Use of Fuller's Technique to Reduce Measurement Error in the Returns/Earnings Association

Empirical estimates of the earnings response coefficient have consistently been lower than theory predicts. This may be because empirical proxies for unexpected earnings contain measurement error. I demonstrate and evaluate the use of a recently developed technique by Fuller that yields consistent parameter estimates in the presence of measurement error. The empirical results indicate that this technique is successful at mitigating measurement error bias in the earnings response coefficient. The earnings response coefficient increases by as much as 52%. In contrast, replication of the techniques performed in previous studies increases the earnings response coefficient by only 8%.