PERMANENT VERSUS TRANSITORY COMPONENTS OF ANNUAL EARNINGS AND ESTIMATION ERROR IN EARNINGS RESPONSE COEFFICIENTS

Previous research has generally estimated unexpected earnings as the change relative to the previous year, assuming that shocks to annual earnings are purely permanent. In the presence of transitory components of annual earnings, we predict that using earnings changes as a proxy for unexpected earnings causes earnings response coefficients to be understated and the estimation error to be negatively cross-sectionally correlated with persistence. The negative correlation causes the association between earnings response coefficients and persistence to be overstated. We use the IMA (1, 1) model to capture transitory components of annual earnings and obtain results that are consistent with our predictions.