Price elasticities and tax reform in Mexico

Abstract

Price responses are usually estimated for the average household. However, different households are unlikely to respond in a similar way to movement in prices. Consequently, relying on averages may be misleading when examining the behaviour of a particular group of households such as the poor. This article uses six household surveys collected in Mexico between 1989 and 2000 to derive price responses for 10 product groups and for five levels of income households. The estimated price elasticities are then fed into a micro simulation model to measure the effect of a marginal tax reform. The results find that that poorer households tend to react substantially more to movement in prices, suggesting the usefulness of estimating elasticities that reflect the behavioural responses of the poor rather than of the entire population. The micro simulation results indicate that reducing the taxes on maize, alcoholic beverages and vegetables would be both more equitable and more efficient in terms of social welfare. Meanwhile, a reduction in the tax on legumes, sugar, and oils and fats, while inefficient, would contribute to reduce inequality.

Notes

¹ Deaton's methodology introduced a technique, based on the variation of prices across space to identify behavioural responses.

² Especially for developing countries with high inequality, households’ responses to changes in prices are likely to vary according to the income level of the household. Therefore, the use of aggregate price elasticities may be misleading.

³ For example, meat includes beef, chicken and pork as well as different qualities of cut.

⁴ Quality elasticities refer to the change in the quality of purchased products corresponding to a change in household income. This involves correcting unit values for quality. Chung (2006) examines the bias in the estimation price elasticites when ignoring quality.

⁵ Ayadi et al . (2003) measure the price elasticities using a series of household surveys for Tunisia.

⁶ The estimation includes a series of variables to capture household characteristics. These consist of the logarithm of household size, a series of age–gender ratios, a dummy variable for skilled workers, and a rural indicator. Because time and regional differences in tastes may result in demand not being the same across time and across region, the covariates are augmented with regional and time dummies. The model is estimated by allowing the response coefficient to vary across the surveys.

⁷ This follows the methodology of Hall (1995).

⁸ The price elasticities are symmetry-constrained estimates purged from the quality effect.

⁹ Cross-price elasticities were estimated in all cases, but in every case besides that of the total households their large SE resulted in no cross-price elasticity different from zero.

¹⁰ Price and cross-price elasticities enter into the calculation of the government revenues, and therefore the benefits of the tax reforms.

¹¹ Social welfare is defined as the sum of each households’ welfare as in W = V(u ₁,u ₂,…,u _n ), with u _h  = ψ(x _h ,p).

¹² These are measured by the producer and consumer support estimates, OECD Database 1986–2004.