Unconventional Stock Market Predictors And the Unstable Performance of Traditional Stock Market Predictors

Abstract: Stock market predictability is a contradictory topic within research in finance. Despite the fact that numerous articles presenting evidence on stock market predictability have been published in the last two decades, Goyal and Welch (2008) argue that most previously suggested models have performed poorly out-of-sample during the last 35 years or so. I confirm these results, and highlight the years around the IT-bubble as a particularly problematic period for traditionally used predictors. Furthermore, I have in this paper examined the performance of some unconventional predictors which measure aspects such as investor sentiment, consumer confidence, development in the manufacturing sector, and early indications of business cycle fluctuations. Analyzing the period of 1978-2010 in search of predictive ability of 1-month, 1-year, and 5-year returns I find that the unconventional predictors in general provide better results both in-sample and out-of-sample, compared to the traditional predictors. Two particular variables stand out - a leading indicator of the business cycle and a consumer confidence index.