By Elaine Kub
DTN Contributing Analyst
Predictions can be tricky things. Originating from eastern European immigrants to the United States (I think), there is a strange, rare tradition of letting an onion predict the year's precipitation pattern. Here's what you do: On New Year's Eve, take a locally grown onion, split it down its axis, then pull apart the layers of the two halves into cups. Align 12 cups in two rows representing the 12 months of the upcoming year. Put an equal amount of salt in each cup, then, when morning comes, observe which "months" have grown the wettest overnight. Those are the months that will receive large amounts of precipitation in the new year. The driest cups, on the other hand, represent months that will be dry.
Or so the theory goes. I gave it a try this past New Year's Eve with the thought of doing a year-end review comparing the onion's nonsense to 2015's real precipitation totals sometime in December. But now it's March, and I can already tell you that my onion was way off track. A great pool of water sat in the February cup on New Year's morning -- predicting that February would by far be the wettest month of 2015. We got less than three inches of snow in February, and if that's going to be the wettest month of the year, well ...
Other predictions -- ones with better basis in sound, scientific reasoning or sensible economic theory -- are a big deal this time of year. The annual Prospective Plantings report from USDA, for instance, predicts U.S. farmers will plant 89.2 million acres of corn in 2015, a record 84.6 million acres of soybeans, 55.4 million acres of wheat, and 7.9 million acres of sorghum, but we all know that the final numbers may be quite different than that. Maybe spring will suddenly turn wet enough to prevent some planting intentions from reaching fulfillment in some areas, or market prices will suddenly change farmers' economic motivations, or a lack of access to an inputs like seed or fertilizer will stymie some intentions.
As of this moment, none of those things looks very problematic in 2015 yet (with the exception of some struggles to source 10-34-0; see Russ Quinn's Weekly Retail Fertilizer Trends column from March 31 "10-34-0 Streaks Higher Again"). But that's the thing about predictions -- you just never know.
It's more interesting, from a mathematical standpoint, to examine yield predictions for 2015. Where could an analyst even begin, without being crippled by dread, knowing that summer weather could wildly throw off even the best-crafted springtime predictions?
Well, a sound place to start would be the historical facts. In Fundamentally Speaking blog posts this week, DTN Contributing Analyst Joel Karlin looked at the past 25 years of U.S. yield data. Drawing straight trend lines along just the past 15 years of that timeframe, which included some exceptionally great yields, and weighting the most influential production states, he concluded 2015's corn yield may be 165 bushels per acre and soybean yield may be 45 bushels per acre. Karlin remarked to me that "the length of time [for the trendline] is probably the biggest question," and I encourage you to see his posts for the finer variations on these numbers, depending how far back into history one looks.
An additional layer of prediction comes when one considers not just the historical trend, but also the weather conditions that overlay the present scenario as planting starts to get underway. Chris Anderson, with the weather forecasting corporation Freese-Notis, has made yield forecasts for U.S. corn at 162 bushels per acre and soybeans at 44 bushels per acre, starting with the 25-year trend as a first guess and then including "real-time climate indicators, not forecast conditions." He said, "Given the February climate indicator, we can look at historical records to evaluate what summer outcomes have happened and how yield responded." Specifically, he considered how yield when the U.S. experienced above average summer temperature and spotty drought, as in 2003 and 2005.
It seems very daunting to me to publicly put a number out there for a final yield that won't be known until fall, and which could be affected by any number of strange variables in the meantime. But Anderson handles the pressure with aplomb, noting that his forecast is free to change as conditions warrant.
"I think it is honest business to acknowledge that prior to the growing season the forecasts are highly variable. I interpret that to mean the most reliable information is historical information. It's not clear that the forecast consistently provides more useful information than historical data. Because, for example, which forecast do you put more stock in? Should I believe the January forecast this year, which had 45% chance of summer El Nino, or the mid-March forecast, which had >70% of summer El Nino? And, do they mean a weak summer El Nino or a strong summer El Nino?
"Using historical climate data alone, the forecast yield might be more conservative than using forecast data -- it certainly is less volatile. This historical data-driven approach is a more informed yield forecast than not using climate information at all."
Anderson further noted, "The most likely month for a significant move is May. If El Nino becomes very strong in May-June, the yield forecast will go up significantly. If El Nino remains weak -- at the level it is now -- the yield forecast will deteriorate."
In other words, predictions are tricky -- and even trickier the farther out you make them. Just ask my foolish onion.
Elaine Kub is the author of "Mastering the Grain Markets: How Profits Are Really Made" and can be reached at firstname.lastname@example.org, or on Twitter @elainekub.
© Copyright 2015 DTN/The Progressive Farmer. All rights reserved.