Although precipitation is also an important risk factor in crop yields, we restrict
analysis to temperature derivatives due to the higher potential for liquidity in temperature
derivatives. For instance, from October 14, 1997 to April 15, 2001, temperature
derivatives represented over 98% of all WDs (Brockett, Wang and Yang 2005). The use
of temperature derivatives may not be a major shortcoming as atmospheric flow patterns
that control much of the North American climate tend to be persistent (Namias 1986). In
particular, during extreme drought events—those most likely to result in widespread crop
losses—this persistence phenomenon causes heat and precipitation conditions to interact
causing a self-perpetuating event. On a large scale, average temperature and precipitation
conditions for a given region are likely highly negatively correlated in extreme events.
Figure 2 displays aggregate detrended Illinois state corn yields for 1971-2002,
with the x-axis ordered by summer season ACDD’s. The hottest years, those in which
ACDD’s exceeded approximately 900, corresponded roughly to the driest years. In fact,
four of the five hottest years were also drought years. Furthermore, all droughts
corresponded to temperatures in excess of 900 ACDD’s. Thus, it appears that
temperature derivatives may act as a suitable substitute in hedging precipitation risk when
it is most needed. The use of an accumulated index is further motivated by the fact that
in the U.S. corn growing regions, month-to-month temperatures are typically
autocorrelated (Jewson and Brix 2005).
Hedging yield risk with WDs becomes a difficult problem for two reasons. First,
there is a high degree of yield variability that cannot be attributed to potentially tradable
weather indexes. For instance, while yields are systemically related to summer ACDD’s,
there is still considerable yield variability that cannot be ascribed to ACDD’s. For
12