Another Seeking Alpha contributor has written in USDA Report Bearish For Corn about the rise in the USDA's yield estimate to 166 bushels per acre for the coming corn crop in the US.
The USDA gives this estimate based on :
Projected corn yield based on the simple linear trend of the national average yield for 1990-2010 adjusted for 2012 planting progress.
This estimate, if it proves accurate, would be an all time high for US corn crop yields. Below is a table and chart of the data the USDA used to arrive at this forecast, which is available for download here.
(click to enlarge)
One finds that the time trend is more statistically significant than the progress of planting; having a correlation coefficient of .8419 for the time trend and .5309 for planting progress.
In 2012 the USDA decided to use only data from 1990-2010 to estimate the crop yield. I found it odd that the USDA did not include 2011's data in the regression. From 2008 until 2011 the USDA included a regression from 1990 until the data of the previous year. Before 2008 the USDA used a regression model starting from 1960.
So I decided to run a regression that went from 1990-2010 and also from 1990-2011 and then forecast 2012. These models both give estimates below the USDA's 166 a bushel per acre.
One would think that this is precisely the USDA's model; look again at the quote above, it has both a time trend and is plant progress adjusted. However, if you plug in the numbers for the variables to forecast 2012, it comes out at 164.7; which is still short of 166.0, and would reduce the corn crop by 115 mm bushels and give 12.8% stocks to use ratio.
Another interesting thing happens when we include the data for 2011, as was the usual practice from 2008 until 2012; that is of including all of the data from 1990 through the most previous crop year. The forecasted yield in 2012 then drops back down to 163.6 bushels, which translates into a 213 mm bushel drop in production, and a stocks to use ratio of 12.1%.
One can see that this model has just a slightly lower adjusted R^2, -.2%, but also notice that standard error of the regression has dropped in favor of this newer model and that the statistical significance of each variable has increased, as shown by the raising t-statistics.
In conclusion this research has led me to guess that the USDA set its crop yield estimate too high. Weather conditions over the summer can still increase or decrease the USDA's estimate. However, only once over that last four years have crop yields been above 160 bushels per acre, in 2009, with a yield of 164.7.
Qualitative analysis also suggests to consider the large increase in corn acreage this year. Economic theory would suggest that producers bring on marginal resources. A farmer most likely plants his best land and lets the less productive land lay fallow. Now this more marginal land is being put into production. Sure, some of these acres of corn are taken from other crops in prime areas, such as soybeans; but still, some must be from land that was marginal in the past.
Bears have the edge in the corn market right now and a world economic slowdown could send money out of commodities. However, people will still eat and world grain stocks are still relatively low. That said, corn bulls should wait for all this bearish news to get priced in. Bears have the USDA data on their side right now, but I would take it with a grain of salt; the bar might have been set to high with this yield estimate.
Traders who do not have a futures account can use the exchange traded fund, CORN, to establish positions in the corn market, as I have previously written in an article. I plan to have an article about estimated corn yields before the June crop report release; which should have data about the actual condition of the crop and not just plantings and a time trend line.
Additional disclosure: I have positions in the corn futures market