Biases from Uncontrolled Quality When Unit Values Are Used To Estimate Quantity Demand Give Grounds for Caution on Fat and Sugar Taxes

---

• 研討會日期 : 2015-11-17
• 時間 : 14:30
• 主講人 : Professor Bonggeun Kim
• 主持人 : Professor Kamhon Kan
• 地點 : Conference Room B110
• 演講者簡介 : Professor Kim received his Ph.D. in Economics from University of Michigan in 2002. He is currently a Professor at Seoul National University. His research interests are Labor, Development, and Public Economics.
• 演講摘要 : Applied demand studies increasingly inform public health interventions like taxes on unhealthy foods and drinks. Quantity consumed is of interest (since sugar and fat intake is proportionate to this) but most econometric models are for expenditures, typically in the form of budget shares. Most studies are estimated from household surveys, which are the only source of data with age-, income- and ethnic-group specific responses to price changes. But household surveys do not collect data on the consumption of elementary goods, and instead focus on groups of similar goods. There are many varieties within a group so group expenditure is due to price, quantity, and quality (average expenditure per unit). If quality responses to price are not controlled for, what is estimated as a price elasticity of quantity demand conflates both quantity and quality responses and will make taxes on unhealthy foods seem more effective at moderating consumption than they truly are. A further bias occurs because household surveys are rarely linked to spatially and product-wise disaggregated price data, and researchers instead use unit values – group expenditures divided by group quantity – as a proxy for price. A method to recover price elasticities with such data was proposed by Deaton (1990) and uses separability restrictions to obtain the response of quality to price changes. In this study we use especially collected data from Vietnam to directly estimate the response of quality to price changes, and compare with indirect estimates made using weakly separable preference restriction. We also test the other condition needed for unit values to be a key element of the model of aggregated demand and quality substitutions to accurately proxy for market prices, which is that inter-area price variation dominates within-group relative price differences. The results show widespread failure of both restrictions. Consequently, estimated price elasticities of quantity demand over-state the likely efficacy of tax-induced price rises in reducing consumption. There may even be perverse effects if price rises cause consumers to switch to lower quality items within a group and these lower quality items are less healthy. We illustrate this effect using a hypothetical fat tax which causes some consumers in Vietnam to switch from cooking oils to lard.