Effects of Nonlinearity and Multi-Period Lags on Basu (1997) Measure

This paper examines the impacts of two attributes of accounting recognition, i.e., nonlinearity and multi-period lags, on Basu (1997) measure of asymmetric timeliness. First, we predict and find that asymmetric timeliness is affected by nonlinear earnings responses and more pronounced when news of large absolute magnitudes occurs. Results also show that the time trend in asymmetric timeliness may be confounded by nonlinearity. Next, we examine lagged earnings responses to good news and bad news and hypothesize that positive differential earnings responses exist in the current and sufficiently short following periods, indicating differential earnings responses do not only exist concurrently. We also predict differential earnings responses turn negative, showing positive differential earnings responses reverse as lag increases. Besides, we hypothesize that differential earnings responses decay as lag increases. Empirical results are all consistent with our expectations. Our results are robust to several changes in model or sample specifications. Possible methods to compare the extents of conservatism when considering nonlinearity and multi-period lags are also provided in our study. Collectively, our empirical results suggest that nonlinearity in earnings responses and multi-period earnings responses rather than a single-period asymmetric timeliness measure should be considered to measure conservatism more validly.