New research reveals a conundrum for public innovation policy: while it might be socially desirable to support investment in basic research during economic downturns and in innovation by large and old firms with a history of success, public support is less likely to generate additional investment in research and development (R&D) under these conditions.
Analysing data on 43,650 British firms over the period 1998-2012, the study by Mehmet Ugur and Eshref Trushin finds that the effect of R&D subsidies on business investment varies by firm type and R&D type. On the bright side, R&D subsidies generate R&D ‘additionality’ among small and young firms, and firms further away from the innovation frontier in the industry. But these firms receive only 2-15% of the subsidy allocations.
In contrast, R&D subsidies have no effect on R&D investment by older and larger firms, and those closer to the innovation frontier – firms that accounts for 85-98% of the subsidies. Furthermore, R&D subsidies are less effective in generating additionality when the investment is in basic research or undertaken during crisis episodes.
The case for public funding of business R&D is based on the existence of market failures that drive a gap between the actual and socially desirable levels of R&D investment at the firm, industry, or country levels. One type of market failure results from knowledge spillovers and return uncertainty that reduce the incentives for business investment in R&D. Another type reflects capital market imperfections, which leave innovative firms financially constrained. Therefore, governments have tended to support business R&D through direct measures such as subsidies and indirect measures such as R&D tax credits.
Nevertheless, economic theory also suggests that firms differ with respect to the R&D gap they face. The gap depends on the balance between the investment-deterring effects of market failures and the investment-inducing effects of the competition in product or technology markets.
Therefore, one aim of this study is to establish whether subsidy allocations in the UK have been effective in generating R&D investment additionality. A second aim is to identify the firm types and R&D investment types where additionality effects are more or less likely. A third aim is to verify if subsidy allocations are optimal – that is, if the R&D subsidies are allocated to firm and R&D types where additionality effects are more likely.
Data and method
The study analyse Office for National Statistics (ONS) data for 43,650 British firms observed from 1998 to 2012. The data are from the Business Research and Development Database (BERD), which contains survey data for a sample of R&D-active firms that account for about 80% of the R&D undertaken by UK firms.
The researchers estimate the effect of subsidy on business R&D investment through an entropy balancing (EB) methodology that minimises the dissimilarity between the probability distributions for the subsidised and non-subsidised firms. The distributions of the two firm types are balanced along 139 characteristics at the firm, industry and country levels – including the changes in the R&D tax regime that also affect business R&D investment.
The EB methodology is demonstrated to outperform the propensity score matching used in similar literature. The advantages include consistency with a doubly-robust estimator where both the selection and outcome models are specified correctly; smaller bias than conventional doubly robust estimators; higher precision; full use of the information in the non-subsidised sample; and better performance that propensity-score-based matching estimators.
The effect of R&D subsidies on business R&D investment varies by firm type and R&D type. On the bright side, R&D subsidies generate R&D additionally among small and young firms and firms further away from the innovation frontier in the industry. Nevertheless, these firms receive only 2-15% of the subsidy allocations.
In contrast, R&D subsidies have no effect on R&D investment by older and larger firms and firms closer to the innovation frontier. Yet the latter accounts for 85-98% of the R&D subsidies during the analysis period. Furthermore, R&D subsidies are less effective in generating additionality when the investment is in basic research or undertaken during crisis episodes.
These findings reveal a set of conundrums for public innovation policy: it is socially desirable to support investment in basic research, during economic downturns, and innovation by large and old firms with a history of success; but public support is less likely to generate additional R&D investment under these conditions.
INFORMATION ASYMMETRY, RISK AVERSION AND R&D SUBSIDIES: EFFECT-SIZE HETEROGENEITY AND POLICY CONUNDRUMS