# The Econ 101 of Agriculture

Wolfram Schenker from Columbia University gave a talk about the impact of climate change on agricultural production at ZEW. The part which struck me the most, I found out later, is part of a paper published in the American Economic Review (Roberts and Schlenker, 2013). Seems like I have a good taste…

In recent years, especially before 2010, we saw a tremendous rise in agricultural commodity prices. The news was full of it. Between 2005 and 2008, corn prices went up by a factor of 4, from about \$2 to \$8 per bushel. One of the factors which contributed to this development was the US ethanol policy*, which encouraged the use of ethanol as a fuel substitute in order to reduce CO$_2$. Because of these incentives, farmers substituted food production with the production of ethanol. As a consequence, approximately one third of US corn production is nowadays used in the ethanol industry. And since the US market for agricultural products is so large (around 23% of the world market), this translates into a 5% decrease in the combined production of corn, soybeans, wheat, and rice worldwide.

How does this drop in food production affect prices? For that we need to know the elasticities of demand and supply. However, as students of empirical industrial organization know, estimating demand and supply functions is not trivial since observed market prices and quantities are the result of an equilibrium condition. If there is a spike in prices, for example, we cannot be sure whether it is caused by a decrease in supply, or whether supply actually remained constant but demand has increased. This is the standard textbook example of an endogeneity bias in simultaneous equation models.

We can solve this problem with instrumental variable techniques. We need some variables of which we can be sure that they only affect one curve, either demand or supply, to trace out the shape of the other. In their paper, Roberts and Schlenker use weather conditions as a supply shifter. Weather clearly has an impact on the yield of crops and thus supply. But it leaves demand unaffected. We still have to eat even when the sun is not shining.

This empirical strategy is pretty standard. In my empirical IO course we used the data in Angrist, Graddy, and Imbens (2000) to estimate the demand of fresh fish at the Fulton fish market by using weather conditions at the sea. The nice thing is that weather is a local phenomenon. So weather at sea, which affects catch rates and the supply of fish in Fulton, is not the same as for consumers at land. Therefore, it is unrelated to demand. Porter’s famous study on cartel stability (Porter, 1983) is another example for the use of weather as an instrument.

The new contribution of Roberts and Schlenker is now, that they also use weather as an instrument to estimate the supply function. How is that possible? Think of supply shocks in the past, let’s say a bad harvest one year ago. Agricultural commodities are storable. Hence, a bad harvest causes us to reduce our stock of commodities. Today supply may be back at its normal level, but we still want to refill the stocks that we consumed last year. This creates extra demand today. A supply shock in the past thus serves as demand shifter in this period.

Usually, I’m not a huge fan of using lagged variables to solve endogeneity problems. In most cases the problem remains as unobservables, which cause the endogeneity, are correlated over time. But in this case it seems plausible. It’s hard to imagine that weather last year is related to weather conditions today**. Thus, weather today is a supply shifter. But weather in the past period is a demand shifter because we have a storable commodity. Pretty cool idea, if you ask me.

Roberts and Schlenker find a supply elasticity which is considerably higher than for demand. This makes sense. We all have to eat, even when food becomes more expensive. But this implies that a decrease in supply, caused by the US ethanol policy, will have a large effect on prices. You can see this in Figure 1 in which the demand function is very inelastic and steep. Decreasing world production by 5% (from the red to the green line) increases prices, according to the authors’ results, by around 30%. And because of the old rule from my intermediate micro class, “the inelastic loses”, consumer surplus shrinks. The largest part of this lost consumer surplus gets transferred to suppliers via the ethanol subsidies. If you think of a strong farming lobby you might answer yourself the question why we have an ethanol policy in the first place.

Of course, a 30% price increase is much less than the 400% mentioned earlier. There are definitely other important factors at play. But the negative effect of the ethanol policy is nevertheless substantial. In addition, the policy probably even has diametral effects on CO$_2$ emissions. Food production decreases, but farmers expand overall production of crops because of the subsidy incentives. To achieve this additional production you have to increase land use, which is mainly achieved by deforestation. The negative effects of deforestation on CO$_2$ emissions are likely to be so large that they offset any gain from more environmental-friendly biofuel.

* Europe has quite similar policies in place.

** Meteorologist shall prove me wrong!

References

Angrist, J. D., Graddy, K., and G. W. Imbens (2000): “The interpretation of Instrumental Variables Estimators in Simultaneous Equations Models with an Application to the Demand for Fish,” 67(3), 499–527.

Porter, R. H. (1983): “A Study of Cartel Stability: The Joint Executive Committee, 1880-1886,” The Bell Journal of Economics, 14(2), 301–314.

Roberts, M. J. and W. Schlenker (2013): “Identifying Supply and Demand Elasticities of Agricultural Commodities: Implications for the US Ethanol Mandate,” American Economic Review, 103(6), 2265–95.