Long-term over-reaction in the UK stock market and size adjustments

Publisher Website
Google Scholar
Abstract
This study tests for long-term reversals in the abnormal returns of UK companies classified as ‘winners’ and ‘losers’ over the period from January 1979 to December 1990 using publicly available data from the London Business School (LBS) Risk Measurement Service. In the first part of the study we find that in the 12 months following portfolio formation ‘loser’ companies continued to experience negative abnormal returns and ‘winner’ companies persisted in generating positive abnormal returns, thus appearing to contradict the findings of US studies which support the ‘winner-loser’ effect. The possible influence of firm size was examined by splitting the ‘winner’ and ‘loser’ portfolios into groups based on equity market capitalization. It was found that the very smallest ‘loser’ companies did experience a reversal in their abnormal returns over the following 12 months, but that no such reversal existed for the smallest ‘winner’ companies. When the study was extended to cover the five-year period following portfolio formation, it was found that a reversal in the abnormal returns of winner and loser portfolios was experienced over each of years 2-5, thus lending support to the winner-loser effect. Returns of sufficient magnitude were generated from the strategy of short-selling winners and buying losers over this extended period to suggest that the winner-loser effect is an exploitable anomaly. The excess returns remained when companies were matched with control portfolios of similar size, thus suggesting that the long-term ‘over-reaction phenomenon’ in the UK is not simply a manifestation of the well documented size effect. However, this latter result is also a function of the length of sample selection period. Finally, we report the results of tests which suggest that the phenomenon is seasonal in nature, with the most significant return reversals occurring in January and April for small companies in the loser portfolio. The latter result provides support for the tax-loss selling hypothesis.

This publication has 31 references indexed in Scilit: