this can be effectively managed using WDs. The ES measure leads to similar
conclusions. For instance, the full-sample results indicate an ES 6% increase of $59.91
for the State portfolio, versus $46.57 for the averaged district portfolios.
The hedging effectiveness results were also stronger for the “State (Aggregated)”
portfolio than for any of the individual districts. For instance, Table 2 shows that the
reduction in RMSL was greater for the State portfolio, 43.31%, than for any of the
individual district portfolios, the next closest being 41.76% for D90. Also, the hedging
effectiveness for the individual districts varied widely across districts with reductions in
RMSL ranging from 41.76% for D90 to 13.81% for D80.
The out-of-sample results lead to similar conclusions.17 The out-of-sample
estimates for the 2nd half (Table 3) of the sample period show reductions in RMSL of
25.66% for the State portfolio, versus 16.85% for the averaged district portfolios. The
change in ES as well as the level of ES was greater for the State portfolio in all out-of-
sample cases. For instance, Table 3 shows that the ES 6% (9%) was $291.85 ($296.17)
for the State portfolio versus $268.41 ($279.99) for the averaged district portfolios. On
average, the hedging effectiveness for the out-of-sample results in this study, which
employs simple seasonal temperature contracts, are comparable to those obtained by
VB’s (2004) analysis which employs complex combinations of monthly precipitation and
temperature derivatives. This suggests that although substantial amounts of yield risk can
be hedged using WDs, the marginal risk of overfitting weather hedges increases
substantially as more complex instruments are employed.
The findings suggest that aggregating individual production exposures has the
effect of reducing idiosyncratic yield risk, leaving a greater proportion of the aggregated
19