Food Prices and Food Riots: How High — Not Volatile — Food Prices Cause Food Riots

My article “Rising Food Prices, Food Price Volatility, and Social Unrest” was finally made available online (click here if you don’t have an institutional subscription and would like an ungated copy identical to the published version) last week by the American Journal of Agricultural Economics, roughly three years after I wrote the first draft of it. Here is the abstract:

Can food prices cause social unrest? Throughout history, riots have frequently broken out, ostensibly as a consequence of high food prices. Using monthly data at the international level, this article studies the impact of food prices – food price levels as well as food price volatility – on social unrest. Because food prices and social unrest are jointly determined, data on natural disasters are used to identify the causal relationship flowing from food price levels to social unrest. Results indicate that for the period 1990–2011, food price increases have led to increases in social unrest, whereas food price volatility has not been associated with increases in social unrest. These results are robust to alternative definitions of social unrest, to using real or nominal prices, to using commodity-specific price indices instead of aggregated price indices, to alternative definitions of the instrumental variable, to alternative definitions of volatility, and to controlling for non-food-related social unrest.

In other words, contrary to the popular opinion, what drives social unrest is the food price level, and not the volatility of food prices. That is, what matters is the mean — not the variance — of the food price distribution, and this is robust to a variety of alternative specifications.*

If you have a good working knowledge of microeconomic theory, this should not be surprising: food price increases directly hurt consumers and benefit producers. Increases in food price volatility, which are analytically equivalent to an increase in food price uncertainty, directly hurt producers (as per Sandmo, 1971), but for consumers, it really isn’t clear what the effect on welfare is. Since riots tend to break out in cities, where the vast majority of people are net consumers of food, it should not be a surprise that what drives social unrest is high rather than volatile food prices.

This is consistent with the micro level results in my 2013 article with Chris Barrett and David Just, where we look at the food price volatility (holding food price levels constant) on the welfare of rural Ethiopian households and find that it is disproportionately wealthier rural households — those households who are much more likely to be net producers of food — who get hurt by food price volatility.

Now, for the enterprising young researchers out there, this suggests an interesting research question: We have a pretty good idea that peasant uprisings are more likely when food prices are low, but is it true that they are more likely when food prices are more uncertain?

*One robustness check that is not in the paper nor the appendix is a specification in which the food price level is interacted with food price volatility, holding each individual component of that interaction constant, both in an OLS and in an IV setting. I checked last week just for fun, and in no case is that interaction term significant at any of the conventional levels. I might write a blog post about this, given that it is important to know whether it is high and volatile food prices that cause food riots, but I doubt anyone would want to publish that null finding…

No related content found.


  1. Michael W

    “We have a pretty good idea that peasant uprisings are more likely when food prices are low”

    Do you mean “when food prices are high”?

  2. Marc F. Bellemare

    No, I meant “low,” under the implicit assumption that peasants mainly riot when they are net sellers of food. But of course, if they are net buyers of food, the same reasoning applies as in the case of urban, net-food-consumer households. Still (and bear in mind I’m not as knowledgeable about peasant uprisings as I am on food riots), I think the textbook peasant uprising occurs when the state taxes rural households in order to subsidize the food consumption of urban households–which can occur when food prices are kept artificially low via commodity boards.

  3. Stu

    I haven’t looked at the paper, but I have to ask — is there any subjective method of measuring the amount of social unrest? Riots would be very broad strokes, but there are small riots and large ones, and other contributing factors, like civil war.

  4. Marc F. Bellemare

    Thank for your question. You are right, riots and social unrest are difficult to measure, given that they are hard to define to begin with. What I use here is a count of news stories (from LexisNexis, then from Factiva for robustness) about social unrest in the English-language media. It’s not ideal (some riots are counted twice, because both the New York Times and the Washington Post talk about it, and this is a count of news stories, so it’s not clear how one should interpret the effect of a marginal change in the price of food), which is why I test about every imaginable combination of variables to make sure that my results are robust. Then I supplement this using the count of food riots in Africa for 1990-2011, which Cullen Hendrix and Idean Salehyan have in their Social Conflict in Africa Data (SCAD).

  5. Pingback: Food Prices and Food Riots: How High — No...
  6. Pingback: Food Prices and Food Riots: How High — Not Volatile — Food Prices Cause Food Riots « Economics Info
  7. am

    I suppose it is difficult to know what really starts the riot, sometimes. Food may be the last thing in a chain of events that causes it.

    Another thing is like situations not having the same effect. Some people, I suppose, riot more than others in the same circumstances. It must be a difficult subject to sift through.