Originally published on VoxEU.org on 02 February 2018
The majority of obese adults are now found in developing countries. This column presents new evidence on the effects of trade on obesity in Mexico. The results indicate that across Mexican states, a one standard deviation increase in the unhealthy share of food imports from the US increases the likelihood of individuals being obese by about 5 percentage points. As developing countries around the world open up their food markets to industrialised countries, they may be accelerating their ongoing nutrition transition and imposing high future costs on their health systems.
Obesity is not the first health challenge that comes to mind when thinking about the global south. Obesity is rather associated with the Global North, particularly the US (think soda drinks, fast food, and lack of exercise). But this conventional wisdom is passé. The majority of obese adults – those with a body mass index of 30 and more – are now found in developing countries (Ng et. al 2014). The Global South is in the midst of a health and nutrition transition (Popkin and Gordon-Larsen, 2004). While communicable diseases and undernutrition are on a (slow) decline, non-communicable diseases and overnutrition are taking hold of populations, and they are doing so rapidly.
Given the known health risks (such as diabetes and cardiovascular diseases) and economic costs of obesity, what can policymakers in the Global South do to prevent obesity in epidemic proportions? Important lessons may come from countries that have already undergone this transition and from examining potential drivers that are amenable to public policy. The much-discussed case of Mexico is ideal for such an exercise.
Obesity and trade: The case of Mexico
Mexico’s obesity rates increased from 10% to 35% over the period 1980-2012 (according to our analysis sample including adult females). And among the already obese OECD countries, Mexico ranked second in 2015, surpassed only by the US (OECD 2017).
Coinciding with these profound changes in population health, Mexico has opened itself to trade in foods mostly with the US. Currently over 80% of Mexican food imports are American. In Figure 1, we show the evolution of Mexican imports of foods and beverages from the US over time. While overall food imports have increased dramatically, food that is typically considered unhealthy has surged quite spectacularly. Notably, exports of ‘food preparations’ are 23 times larger in 2012 compared to 1989.
Figure 1 Mexican imports of food and beverages from the US over time
In Figure 2, we classify Mexican imports from the US as unhealthy or healthy food using the United States Department of Agriculture (USDA) Dietary Guidelines (for instance, ‘dark green vegetables’ are recommended for increased consumption, while ‘refined flour and mixes’ are advised for reduced consumption). US exports to Mexico of both food groups have increased since the late 1980s, but unhealthy food groups at a much faster rate.
Figure 2 Unhealthy and healthy Mexican food and beverage imports from the US
Such trends naturally raise the suspicion of a possible, causal link running from greater consumption of US foods to rising obesity prevalence (e.g. Jacobs and Richtel 2017, Rogoff 2017). However, to date no paper has attempted to estimate a direct causal relationship between obesity and trade.
Estimating weight gains from trade in foods
In a new working paper, we quantify the impact of US food exports on individuals’ likelihood of being obese across Mexican states over the period 1988 to 2012 (Giuntella et al. 2017). To this end, we match several rounds of anthropometric and household expenditure surveys with product-level food trade data. Our main results are based on female adults for which data are available over this long time span.
We calculate the unhealthy share of food imports from the US by differentiating food items using the USDA Dietary Guidelines for Americans. We then allocate these aggregate food imports (healthy, unhealthy) to Mexican states. More specifically, we exploit variation in Mexican states’ historical expenditure by food products prior to trade integration. Our identification strategy assumes that aggregate trade shocks heterogeneously impact sub-national units as a function of time-invariant or ‘baseline’ variables (e.g. Dix-Carneiro and Kovak 2017, Autor et al. 2013). Note that there is substantial heterogeneity across Mexican states in obesity rates and historical food expenditure patterns, which further motivates our modelling approach.
Our empirical models also control for a host of state (such as food prices, GDP, FDI, migration) and individual covariates, as well as state fixed effects and state-specific time trends. In a second empirical strategy, we relate long-run differences in obesity rates at the state level to changes in unhealthy food imports conditional on baseline covariates. We instrument US exports of unhealthy foods to Mexico with corresponding US exports to other countries. And alternatively, we use ‘gravity residuals’ singling out the comparative advantage of the US in unhealthy food production relative to Mexico (akin to Autor et al. 2013).
Quantifying weight gains from trade in foods
We find that a one standard deviation increase in the unhealthy share of imports (equivalent to a 14 percentage point increase) raises the likelihood of obesity by about 5 percentage points. The effect amounts to 18% of the sample average in obesity. Findings using long-run difference models and IV estimates, as well as gravity residuals, are qualitatively similar – pointing to a plausibly causal effect.
Our main finding passes a series of robustness and placebo checks:
- Plausibly unrelated imports from the US (such as apparel products) do not impact obesity.
- The effects associated with food imports from the rest of the world are insignificant and small, underlining the specific importance of US foods for obesity.
- Likewise, unhealthy Mexican food exports to the US are not correlated with obesity.
- Similar patterns emerge if we employ food imports from the US for final demand.
- Overall (the sum of healthy and unhealthy) food imports do not correlate with obesity, highlighting the importance of differentiating ‘unhealthy’ and ‘healthy’ US foods.
- Our main results are robust to dropping Mexican states one by one.
- Similar patterns are obtained using body mass index (in quantile regressions) or overweight as outcome variables.
Health inequality and trade
Weight gains due to trade vary across socioeconomic groups. As illustrated in Figure 3, women with low levels of education face a greater risk of trade-induced obesity – their obesity risk is 5 percentage points higher than that of more educated women in a Mexican state with average exposure to unhealthy food imports. This differential goes up to 8 percentage points as the state’s trade exposure rises by 14 percentage points (one standard deviation). This interaction effect between education and trade is robust to the inclusion of state-time fixed effects (that is, purging the main local effect of trade exposure). The results are consistent with the well-known hypothesis that more educated individuals are more efficient producers of health investment than less educated ones. This educational gradient may be exacerbated in food environments where individuals are faced with more unhealthy food choices (Mani et al. 2013, Mullanaithan 2011, Dupas 2011).
Figure 3 Inequality between education groups in obesity risk and unhealthy food imports
Income, prices, and tastes
Having established a direct impact of US food exports on obesity prevalence in Mexico, we next turn to exploring possible mechanisms. Trade impacts incomes, prices, and tastes (due, for instance, to exposure to a foreign lifestyle and advertisement), all of which may drive the observed impacts on obesity. First, note that our main effect is robust to controlling for state GDP per capita, the total share of expenditures on unhealthy foods, as well as the relative prices of healthy versus unhealthy goods. Second, estimating demand equations over healthy and unhealthy food groups, we find that exposure to unhealthy foods from the US has redirected overall expenditure towards unhealthy foods. This observed shift is robust to controlling for real income and prices (for a similar empirical strategy, see Atkin 2013). In other words, trade with the US appears to influence tastes for relatively unhealthy foods. Increased variety of unhealthy foods boosts demand. These patterns are in line with a shift to ‘Western’ food consumption and increases in body weight among East Germans following the fall of the Berlin Wall (Dragone and Ziebarth 2017).
Nations have a lot to gain from trade. But weight gains and the associated health losses from trade in foods have been largely omitted from the equation. As developing countries around the world open up their food markets vis-à-vis industrialised countries – which tend to have a comparative advantage in more processed and less healthy foods – they may accelerate their ongoing nutrition transition. Obesity may put high costs on future health systems and the economies of the Global South.
Undoing the nutrition transition is likely harder than moderating it in the first place. Obesity and unhealthy food habits tend to be persistent. The Mexican experience is informative for countries in the Global South. Integrating nutrition and other health concerns into the formulation of food trade policies is paramount, and such concerns should feature high up on the agenda in future trade negotiations.
Our findings suggest that differentiating between healthy and clearly unhealthy imports may help slow down secular trends in obesity around the world.