According to this new paper of mine titled “Rising Food Prices, Food Price Volatility, and Political Unrest,” the answer is a qualified yes:
“Do food prices cause political unrest? Throughout history, riots appear to have frequently broken out as a consequence of high food prices. This paper studies the impact of food prices on political unrest using monthly data on food prices at the international level. Because food prices and political unrest are jointly determined, the incidence of natural disasters in a given month is used in an attempt to identify the causal relationship between food prices and political unrest. Empirical results indicate that between January 1990 and January 2011, food price increases have led to increased political unrest, whereas food price volatility has been associated with decreases in political unrest. These findings are consistent with those of the applied microeconomics literature on the welfare impacts of food prices.”
In other words, there is good evidence that rising food prices cause political unrest, but the evidence suggests that food price volatility does the exact opposite, although this latter effect cannot be argued to be causal and could only be a correlation.
If you would like a copy of the paper, you can download it from SSRN or on MPRA (opens a .pdf document). It goes without saying that I am grateful in advance for reader comments and suggestions.
Nice work, Marc. I guess the first thing that comes to mind is whether the reduced form of the natural disasters and unrest regression is measuring the same relationship as Miguel et al.’s rainfall and conflict estimates. Can you demonstrate that they are not the same thing? Apologies if I missed it in there, I read through quickly.
Thanks for your comment, Cyrus. That is a good question. The difference lies in the way conflict and unrest are defined in Miguel et al. and in this paper, respectively.
For Miguel et al., conflict is “‘a contested incompatibility which concerns government and/or territory where the use of armed force between two parties, of which at least one is the government of a state, results in at least 25 battle‐related deaths.'” They also write that “[their] database (…) does not include conflict information at the subnational level or by month within each year; nor does it provide the exact number of conflict deaths, which by necessity limits certain aspects of the empirical analysis.”
Here, political unrest is defined as food-related disturbances and is based on whether one of five food-related keywords occurs jointly with five riot-related keywords in the same news article. As such, it is very much subnational, and it need not involve deaths. The natural disasters I use are less predictable than rainfall (at least within a given month). The idea behind the two relationships is similar, they are not one and the same. As for demonstrating that they are not the same thing, I’m not sure what you had in mind.
Marc, to clarify what I mean by “demonstrating”: I suppose one could examine the correlation between the rain measure and the natural disaster measure, as well as between the unrest and civil war measures. One might also try to decompose the natural disaster measure to assess the relative contributions of rain-related events versus other sources of disaster (e.g. earthquakes). The reason in bring it up is just because the food price, Miguel et al agricultural output, and other rainfall related mechanisms are all slightly different, and so reviewers might raise objections about exclusion restrictions (on the basis that multiple mechanisms/causal paths imply exclusion violation).
This is just in the spirit of anticipating critiques. The work is really nice.
Thanks, Cyrus. This will be very helpful as I revise the paper at the end of summer in view of submitting before the academic year begins.
Nice paper on an important topic!
My main concern is that the exclusion restriction could be violated in a number of quite important ways. Why do we think that the only way natural disasters affect political unrest is through food prices. There are certainly many other channels from disasters to political unrest that have little to do with food. For example, what about destruction of capital leading to job losses and an increase in the ease of recruiting potential combatants now facing lower reservation wages? There are surely numerous other channels one could imagine. Ruling out a few of these as directly as possible would help convince skeptics like me.
Other issues:
-I think you need to use Newey-West standard errors if you’re not already. Serial correlation could be driving some of the results.
-If food prices follow a unit root, which I suspect they do, then it would probably be more appropriate to express prices in (log) differences rather than (log) levels.
Thanks for these excellent comments, Sami! I will definitely keep them in mind when I revise the paper.
[…] policy issue in the US. This also makes it easy to see why sharp increases in food prices may well cause people to riot, especially in places where their freedoms are considerably […]