1. USDA dropped US corn yield by 1.8bpa on Friday in the Nov Crop Report. Does that tell us anything about the final revision in the January report? History says yes it does.
2. This is a "contingency table" for Oct-Nov USDA corn yield changes vs. Nov-Jan changes. Just a fancy term for filling in the cells as follows: up-up, up-down, down-up. and down-down. Key observation is that in total more entries in the up-up and down-down cells. 
3. I like to convert the cell counts to conditional probabilities. In other words, given that we know direction of Oct-Nov change, what is chance of Nov-Oct being up or down?
4. For example, over 1988-2018 Oct-Nov change has been down 10 times. In 7 of those 10 years, the Nov-Jan change was also down. this means that historical odds are 70% that the Nov-Jan USDA corn yield change will be negative when Oct-Nov is negative.
5. Notice that the directional bias is much less pronounced when the Oct-Nov change is positive. But we know the change this year was negative so that is not relevant.
6. My takeaway: the odds definitely favor another downward revision in the USDA estimate of final avg US corn yield in January. Definitely not certain, but that is the tendency when Oct-Nov is down. Magnitude of expected drop is much harder to predict.
7. I will now use this as a shameless opportunity to promote research I have done over the years with co-authors about "smoothing" in USDA corn, soybean, and wheat production and yield estimates.
8. First paper was in the AJAE in 2006. One of the first to document the significant tendency of USDA crop production estimates to change in the same direction. What I like about this paper is we actually asked the people in NASS why this happened. academic.oup.com/ajae/article-a…
9. In this 2013 article we tackled the question whether USDA smoothing in corn and soybean production estimates was conditional on crop size. Do big crops get bigger and small crops get smaller? The answer is no, at least in any predictable way. ageconsearch.umn.edu/record/143639
10. The last of the trilogy tackled the question of whether the market anticipates the degree of smoothing in changes to USDA crop production estimates. We find that the market is indeed pretty smart and largely anticipates the smoothing. academic.oup.com/aepp/article/3…
