acrual models.pdf

According to the neoclassical theory of investment, if firms' accruals are a form of short-term investment they should be greatly influenced by the shadow price of capital, namely Tobin's q. In the presence of financial market imperfections, cash-flows should also impact accruals since they proxy for liquidity constraints. In this paper, we test a new version of the cash-flows augmented accrual model featuring a proxy for Tobin's q, and compare it to the most common models found in the literature. To deal with the measurement errors often encountered in accounting data and Tobin's q empirical proxies, we rely on a modified version of the Hausman artificial regression, and find that all the key parameters of the accrual models are indeed systematically biased with measurement errors. More importantly, our findings largely qualify the accrual investment perspective, as both cash-flows and Tobin's q are found strongly significant regressors of firms' accruals. Interestingly, we find that the Tobin's q augmented model is able to isolate discretionary accruals, and to deliver residuals quite close to zero on average.