We construct a novel measure of the closing of short positions using the identity:
Short Covering_t−1 = Short Volume_t−1 − (Short Interest_t − Short Interest_t−1 ).
Our measure provides the first empirical evidence on short covering in U.S. equities which allows us to estimate the profitability of short sales. We find that short transactions are profitable on average, however they exhibit poor Sharpe ratios. Further, we find that periods with high short covering are associated with worse price efficiency. Our findings help reconcile existing evidence that short sellers are informed traders with the poor performance of short-biased hedge funds.