Capturing the Impact of Unobserved Sector-Wide Shocks on Stock Returns with Panel Data Model

Unobserved sector-wide common shocks cause the issue of cross- sectional dependence (CSD) in panel data modelling of stock returns. In this study we apply two econometric techniques: the seemingly unrelated regression approach and a Bayesian estimator for panel data models with factor structural errors, to allow for CSD within a particular sector. By applying these models to monthly stock returns of S&P100 companies from six sectors over 10 years, we can capture and measure the heterogeneous impacts of not only observed individual company accounting fundamentals and market- wide common shocks, but also unobservable sector-wide common shocks. Results from the empirical study show that the impacts from both observed factors and unobserved sector-wide common shocks vary markedly across companies. After controlling for observed accounting fundamentals and market-wide common factors, a considerable proportion of the variation in stock returns can be attributed to unobservable sector-wide common shocks.