Three essays on the demand system

This first essay investigates how scanner data affect demand elasticity estimates and develops methods for scientists to adapt estimated elasticities to analyses of specific policies. We conduct a meta-analysis of U.S. demand elasticities and find evidence that scanner data generate statistically different elasticities, with more elastic demand than other data types. Own-price elasticity estimates from household scanner quantity data appear to be more elastic than other quantity types. Household-level estimates using retail scanner price data, as proxies for prices, tend to be more price-elastic than other price types. These results suggest caution or adjustment when selecting elasticities for policy analysis. The second essay proposes an approach to meta-analysis that is consistent with demand theory. Meta-analyses of demand elasticity do not impose theoretical demand constraints even though elasticities are derived from a demand system. Specifically, we estimate the fundamental parameters elasticity formula of the Quadratic Almost Ideal Demand System using nonlinear least squares augmented by a wild bootstrap procedure for bias correction and to allow for heteroskedasticity across both studies and food categories. Furthermore, using the estimated fundamental parameters and historical market data, we estimate price and income elasticities for the eight major food categories in the United States. The most significant implication of this analysis is that our empirical framework aligns meta-analysis with demand theory, facilitating welfare calculations. Additionally, this method enables the easy computation of demand elasticities using relevant market data. This new approach could become a valuable tool for applied economists. The third essay estimates food demand elasticities across countries using the Exact Affine Stone Index model. The primary objective of this study is to directly estimate a cross-country demand system for aggregated commodities using historical data and produce theoretically consistent demands for applied analysis. We use annual country-level data from the Aglink-Cosimo model data of the OECD-FAO and include 11 regional groups including 45 countries and regions. For the analysis, we extend the EASI model by including the regional group fixed effects and estimate the quartic Engel curve based on the likelihood ratio test result. Finally, using the estimates from the EASI model, we calculate unconditional price and income elasticities by each regional group.