Grain Markets: Don’t Tunnel Vision on Yield Estimates, More Factors Are at Play – DTN

Corn harvest. Photo: Texas A&M AgriLife

As we get ready to turn out the lights on August 2017, we have to admit this has been a bearish time for corn prices. As of Monday, DTN’s national index of cash corn prices had dropped more than 25 cents from the end of July to near $3.00, the lowest prices in nine months.

For anyone following DTN’s daily market comments, this year’s crop estimates are well known by now and there has already been much discussion of crop ratings, ear weights, kernel counts, etc … And these conversations will be revived with every monthly WASDE report until we’re scooping snow in January.

I don’t mean any disrespect to these analytical efforts — I participate in them myself — but I find it important to remind our customers that all production estimates at this stage of the game still have wide margins of error. Even USDA’s September WASDE estimate of U.S. corn production has a 90% confidence interval of plus or minus 1 billion bushels (+/-7.7%).

Margins of error aside, a bigger concern for anyone wanting to understand the corn market is that we get so wrapped up counting kernels we forget that corn prices are influenced by many factors, only some of which have to do with the size of this year’s crop.

I realize that may sound like heresy across the Corn Belt, but allow me some leeway to explain. Give a guy some data, DTN’s ProphetX software, and a spreadsheet, and there is no limit to the amount of trouble he can stir up.

I recently put together a comparison of corn’s monthly spot futures prices with the prices of other commodity futures that ranged from January 1997 to the present. The list included soybeans, wheat, copper, crude oil, gold, the Dow Jones Industrial Average, T-bond futures, and the U.S. dollar index. The results were interesting enough that I decided to create a grid of correlation coefficients, which showed how closely each of the assets tracked with each other.

A correlation coefficient is a statistical tool that measures similarities between two or more different sets of price data. Coefficient values fall between +1.0 and -1.0 and a reading of +0.7 or higher means that there is a strong positive correlation between the two. A reading between +0.3 and -0.3 generally means the two variables have little in common.

It is important to understand a strong statistical correlation suggests the two variables share causative factors, but taken by itself, does not prove cause and effect. Correlation coefficients also change over different periods of time. I deliberately picked a range of 20 years to avoid the risk of being thrown off by a short-term aberration.

With explanations noted, here are the high points of what I found. First, it was no surprise that corn prices showed high correlations to soybeans (+0.92) and wheat (+0.89). Except for their specific differences, corn and soybeans in particular, share the same seasons, the same geography, and mostly, the same weather.

Grain News on AgFax

The surprise of the day came from the three non-grain assets. Versus corn, copper and crude oil both had correlation coefficients of +0.80 and gold came in at +0.85. For assets that have nothing to do with food, are produced year-round, and are not weather sensitive, those coefficients were remarkably high.

To put those results in perspective, I also tested the correlation of DTN’s national index of cash corn prices to USDA’s monthly estimates of U.S. ending corn stocks-to-use ratios — the golden calf of traditional analysis for corn prices.

The result was a correlation coefficient of -.67. The negative value simply means that corn prices go down as ending stocks-to-use ratios go up, but the fact that corn prices showed a weaker correlation to their own stocks-to-use ratios than they did to three non-grain assets with which they seem to have little in common should come as a surprise to many.

I even wondered if adjusting corn prices for inflation over the 20-year period would make a difference and there was a small improvement. Inflation-adjusted corn prices compared to USDA’s monthly estimates of U.S. ending stocks-to-use ratios showed a correlation coefficient of -.70, but was still not as strong as corn’s relationship to the three physical commodities mentioned above.

This topic deserves a part two next week, but for now, I would suggest market prices around the world are much more inter-related than many of us realize and keeping an eye on other commodity markets is every bit as important as straining over yield estimates. If you heard me mention spot copper prices reached a new two-year high in last Tuesday’s closing market video, you know the world is changing and this is no time for tunnel vision.

Todd Hultman can be reached at

Follow him on Twitter @ToddHultman1

The Latest

Send press releases to

View All Events

Send press releases to

View All Events