Are Markets “Surprised” by USDA Information?

Are Markets “Surprised” by USDA Information?

The concept of “market surprise” focuses on the “new” information in the USDA reports by calculating the difference from the previously released information in either earlier reports or private industry expectations.  Private industry expectations are typically obtained from either Bloomberg or Thompson Reuters surveys in more recent studies, or Conrad Leslie and Sparks in the past.  This group of studies, shown in table 5, also follow an event study framework, but measure market reaction not to the fact of report release (as in the first group of studies, measured by the date and time), but to the “market surprise” with the goal of measuring “the event” more precisely. The rationale behind this approach is that in an efficient market prices should react only to unanticipated information (Fama, 1970). These studies regress the unanticipated information in a USDA report on the price change immediately after the release of the report and focus on reports that make it possible to measure the unanticipated information, such as Crop Production reports (e.g., Karali, Irwin, and Isengildina-Massa, 2020; Karali et al, 2019; Adjemian and Arnade, 2017; McKenzie, 2008; Good and Irwin, 2006; Garcia et al, 1997; Orazem and Falk, 1989), Grain Stocks reports (e.g., Karali et al, 2019; Irwin, Good and Sanders, 2016; Irwin, Sanders and Good, 2014), Export Inspections reports (e.g., Colling, Irwin and Zulauf, 1996) and WASDE reports (Plante and Dhaliwal, 2017).

In general, this approach reveals an even stronger market reaction to USDA reports by removing the “noise” of other information that may hit the market on the same day and focusing on the “news” component of the reports.  For example, Karali et al (2019) show the average size of market surprises across different reports (figure 7) and demonstrate that Grain Stocks reports had the largest surprises, followed by August crop production reports. Furthermore, Karali et al (2019) demonstrate that in the most recent years, market reaction to October and November crop production reports was the strongest, as shown in figure 8.  For grain stocks reports, corn market reaction was the strongest for January and March releases in recent years, as shown in figure 9.

Other studies in this group focused on additional elements of market reaction that are possible to isolate using this approach.  For example, McKenzie (2008) used the Hamilton-type approach to derive statistically optimal weights to be placed on a number of different sources of information, including the “news” element of USDA crop production reports. He found that “the August reports along with ex post prices do contain valuable information, and this helps to explain the puzzle of why futures prices continue to react to the release of USDA reports.” (p.365).  Adjemian and Arnade (2017) demonstrated that USDA crop production reports affect not only the US, but international corn markets as well, as shown in figure 10.  Plante and Dhaliwal (2017) examine the effects of oil and grain inventory shocks (measured as surprises for ending stocks estimates in WASDE reports) on oil, ethanol, corn and soybean futures prices.  This study demonstrates that while corn and soybean prices react only to grain inventory shocks and energy futures prices react only to oil inventory shocks, ethanol futures prices react to both oil and grain inventory shocks, thus serving as a link between two sectors.

While these studies tend to measure information content of USDA reports more precisely, a recent study by Karali, Irwin and Isengildina Massa (2020) warns that the traditional measure of “surprise” may be riddled with measurement error and the findings of these studies should be interpreted with care.