, 70 cents, 100 cents and 120 cents. The questionnaire also elicited respondents beliefs about the likelihood of hypothetical bias. Subjects were asked to indicate: i) given that people tend to exaggerate responses in hypothetical surveys when there is no real economic exchange, how likely it is that the same thing happened in this survey ii) how likely it is that other respondents will exaggerate their responses. Responses were recorded on a 5-point Likert scale anchored by extremely unlikely and extremely likely. In addition, given that a fair labor label carries a socially desirable attribute and subjects may be inclined to state a higher valuation, we also elicited the social desirability scale (SDS) of Stöber (2001). The final instrument is composed of 16 questions (e.g., I always admit my mistakes openly and face the potential negative consequences, I occasionally speak badly of others behind their back etc.). Subjects were asked to state their level of agreement with each of the statements. The statements are then individually scored and summed to form a single index of social desirability. The higher the score on the SDS scale, the more likely it is the person falls prey to social desirability bias. In addition, the questionnaire also included a political ideology question adopted from the European Social Survey. 18 Recently, research in the area of non-market valuation has identified political affiliation and means of provision of the good i.e., privately or collectively, as a determinant of willingness to pay for public goods (e.g., Dupont and Bateman, 2012). This mainly originates from a body of research that has found significant differences in the degree of concern for environmental goods expressed by individuals with different political affiliations. While our valuation exercise does not concern a public good, aspects of the fair labor label may appeal to political ideology e.g., the minimum wage provision. There was an additional section in the questionnaire that concerned a Choice Experiment which was discussed in the previous chapter. A set of demographic questions followed asking about price sensitivity with respect to grocery shopping, purchase frequency of strawberries, age, gender, education, household size and perceived income position of the household. 18 Questionnaires from the European Social Survey can be found at europeansocialsurvey.org/methodology/questionnaire/. 28

31 3.2 Randomization to treatment As in every experiment, it is critical to explore whether randomization to treatment was successful by testing if observable characteristic are balanced across the between subjects treatments. Table 3.2 shows basic descriptive statistics for a set of observable characteristics. With respect to demographic variables there is no significant difference between treatments for gender (Pearson s χ 2 = 5.132, p-value=0.644), education (Pearson s χ 2 = , p-value=0.979), income (Pearson s χ 2 = , p-value=0.076), age (Kruskal-Wallis χ 2 = 5.599, p-value=0.587) and household size (Kruskal-Wallis χ 2 = 3.341, p-value=0.852). We also find no difference between treatments with respect to the Social Desirability score (Kruskal-Wallis χ 2 = 6.128, p-value=0.525), political ideology (Kruskal-Wallis χ 2 = 8.036, p-value=0.329), purchase frequency of strawberries (Pearson s χ 2 = , p-value=0.671), price sensitivity in purchasing decisions (Pearson s χ 2 = , p- value=0.767), perceived likelihood of hypothetical bias for the respondent (Pearson s χ 2 = , p-value=0.276) and perceived likelihood of hypothetical bias for other respondents (Pearson s χ 2 = , p-value=0.702). Thus, we can safely conclude that randomization to treatment on a wide set of observable characteristics was successful. 3.3 Descriptive data analysis Before we proceed with the econometric analysis we can gain interesting insights by looking at the raw data. Figure 3.1 graphs the percentage of Yes/No responses elicited with the CVM in the discrete choice question per each of the four script treatments. Two observations are in place. First, it appears that in all treatments elicited values are responsive to prices (bid amounts); we observe a decline of yes responses when the bid amount increases. The 40 cents bid yields statistically significantly lower proportion of Yes responses as compared to the 20 cent bid (p-value<0.001); the 70 cent bid yields significantly lower responses than the 40 cent bid (p-value=0.001); the 100 cent bid is also lower and statistically different than the 70 cent bid (p-value=0.015) while the 120 cent bid does not differ with respect to the 100 cent bid (p-value=0.829). In addition, a significant percentage of subjects states a high value for the fair labor certification label even when the bid is up to 120 cents. This is interesting since the amount of 120 cents represents a 171% increase in the price of strawberries (with respect to a price of 70 cents/500 gr). Second, it looks as if the different scripts are not effective in changing response patterns with respect to the control (no script) treatment. This is confirmed by proportion tests where we find that the proportion of Yes responses in the Control treatment is not significantly different than the Consequentiality script treatment (p-value=0.225); or the Cheap talk script treatment (p-value=0.720); or the Cheap talk & Consequentiality script 29

32 Table 3.2: Descriptives statistics of subjects observable characteristics Variable name and description Variable levels N Mean S.D. Gender: Male dummy Age Hsize: Household size Shopper: Dummy for major grocery shopper SDS: Social Desirability Scale Political: Political ideology scale Up to primary school 3.61% Up to Junior Hi-school 4.75% Educ: Education level Up to Senior Hi-school 22.92% 3708 Some college or university student 21.90% University graduate 37.30% Post-graduate studies 9.52% Bad or Very bad 5.73% Income: Household s economic position HBiasOwn: Perceived likelihood of hypothetical bias for the respondent HBiasOther: Perceived likelihood of hypothetical bias for other respondents PurchFreq: Purchase frequency (when strawberries are available) PriceSens: Price sensitivity Below average 11.16% Average % Above average 19.75% Good or Very good 15.96% Not likely at all 40.70% Unlikely 31.03% Neither likely, nor unlikely % Likely 11.49% Very likely 1.69% Not likely at all 4.26% Unlikely 20.22% Neither likely, nor unlikely % Likely 33.67% Very likely 9.57% Never 14.32% Once a month 25.06% 2-3 times a month 21.12% 3707 Once a week 28.16% 2-3 times a week 9.55% More than 2-3 times a week 1.78% Not important at all 1.14% Not important 3.82% Neither important, nor unimportant % Important 42.14% Very important 37.70% Notes: S.D. stands for standard deviation. For variables with levels, the percentage of each level is denoted instead of the mean. For each variable the sample was restricted to observations for which subjects gave responses to both the CV and IV methods. 30

33 treatment (p-value=0.082). Figure 3.1: Percentage of Yes/No responses in the CVM per treatment Figure 3.2 graphs proportion of responses in the discrete choice question for the IVM. The findings are similar to CVM: we find no effect of the scripts on responses (pvalue=0.285, and when we compare the Control treatment with the Cheap talk script, Consequentiality script and Cheap talk & Consequentiality script treatments, respectively). In addition, while proportion of Yes responses declines for the first two bid amounts, it is flat after 70 cents. Proportion tests indicate that the 40 cents bid amount yields significantly lower proportion of Yes responses when compared with the 20 cents bid (p-value<0.001) and the 70 cents bid yields lower Yes proportion than the 40 cents bid (p-value<0.001). However, the 100 cents and 120 cents do not yield different responses when compared with the 70 cents bid (p-value=0.443 and 0.672, respectively). What is even more interesting is that the IVM seems to work as advertised. Figure 3.2 shows that for all bid amounts the IVM yields a significantly lower proportion of Yes responses when compared with CVM shown in Figure 3.1. Proportion tests show that differences are highly significant for all bid amounts (p-values<0.001 in all cases). The difference in the proportion of Yes is larger than 30% for all bid amounts. This is an indication that the IVM was effective in mitigating social desirability and hypothetical bias and thus, valuations from this method may be more valid for predicting market outcomes. Figures 3.3 and 3.4 compare the elicited value measures (Equivalent loss vs. Willingness to pay) for the CVM and IVM, respectively. It is obvious that since the proportion of Yes responses is higher for WTP than EL then it is implied that values elicited under WTP are larger than values elicited under EL. Proportions tests confirm the insight that the 31

34 Figure 3.2: Percentage of Yes/No responses in the IVM per treatment proportion of Yes for WTP is larger than EL under the CVM (p-value=0.017) and under the IVM (p-value<0.001). 3.4 Econometric analysis To check whether the results obtained above hold under conditional analysis, we estimated interval regression models with clustered standard errors at the individual level to account for the fact that each person provided responses under both CVM and IVM. Table 3.3 shows coefficient estimates from several specifications. Model (1) is our basic specification where only the treatment variables are included in the regression. We confirm that none of the scripts (Cheap talk, Consequentiality or their combination) has any effect on elicited valuations. On the other hand, the IV dummy is highly significant indicating that subjects under the IVM are stating lower valuations which is an indication that this method successfully mitigates social desirability and hypothetical bias. The WTP dummy is also highly significant indicating that elicited valuations are higher when the valuation question is framed as a WTP question rather than an EL question. Our result supports Tversky and Kahneman s (1991) no loss in buying hypothesis which states that buyers do not value the money they give up in a transaction as a loss. Finally, there are evidence of significant order effects indicating that when the IV question was asked first, then subjects tended to align their response in the CV question. This could be due to the fact that answering the IV valuation question first and the CV question second, made subjects think of their own biases. 32

35 Figure 3.3: Percentage of Yes/No responses in the CVM per value measure Figure 3.4: Percentage of Yes/No responses in the IVM per value measure 33

36 Table 3.3: Interval regression estimates Without certainty correction With certainty correction Full sample Consequential Incosequential Full sample Consequential Incosequential (1) (2) (3) (4) (5) (6) Cheap talk (7.976) (13.602) (9.624) (9.003) (15.722) (10.499) Consequentiality (8.008) (13.981) (9.664) (9.093) (15.959) (10.667) Cheap talk & Consequentiality (7.980) (13.779) (9.702) (9.103) (15.839) (10.716) IV (12.278) (25.748) (12.853) (12.316) (26.192) (12.622) WTP (6.002) (11.262) (6.865) (6.753) (12.285) (7.658) Order (6.177) (11.427) (7.206) (7.083) (13.855) (7.948) Constant (12.053) (27.148) (11.885) (10.835) (26.629) (10.338) ln(σ) (0.077) (0.130) (0.094) (0.088) (0.152) (0.105) N AIC BIC Clustered standard errors in parentheses * p<0.1, ** p<0.05, *** p<

37 As we have detailed in previous sections, recent literature has argued that consequentiality is needed for incentive compatibility. We therefore explore differences in results when consequentiality is violated. Based on the five point Likert consequentiality question, we divided subjects into consequential and inconsequential. We defined the consequential sample as subjects that stated that they believed their answers will be taken into account by producers, traders and retailers on a medium or higher likelihood. The inconsequential sample was defined as those that perceived that the survey would have little or no effect to producers, traders and retailers. Columns (2) and (3) in Table 3.3 show estimates from the consequential and inconsequential subsamples, respectively. The results are roughly similar with those from the full sample with one exception. A few script treatment variables are now statistically significant (albeit only at the 10% level). However, this not is not likely a robust result. Table C in Appendix C shows estimates of the consequential and inconsequential subsamples with the addition of demographic and attitudinal variables in the list of independent variables. As evident, the script treatment variables are no longer statistically significant. Thus, we conclude that there are no differential effects with respect to the script treatment variables between the subsamples. A final piece of robustness checks for our results comes from certainty corrections. Following previous literature (Morrison and Brown, 2009), we recode Yes responses in the CV question using a rating of 7 in the certainty question as the cut-off. Columns (4), (5) and (6) in Table 3.3 present coefficients for the full sample, the consequential and inconsequential subsamples, respectively. None of our conclusions changes. The only difference we observe with the addition of demographics (estimates are shown in Table C in the Appendix C) is that for the incosequential subsample, framing of the valuation question does not matter i.e., we observe that willingness to pay equals equivalent loss. We can then proceed in graphing the aggregate demand curves for strawberries with fair labor certification. Figure 3.5 does exactly that for valuations elicited with CV and IV. Given that IV mitigates social desirability bias it is advised that producers and retailers base their marketing decisions on the IVM curve. Each point of this curve indicates the percentage of the respondents that would buy half a kilo of strawberries with fair labor certification at the premium projected on the Y axis. The fact that the IVM curve intersects the X axis indicates that a percentage of consumers would not buy strawberries with fair labor certification even when they are offered at no premium at all. 35

38 Figure 3.5: Aggregate demand curves for the CV and IV elicitation methods 36

CHAPTER 11. AN OVEVIEW OF THE BANK OF ENGLAND QUARTERLY MODEL OF THE (BEQM)

CHAPTER 11. AN OVEVIEW OF THE BANK OF ENGLAND QUARTERLY MODEL OF THE (BEQM) 1 CHAPTER 11. AN OVEVIEW OF THE BANK OF ENGLAND QUARTERLY MODEL OF THE (BEQM) This model is the main tool in the suite of models employed by the staff and the Monetary Policy Committee (MPC) in the construction

More information

Moral Hazard. Itay Goldstein. Wharton School, University of Pennsylvania

Moral Hazard. Itay Goldstein. Wharton School, University of Pennsylvania Moral Hazard Itay Goldstein Wharton School, University of Pennsylvania 1 Principal-Agent Problem Basic problem in corporate finance: separation of ownership and control: o The owners of the firm are typically

More information

The Study on the Value of New & Renewable Energy as a Future Alternative Energy Source in Korea

The Study on the Value of New & Renewable Energy as a Future Alternative Energy Source in Korea , pp.26-31 http://dx.doi.org/10.14257/astl.2015.86.06 The Study on the Value of New & Renewable Energy as a Future Alternative Energy Source in Korea Woo-Jin Jung 1, Tae-Hwan Kim 2, and Sang-Ying Tom Lee

More information

Do Experience and Cheap Talk influence Willingness to Pay in an Open-Ended Contingent Valuation Survey?

Do Experience and Cheap Talk influence Willingness to Pay in an Open-Ended Contingent Valuation Survey? Do Experience and Cheap Talk influence Willingness to Pay in an Open-Ended Contingent Valuation Survey? Fredrik Carlsson A Peter Martinsson B Working Papers in Economics no. 190 January 2006 Department

More information

Marketing Mix Modelling and Big Data P. M Cain

Marketing Mix Modelling and Big Data P. M Cain 1) Introduction Marketing Mix Modelling and Big Data P. M Cain Big data is generally defined in terms of the volume and variety of structured and unstructured information. Whereas structured data is stored

More information

The Cheap-talk Protocol and the Estimation of the Benefits of Wind Power

The Cheap-talk Protocol and the Estimation of the Benefits of Wind Power The Cheap-talk Protocol and the Estimation of the Benefits of Wind Power Todd L. Cherry and John Whitehead Department of Economics Appalachian State University August 2004 1 I. Introduction The contingent

More information

LOCAL FOOD CONSUMERS: HOW MOTIVATIONS AND PERCEPTIONS TRANSLATE TO BUYING BEHAVIOR

LOCAL FOOD CONSUMERS: HOW MOTIVATIONS AND PERCEPTIONS TRANSLATE TO BUYING BEHAVIOR 1st Quarter 2010 25(1) LOCAL FOOD CONSUMERS: HOW MOTIVATIONS AND PERCEPTIONS TRANSLATE TO BUYING BEHAVIOR Yuko Onozaka, Gretchen Nurse, and Dawn Thilmany McFadden Emerging market demand for local foods

More information

Food Demand Survey (FooDS) Technical Information on Survey Questions and Methods. May 22, 2013. Jayson L. Lusk

Food Demand Survey (FooDS) Technical Information on Survey Questions and Methods. May 22, 2013. Jayson L. Lusk Food Demand Survey (FooDS) Technical Information on Survey Questions and Methods May 22, 2013 Jayson L. Lusk The purpose of FooDS is to track consumer preferences and sentiments on the safety, quality,

More information

Community Life Survey Summary of web experiment findings November 2013

Community Life Survey Summary of web experiment findings November 2013 Community Life Survey Summary of web experiment findings November 2013 BMRB Crown copyright 2013 Contents 1. Introduction 3 2. What did we want to find out from the experiment? 4 3. What response rate

More information

Uncovering Complexity-induced Status Quo Effects in Choice Experiments for Environmental Valuation

Uncovering Complexity-induced Status Quo Effects in Choice Experiments for Environmental Valuation Uncovering Complexity-induced Status Quo Effects in Choice Experiments for Environmental Valuation Malte Oehlmann*, Jürgen Meyerhoff, Priska Weller - Preliminary version - Abstract This study investigates

More information

Price Discrimination: Part 2. Sotiris Georganas

Price Discrimination: Part 2. Sotiris Georganas Price Discrimination: Part 2 Sotiris Georganas 1 More pricing techniques We will look at some further pricing techniques... 1. Non-linear pricing (2nd degree price discrimination) 2. Bundling 2 Non-linear

More information

This PDF is a selection from a published volume from the National Bureau of Economic Research

This PDF is a selection from a published volume from the National Bureau of Economic Research This PDF is a selection from a published volume from the National Bureau of Economic Research Volume Title: Fiscal Policy and Management in East Asia, NBER-EASE, Volume 16 Volume Author/Editor: Takatoshi

More information

Organizing Your Approach to a Data Analysis

Organizing Your Approach to a Data Analysis Biost/Stat 578 B: Data Analysis Emerson, September 29, 2003 Handout #1 Organizing Your Approach to a Data Analysis The general theme should be to maximize thinking about the data analysis and to minimize

More information

Ratemakingfor Maximum Profitability. Lee M. Bowron, ACAS, MAAA and Donnald E. Manis, FCAS, MAAA

Ratemakingfor Maximum Profitability. Lee M. Bowron, ACAS, MAAA and Donnald E. Manis, FCAS, MAAA Ratemakingfor Maximum Profitability Lee M. Bowron, ACAS, MAAA and Donnald E. Manis, FCAS, MAAA RATEMAKING FOR MAXIMUM PROFITABILITY Lee Bowron, ACAS, MAAA Don Manis, FCAS, MAAA Abstract The goal of ratemaking

More information

REVIEW OF STATED PREFERENCE AND WILLINGNESS TO PAY METHODS. Introductory note by the Competition Commission

REVIEW OF STATED PREFERENCE AND WILLINGNESS TO PAY METHODS. Introductory note by the Competition Commission REVIEW OF STATED PREFERENCE AND WILLINGNESS TO PAY METHODS Introductory note by the Competition Commission 1. The Competition Commission (CC) commissioned the market research organization, Accent, in association

More information

The Elasticity of Taxable Income: A Non-Technical Summary

The Elasticity of Taxable Income: A Non-Technical Summary The Elasticity of Taxable Income: A Non-Technical Summary John Creedy The University of Melbourne Abstract This paper provides a non-technical summary of the concept of the elasticity of taxable income,

More information

University of New South Wales Group. Group leader: John Piggott

University of New South Wales Group. Group leader: John Piggott University of New South Wales Group Group leader: John Piggott Contributing members: Hazel Bateman, Suzanne Doyle, Sachi Purcal Executive summary - January 2002 One of the economic imperatives of the new

More information

Doctor of Philosophy in Economics (English Program) Curriculum 2006

Doctor of Philosophy in Economics (English Program) Curriculum 2006 Doctor of Philosophy in Economics (English Program) Curriculum 2006 1. Program Title Doctor of Philosophy Program in Economics (English Program) 2. Degree Title Doctor of Philosophy (Economics) Ph.D. (Economics)

More information

An Improved Measure of Risk Aversion

An Improved Measure of Risk Aversion Sherman D. Hanna 1 and Suzanne Lindamood 2 An Improved Measure of Risk Aversion This study investigates financial risk aversion using an improved measure based on income gambles and rigorously related

More information

THE IMPACT OF SOLAR PANELS ON THE PRICE AND SALEABILITY OF DOMESTIC PROPERTIES IN OXFORD. By Charlie Morris-Marsham and Dr Gemma Moore

THE IMPACT OF SOLAR PANELS ON THE PRICE AND SALEABILITY OF DOMESTIC PROPERTIES IN OXFORD. By Charlie Morris-Marsham and Dr Gemma Moore THE IMPACT OF SOLAR PANELS ON THE PRICE AND SALEABILITY OF DOMESTIC PROPERTIES IN OXFORD By Charlie Morris-Marsham and Dr Gemma Moore Energy efficiency is no replacement for a good location, being an appropriate

More information

Diversification of the marketing chains among organic producers

Diversification of the marketing chains among organic producers Diversification of the marketing chains among organic producers Alessandro Corsi a Patrizia Borsotto b Ilaria Borri b Steinar Strøm a a Dept. of Economics, University of Torino, Italy; b INEA (National

More information

CHALLENGES AND OPPORTUNITIES THE INSURANCE INDUSTRY FACING WITH IN RELATION TO CLIMATE CHANGE

CHALLENGES AND OPPORTUNITIES THE INSURANCE INDUSTRY FACING WITH IN RELATION TO CLIMATE CHANGE CHALLENGES AND OPPORTUNITIES THE INSURANCE INDUSTRY FACING WITH IN RELATION TO CLIMATE CHANGE Fekete Mária FARKASNÉ Szent István University 2100 Gödöllő, Páter Károly u. 1. E-mail: Farkasne.Fekete.Maria@gtk.szie.hu

More information

MASTER OF SCIENCE FINANCIAL ECONOMICS ABAC SCHOOL OF MANAGEMENT ASSUMPTION UNIVERSITY OF THAILAND

MASTER OF SCIENCE FINANCIAL ECONOMICS ABAC SCHOOL OF MANAGEMENT ASSUMPTION UNIVERSITY OF THAILAND MASTER OF SCIENCE FINANCIAL ECONOMICS ABAC SCHOOL OF MANAGEMENT ASSUMPTION UNIVERSITY OF THAILAND ECO 5001 Mathematics for Finance and Economics The uses of mathematical argument in extending the range,

More information

NCSS Statistical Software Principal Components Regression. In ordinary least squares, the regression coefficients are estimated using the formula ( )

NCSS Statistical Software Principal Components Regression. In ordinary least squares, the regression coefficients are estimated using the formula ( ) Chapter 340 Principal Components Regression Introduction is a technique for analyzing multiple regression data that suffer from multicollinearity. When multicollinearity occurs, least squares estimates

More information

ECON 201 Section 002 Principles of Microeconomics Fall 2014 Tuesday & Thursday 1-2:15, Cuneo, Room 002

ECON 201 Section 002 Principles of Microeconomics Fall 2014 Tuesday & Thursday 1-2:15, Cuneo, Room 002 Dr. Roy Gobin Office: 773-508-8499 E-mail: rgobin@luc.edu Crown Center Rm#434 cubicle F Office Hours: Tues/Thu 3:50pm 4:20pm http://www.luc.edu/quinlan/faculty/terrygobin Catalog Description ECON 201 Section

More information

Love the One You re With: The Endowment Effect in the Dating Market

Love the One You re With: The Endowment Effect in the Dating Market Love the One You re With: The Endowment Effect in the Dating Market Colette Nataf, Thomas S. Wallsten University of Maryland cnataf@umd.edu, tswallst@umd.edu Keywords: endowment effect; dating market;

More information

SHORT-SELLING AND THE WTA- WTP GAP. Shosh Shahrabani, Tal Shavit and Uri Ben-Zion. Discussion Paper No. 07-06. July 2007

SHORT-SELLING AND THE WTA- WTP GAP. Shosh Shahrabani, Tal Shavit and Uri Ben-Zion. Discussion Paper No. 07-06. July 2007 SHORT-SELLING AND THE WTA- WTP GAP Shosh Shahrabani, Tal Shavit and Uri Ben-Zion Discussion Paper No. 07-06 July 2007 Monaster Center for Economic Research Ben-Gurion University of the Negev P.O. Box 653

More information

INCORPORATING SMALL PRODUCERS INTO FORMAL RETAIL SUPPLY CHAINS SOURCING READINESS CHECKLIST 2016

INCORPORATING SMALL PRODUCERS INTO FORMAL RETAIL SUPPLY CHAINS SOURCING READINESS CHECKLIST 2016 INCORPORATING SMALL PRODUCERS INTO FORMAL RETAIL SUPPLY CHAINS SOURCING READINESS CHECKLIST 2016 LSteinfield/Bentley University Authors: Ted London Linda Scott Colm Fay This report was produced with the

More information

The Financial Services Trust Index: A Pilot Study. Christine T Ennew. Financial Services Research Forum. University of Nottingham

The Financial Services Trust Index: A Pilot Study. Christine T Ennew. Financial Services Research Forum. University of Nottingham The Financial Services Trust Index: A Pilot Study Christine T Ennew Harjit Sekhon Financial Services Research Forum University of Nottingham The Financial Services Trust Index: A Pilot Study Executive

More information

TCO Certified Self-assessment Questionnaire

TCO Certified Self-assessment Questionnaire ! TCO Certified Self-assessment Questionnaire A.7.2 Senior Management Representative, Socially Responsible Manufacturing Introduction: Completion of this Self-assessment Questionnaire is required under

More information

A. Framework and compilation

A. Framework and compilation Framework for data integration in support of SNA compilation and modeling: Exercise for use of SNA in early estimates and projections in Central America By Jan W. van Tongeren, IVO. April 2006. A. Framework

More information

Optimal Consumption with Stochastic Income: Deviations from Certainty Equivalence

Optimal Consumption with Stochastic Income: Deviations from Certainty Equivalence Optimal Consumption with Stochastic Income: Deviations from Certainty Equivalence Zeldes, QJE 1989 Background (Not in Paper) Income Uncertainty dates back to even earlier years, with the seminal work of

More information

Students' Opinion about Universities: The Faculty of Economics and Political Science (Case Study)

Students' Opinion about Universities: The Faculty of Economics and Political Science (Case Study) Cairo University Faculty of Economics and Political Science Statistics Department English Section Students' Opinion about Universities: The Faculty of Economics and Political Science (Case Study) Prepared

More information

LOGIT AND PROBIT ANALYSIS

LOGIT AND PROBIT ANALYSIS LOGIT AND PROBIT ANALYSIS A.K. Vasisht I.A.S.R.I., Library Avenue, New Delhi 110 012 amitvasisht@iasri.res.in In dummy regression variable models, it is assumed implicitly that the dependent variable Y

More information

Topic 1 - Introduction to Labour Economics. Professor H.J. Schuetze Economics 370. What is Labour Economics?

Topic 1 - Introduction to Labour Economics. Professor H.J. Schuetze Economics 370. What is Labour Economics? Topic 1 - Introduction to Labour Economics Professor H.J. Schuetze Economics 370 What is Labour Economics? Let s begin by looking at what economics is in general Study of interactions between decision

More information

Usefulness of expected values in liability valuation: the role of portfolio size

Usefulness of expected values in liability valuation: the role of portfolio size Abstract Usefulness of expected values in liability valuation: the role of portfolio size Gary Colbert University of Colorado Denver Dennis Murray University of Colorado Denver Robert Nieschwietz Seattle

More information

Challenger Retirement Income Research. How much super does a retiree really need to live comfortably? A comfortable standard of living

Challenger Retirement Income Research. How much super does a retiree really need to live comfortably? A comfortable standard of living 14 February 2012 Only for use by financial advisers How much super does a retiree really need to live comfortably? Understanding how much money will be needed is critical in planning for retirement One

More information

Equity Monitoring Report

Equity Monitoring Report The Judiciary in Northern Ireland 2014 Equity Monitoring Report C ontents Page Executive i 1. About the report 1 2. Overall Composition: Gender 2 Community Background 3 Age 4 Declared Disability 5 Ethnicity

More information

5 August 2009. The Social Value of the Post Office Network Report for Postcomm

5 August 2009. The Social Value of the Post Office Network Report for Postcomm 5 August 2009 The Social Value of the Post Office Report for Postcomm Project Team NERA Stuart Holder Paul Metcalfe Helen Webb Orjan Sandewall Kenneth Train Accent Rob Sheldon Alison Lawrence Peer Review

More information

CATALYST REPORT DebateHub and Collective Intelligence Dashboard Testing University of Naples Federico II

CATALYST REPORT DebateHub and Collective Intelligence Dashboard Testing University of Naples Federico II CATALYST REPORT DebateHub and Collective Intelligence Dashboard Testing University of Naples Federico II 1. Overview: Information about Community Partner Involved 1.1. Resumé This test was carried out

More information

Lecture 6: Price discrimination II (Nonlinear Pricing)

Lecture 6: Price discrimination II (Nonlinear Pricing) Lecture 6: Price discrimination II (Nonlinear Pricing) EC 105. Industrial Organization. Fall 2011 Matt Shum HSS, California Institute of Technology November 14, 2012 EC 105. Industrial Organization. Fall

More information

NON-PROBABILITY SAMPLING TECHNIQUES

NON-PROBABILITY SAMPLING TECHNIQUES NON-PROBABILITY SAMPLING TECHNIQUES PRESENTED BY Name: WINNIE MUGERA Reg No: L50/62004/2013 RESEARCH METHODS LDP 603 UNIVERSITY OF NAIROBI Date: APRIL 2013 SAMPLING Sampling is the use of a subset of the

More information

Methodology. Discounting. MVM Methods

Methodology. Discounting. MVM Methods Methodology In this section, we describe the approaches taken to calculate the fair value of the insurance loss reserves for the companies, lines, and valuation dates in our study. We also describe a variety

More information

Sharing Online Advertising Revenue with Consumers

Sharing Online Advertising Revenue with Consumers Sharing Online Advertising Revenue with Consumers Yiling Chen 2,, Arpita Ghosh 1, Preston McAfee 1, and David Pennock 1 1 Yahoo! Research. Email: arpita, mcafee, pennockd@yahoo-inc.com 2 Harvard University.

More information

BACKGROUND ON THE SURVEY PROCESS

BACKGROUND ON THE SURVEY PROCESS Chapter Two BACKGROUND ON THE SURVEY PROCESS In this chapter, we present an overview of the various aspects of the research survey process. 1 We emphasize that surveying should first be thought of as a

More information

Solving Insurance Business Problems Using Statistical Methods Anup Cheriyan

Solving Insurance Business Problems Using Statistical Methods Anup Cheriyan Solving Insurance Business Problems Using Statistical Methods Anup Cheriyan Ibexi Solutions Page 1 Table of Contents Executive Summary...3 About the Author...3 Introduction...4 Common statistical methods...4

More information

HELP Interest Rate Options: Equity and Costs

HELP Interest Rate Options: Equity and Costs HELP Interest Rate Options: Equity and Costs Bruce Chapman and Timothy Higgins July 2014 Abstract This document presents analysis and discussion of the implications of bond indexation on HELP debt. This

More information

Financial Instruments Traded on the Commodity Exchange

Financial Instruments Traded on the Commodity Exchange Financial Instruments Traded on the Commodity Exchange Using the Financial Markets A cotton trader will almost always prefer to manage his price risk through physical positions and contracts for cotton,

More information

MEASURING INCOME DYNAMICS: The Experience of Canada s Survey of Labour and Income Dynamics

MEASURING INCOME DYNAMICS: The Experience of Canada s Survey of Labour and Income Dynamics CANADA CANADA 2 MEASURING INCOME DYNAMICS: The Experience of Canada s Survey of Labour and Income Dynamics by Maryanne Webber Statistics Canada Canada for presentation at Seminar on Poverty Statistics

More information

The third way: A hybrid model for pensions October 2015

The third way: A hybrid model for pensions October 2015 The third way: A hybrid model for pensions October 2015 The Club David Villa Views expressed here are those of the author, who is solely responsible for any errors and omissions. Content: Introduction

More information

THE CAYMAN ISLANDS LABOUR FORCE SURVEY REPORT SPRING 2015

THE CAYMAN ISLANDS LABOUR FORCE SURVEY REPORT SPRING 2015 THE CAYMAN ISLANDS LABOUR FORCE SURVEY REPORT SPRING 2015 Published September 2015 Economics and Statistics Office i CONTENTS SUMMARY TABLE 1. KEY LABOUR FORCE INDICATORS BY STATUS... 1 SUMMARY TABLE 2.

More information

Chapter 3 Local Marketing in Practice

Chapter 3 Local Marketing in Practice Chapter 3 Local Marketing in Practice 3.1 Introduction In this chapter, we examine how local marketing is applied in Dutch supermarkets. We describe the research design in Section 3.1 and present the results

More information

1 Uncertainty and Preferences

1 Uncertainty and Preferences In this chapter, we present the theory of consumer preferences on risky outcomes. The theory is then applied to study the demand for insurance. Consider the following story. John wants to mail a package

More information

GUIDELINES FOR ASKING THE NATIONAL STUDENT AND PARENT SCHOOL OPINION ITEMS

GUIDELINES FOR ASKING THE NATIONAL STUDENT AND PARENT SCHOOL OPINION ITEMS GUIDELINES FOR ASKING THE NATIONAL STUDENT AND PARENT SCHOOL OPINION ITEMS Guidelines for sampling are provided to assist schools in surveying students and parents/caregivers, using the national school

More information

PAY AND REWARD. Think Business, Think Equality

PAY AND REWARD. Think Business, Think Equality PAY AND REWARD Think Business, Think Equality CONTENTS INTRODUCTION WHAT TO CONSIDER WHEN DECIDING HOW MUCH TO PAY STAFF DEVELOPING PAY SYSTEMS INTRODUCING A NON-DISCRIMINATORY PAY RATE SYSTEM INTRODUCING

More information

Statistics Notes Revision in Maths Week

Statistics Notes Revision in Maths Week Statistics Notes Revision in Maths Week 1 Section - Producing Data 1.1 Introduction Statistics is the science that studies the collection and interpretation of numerical data. Statistics divides the study

More information

A Basic Introduction to Missing Data

A Basic Introduction to Missing Data John Fox Sociology 740 Winter 2014 Outline Why Missing Data Arise Why Missing Data Arise Global or unit non-response. In a survey, certain respondents may be unreachable or may refuse to participate. Item

More information

1. What is your primary motive for attending this farmers market? (choose only one) Purchase produce. Events/activities Purchase packaged goods

1. What is your primary motive for attending this farmers market? (choose only one) Purchase produce. Events/activities Purchase packaged goods Conducting Market Research Using Primary Data Kynda R. Curtis, Ph.D. Assistant Professor and State Extension Specialist Department of Resource Economics, University of Nevada, Reno WEMC FS#7-08 Overview

More information

Working Paper Series No. 12

Working Paper Series No. 12 Working Paper Series No. 12 Hypothetical, real, and predicted real willingness to pay in open-ended surveys: experimental results Anabela Botelho Lígia Costa Pinto September 2001 Published in Applied Economics

More information

Paid and Unpaid Labor in Developing Countries: an inequalities in time use approach

Paid and Unpaid Labor in Developing Countries: an inequalities in time use approach Paid and Unpaid Work inequalities 1 Paid and Unpaid Labor in Developing Countries: an inequalities in time use approach Paid and Unpaid Labor in Developing Countries: an inequalities in time use approach

More information

Chapter 5: Analysis of The National Education Longitudinal Study (NELS:88)

Chapter 5: Analysis of The National Education Longitudinal Study (NELS:88) Chapter 5: Analysis of The National Education Longitudinal Study (NELS:88) Introduction The National Educational Longitudinal Survey (NELS:88) followed students from 8 th grade in 1988 to 10 th grade in

More information

POST DISTRIBUTION MONITORING: - Guidelines to Monitor processes, outputs and outcomes

POST DISTRIBUTION MONITORING: - Guidelines to Monitor processes, outputs and outcomes POST DISTRIBUTION MONITORING: - Guidelines to Monitor processes, outputs and outcomes A guide for the Afghanistan Cash & Voucher Working Group George Bete: - August 2013 Funded by European Community Humanitarian

More information

2. Information Economics

2. Information Economics 2. Information Economics In General Equilibrium Theory all agents had full information regarding any variable of interest (prices, commodities, state of nature, cost function, preferences, etc.) In many

More information

Decision making in civil disputes: The effects of legal role, frame, and perceived chance of winning

Decision making in civil disputes: The effects of legal role, frame, and perceived chance of winning Judgment and Decision Making, Vol. 3, No. 7, October 2008, pp. 512 527 Decision making in civil disputes: The effects of legal role, frame, and perceived chance of winning Victoria Gilliland and John C.

More information

AP Psychology 2000 Scoring Guidelines

AP Psychology 2000 Scoring Guidelines AP Psychology 2000 Scoring Guidelines The materials included in these files are intended for non-commercial use by AP teachers for course and exam preparation; permission for any other use must be sought

More information

The Conference on Computers, Freedom, and Privacy. 'Overseeing' the Poor: Technology Privacy Invasions of Vulnerable Groups

The Conference on Computers, Freedom, and Privacy. 'Overseeing' the Poor: Technology Privacy Invasions of Vulnerable Groups 'Overseeing' the Poor: Technology Privacy Invasions of Vulnerable Groups Birny Birnbaum Center for Economic Justice www.cej-online.org Credit Scoring: 21 st Century Redlining and the End of Insurance Why

More information

Transportation costs make up a

Transportation costs make up a The Cost and Demographics of Vehicle Acquisition LAURA PASZKIEWICZ Laura Paszkiewicz is an economist in the Branch of Information Analysis, Division of Consumer Expenditure Surveys, Bureau of Labor Statistics.

More information

Summary value willingness to pay

Summary value willingness to pay 1 Summary Second-generation mobile coverage in the UK is high. Figures from the Ofcom 2013 Infrastructure Report 1 show that 99.6 per cent of UK premises and 87 per cent of the UK land mass have 2G coverage

More information

Identifying Requirements of Agricultural Mobile Marketing from Experts Perception

Identifying Requirements of Agricultural Mobile Marketing from Experts Perception , pp.113-118 http://dx.doi.org/10.14257/ijunesst.2014.7.1.11 Identifying Requirements of Agricultural Mobile Marketing from Experts Perception Maryam Omidi Najafabadi, Mahmood Ahmadi and Farhad Lashgarara

More information

Sourcing Products From Suppliers Certified Under Other Fair Trade Schemes

Sourcing Products From Suppliers Certified Under Other Fair Trade Schemes Sourcing Products From Suppliers Certified Under Other Fair Trade Schemes A Guide for Fair for Life certified Handlers Version April 2014 Table of Content Overview Fair for Life Product Sourcing... 1 Buying

More information

How Much Should I Save for Retirement? By Massi De Santis, PhD and Marlena Lee, PhD Research

How Much Should I Save for Retirement? By Massi De Santis, PhD and Marlena Lee, PhD Research How Much Should I Save for Retirement? By Massi De Santis, PhD and Marlena Lee, PhD Research June 2013 Massi De Santis, a senior research associate at Dimensional, is a financial economist with more than

More information

Section A. Index. Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting techniques... 1. Page 1 of 11. EduPristine CMA - Part I

Section A. Index. Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting techniques... 1. Page 1 of 11. EduPristine CMA - Part I Index Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting techniques... 1 EduPristine CMA - Part I Page 1 of 11 Section A. Planning, Budgeting and Forecasting Section A.2 Forecasting

More information

Implicit versus explicit ranking: On inferring ordinal preferences for health care programmes based on differences in willingness-to-pay

Implicit versus explicit ranking: On inferring ordinal preferences for health care programmes based on differences in willingness-to-pay Journal of Health Economics 24 (2005) 990 996 Implicit versus explicit ranking: On inferring ordinal preferences for health care programmes based on differences in willingness-to-pay Jan Abel Olsen a,b,,

More information

A Comparison of Training & Scoring in Distributed & Regional Contexts Writing

A Comparison of Training & Scoring in Distributed & Regional Contexts Writing A Comparison of Training & Scoring in Distributed & Regional Contexts Writing Edward W. Wolfe Staci Matthews Daisy Vickers Pearson July 2009 Abstract This study examined the influence of rater training

More information

Complexity as a Barrier to Annuitization

Complexity as a Barrier to Annuitization Complexity as a Barrier to Annuitization 1 Jeffrey R. Brown, Illinois & NBER Erzo F.P. Luttmer, Dartmouth & NBER Arie Kapteyn, USC & NBER Olivia S. Mitchell, Wharton & NBER GWU/FRB Financial Literacy Seminar

More information

Labor Standards Performance in Your Supply Chain

Labor Standards Performance in Your Supply Chain Labor Standards Performance in Your Supply Chain 6 68 96 Introduction Labor standards in the global supply chain is gaining recognition as an important issue. Consumers are becoming more aware of where

More information

Demand for Health Insurance

Demand for Health Insurance Demand for Health Insurance Demand for Health Insurance is principally derived from the uncertainty or randomness with which illnesses befall individuals. Consequently, the derived demand for health insurance

More information

HOW CLOSELY DO HYPOTHETICAL SURVEYS AND LABORATORY EXPERIMENTS PREDICT FIELD BEHAVIOR?

HOW CLOSELY DO HYPOTHETICAL SURVEYS AND LABORATORY EXPERIMENTS PREDICT FIELD BEHAVIOR? HOW CLOSELY DO HYPOTHETICAL SURVEYS AND LABORATORY EXPERIMENTS PREDICT FIELD BEHAVIOR? JAE BONG CHANG,JAYSON L. LUSK, AND F. BAILEY NORWOOD We compare the ability of three preference elicitation methods

More information

REMITTANCE TRANSFERS TO ARMENIA: PRELIMINARY SURVEY DATA ANALYSIS

REMITTANCE TRANSFERS TO ARMENIA: PRELIMINARY SURVEY DATA ANALYSIS REMITTANCE TRANSFERS TO ARMENIA: PRELIMINARY SURVEY DATA ANALYSIS microreport# 117 SEPTEMBER 2008 This publication was produced for review by the United States Agency for International Development. It

More information

COMPARISONS OF CUSTOMER LOYALTY: PUBLIC & PRIVATE INSURANCE COMPANIES.

COMPARISONS OF CUSTOMER LOYALTY: PUBLIC & PRIVATE INSURANCE COMPANIES. 277 CHAPTER VI COMPARISONS OF CUSTOMER LOYALTY: PUBLIC & PRIVATE INSURANCE COMPANIES. This chapter contains a full discussion of customer loyalty comparisons between private and public insurance companies

More information

Summary In the introduction of this dissertation, three main research questions were posed. The first question was: how do physical, economic, cultural and institutional distance act as barriers to international

More information

Exploratory Research Design. Primary vs. Secondary data. Advantages and uses of SD

Exploratory Research Design. Primary vs. Secondary data. Advantages and uses of SD Exploratory Research Design Secondary Data Qualitative Research Survey & Observation Experiments Företagsakademin, Henriksgatan 7 FIN-20500 Åbo Primary vs. Secondary data Primary data: originated by the

More information

Nonmarket Valuation Methods Theory and Applications. czajkowski@woee.pl

Nonmarket Valuation Methods Theory and Applications. czajkowski@woee.pl Nonmarket Valuation Methods Theory and Applications Mikołaj Czajkowski czajkowski@woee.pl Environmental Protection and Economics Value of environmental goods? Human activity Environment Optimum Human activity

More information

Chapter 21: The Discounted Utility Model

Chapter 21: The Discounted Utility Model Chapter 21: The Discounted Utility Model 21.1: Introduction This is an important chapter in that it introduces, and explores the implications of, an empirically relevant utility function representing intertemporal

More information

The Real Business Cycle model

The Real Business Cycle model The Real Business Cycle model Spring 2013 1 Historical introduction Modern business cycle theory really got started with Great Depression Keynes: The General Theory of Employment, Interest and Money Keynesian

More information

Anchoring Bias in Forecast Information Sharing in a Supply Chain

Anchoring Bias in Forecast Information Sharing in a Supply Chain Anchoring Bias in Forecast Information Sharing in a Supply Chain N. Karthikram R.K. Rajagopal G. Janarthanan RKAmit Abstract This paper investigates the behavioral aspects of decision making of retailers,

More information

4. Work and retirement

4. Work and retirement 4. Work and retirement James Banks Institute for Fiscal Studies and University College London María Casanova Institute for Fiscal Studies and University College London Amongst other things, the analysis

More information

Intro to Data Analysis, Economic Statistics and Econometrics

Intro to Data Analysis, Economic Statistics and Econometrics Intro to Data Analysis, Economic Statistics and Econometrics Statistics deals with the techniques for collecting and analyzing data that arise in many different contexts. Econometrics involves the development

More information

Data Mining Introduction

Data Mining Introduction Data Mining Introduction Bob Stine Dept of Statistics, School University of Pennsylvania www-stat.wharton.upenn.edu/~stine What is data mining? An insult? Predictive modeling Large, wide data sets, often

More information

The relationship between exchange rates, interest rates. In this lecture we will learn how exchange rates accommodate equilibrium in

The relationship between exchange rates, interest rates. In this lecture we will learn how exchange rates accommodate equilibrium in The relationship between exchange rates, interest rates In this lecture we will learn how exchange rates accommodate equilibrium in financial markets. For this purpose we examine the relationship between

More information

An Empirical Analysis of Insider Rates vs. Outsider Rates in Bank Lending

An Empirical Analysis of Insider Rates vs. Outsider Rates in Bank Lending An Empirical Analysis of Insider Rates vs. Outsider Rates in Bank Lending Lamont Black* Indiana University Federal Reserve Board of Governors November 2006 ABSTRACT: This paper analyzes empirically the

More information

PME Inc. Final Report. Prospect Management. Legal Services Society. 2007 Tariff Lawyer Satisfaction Survey

PME Inc. Final Report. Prospect Management. Legal Services Society. 2007 Tariff Lawyer Satisfaction Survey PME Inc. Prospect Management 2007 Tariff Lawyer Satisfaction Survey Final Report October 2007 Table of Contents 1. EXECUTIVE SUMMARY...6 2. BACKGROUND AND OBJECTIVES...9 3. METHODOLOGY...10 4. DETAILED

More information

HOW DO EMOTIONS INFLUENCE SAVING BEHAVIOR?

HOW DO EMOTIONS INFLUENCE SAVING BEHAVIOR? April 2009, Number 9-8 HOW DO EMOTIONS INFLUENCE SAVING BEHAVIOR? By Gergana Y. Nenkov, Deborah J. MacInnis, and Maureen Morrin* Introduction Employers have moved away from traditional defined benefit

More information

Imperfect information Up to now, consider only firms and consumers who are perfectly informed about market conditions: 1. prices, range of products

Imperfect information Up to now, consider only firms and consumers who are perfectly informed about market conditions: 1. prices, range of products Imperfect information Up to now, consider only firms and consumers who are perfectly informed about market conditions: 1. prices, range of products available 2. characteristics or relative qualities of

More information

Freight transport distance and weight as utility conditioning effects on a stated choice experiment

Freight transport distance and weight as utility conditioning effects on a stated choice experiment Journal of Choice Modelling, 5(1), 2012, pp 64-76 www.jocm.org.uk Freight transport distance and weight as utility conditioning effects on a stated choice experiment Lorenzo Masiero 1,* David A. Hensher

More information

Chapter #9: Valuation of Environmental Benefits

Chapter #9: Valuation of Environmental Benefits Chapter #9: Valuation of Environmental Benefits Contents: General Overview Types of Benefits Concepts in Benefits Measurement Uncertainty, Expected Benefits, and Risk Aversion Methodologies for Benefits

More information

A CHARTER OF FAIR TRADE PRINCIPLES

A CHARTER OF FAIR TRADE PRINCIPLES A CHARTER OF FAIR TRADE PRINCIPLES January 2009 2009 World Fair Trade Organization and Fairtrade Labelling Organizations International. All Rights Reserved. A CHARTER OF FAIR TRADE PRINCIPLES Contents

More information

An Analysis of the Health Insurance Coverage of Young Adults

An Analysis of the Health Insurance Coverage of Young Adults Gius, International Journal of Applied Economics, 7(1), March 2010, 1-17 1 An Analysis of the Health Insurance Coverage of Young Adults Mark P. Gius Quinnipiac University Abstract The purpose of the present

More information

Joint Product Signals of Quality. James E. McClure and Lee C. Spector. Ball State University Muncie, Indiana. September 1991.

Joint Product Signals of Quality. James E. McClure and Lee C. Spector. Ball State University Muncie, Indiana. September 1991. Joint Product Signals of Quality By James E. McClure and Lee C. Spector Ball State University Muncie, Indiana September 1991 Abstract In the absence of other information about the quality of an experience

More information

Table 1. Active User Projections for Planned M-Banking Service

Table 1. Active User Projections for Planned M-Banking Service Market Research MARKET SIZING OVERVIEW Market sizing is traditionally defined as estimating the number of buyers of a particular product, or users of a service. Because of the relative newness of mobile

More information