The Relative Performance of In-Sample and Out-of-Sample Hedging Effectiveness Indicators

Hedging effectiveness is the proportion of price risk removed through hedging. Empiricalhedging studies typically estimate a set of risk minimizing hedge ratios, estimate the hedgingeffectiveness statistic, apply the estimated hedge ratios to a second group of data, and examinethe robustness of the hedging strategy by comparing the hedging effectiveness for this "out-of-sample"period to the "in-sample" period. This study focuses on the statistical properties of thein-sample and out-of-sample hedging effectiveness estimators. Through mathematical andsimulation analysis we determine the following: (a) the R2 for the hedge ratio regression willgenerally overstate the amount of price risk reduction that can be achieved by hedging, (b) theproperly computed hedging effectiveness in the hedge ratio regression will also generallyoverstate the true amount of risk reduction that can be achieved, (c) hedging effectivenessestimated in the out-of-sample period will generally understate the true amount of risk reductionthat can be achieved, (d) for equal numbers of observations, the overstatement in (b) is less thatthe understatement in (c), (e) both errors decline as more observations are used, and (f) the mostaccurate approach is to use all of the available data to estimate the hedge ratio and effectivenessand to not hold any data back for hedge strategy validation. If structural change in the hedgeratio model is suspected, tests for parameter equality have a better statistical foundation that dotests of hedging effectiveness equality.