The predictive content of financial ratios and macroeconomic factors for stock return in the EU Stock Market

Abstract

This master’s thesis studies stock returns forecasting power of microeconomic and macroeconomic variables for European listed companies. Listed companies are divided into six industries and we conduct an in-sample estimation with the Lasso and Elastic Net regression for α=0.5 and α=0.25 in order to compare the selection and the in-sample performance for the models built with the two regularization techniques. In a second step, we study the out-of-sample accuracy of the created models with several statistical tools with two time periods: one including the Covid-19 pandemic period and the other one without this time period. The results showed an insignificant relationship between stock returns and the predictive variables for both the training and testing data. Two possible explanations are either there a linear regression cannot forecast stock return as it is too volatile, or it is due to the huge number of outliers in our dataset. In conclusion, the return on equity, return on assets, net profit margin, debt-equity, earnings per share, price-to-earnings, earnings yield, dividend yield, dividend payout, book-to-market, inventory turnover, quick ratio, current ratio, inflation, long term yield and the GDP growth rate do not have prediction power on stock returns for European listed companies in the last decade.