R&D Subsidies in Dynamically Competitive Industries

Dagoberto, Giat, Hackman and Peled

R&D Subsidies in Dynamically Competitive Industries

1N. Dagoberto, 2Y. Giat, 3S. Hackman, 4D. Peled

1Monterrey University, Mexico

2Department of Ind.Eng.,JerusalemCollege of Technology, Jerusalem

3ISyE, GeorgiaInstitute of Technology, Atlanta, GA, USA

4Department of Economics, HaifaUniversity, Har Hacarmel, Haifa

ABSTRACT

Government R&D subsidies are evaluated within a computational model of dynamically competitive industries. The focus of analysis is an industry with differentiated products, where firms interact repeatedly over time by choosing their output levels and investments in capacity expansion and various forms of R&D. We find that fiscal incentives stimulate R&D expenditures by more than the subsidy amount, (i.e. positive R&D additionality). More general welfare measures that take into account government, producers and consumers surpluses also exhibit positive returns on average to such subsidies. Interestingly, mid-range R&D intensive sectors typically exhibit higher social returns than either high or low R&D intensive sectors. We also find that R&D subsidy schemes that leave it up to the firms to decide how to use the subsidy, similar to R&D Tax Credits, spur more R&D than support programs such as R&D grants.

1. INTRODUCTION AND LITERATURE REVIEW

It is widely recognized that social returns on R&D activities are higher than private returns, and hence that such activities need to be subsidized in order to compensate for under provision of R&D by free markets, (see Hall et al. [1] for a recent survey). Indeed, many industrialized countries are now offering generous fiscal incentives corporate R&D, and the number of countries offering such programs is growing. However, there is large variation in how such programs are designed and in the extent of their support.Additionaly, there is no consensus on how effective these measures are even for the narrowly defined goal of fostering R&D activities beyond what the companies would have undertaken absent subsidies. More ambitious evaluations of the welfare effects of R&D subsidies on all affected stakeholders – the public at large, the producers, and the government – have never been performed.

Empirical studies that attempt to resolve these issues come up with the following results:

(i)The long-run elasticity of R&D expenditures with respect to its own price is around -1, meaning that a % reduction in the cost to the firm would result in % increase in its R&D expenditures in the long run, (Bloom et al. [2] and Hall et al. [3]).

(ii)The R&D Additionality, i.e. the increase in R&D expenditures per $1 of subsidy is estimated by Lach [4]as $2, and by Lach et al. [5] as $2.28 in manufacturing in general, and $2.81 in certain high-tech sectors.

Studies that examine the effect of competition among firms on their tendency to innovate and spend resources on R&D (see review by Vives [6]) find conflicting results. Particularly, Aghion et al. [7] finds that competitiveness effect on innovation is non-monotone.

Our contribution to this issue is to study the impact of R&D subsidies in a dynamically competitive equilibrium model that allows us to consider heterogeneity in technological intensities of different industries, different intra- and inter-industry competitive pressures, and different designs of R&D support programs. We allow for R&D spillovers across firms, and demand spillovers across markets where consumers optimally allocate their budget between goods from this industry and another good, depending on goods' qualities and prices.

We find that fiscal incentives stimulate R&D expenditure by more than the subsidy amount. More general welfare measures that take into account government, producers and consumers surpluses also exhibit positive returns on average to such subsidies. Interestingly, mid-range R&D intensive sectors typically exhibit higher social returns than either high or low R&D intensive sectors. However, in some cases competing firms in the industry are pushed to perform so much R&D in equilibrium that their own value is hurt by the subsidy. We also find that unrestrictive R&D subsidy schemes (similar to R&D Tax Credits), spur more R&D than support programs such as R&D grants.

2. THE MODEL

Two firms in an industry produce differentiated goods. Each firm can invest in R&D to improveproduct quality and reduceproduction costs, and can also invest in expanding its production capacity. The firms in the industry compete on consumers’ spending with each other, and with all other industries.

Each period, consumers maximize their utility by allocating their income between the industry goods and all the other goods according to the prices and qualities of the goods. R&D and production capacity investments in previous periods determine the current quality of the product and cost structure of each firm. Equilibrium prices and firms' profits each period are determined in a Cournot competition. Firms set their reinvestments for the next period in product quality, cost reductionand capacity, which determine their relative position in the next period competition, and so on. Past R&D investments of each firm spillover to benefit its rival, and the quality of both industry goods influence how much of their income consumers allocate to that industry in the current period.

This competitive interaction is repeated over a finite horizon. Nash equilibrium in reinvestment strategies, including its breakdown to product quality improvement, cost reduction, and capacity buildup, are computed numerically, given parametric choices that determine the costs and productivity of R&D activitities, production function and the extent of the competition among the industry firms. We randomly generated large numbers of such parameters, and computed the equilibrium for each of them, under different subsidy policies.

R&D subsidies come in two alternative forms: (i) an R&D Tax Credit, which refunds a certain percentage of the firm's pervious period qualifying R&D expenditures, but does not impose any restriction on how the firm can utilize the refund;(ii) an R&D Grant, which also refunds the same fraction of qualifying R&D expenditures from the previous period, but mandates that such refunds be used to finance R&D in the next period. We refer the reader to Dagoberto et al. [8] for detailed descriptions of the model and its solution.

3. RESULTS

In the tables below we report averages for the sample of "industries”, which is how we refer to each set of randomly chosen parameter values for which computed the equilibrium. Corresponding to most empirical studies of the effectiveness of R&D subsidies, we stratify such industries by the their so called technological intensity, i.e.the ratio of R&D expenditures to sales in each manufacturing industry. We associate each industry in our sample to one of three equally sized groups (High, Med, Low) according to their equilibrium R&D Intensity (RDI) ratios in equilibrium without subsidies.

Each of the measures below is computed by first summing up the long run equilibrium impact over the planning horizon in each industry, and then averaging these long-run impacts across all industries in the technology intensity subclass.

We look at two broad categories of impacts of R&D subsidies. One includes direct impact on R&D activities, which can be measured or at least estimated in practice. These measures are: R&D Intensity: ratio of R&D expenditures to sale; R&D Additionality: how many additional dollars of R&D expenditures are generated by each dollar of subsidy; R&D Expenditures Elasticity: the elasticity of R&D expenditures to the subsidized cost of R&D; Subsidy Cost: The subsidy cost (outlays) to the government. Table 1 below presents the results of these measures for a 30% subsidy rate, under the two alternative subsidy schemes described earlier.

The second category consists of impacts that are harder to measure, and encompasses much broader welfare measures, attempting to evaluate what are the net benefits to various stakeholders in the society from subsidized R&D activities. The welfare measures are presented in Table 2 and include:Consumer Surplus - a measure of the additional income consumers would need in equilibrium without R&D subsidies to enjoy the same utility they have in equilibrium with subsidies; Producer Surplus measures the change in firms equilibrium values, with and without subsidies; Tax surplus measures the change in government corporate tax revenues, with and without R&D subsidies; Net Social Welfare Gain (NSWG) is the sum of the three measures above minus the subsidy cost.

4. DISCUSSION AND CONCLUSIONS

We find that (1) R&D subsidies produce positive results in every measure we have considered.(2) Commonly considered metrics, such as R&D Intensity, R&D Additinality and R&D Elasticity, tend to be higher in sectors that perform less R&D to begin with. (3) In contrast, welfare measures display an inverted U-shaped pattern with respect to technological intensities. High- and low-technology intensity sectors display much smaller welfare gains from subsidies than medium intensity sectors. (4) R&D Tax Credit appear to be more effective than R&D grants in very measure we have considered. This is surprising given that Tax credit does not force firms to reinvest. (5) Interestingly, subsidies can be “too much of a good thing” to producers due to the competitive pressure among low-tech producers that pushes them to increase their R&D by so much that their value decreases!

In conclusion, we have constructed a model to evaluate the effectiveness of R&D subsidies, which seem to be increasing in scope and popularity as a policy tool in industrialized countries. Our theoretical results almost uniformly indicate that R&D subsidies are effective in fostering R&D, enhance welfare, and are of the same order of magnitude as the scant empirical estimates that were forced to abstract from many of the features that can be controlled for in our theoretical model.

6. REFERENCES

1. B. Hall, J. Mairesse, and P. Mohnen, Measuring the Returns to R&D, NBER Working Paper 15622, December 2009.
2. N. Bloom, R. Griffith and J. Van Reenen, Do R&D tax credits work? Evidence from a panel of countries 1979-1997, Journal of Public Economics, 85, pp.1-31, 2002.
3. B. Halland J. Van Reenen, How effective are fiscal incentives for R&D? A review of the evidence, Research Policy, 29, pp.449-69,2000.
4. S. Lach, Do R&D Subsidies Stimulate or Displace Private R&D? Evidence from Israel, The Journal of Industrial Economics, L(4), p.360-90, 2002.
5. S. Lach, S. Parizat, and D. Wasserteil, The Impact of Government Support to Industrial R&D on the Israeli Economy, Report commissioned by the Ministry of Finance and the Ministry Trade, Industry and Labor, Israeli Government,2008.
6. X. Vives, Innovation and Competitive Pressure, The J. of Ind.Econ., LVI, 419-69,2008.
7. P. Aghion, N. Bloom, R. Blundell, R. GriffithandP. Howitt, Competition and Innovation: An Inverted-U RelationshipThe Quarterly J. of Economics,120, pp. 701-28, 2005.
8. N. Dagoberto,Y. Giat, S. Hackman, D. Peled, R&D Subsidies in Dynamically Competitive Industries, Working Paper, 2010.

1