Direct Tests of Index Arbitrage Models

Previous tests of stock index arbitrage models have rejected the no-arbitrage constraint imposed by these models. This paper provides a detailed analysis of actual S&P 500 arbitrage trades and directly relates these trades to the predictions of index arbitrage models. An analysis of arbitrage trades suggests that i) short-sale rules are unlikely to affect the cash-futures mispricing, ii) the opportunity cost of arbitrage funds exceeds the Treasury bill rate, and iii) the average price discrepancy captured by arbitrage trades is small. Tests of the models provide some support for a version of the arbitrage model that incorporates an early liquidation option. The ability of these models to explain arbitrage trades, however, is surprisingly low.