Weekly returns of stock portfolios exhibit substantial autocorrelation. Analytical studies suggest that nonsynchronous trading is capable of explaining from 5% to 65% of the autocorrelation. The varying importance of nonsynchronous trading in these studies arises primarily from differing assumptions regarding nontrading periods of stocks. We simulate the effects of nonsynchronous trading by sampling stock returns from a return generating process using transactions data to obtain the precise

more »

... of each stock's last trade. We find that simulated weekly portfolio returns exhibit autocorrelations that are roughly 25% that of their observed (CRSP) weekly returns. It is well documented that observed short-horizon returns of portfolios exhibit positive autocorrelation. For example, Boudoukh, Richardson, and Whitelaw (1994) report first-order autocorrelations of .23 for weekly returns of an equally weighted index and .36 for weekly returns of a small-stock portfolio. 1 The source of the autocorrelation, however, remains the subject of much debate. Explanations include nonsynchronous trading, market-maker inventory control, transactions costs, time-varying expected returns, and market inefficiency. 2 Nonsynchronous trading -the tendency for prices recorded at the end of the day to represent the outcome of transactions that occur at different points in time for different stocks -is perhaps the most widely recognized source of autocorrelation in portfolio returns. To the extent that nonsynchronous trading is responsible, the autocorrelations are an artifact of the data sampling process and have no economic meaning. There We are grateful for many helpful comments from