The U.S. Biofuel Mandate and the Supply and Demand for Calories
I'm buried with teaching and research and travel these days, so even fewer postings than usual.
Here are my slides for my talk today at Cornell. This is stuff I've been working on for awhile with Wolfram Schlenker at Columbia. The key conceptual contribution is showing how to use weather shocks to estimate the supply of agricultural commodities. We also estimate demand using the weather, which isn't new in concept, but surprisingly hasn't been done on a large scale.
We use our estimates to consider the effect of U.S. ethanol policy:
(1) about a one-third increase in the price of calories from raw commodities: corn, soybeans, wheat and rice.
(2) an annual reduction in calories used for food production equal to the annual caloric requirements of over 200 million people.(*)
While we do not yet estimate where the calories in (2) are coming from, we expect most of the demand response comes from poor countries due to a larger income effect.
(*) For rich people eating meat one calorie not used in food production is about five to 10 times the calories consumed by people; for poorer people living mainly on grains the ratio of calories produced to consumed is much closer to one.
Update: Sorry, these slides translated terribly when I uploaded to Zoho. I'll see if I can fix that. And I'll try to post a copy of the working paper too. I expect I'll be juggling a lot today so we'll see how it goes.
Update 2: Most of my day was filled with meetings so I didn't figure out a way to post the slides until very late. But I think this alternative works better than zoho. You can find a copy of the working paper on my personal website (here). That's not quotable yet, but this earlier meetings paper is.
Update 3: It seems the largest question/concern people have about this research is aggregation: we estimate supply and demand for the whole world and for the caloric content of all four key crops combined.
Why aggregate? Several reasons:
1. By aggregating we can cut directly to the questions that are most interesting and most important.
2. By aggregating we can show in a very clear and transparent manner the variations that allow us to identify supply and demand.
3. By aggregating we increase statistical power a lot.
4. If we disaggregate countries or crops, identification becomes difficult or impossible, and far less transparent. We have more intrumental variables (so more bias), fewer degrees of freedom (bigger standard errors and more bias), and far less transparency.
In short, aggregation allows us to simplify the problem to the point where we can get clear, defensible answers to the biggest questions.
What are the costs of aggregation?
Well, we cannot identify heterogeneity in supply response across crops and countries. This would be nice, and we do plan to try this in the future, but we're not going to have much statistical precision. And I don't see how it would change the aggregate numbers that are most interesting and most important.
Here are my slides for my talk today at Cornell. This is stuff I've been working on for awhile with Wolfram Schlenker at Columbia. The key conceptual contribution is showing how to use weather shocks to estimate the supply of agricultural commodities. We also estimate demand using the weather, which isn't new in concept, but surprisingly hasn't been done on a large scale.
We use our estimates to consider the effect of U.S. ethanol policy:
(1) about a one-third increase in the price of calories from raw commodities: corn, soybeans, wheat and rice.
(2) an annual reduction in calories used for food production equal to the annual caloric requirements of over 200 million people.(*)
While we do not yet estimate where the calories in (2) are coming from, we expect most of the demand response comes from poor countries due to a larger income effect.
(*) For rich people eating meat one calorie not used in food production is about five to 10 times the calories consumed by people; for poorer people living mainly on grains the ratio of calories produced to consumed is much closer to one.
Update: Sorry, these slides translated terribly when I uploaded to Zoho. I'll see if I can fix that. And I'll try to post a copy of the working paper too. I expect I'll be juggling a lot today so we'll see how it goes.
Update 2: Most of my day was filled with meetings so I didn't figure out a way to post the slides until very late. But I think this alternative works better than zoho. You can find a copy of the working paper on my personal website (here). That's not quotable yet, but this earlier meetings paper is.
Update 3: It seems the largest question/concern people have about this research is aggregation: we estimate supply and demand for the whole world and for the caloric content of all four key crops combined.
Why aggregate? Several reasons:
1. By aggregating we can cut directly to the questions that are most interesting and most important.
2. By aggregating we can show in a very clear and transparent manner the variations that allow us to identify supply and demand.
3. By aggregating we increase statistical power a lot.
4. If we disaggregate countries or crops, identification becomes difficult or impossible, and far less transparent. We have more intrumental variables (so more bias), fewer degrees of freedom (bigger standard errors and more bias), and far less transparency.
In short, aggregation allows us to simplify the problem to the point where we can get clear, defensible answers to the biggest questions.
What are the costs of aggregation?
Well, we cannot identify heterogeneity in supply response across crops and countries. This would be nice, and we do plan to try this in the future, but we're not going to have much statistical precision. And I don't see how it would change the aggregate numbers that are most interesting and most important.
Hi Michael,
ReplyDeleteIs there a paper that accompanies this presentation? I think that I'm missing some key points in the slides.
Ok... I read the AJAE-Conference version and was not clear in how storage & inventories link past weather to current supply. At the end, my understanding was that you use the same first-step for both demand and supply, only that in the latter, you ussed lagged prices and weather while in the former the instrumentation was using contemporaneous values. I'll read the newer version to see if I clear this in my mind. Nice paper.
ReplyDeleteNelson,
ReplyDeleteSorry I missed your comment earlier. i've been out of commission recently.
The link between past yield shocks and current prices does confuse a lot of people. We're working on making this clearer, both rhetorically and in theory. In a nutshell, past weather shocks randomly affect current inventories. And inventories affect current price, holding all else the same. So we see bad weather events in the past followed by land expansion in the future. We believe this stems mainly from the effect of bad past weather shocks on current prices. I think this is way more compelling and exogenous than traditional Nerlove-type model. Indeed, I think it takes Nerlove's initial insights to their logical next step.