The Profitable Model Failure: Using Forecasts to Settle Temperature Derivative Contracts

The trading of derivative contracts is used by businesses to hedge against lower-impact weather risk which causes large volatility in annual profits. A common example is a swap contract where the payoff increases or decreases linearly with the number of heating or cooling degree days. Settlement using observed temperatures addresses the weather's effects on demand, but not on cost, which can be substantial for energy companies. This issue will be most significant when large model failures occur, such as happens repeatedly in the chinook belt of the United States.

A case study examines the possible benefits of settling these contracts using forecast temperatures. First, a sample swap is created for February HDD's in Billings, MT based on climatology. In February 2019, this station accumulated 980 HDD's, which is near the point of zero payoff. Nearly one-fifth of this total was from a four day period on 4-7 February where only a brief chinook on the 5^th pushed temperatures above freezing and overnight lows were all in the single digits. However, the GFS MOS forecast for these days spread this chinook over a much longer period resulting in five over-forecasts of 10+ degrees F and nearly a 20% drop in HDD totals. A derivative contract using the forecast temperatures would result in a $180,000 payoff increase, which compares favorably with possible losses on the energy trade market, but other considerations on the use of forecasts settlements in general are also discussed.