Sebastian Olascoaga

Why it matters

How the 2009 contraction reshaped Mexican household food security.

Mexico's GDP contracted 6.7% in 2009 as the global financial crisis worked through the Mexican economy. The aggregate effects were well documented. The household-level effects on food security were not. Standard inequality and poverty statistics could not say whether households able to meet their food needs in 2008 were still doing so in 2010.

Few studies investigated this for middle-income countries during the 2008 crisis, and none for Mexico. ENIGH 2008 was the first wave to include the Mexican Food Security Scale (EMSA), so the 2008+2010 comparison provides a consistent food-security measure across the crisis.

How we did it

Pooled ENIGH 2008 and 2010 with the EMSA food-security scale.

The analytical sample pooled 29,468 households from ENIGH 2008 with 27,654 households from ENIGH 2010. Food security was measured at the household level using the EMSA, a 12-item scale that classifies households into four ordered categories. Food secure households gave zero affirmative answers. Mildly food insecure households gave 1 to 3. Moderately food insecure households gave 4 to 7. Severely food insecure households gave 8 to 12.

We estimated three regression models on the pooled 2008+2010 cross-section. A generalized ordered logistic regression evaluated the experience-based measure across the four EMSA categories. An OLS regression on the share of household disposable income spent on food evaluated the expenditure-based measure. A quantile regression on the same continuous outcome allowed the effect of the crisis to vary across the income distribution. All three models adjusted for rural/urban residence, total household income quintile, household size, the head's gender and educational attainment, and indicators for social security, social health insurance, and participation in Oportunidades (the Mexican cash-transfer program).

What we found

Severe food insecurity grew between 2008 and 2010 while the total food-insecure population stayed flat.

Experience-based measure. The crisis significantly reduced the odds of being food secure (relative to severely food insecure), OR = 0.74 (p < .05). The corresponding reduction in odds of being severely or moderately food-insecure (relative to mildly food-insecure or food-secure) was OR = 0.81. Households were materially more likely to be classified as severely food insecure after the crisis than to remain in any of the milder categories.

Expenditure-based measure. The OLS regression showed the crisis significantly increased the share of total income spent on food (β coefficient of 0.02, p < .05). Households spent a larger share of their budget on food after the crisis than before.

Distributional measure. The quantile regression showed the crisis effect on food-expenditure share was more pronounced at lower quantiles. Households already spending a large fraction of their income on food before the crisis spent an even larger fraction after.

Across measures. The experience-based and expenditure-based food-security measures agreed in sign and gradient. Households more vulnerable before the crisis showed the largest deterioration after, across both self-reported food insecurity and the share of income spent on food.