Graph by Shutterstock.com
In Cummins, Hassett, and Oliner (2006), we assessed the influence of cash flow on investment when controlling explicitly for measurement error in Tobin's Q. We showed that the cash-flow effects often found in other studies disappeared after controlling for this measurement error. We also showed that a measure of Q based on analyst forecasts of firm-level earnings does a better job of capturing the fundamentals that drive investment spending than does the usual measure of Q based on stock prices.
Agca and Mozumdar (2012), henceforth AM, assert that errors in our data and econometric methodology are responsible for both of our key findings. AM identify three alleged problems with our work: (a) that we may have misdated the cash-flow series in our dataset, (b) that the set of instruments we use is too restrictive, and (c) that our analyst-based measure of Q is flawed.
As discussed below, we do not find AM's critique to be compelling. First, AM provide no convincing evidence to support the claim that our cash-flow series was misdated. Second, even when AM modify our data to correct this alleged problem, use their preferred instruments, and substitute a different analyst-based measure of Q, their results do not overturn the central findings in our paper. We cannot emphasize this point strongly enough: When AM use our dataset, modified as they see fit, they do not reverse our results. Only when they construct their own dataset do their results differ substantially from those in our paper. Those results may be of interest in their own right, but they do not constitute a rebuttal of our findings.