Aggregate Stock Return Predictability with Unadjusted and Adjusted Financial Ratios

Abstract

This study applies and expands the model of Lacerda and Santa Clara (2010) in several ways. It is applied to the USA, the UK and Japan. I use unadjusted and adjusted variations of dividend-price and earnings-price ratios. The sample period is after the Second World War to 2012 using yearly data. On the one hand, the main finding is that the model of Lacerda and Santa Clara (2010) works well for the USA, partially for the UK and does not apply to Japan. I find that results remain stable after applying different specifications of dividend growth rates and time horizons. Also, the adjusted and the unadjusted dividend price multiple for the USA can predict the returns of the UK and Japan. In the case of dividend-price ratio, the adjusted versions outperform the unadjusted version for the USA and the UK. The adjusted version of the dividend-price multiple for the USA and the UK is statistically significant. The unadjusted dividend-price multiple is significant as well for both countries. On the other hand, the unadjusted earnings-to-price is important for the three countries in-sample. Also, the adjusted earnings-price ratio outperforms the unadjusted version only for the UK out-of sample. Finally, the parameters have the correct sign.