There is a large body of literature that suggests that because of information asymmetries and capital market imperfections, corporate investment expenditures are strongly influenced by a firm’s ability to internally generate cash. In an influential study, Fazzari, Hubbard, and Petersen (1988) show that firms that pay no dividends demonstrate higher sensitivity of investment to cash flow and suggest that investment-cash flow sensitivity reflects the tighter liquidity constraints faced by these firms. There has been subsequent empirical evidence supporting this finding, particularly for firms that are young, small, pay no dividends, have no bond rating. In another prominent study, Hoshi, Kashyap, and Scharfstein (1991) show that investment is much more sensitive to internal funds for independent Japanese firms compared to those that have close ties with banks and are therefore less likely to be financially constrained. In general, a strong positive effect of internal funds on investment has been interpreted as evidence reflecting the difference in the costs of external versus internal financing.
There is also a growing parallel literature that addresses some of methodological problems that were originally outlined in Poterba’s (1988) discussion of the Fazzari, Hubbard, and Petersen (1988) paper. The first strand of this literature questions how firms are classified into financially more or less constrained groups. For example, Kaplan and Zingales (1997), applying an alternative approach to classifying firms into financially more or less constrained, report that the sensitivity of investment to cash flows is not monotonic with respect to financial constraints. In particular, it is the lowest for firms that they classify as being the most likely to be financially constrained. The second strand of the literature addresses issues relating to measurement error in Tobin’s Q. Specifically, if investment opportunities are not measured properly, then cash flows, in addition to conveying information about internal liquidity, may also reflect information about future investment opportunities that are not captured by proxies for Q.