How does household welfare vary in response to changes in food prices? Poor vs. non-poor households

ABSTRACT

In this paper, we focus on poor households vs. non-poor households, using objective and subjective definitions of poverty. We evaluate the various responses of these households to changes in food expenditures, income, and prices, and simulate the welfare losses of food price changes across poverty definitions. We use the QUAIDS model to estimate food elasticities and rely on the National Survey of Expenditure and Household Income, from Uruguay 2016/2017, because it contains novel information on the subjective poverty status of each household. Our results show important differences across poverty definitions. In addition, we find that price increases may lead to larger welfare losses in poor households. On average, the percentage of income needed to avoid a loss in the economic welfare of poor households, defined by the objective method, is twice that required by the non-poor households, for all price changes. Differences are much smaller when using the subjective approach. Our main contribution is the first evidence of consumption behaviour and welfare analysis under different poverty measures. Our findings highlight the need for policies that mitigate the negative effects of price shocks.

KEYWORDS:

• Food consumption
• Demand system
• Poverty
• D12

JEL CLASSIFICATION:

• D12
• I31
• I32

Acknowledgements

We are grateful to the participants of the 2021 meeting of the Society of Economics of the Household (Boston, US) for helpful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Availability of data and materials

ENGIH data set can be downloaded from the website of the National Institute of Statistics (INE) of Uruguay, https://www.ine.gub.uy/engih2016

Notes

¹ Given the definition of our food categories, as in Wood, Nelson, and Nogueira (2012), we do not face the problem of zero expenditures. The percentage of zero expenditure in each of the five food categories is sufficiently small.

² In Table A2. of the Appendix A we report the estimates of the quadratic term (${\lambda }_i$) and the residuals of the first stage of the control function included in the system (${\nu }_i$) to show that the quadratic specification provides a good fit and that total food expenditure is indeed endogenous.

³ We focus on own-price elasticity because these elasticities are used to estimate welfare losses due to changes in price.

⁴ To ensure that the modified cost function maintains the homogeneity property, an additional constraint is imposed (Perali 2003).

⁵ We also use an index for access to basic amenities as alternative instrument, as in Caro et al. (2017). Even though system parameters are robust, total income is a stronger instrument in the context of our data.