In spite of the increasing availability of high-quality data and the possibility of obtaining direct commodity trading costs in recent periods, historical series of transaction costs still require bid-ask spread estimation. In this work, we verify whether the popular high-low spread estimator performs well in commodity markets so that it can be used to construct long-term cost estimates. We find that the estimator suffers both from measurement error increasing in the volatility-to-spread ratio and consistently positive error in a variety of empirical and experimental settings. As the measurement error in the high-low estimator depends on ex-ante knowledge about the usually unobserved true spread level, we conclude that the spread measure is not well-behaved and should be avoided in commodity markets.
【 预 览 】
附件列表
Files
Size
Format
View
The high–low spread estimator is not well–behaved in commodity markets