The Out-Of-Sample Prediction of Annual Operating Cash Flow: a Comparison of Regression and Naïve Forecast Models

This study proposes that a no-change naive forecast of (operating) cash flow is as accurate as time-series (firm-specific) and cross-sectional regression forecasts of cash flow. The study first demonstrates that cross-sectional regression forecasts of cash flow with firm-size controls are as accurate as time-series regression forecasts. Next the study confirms the expectation that a naive forecast is as accurate as the regression model forecasts. Finally, the study identifies apparent misapplications of Theil's U-statistic, which overstate the ability of regression forecast models to outperform a naive forecast model.