For Agricultural Commodities the September WASDE is an Important Benchmark, But How Accurate Is It?

Photo by Tomasz Filipek on Unsplash

The September USDA World Agricultural Supply Demand Survey, known as the September WASDE, is widely considered to provide benchmark predictions for the report’s respective crop year. How do its projected measures stand up to observed data for corn and soybeans? 

U.S. farmers and ag market participants look to the report to guide major decisions about when to buy or sell corn and soybeans. Each September the report projects yield, production, harvested acreage, and planted acreage (among other measures) for the year’s new crop of domestic corn and soybeans, which are harvested in the autumn in most of the country and stored and sold throughout the coming year. Although the September WASDE is released during the time of harvest, its projections are based on predictive factors gleaned throughout the planting season and can thus differ from the measurements observed post-harvest and production, as published in the USDA Feed Grains Yearbook for corn and USDA Oil Crops Yearbook for soybeans. 

Many watch the numbers closely, likely including companies that can be greatly affected such as Bunge Ltd BG, Archer-Daniels-Midland Co ADM, and Ingredion Inc (NYSE: INGR).

Projected vs. Realized Crop Numbers

Over the past decade, discrepancies between the projected September WASDE and observed statistics as reported in USDA yearbooks for harvested acreage, planted acreage, production (million bushels), and yield (bushels per acre) have ranged from 6.49%, when the WASDE overestimated corn yield in 2010, to -13.41%, when the WASDE underestimated soybean production in 2012.

WASDE measures of production and yield experienced greater predictive error for both corn and soybeans than did measures of harvested and planted acreage. Planted acreage is more easily predicted for the upcoming crop year than are production and yield because at the time of projection, planted acreage has essentially already been established, while unforeseen factors may yet affect production and yield.

Figures 1 and 2, below, show the difference between September WASDE’s forecast and the final Yearbook numbers by year for corn and soybeans, respectively, over the marketing years of 2010-2020. Error is calculated as the percent difference between the September WASDE projected metric and the observed metric as published in the respective USDA yearbook.

For example, the September 2010 WASDE projected production of 13,160 million bushels of corn for 2010/2011 marketing year. The USDA Grains and Feed Yearbook later published observed production of 12,425 million bushels of corn that year, marking a 5.91% overestimation by the September WASDE’s projection. This measurement is represented by the third bar in the 2010/2011 panel of Figure 1, below. 

As represented in Figure 1 above, the marketing years of 2010/2011, 2018/2019, and 2020/2021 experienced significant overestimations of corn production and yield by the respective September WASDE, while 2017/2018 saw a notable underestimation by the report. In 2010, high temperatures in corn-growing regions at the end of the growing season brought yields down greater than expected. The 2017/2018 marketing year was underestimated by the September WASDE when cool temperatures resulted in an unexpected record yield, while the next year, heavy rain slowed harvests more than had been accounted for by the September WASDE. In the 2020/2021 harvest year, a derecho in Iowa followed by unusually dry weather similarly resulted in lower-than-expected yield. 

Figure 2 above, depicting the forecasting error by September WASDEs for the U.S. soybean crop, shows radical underestimation in production and yield by the September 2012 report, with significant overestimation in 2018/2019. The southern United States experienced historic drought following two years of La Niña weather patterns, creating an expectation for significantly depressed soybean yields for the 2012/2013 marketing year. In the 2018/2019 marketing year, soybeans suffered, with corn, from excessive rain. 

A Closer Look at Yield and Production

Figure 3 below compares the WASDE projections and USDA observations for corn and soybean yield over the period 2010-2020. In four of the eleven marketing years (2010/11, 2018/19, 2019/20, and 2020/21), the WASDE overestimated both corn and soybean yield. In an additional four years (2012/13, 2013/14, 2015/16, 2016/17), the WASDE underestimated both corn and soybean yield, while in only three of the eleven years (2011/12, 2014/15, 2017/18), did the WASDE overestimate the yield of one commodity while underestimating the other. That is, in eight of the eleven years, the direction of WASDE forecasting error (either overestimation or underestimation) was the same across corn and soybeans.

Comparing WASDE production projections across corn and soybeans tells a similar story. Among the eleven marketing years between 2010-2020, there exists a trend of the two commodities measures seeing directionally similar forecasting error. Figure 4 below shows that along with yield, forecasting error for production differed directionally in the 2011/12, 2014/15, and 2017/18 marketing years. Additionally, the marketing years 2013/14 and 2015/16 saw directionally distinct forecasting error across the two commodities, though in these two marketing years, forecasting error was near zero for at least one of the commodities: in 2013/14, corn’s forecasting error was only 0.09% and in 2015/16, the forecasting errors were only -0.12% and 0.21% for corn and soybeans, respectively. 

 Table 1 below shows the September WASDE numbers compared to the August projection for the respective upcoming marketing year. September 2021’s WASDE estimated corn production of 14,996 million bushels, an upward revision of 1.67% compared to August 2021’s estimation of 14,750 for the 2021/2022 marketing year. Soybean production for the 2021/2022 marketing year was revised upward 0.81% in September 2021 compared to the previous month’s projections. 

The September WASDE saw an upward month-on-month revision in projected corn production similar to what the September 2018 WASDE projected relative to the previous month’s report. As seen in Figure 1, the September 2018 WASDE ultimately overestimated corn production by 3.39%. The September 2021 WASDE projected soybean production revision of 0.81% was the second smallest over the last 11 marketing years among September soybean production revisions. 

Current Conditions

General market sentiment estimated that the August WASDE projection for the 2021/2022 soybean crop was too low. Though the USDA increased their estimates from August to September due to good, rainy weather, early indications show that bean yields are coming in slightly lower than forecasted.

Similar to soybeans, the USDA increased corn production estimates from August to September due to higher planted acreage projections. The USDA numbers would suggest that the United States will see the second-largest corn harvest ever. Though this production estimate was higher than some market participants expected, early yield numbers suggest that some areas are already bringing in record yield.

The yearbook will again tell the final story about corn and soybean production. In the meantime, the outlook provided by the September WASDE will remain a key measure for producers and traders.