Business fixed investment spending: a critical survey of modeling strategies, empirical results, and policy implications

This study has offered a critical review of the literature on business fixed investment spending, and has assessed the current state of knowledge and future research agenda. To place some structure on this vast literature, the survey has been organized according to two principles. The first sorted models by whether dynamics were introduced into the econometric equation implicitly or explicitly. Benchmark models were developed for all of the models discussed in this survey. The second organizing principle focused on the four important issues (listed in Section I) that have been faced repeatedly by investment researchers. A number of these issues have been addressed reasonably well, and most recent models are theoretically consistent and isolate the effects of expectations and technology on the econometric equation. Such success has been purchased partly by maintaining a number of uncomfortable restrictions, and the research agenda aims to expand our view of the firm and the margins along which it operates. A final issue concerns the relative importance of prices vs. quantities vs. shocks as determinants of investment. While there is clearly no uniformity in the results and the role of shocks remains to be assessed, it appears to this author that, on balance, the response of investment to prices tend to be quite small and unimportant relative to quantity variables. The fundamental problem facing the applied econometrician is how to generate and interpret econometric evidence when the available data are non-experimental and have limited and noisy variation. The most direct solution would be to obtain better data, but collecting comparable data for firms is a rather difficult and expensive task. An alternative research strategy would use sophisticated statistical techniques to attempt to correct for various difficulties. Since these procedures are based frequently on large samples of spotlessly measured data, doubts exist about their usefulness for applied work. Statist(This abstract was borrowed from another version of this item.)