With the 2015 Kansas Wheat Tour (http://wheatqualitycouncil.org/) just completed and the May 12, 2015 USDA NASS Crop Production report (here) coming soon, an evaluation of the forecast accuracy of these information sources is in order. Following is a summary of the accuracy of both of the Kansas Wheat Tour and the USDA May Crop Production reports over the 2000-2014 and 2008-2014 periods.
The upshot of all of these technical calculations is that these forecast sources have provided valuable information on the size of the Kansas wheat crop in recent years. However, there is a track record of significant variability between the early May forecasts from both of these sources and the final outcome of the Kansas wheat crop over the 2000-2014 period.
The final, complete version of this article will be available on the Kansas State University Agmanager website (www.agmanager.info) on Monday, May 11, 2015.
Following is a summary of the article, along with the key charts and tables of the paper.
Kansas Wheat Crop Forecast Accuracy in Early May, 2015
Daniel M. O’Brien – Extension Agricultural Economist, Kansas State University
May 11, 2015
The 2015 Kansas Wheat Tour occurred May 4-7, 2015 in Kansas, and parts of the adjoining states of Nebraska, Colorado, and Oklahoma. Following are technical comments on the accuracy of the forecasts of Kansas wheat production, yields, and harvested acreage since 2000 associated with the Kansas Wheat Tour and with the USDA May Crop Production report. These results indicate opportunity for variation from these early May forecasts in regards to the final production of wheat in Kansas in 2015.
2015 Kansas Wheat Production
On Thursday, May 7th the 2015 Kansas Wheat Tour forecast that 2015 Kansas wheat production would be 288.5 million bushels (mb). This projection for 2015 was up 17% or 42.1 mb from the low 2014 Kansas wheat crop of 246.4 mb as reported by the USDA. The 2015 projection of 288.5 mb is much smaller than the largest five Kansas crops that have occurred since year 2000 (360 to 480 mb). Using the 2008-2014 measure of accuracy, the 67% confidence interval is 265.0 to 312.0 mb (+/- 23.1 mb).
The May 12, 2015 USDA Kansas wheat production forecast 67% confidence interval may be calculated by subtracting and adding +/- 43.0 mb using the 2000-2014 measure of forecast accuracy. For the 2008-2014 period, the 67% confidence interval may be calculated by subtracting and adding +/- 16.9 mb for the lower and upper bounds of the range.
2015 Kansas Wheat Yields
Kansas wheat yields in 2015 were forecast by the Kansas Wheat Tour to be 35.9 bushels per acre (bu/ac), up from the low USDA estimate of 28.0 bu/ac in 2014. Kansas wheat average yields since year 2000 have ranged from a low of 28.0 bu/ac in 2014 up to a high of 48.0 bu/ac in 2003. Using the 2008-2014 measure of forecast accuracy, the 67% confidence interval is 31.7 to 40.1 bu/ac (+/- 4.2 bu/ac).
The 67% confidence interval for the May 12, 2015 USDA Kansas wheat yield forecast can be calculated by subtracting and adding +/- 4.1 bu/ac, based on forecast accuracy during the 2000-2014 period. Similarly, using a 2008-2014 measure of forecast accuracy, 2.0 bu/ac would subtracted and added to calculate the lower and upper bounds of the range for the 67% forecast confidence interval.
2015 Kansas Wheat Harvested Acreage
Kansas wheat harvested acreage in 2015 is implicitly projected to be 8.036 million acres (ma), down from the USDA estimate of 8,800,000 ma in 2014, and the lowest amount since 8,000,000 ma in 2010, and 7,900,000 ma in 2011. Using the 2008-2014 measure of accuracy, the 67% confidence interval is 7,314,876 to 8,757,547 acres (+/- 721,336 acres). Harvested acreage of wheat in Kansas is typically not directly projected publically by the Kansas Wheat Tour, but rather implied from the Tour’s production and yield forecasts.
The 67% confidence interval for the May 12, 2015 USDA Kansas wheat harvested acres forecast would first be calculated by subtracting and adding +/- 368,782 acres using the 2000-2014 measure of forecast accuracy. For a 2008-2014 measure of forecast accuracy, the 67% confidence interval would be calculated by subtracting and adding +/- 309,377 acres for the lower and upper bounds of the range.