Industry effects in Firm Profitability Forecasting

A substantial body of research stresses the importance of industry effects on firm profitability and performance. It has also shown that stock returns are closely connected to industry membership.

Yet the view that industry membership is important to understanding firm performance does not go unchallenged. A recent study focused on a large sample of US firms shows that industry-specific models do not improve firm profitability forecasts relative to economy-wide models. It concludes that there is no industry effect in profitability forecasting.

Recognising that these conflicting perspectives exist, this study introduces two novel aspects to the analysis which have been previously unexplored.

Firstly, it considers that many firms are diversified into various industries, with activities generally organised in separate business segments. When segments belong to different industries, no single industry can accurately represent the whole firm. A firm-level industry-specific forecasting model is therefore unable to capture the industry effects in profitability forecasting in these cases. Put differently, the lack of industry effects can be explained by the aggregated reporting of multiple-segment firms at the firm level. However, industry effects in profitability forecasting should reappear when confining the analysis to single-segment firms.

Second, the researchers follow the insights of the forecasting literature to determine a better trade-off between the advantage of economy-wide models (high estimation reliability) and industry-specific models (less estimation bias). To reliably extract industry patterns from the data, the industry classifications have to be sufficiently broad - otherwise industry-specific profitability forecasts are too noisy to accurately predict future profitability.

This paper examines the incremental advantage of industry-specific models. It finds considerable industry effects in profitability forecasting. However, the effects are only visible for focused firms. For diversified firms, aggregated reporting at the firm level prevents the effects from being observed. Furthermore, to reliably extract industry patterns from the data, industry classifications have to be sufficiently broad - otherwise industry-specific profitability forecasts are too noisy to improve forecast accuracy.

The results have a number of implications for academics and practitioners:

• Industry effects in profitability forecasting can be exploited by market participants. A long/short trading strategy based on the firms' industry-specific profitability forecasts can yield significant returns.
• The research findings, such as that information contained in segment-level data can help to improve a firm's profitability forecasts, are relevant to the accounting disclosure literature.

A pre-published version of the research paper is available for download below.