T. Lance Toler and Dr. Farrell E. Jensen, Economics
The United States Department of Agriculture (USDA) prepares reports on the income and financial stability of American farmers and agricultural organizations. If the information in USDA reports on the financial stability of American agriculture is already known by the markets, then the prices of Federal Farm Credit System (FCS) Bank bonds, which are related to the credit risk associated with American agriculture, will not be affected by the release of the reports. If the information provided in the reports is new to the market, however, significant changes in FCS bond prices should immediately be seen as the new information is absorbed into market prices.
To analyze this issue an event study framework was used to test the effects of the release of the USDA’s “Agricultural Income and Finance Report” on FCS bond prices. An original yield-to-maturity (YTM) application of traditional event study methodology was applied. Several distinct tests were used to determine the existence and significance of abnormal YTM changes on the day of the report announcement.
Event studies are often used in financial research to determine the effects of certain events on the value of a firm. Assuming efficient markets, event studies provide an ideal tool for examining the information content of disclosures and announcements. Only unexpected information contained in events should cause the asset price to change around event dates. In this study, the unexpected element of the USDA reports are the only element of the reports that should affect FCS bond prices.
The data used in this study include the YTMs for five-year U.S. Treasury bonds and five-year FCS bonds. The data provide the daily closing yields to maturity for the FCS bonds from October 29, 1993 – June 28, 2002. This time period includes 2,157 observations. A total of 28 “Agricultural Income and Finance” reports were issued on different dates during the time period covered by the data set.
The basic empirical model to determine the existence of an effect on a common stock security from an event was first outlined in Fama et al (1969) and has since undergone some modifications. Because of differences in the characteristics between stock and bond markets, the original empirical model was modified according to recommendations found in modern finance literature. Essentially, normal returns for an equity or debt security were calculated over an estimation window through ordinary least squares regression on the returns of a market index. Abnormal returns during the event window, or the time the asset return is affected by an event, are then calculated, and summed to come up with a cumulative abnormal return. Standardized cumulative abnormal returns can be aggregated and manipulated to determine if the returns are significantly different from zero, and thus, if the event had a significant impact from the normal return on the debt or equity security. The full empirical model, though prohibitively lengthy, allowed test statistics to be found to test a null hypothesis. The null hypothesis being tested is that the abnormal returns of FCS bonds on the day of announcement of the USDA’s “Agricultural Income and Finance” report are not statistically different from zero, or that the event has no effect on the prices of FCS bonds.
The first steps of the empirical model are iterated 27 times for the available report dates and the days were aggregated to yield a single non-significant statistic in an “Aggregated Test.” These results suggest that, in aggregate, the release of the USDA reports did not have a significant up or down effect on the prices of FCS bonds.
In the “Separated Test by ‘Good’ and ‘Bad News’ Criteria,” the “Agricultural Income and Finance” reports were examined to determine which reports gave “good news” for bond markets and which gave “bad news.” The empirical model was run again with these two different data sets. The results suggested a slightly abnormal increase in price of FCS bonds after the announcement of “good news.” The results also suggested a significant decrease in the price of FCS bonds relative to U.S. Treasuries on the day of announcement. Further analysis of these results shows that the effect is long-lasting and semi-permanent.
Because of the difficulties surrounding the inexactness of choosing “good” and “bad news” reports, an approach in which the market is allowed to decide which reports are “good” and “bad” was used in a “Market-Determined News Test.” To do this, all of the report dates with a negative effect on bond prices, corresponding to “bad news” as the market determined, were evaluated together. All 14 dates corresponding to “good news” were also evaluated together. The results were compelling and show that the bond prices change semi-permanently on the day of the event. The market seems to recalculate appropriate default risk premiums into the credit spreads of the FCS bonds and apply them accordingly until new information is received in the marketplace.
The results of an analysis on the credit spreads of FCS bonds on the event days in a “Credit Spread Test” show that, when evaluated separately, the credit spreads on the day of the event are usually not statistically different than the credit spreads of the FCS bonds on the days within the estimation window. Except for three cases, the bond price movement was not statistically different from the average bond price. This strengthens the assertion that when the event dates are evaluated individually, the reports don’t seem to have a significant effect on bond prices, but when they are looked at together, as the tests above did, the significance of the results is unquestionable.
Evidence was found that suggests that “good news” in the “Agricultural Income and Finance Report” is associated with abnormal drops in the YTM of FCS bonds and “bad news” is associated with abnormal rises in the YTM. Evidence was also found showing the semi-permanent nature of the risk premium changes in the credit spreads of FCS bonds that the information in the reports causes. Further research is suggested, perhaps with other USDA research reports, to determine the robustness and significance of these results.