Is the Overproduction Proxy a Valid Measure?

Abstract

Roychowdhury (2006) constructed the measure of abnormal production costs to detect overproduction behaviour, which is widely used as a valid proxy in real earnings management. However, there is little evidence about its validity. This study is intended to fill the gap by focusing on the measure of abnormal production costs and examining its validity. First, the model proposed by Roychowdhury (2006) imposes restrictive linearity assumption about the association between COGS and sales. But in fact, the linearity assumption is not suitably specified. We use analytical methods to derive the non-linear relationship between COGS and sales and find omitted variables in his original model. Then we illustrate why the original production costs model suffer severe model misspecification. Second, after relaxing the linearity assumption we re-calculate abnormal production costs and find that small profit firms no longer exhibit high abnormal production costs compared to their counterparts. These results indicate that above-mentioned misspecification lead to severe biased estimate and previous conclusions are driven primarily by misspecification. Overall, we conclude that the overproduction proxy is not a valid measure.

Keywords: Abnormal Production Costs; Production Costs Model; Misspecification; Validity