Business cycles and research investment

Abstract

This paper examines the impact of uncertainty and the business cycle on US high-tech manufacturing firms’ research investment. Although the reliance on internal financing suggests firms will consider uncertainty and the business cycle when determining their research budget, little is known about how the business cycle and uncertainty influence research investment. Using firm-level data on sales, cash flow, and industry-level indicators of the business cycle, this paper finds that the firm's response to the business depends on the firm's industry and the industry's current location in the business cycle. The data also shows that the business cycle also depresses the firm's reaction to changes in sales and cash flow. Uncertainty clearly reduces research efforts, although non-linearly.

Notes

¹ Given the monthly data, the Hodrick–Prescott smoothing parameter is set at 14 400. The business cycle variable allows each firm to experience business cycles distinct from the aggregate national level cycle. The advantage of this business cycle variable over firm-level indicators, such as stock returns, is that this variable is exogenous to the firm yet is something firms may observe and consider when determining R&D budgets. While some studies interpret firm sales as a measure of demand conditions, the business cycle variable used in this paper adds information beyond that contained in sales. The correlations between firm-level sales and industry-level business cycle variables range from a high of 0.91 to a low of −0.88 with a median correlation of 0.19.

² Bloom et al . (2001) impose a similar restriction in their examination of uncertainty on physical capital investment.

³ Peiro (2004) questions whether the business cycle is asymmetric using a data from seven European countries. Note that this paper only allows for an asymmetric response to the business cycle, and does not require that business cycle itself be asymmetric.

⁴ Consistent with the literature, R&D investment is used in the estimation instead in place of R&D capital stock (see Bond et al ., 2003).

⁵ The cash flow variable is net of R&D. Some studies use gross cash flow, created by adding research expenses back into reported cash flow to reflect the cash available for research and other activities, based on the argument that firms expense rather than capitalize research. In this study, the estimated coefficients on gross cash flow were smaller and less statistically significant than the net cash flow coefficients. Whether net cash flow or gross cash flow is used did not qualitatively affect the other estimated coefficients.