Yield Variability : All or nothing … or somewhere in between
With Monday morning’s round of supply and demand prognostications from USDA came a new and improved set of yield numbers to play with. Given how uncertainty assails farmers from all angles these days — from input markets, from the futures markets for their crops — it’s especially appropriate to examine those yield numbers for signs of even more uncertainty.
For instance, corn yield tends to get all summed up in one number … just one number: 153.8 bushels per acre. But nobody actually gets exactly 153.8 bushels per acre out of a given field, or if they do, it is nothing more than a coincidence. Farmer Joe may see 154 flash momentarily across his yield monitor and ultimately get 149 out of one field, while his neighbor Farmer Chuck gets 160 bpa out of the corn field right over the fence. Their colleagues Farmer Al in Illinois and Farmer George in North Dakota could get 200 and 120 bpa averages, respectively, but the idea is that all across the nation, these results will average out to 153.8 this year.
Well, how unhelpful. That number, taken in conjunction with the nationwide averages for the past 30 years or so, can show a trend of increasing yield potential in corn, but it does nothing to help a given farmer in a given location decide if corn’s production risk is worth the planting costs next year, or if it’s the least risky of all his options.
My attention was drawn to this topic by a column in Dakota Farmer, written by Kevin Hall, a 23-year-old farmer from Hoople, N.D. He wrote, “Would you plant a soybean if you had a 60 percent chance of getting 120 bushels per acre and a 40 percent chance of getting absolutely nothing? … There is a ton of emphasis on yield potential and average yields, but yield consistency is not really measured or talked about. … Measuring standard deviation would give farmers the ability to plant exactly the mix of yield potential and yield consistency they feel comfortable with.”
Unfortunately, that’s a whole other layer of complexity and risk that is in the hands of seed companies to make clear to customers, but we can at least use the recent Crop Production data to examine crops on a state-by-state basis, although not on a variety-by-variety or field-by-field level.
For instance, in 2008, there is expected to be a 156-bushel-per-acre difference between the highest-yielding state, Washington (210 projected average), and the lowest-yielding state, North Carolina (54 bpa). Obviously, the weather factors in any given year will affect that range, so it’s more valuable to compare how states have performed over an extended history, say 30 years, to even out the best and worst years.
Since 1979, Illinois has generally outperformed the national average corn yield by 10 bushels per acre in any given year, but that series of data (how far Illinois yields are above or below the nation’s yields) has a standard deviation of 9.62. That means Illinois’ yields are statistically expected to be at least 0.38 bushels per acre above the nation’s yields about 84 percent of the time, but there’s an equal likelihood of them being either 0.38 bpa over or 19.62 bpa over.
For another example, North Dakota has generally trailed the national average by 33.16 bpa over the past 30 years, with a standard deviation of 12.52. In 2008 when the nation’s average yield is expected to be 153.8, that means any randomly selected corn field in North Dakota has an equal likelihood of yielding 108 bpa (arrived at by 153.8 – (33.16 + 12.52)) or 133 bpa (153.8 – (33.16 – 12.52)), all other things being equal.
Now what does this mean for the future of corn production in the United States? Well, in the states where corn yields have a higher standard deviation, i.e. a greater risk of not performing to expectations, farmers may grow relatively less willing to take on that risk if input costs remain high.
This is especially true of the areas that already have infrastructure in place to deal with alternative crops. Nationwide over the past 30 years, corn yields have generally ranged within 31 bpa above or below their historical yield average, a 25 percent up or down risk. Soybeans, similarly, have a standard deviation over the past 30 years (7.9) that is about 25 percent of their historical yield average (31 bpa). Wheat has a standard deviation (16.2) that is 34 percent of its 30-year yield average (47 bpa), and grain sorghum’s standard deviation is 30 percent of its 30-year yield average.
The real star of yield reliability over the past three decades has been sunflowers. Nationwide, their yields have tended not to veer more than about 200 pounds per acre above or below the 30-year average at 1,225 pounds (about 17 percent). Spoiled for choice of North Dakota elevators that accept flowers, no wonder Mr. Hall has the luxury of demanding more reliability — or at least more information about yield consistency — from his crops.
For the raw data on your own state’s historical crop yields, you can query the National Agricultural Statistics Service’s Quick Stats page at http://www.nass.usda.gov/…. The data can be imported into a spreadsheet and analyzed for averages and standard deviations.
Elaine Kub can be reached at email@example.com
If you don't follow the rules, your comment may be deleted.
User Legend: Moderator Trusted User