Price elasticities of meat, fish and seafood: A meta-analysis
| Název práce v češtině: | Cenové elasticity masa, ryb a mořských plodů: Meta-analýza |
|---|---|
| Název v anglickém jazyce: | Price elasticities of meat, fish and seafood: A meta-analysis |
| Klíčová slova: | meta-analysis, elasticity, price elasticity, food, meat, fish, seafood, prices, heterogeneity, cross-country, publication bias, consumer sensitivity |
| Klíčová slova anglicky: | meta-analysis, elasticity, price elasticity, food, meat, fish, seafood, prices, heterogeneity, cross-country, publication bias, consumer sensitivity |
| Akademický rok vypsání: | 2022/2023 |
| Typ práce: | bakalářská práce |
| Jazyk práce: | angličtina |
| Ústav: | Institut ekonomických studií (23-IES) |
| Vedoucí / školitel: | doc. PhDr. Zuzana Havránková, Ph.D. |
| Řešitel: | skrytý - zadáno a potvrzeno stud. odd. |
| Datum přihlášení: | 16.03.2023 |
| Datum zadání: | 21.06.2023 |
| Datum potvrzení stud. oddělením: | 21.06.2023 |
| Datum a čas obhajoby: | 09.09.2024 09:00 |
| Místo konání obhajoby: | Opletalova, O206, místnost. č. 206 |
| Datum odevzdání elektronické podoby: | 30.07.2024 |
| Datum proběhlé obhajoby: | 09.09.2024 |
| Oponenti: | Mgr. Josef Bajzík, Ph.D. |
| Seznam odborné literatury |
| Bibliography
van Aert, R.C. & M. van Assen (2021). Correcting for publication bias in a meta-analysis with the p-uniform* method. Working paper, Tilburg University & Utrecht University, available online at osf.io/preprints/metaarxiv/zqjr9/download (oned on November 5, 2022). Amini, S.M. & C.F. Parmeter (2012): Comparison of model averaging techniques: Assessing growth determinants. Journal of Applied Econometrics 27(5): pp. 870-876. Andrews, I. & M. Kasy (2019): Identification of and Correction for Publication Bias. American Economic Review 109(8): pp. 2766-2794. Andreyeva, T., Long, M.W. & K.D. Brownell (2010): The Impact of Food Prices on Consumption: A Systematic Review of Research on the Price Elasticity of Demand for Food. American Journal of Public Health 100(2): pp. 216-222. Ashenfelter, O., Harmon, C. & H. Oosterbeek (1999): A Review of Estimates of the Schooling/Earnings Relationship, with Tests for Publication Bias. Labour Economics 6(4): pp. 453-470. Bom, P. R. D. & H. Rachinger (2019): A Kinked Meta-Regression Model for Publication Bias Correction. Research Synthesis Methods 10(4): pp. 497-514. Chen, D., Abler, D., Zhou, D., Yu, X. & W. Thompson (2015): Meta-analysis of Food Demand Elasticities for China. Applied Economic Perspectives and Policy 38(1): pp. 50-72. Cornelsen, L., Green, R., Turner, R., Dangour, A.D., Shankar, B., Mazzocchi, M. & R.D. Smith (2015): What Happens to Patterns of Food Consumption when Food Prices Change? Evidence from A Systematic Review and Meta-Analysis of Food Price Elasticities Globally. Health Economics 24(12): pp. 1548-1559. Green, R., Cornelsen, L., Dangour, A.D., Turner, R., Shankar, B., Mazzocchi, M. & R.D. Smith (2013): The effect of rising food prices on food consumption: Systematic review with meta-regression. BMJ 346: art. f3703. Egger, M., Smith, G.D., Schneider, M. & C. Minder (1997): Bias in meta-analysis detected by a simple, graphical test. British Medical Journal 315(7109): pp. 629–634. Elliott, G., Kudrin, N., & Wuthrich, K. (2022). Detecting p-hacking. Econometrica 90(2): pp. 887-906. Furukawa, C. (2019): Publication Bias under Aggregation Frictions: Theory, Evidence, and a New Correction Method. MIT working paper, Massachusetts Institute of Technology. George, E. I. (2010): Dilution priors: Compensating for model space redundancy. In IMS Collections Borrowing Strength: Theory Powering Applications - A Festschrift for Lawrence D. Brown, volume 6, p. 158-165. Institute of Mathematical Statistics. Hansen, B.E. (2007): Least Squares Model Averaging. Econometrica 75(4): pp. 1175–1189. Havranek T., Irsova, Z., Laslopova, L. & O. Zeynalova (2022): Publication and Attenuation Biases in Measuring Skill Substitution. Review of Economics and Statistics, forthcoming. Hedges, L. V. (1992): Modeling publication selection effects in meta-analysis. Statistical Science 7(2): pp. 246–255. Ioannidis, J.P., T.D. Stanley & H. Doucouliagos (2017): The Power of Bias in Economics Research. Economic Journal 127(605): pp. 236-265. Maier, M., Bartos, F. & E.J. Wagenmakers (2023): Robust Bayesian meta-analysis: Addressing publication bias with model-averaging. Psychological Method, forthcoming. doi: 10.1037/met0000405. Stanley, T.D. (2005): Beyond Publication Bias. Journal of Economic Surveys 19(3): pp. 309-345. Stanley, T. D., S. B. Jarrell, & H. Doucouliagos (2010): Could It Be Better to Discard 90% of the Data? A Statistical Paradox. The American Statistician 64: pp. 70–77 Steel, M.F. (2020): Model averaging and its use in economics. Journal of Economic Literature 58(3): pp. 644-719. |
| Předběžná náplň práce |
| Research question and motivation
How do consumers respond to changes in food prices? This response can be measured by economists through the concept of price elasticity of demand for food. Understanding price elasticity is crucial for assessing the impact of fiscal policies aimed at discouraging the consumption of potentially harmful foods. It also provides policymakers with insights into how tax revenues may change with pricing policies. This research question necessitates a comprehensive investigation, thus, I have identified four previous meta-analyses that focus on own-price food elasticities: Andreyeva et al. (2010), Cornelsen et al. (2015), Green et al. (2013), and Chen et al. (2016). However, these meta-analyses have certain limitations: they do not consider publication bias, overlook endogeneity bias, and only one of them provides a ceteris-paribus analysis while neglecting model uncertainty. Publication selection bias occurs when certain results are more likely to be reported than others (Stanley, 2005). Typically, preference is given to statistically significant results or parameter values that align with well-established theories. The Law of demand is one of the strongest economic theories, stating that an increase in price leads to a decrease in demand. As most types of food are considered ordinary goods, the price elasticity is expected to be negative. Thus, if a researcher gets a positive estimate, they may choose not to write a study based on such results, or to adjust (intentionally or not) their methodology or dataset in order to produce the intuitive outcome. While such adjustments may be justified for individual studies (such as model misspecification, small sample size, or data noise), if done systematically, they distort the entire body of literature and bias the overall mean. For example, Ioannidis et al. (2017) demonstrate that the mean estimate reported in economics is inflated twofold due to this bias. Contribution The aim of this study is to synthesize global evidence on consumer price sensitivity while addressing any potential biases. I intend to employ innovative methodologies that correct for publication bias and isolate the impact of endogeneity in demand models estimated using simple regression, such as ordinary least squares (OLS). We will analyze the interplay between publication and endogeneity biases (as suggested earlier by Ashenfelter et al., 1999, publication bias plagues the instrumental variable estimates, which typically yield higher standard errors, pushing researchers to search for larger estimates to get to the desired level of statistical significance). To analyze why the estimated elasticities vary, we will code for the study design (focusing on the cross-country variation) and employ model averaging techniques. Finally, we will use our results to estimate a synthetic best-practice estimate. Methodology Data collection will involve identifying relevant studies that estimate own-price elasticities by conducting a comprehensive search using a well-designed query in Google Scholar. The empirical research will follow the classical demand model, which entails regressing food quantity on food price. Typically, the price variable will be accompanied by other covariates that influence demand, such as income, prices of alternative substitutes, and socio-demographic variables including education, family size, or urban area. Most researchers use the log-log specification, which provides a direct estimate of the elasticity. To assess publication bias, I will also gather information on the reported standard errors for each elasticity estimate. I will ensure the inclusion of studies from previous meta-analyses and snowball the references of recent studies to minimize the risk of missing relevant literature. To identify publication bias, meta-researchers often utilize a funnel plot, originally proposed by Egger et al. (1997). However, this test is purely visual. To address this limitation, Stanley (2005) introduced a formal analog called the test of funnel asymmetry. The underlying idea is that in the absence of publication bias, there should be no systematic correlation between the estimate and its standard error. Since primary studies report t-statistics assuming the ratio of estimates to standard errors follows a t-distribution, it is expected that estimates and standard errors are statistically independent. However, when published studies selectively report estimates with a specific sign or statistical significance, the estimates become correlated with standard errors. Nevertheless, the funnel asymmetry test has two main drawbacks: it assumes a linear relationship between the estimate and standard error (ignoring potential jumps at conventional critical values) and assumes that the standard error is exogenous to the estimate (disregarding reverse causality, omitted variable bias, and measurement error, as described in Havranek et al., 2022). I will use weighted averaging methods, such as those proposed by Ioannidis et al. (2017), Furukawa (2019) using stem-based approaches, Andrews and Kasy's (2019) selection model, and Bom and Rachinger's (2019) endogenous kink method, which address the limitations of the funnel asymmetry test by dropping the linearity assumption. Additionally, we I will use p-uniform* estimation based on reported p-values for effects, following van Aert and van Assen (2020), which drops the exogeneity assumption. Besides publication bias, I will support my arguments with the analysis of heterogeneity behind the estimates that comes from the different study design. Building upon previous surveys, I will expand the list of coded variables that represent the study design. These variables will include country-level factors, the definition of demand (such as food consumption, household expenditures normalized by demanded quantity, or food sales), the measurement of prices (average price or marginal price), the distinction between short-run and long-run estimates, data aggregation levels (aggregated or micro-level), and whether the elasticities are commercially or household-relevant, among others. The food categories that the research will specifically focus on are meat, fish, dairy products, fruits and vegetables, fats and oils, and cereals. An important aspect of the study design will be the coding for the treatment of potential endogeneity of the price measure, which is inherent in any demand rule. By incorporating multiple aspects of study design as explanatory variables, we can better explain the differences in collected elasticity estimates, including their standard errors. To account for model uncertainty arising from the multitude of explanatory variables, I will use model averaging techniques, utilizing both Bayesian and Frequentist approaches as described by Steel (2020). For the Bayesian approach, I will use the dilution prior suggested by George (2010) to handle multicollinearity, while for the Frequentist approach, my algorithm will be based on Hansen (2010) and Amini and Parmeter (2012). Additionally, I will use the robust Bayesian model averaging (Maier et al. 2023) method, which combines the aforementioned techniques for publication bias detection into one estimate of the true mean value and the extent of publication bias while accounting for between-study heterogeneity. |
| Předběžná náplň práce v anglickém jazyce |
| Research question and motivation
How do consumers respond to changes in food prices? This response can be measured by economists through the concept of price elasticity of demand for food. Understanding price elasticity is crucial for assessing the impact of fiscal policies aimed at discouraging the consumption of potentially harmful foods. It also provides policymakers with insights into how tax revenues may change with pricing policies. This research question necessitates a comprehensive investigation, thus, I have identified four previous meta-analyses that focus on own-price food elasticities: Andreyeva et al. (2010), Cornelsen et al. (2015), Green et al. (2013), and Chen et al. (2016). However, these meta-analyses have certain limitations: they do not consider publication bias, overlook endogeneity bias, and only one of them provides a ceteris-paribus analysis while neglecting model uncertainty. Publication selection bias occurs when certain results are more likely to be reported than others (Stanley, 2005). Typically, preference is given to statistically significant results or parameter values that align with well-established theories. The Law of demand is one of the strongest economic theories, stating that an increase in price leads to a decrease in demand. As most types of food are considered ordinary goods, the price elasticity is expected to be negative. Thus, if a researcher gets a positive estimate, they may choose not to write a study based on such results, or to adjust (intentionally or not) their methodology or dataset in order to produce the intuitive outcome. While such adjustments may be justified for individual studies (such as model misspecification, small sample size, or data noise), if done systematically, they distort the entire body of literature and bias the overall mean. For example, Ioannidis et al. (2017) demonstrate that the mean estimate reported in economics is inflated twofold due to this bias. Contribution The aim of this study is to synthesize global evidence on consumer price sensitivity while addressing any potential biases. I intend to employ innovative methodologies that correct for publication bias and isolate the impact of endogeneity in demand models estimated using simple regression, such as ordinary least squares (OLS). We will analyze the interplay between publication and endogeneity biases (as suggested earlier by Ashenfelter et al., 1999, publication bias plagues the instrumental variable estimates, which typically yield higher standard errors, pushing researchers to search for larger estimates to get to the desired level of statistical significance). To analyze why the estimated elasticities vary, we will code for the study design (focusing on the cross-country variation) and employ model averaging techniques. Finally, we will use our results to estimate a synthetic best-practice estimate. Methodology Data collection will involve identifying relevant studies that estimate own-price elasticities by conducting a comprehensive search using a well-designed query in Google Scholar. The empirical research will follow the classical demand model, which entails regressing food quantity on food price. Typically, the price variable will be accompanied by other covariates that influence demand, such as income, prices of alternative substitutes, and socio-demographic variables including education, family size, or urban area. Most researchers use the log-log specification, which provides a direct estimate of the elasticity. To assess publication bias, I will also gather information on the reported standard errors for each elasticity estimate. I will ensure the inclusion of studies from previous meta-analyses and snowball the references of recent studies to minimize the risk of missing relevant literature. To identify publication bias, meta-researchers often utilize a funnel plot, originally proposed by Egger et al. (1997). However, this test is purely visual. To address this limitation, Stanley (2005) introduced a formal analog called the test of funnel asymmetry. The underlying idea is that in the absence of publication bias, there should be no systematic correlation between the estimate and its standard error. Since primary studies report t-statistics assuming the ratio of estimates to standard errors follows a t-distribution, it is expected that estimates and standard errors are statistically independent. However, when published studies selectively report estimates with a specific sign or statistical significance, the estimates become correlated with standard errors. Nevertheless, the funnel asymmetry test has two main drawbacks: it assumes a linear relationship between the estimate and standard error (ignoring potential jumps at conventional critical values) and assumes that the standard error is exogenous to the estimate (disregarding reverse causality, omitted variable bias, and measurement error, as described in Havranek et al., 2022). I will use weighted averaging methods, such as those proposed by Ioannidis et al. (2017), Furukawa (2019) using stem-based approaches, Andrews and Kasy's (2019) selection model, and Bom and Rachinger's (2019) endogenous kink method, which address the limitations of the funnel asymmetry test by dropping the linearity assumption. Additionally, we I will use p-uniform* estimation based on reported p-values for effects, following van Aert and van Assen (2020), which drops the exogeneity assumption. Besides publication bias, I will support my arguments with the analysis of heterogeneity behind the estimates that comes from the different study design. Building upon previous surveys, I will expand the list of coded variables that represent the study design. These variables will include country-level factors, the definition of demand (such as food consumption, household expenditures normalized by demanded quantity, or food sales), the measurement of prices (average price or marginal price), the distinction between short-run and long-run estimates, data aggregation levels (aggregated or micro-level), and whether the elasticities are commercially or household-relevant, among others. The food categories that the research will specifically focus on are meat, fish, dairy products, fruits and vegetables, fats and oils, and cereals. An important aspect of the study design will be the coding for the treatment of potential endogeneity of the price measure, which is inherent in any demand rule. By incorporating multiple aspects of study design as explanatory variables, we can better explain the differences in collected elasticity estimates, including their standard errors. To account for model uncertainty arising from the multitude of explanatory variables, I will use model averaging techniques, utilizing both Bayesian and Frequentist approaches as described by Steel (2020). For the Bayesian approach, I will use the dilution prior suggested by George (2010) to handle multicollinearity, while for the Frequentist approach, my algorithm will be based on Hansen (2010) and Amini and Parmeter (2012). Additionally, I will use the robust Bayesian model averaging (Maier et al. 2023) method, which combines the aforementioned techniques for publication bias detection into one estimate of the true mean value and the extent of publication bias while accounting for between-study heterogeneity. |