The Role of a Latent Value-Relevant Measure in Tracking and Predicting Stock Returns: A FAVAR Approach

Abstract

This paper attempts to estimate and study the role of 'other information', as posited in the residual income valuation model of Ohlson (1995), for tracking and predicting future returns of the S&P 500. 'Other information' is an unobserved variable and defined as a summary of value-relevant information about events and their effect on future profitability, which is captured in a company's current stock price and returns, but not yet reflected in a company's current financial statements. This suggests a potential to predict subsequent returns. Previous literature has found that traditional valuation metrics (e.g. B/P, E/P, and D/P ratios) have poor predictive power. In this study, we apply a factor augmented vector autoregression (FAVAR) to estimate this value-relevant latent variable and assess its predictive performance. The FAVAR is a suitable model because it enables us to analyze and quantify the linkages of stock market value, profitability, and unobserved factors that are broadly captured by big data. We use a two-step principal components estimation approach to extract the unobserved factors of 78 informational variables from financial market, accounting, investor and consumer sentiment, and macroeconomic data. Our analysis shows that, in comparison to competing measures, the estimated latent value-relevant variable can track contemporaneous stock returns and has statistically reliable power to predict both future real stock returns and excess returns over a Treasury Bill rate, both in- and out-of-sample.