A Survey on Lead-Lag Effect on Small and Large Size Portfolios in Tehran Stock Exchange

In inefficient markets, returns are not distributed normally and they have serial correlations. It is obvious that the price changes are not independent, so there is a pattern in price changes which help investors to gain unusual benefits. One of the patterns which are concerned with an inefficient market is the lead-lag effect. This research investigates the existence of this effect between small and large size portfolios in Tehran Stock Exchange during the period of 2011-2017. This relationship was examined both in short-run by using cross-correlation approach and vector auto-regressive model and long-run by employing cointegration methodology. Cross-correlation matrices imply that there is a lead-lag effect in short-run. Existence profiles and variance decomposition are used for further validation, the results show that all of the shocks were fully absorbed after three weeks but there is no pattern for big portfolios indicating that they absorbed the shocks more rapidly than small portfolios, and also overreaction is observed only in one out of two small portfolios of the sample.With confirming the existence of lead-lag effect in long-run by Cointegration approach, the ability of ECM models in out of sample forecasting is concerned which is measured by root mean squared error and Wilcoxon's signed-rank test. However, the results indicate that the error correction model has superior forecasting performance relative to models without the error correction terms but the Wilcoxon's signed-rank test does not reject the null hypothesis that the two RMSEs are the equal