Association for Public Policy Analysis & Management

Increases in carbonated soft drink (CSD) consumption mirror the increase in the prevalence of obesity in the US and other countries of the world. Nevertheless, we do not know the extent to which CSD consumption has been responsible for the increase in obesity prevalence. Studies of the impact of CSD taxes on obesity rates reveal little to no effect. These taxes, however, have been very small and it is possible that individuals are responsive only to larger-scale changes in prices.

This paper explores whether sharp decreases in CSD prices (up to 30% over three months) in Peru in the late 1990s increased CSD consumption and, as a result, increased obesity prevalence among adult women. The empirical strategy consists of two specifications. In the first specification, I rely on the assumption that within-region changes in CSD prices during the period of analysis were exogenous. I claim that exogeneity came from variations in the supply of CSD generated by the staggered entry of a new CSD producer into different regions of Peru at different points in time, and the responses of incumbent firms that lowered prices sharply. I show that prices did not decrease more in regions where consumption or obesity had already been increasing to a greater degree. I also show that the drop in CSD prices was not correlated with a broader drop in the price of calories.

In my second specification, I exploit differential responses to the decrease in CSD price by access to running water in the houses. This specification relies on the weaker assumption that if there are unobservable variables that are correlated with changes in CSD prices and in weight outcomes, this correlation is not significantly stronger in households without access to running water at home compared to households within the same region and time but with such access.

I find that a 10% decrease in the price of CSD increases BMI of adult women of childbearing age by 0.5%, and obesity rate by 8.5%. These effects are explained by an elasticity of demand of 1.3 units in young families (and of 1.0 in the general population) and no substitution effect on other high-caloric beverages. I do find complementary effects on food prepared outside the home but consumed at home but find no such effect on the number of times the families eat outside the home.

The effects are greatly higher for families without running water access in their houses. These results warn us of the multiplicative effect greater CSD consumption may have on the obesity epidemic in developing countries, where water shortages are widespread.

The good news is that these results suggest that individuals are responsive to large-scale changes in the prices of CSD, and, hence, policies that tax CSD heavily are likely to have an impact on obesity prevention. Furthermore, this study suggests that improving access to drinking water, a significantly less regressive policy, could also serve to prevent obesity.