Investment behavior, observable expectations, and internal funds: a comment on Cummins et al. (AER, 2006)

Cummins et al. (2006) construct a new measure of fundamentals, and show that the positive cash flow effects typically found in investment-Q models disappear when traditional Q is replaced with their new measure. Their results are not robust to small changes in their specification or in the dataset used to estimate their model. The explanatory power of cash flow does not disappear when replacing traditional Q with their new measure of Q; it is never there to begin with. Investment’s lack of sensitivity to cash flow may be because their data is biased towards firms with positive cash flow (it is negative for only 242 observations of 11431). This bias and our results mute their argument that the positive cash-flow effects obtained in such models may reflect a failure to control properly for fundamentals rather than the presence of financial constraints.