Testing the uncertainty–investment relationship using survey data on capital stock disequilibrium

Abstract

This article uses unique survey-based data that record the extent of positive and negative disequilibrium in capital stock at industry level. Change in these disequilibria are hypothesized to take account of planned and revised targets, and the influence of uncertainty on adjustment. We find that increased uncertainty slows the adjustment of fixed capital towards equilibrium levels. That is consistent with the predictions of real options theory and partial irreversibility models.

Notes

¹There have been some attempts to investigate real option theory by exploiting predicted differences between the investment responses of different categories of firm or between firm-level and industry-level responses (see, e.g. Henley et al. (2003); Driver et al. (2006, 2008)).

²A further complication arises when different option types are distinguished. The option to expand and the option to defer may require oppositely signed adjustments to the Jorgenson cost of capital (Abel et al., 1996). Firm surveys show that firms fear both under and over-investment (Driver and Whelan (2001)).

³Our data are industry-level, which is not ideal as aggregation would be expected to smooth out some real option effects. Nevertheless it is shown in Caballero and Engel (1999) that nonlinearities are still a feature of industry and aggregate investment.

⁴If the distribution of actual capacity distribution across firms is assumed to be approximately normal, a logit transformation of the data results in a linear proxy for the utilization variable.

⁵As in Blanchard and Fischer (1989) we will represent the disequilibrium as the gap between capital and expected demand. We abstract here from all strategic and corporate governance influences [Driver (2008); Driver and Temple (2008); Krafft et al. (2008); Pitelis and Teece (2009)].

⁶Note that we are not engaging here with cross-industry variation in the level of the markup; our data refer to rates of change in markups. Although profitability is properly measured by a combination of markup and the capital-output ratio, short-run movement in the latter is already captured by the CU variable.

⁷Given that the CU term is measured from the survey data as the % below capacity, we express it as a logit rather than a log. Note also that the κ term in (7) may be treated as a cyclical term.

⁸See Driver et al. (2006, 2008) for other empirical applications of this measure.

⁹The results here may suggest that in the heterogeneous industries, some of the dispersion is reflecting structural change that is positive for investment and is counteracting the negative effect of uncertainty.