Abstract
This paper investigates autocorrelations of market and industry returns in both short- and long-term holding periods in the Colombo Stock Exchange using a longer sample period to test the efficiency of stock prices. Daily stock market returns are reliably positively autocorrelated in the order of 43 percent. Daily returns are more predictable in rising markets and for large industries such as Bank Finance and Insurance, Manufacturing, Diversified, Plantations, Beverages Food and Tobacco, and Hotels and Travels. Monthly stock market returns have a reliable positive autocorrelation of 21 percent, and for most industries monthly autocorrelations are larger than daily autocorrelations. Monthly returns are also more predictable in rising markets. Consistent with observed price reversals in the Sri Lankan market, the long-horizon returns exhibit large negative autocorrelations suggesting not only that they are predictable but also that they contain a large mean-reverting component. The average predictable variation in 1-4-year returns is 56 percent for market returns, and 49 percent for industry returns. These results, consistent with previous findings, reliably reject the random-walk behavior of prices and the weak-form market efficiency in the Sri Lankan stock market