In order to establish a theory which can be tested empirically with existing macroeconomic data, the neoclassical investment literature assumes that a firm’s capital stock is homogenous, lasts forever and depreciates at a constant rate which is unrelated to economic conditions. These assumptions homogeneity, stability, and exogeneity are critical for estimating the determinants of investment and are by no means innocuous. In practice, investment data concern only capital expenditures gross investment rather than the theoretically appropriate net investment. The difference is depreciation and capital retirement. To the extent that the conventional literature is correct to assume depreciation and capital retirement are exogenous and constant, fluctuations in gross investment are the same as fluctuations in true net investment. If, however, the assumptions about depreciation are not valid and economic factors matter, especially if the factors are the same ones that influence capital expenditures, then the impact that they have on gross investment (estimated in the conventional literature) may be quite different from their impact on true, net investment. This problem will also influence measured productivity.

When capital is heterogeneous it is difficult to believe that depreciation or retirement are constant and exogenous. As discussed in an early literature by Feldstein and Rotschild (1974) and Feldstein and Foot (1971), constant depreciation with differing rates across goods can create “lumpiness” in investment and “echoes” of past investment in future decisions. Boddy and Gort (1971) noted early that there is evidence of capital heterogeneity across sectors and more recent work by Goolsbee and Gross (1997) has demonstrated how important capital heterogeneity for the study of investment at the micro level.