Throughout the world, proposals to reduce carbon emissions include increased research and development (R&D) funding. President Obama proposes spending an additional $150 billion on energy R&D over the next ten years, compared to current funding levels of about $5 billion per year. A 2004 report from the bipartisan National Commission on Energy Policy recommended doubling US government energy R&D spending, while others advocate increasing it by a factor of five or ten, likening the need for new energy technologies to the Manhattan Project’s efforts to create a nuclear weapon (Kammen & Nemet, 2005). A potential concern with such proposals is that dramatic increases in energy R&D may come at a high cost, as these research efforts may draw away research funding and scientists from other productive sectors (Popp, 2006, Yang & Oppenheimer, 2007, Schneider & Goulder, 1997).
Recent efforts to endogenise technological change in integrated assessment models of climate policy highlight the need to properly account for the opportunity cost of environmental R&D investments. Because the social returns to R&D investments are typically higher than the social returns to other types of investment, any new environmental R&D that comes at the expense of other R&D investment will dampen the cost-savings potential of induced technological change. In Popp (2004), I show that assumptions about the magnitude of such crowding out explain much of the variation in results across integrated assessment models with induced technological change. For example, among models that use R&D expenditures as the means by which technology improves, Nordhaus (2002) assumes a fixed supply of R&D labour, so that new energy R&D completely crowds out other R&D. As a result, he finds induced technological change has little effect on welfare or climate policy. In contrast, Buonanno et al. (2003) model energy R&D and other R&D as complements, so that crowding out does not occur. Their model finds a stronger effect for policy-induced R&D than other models.
Where does energy R&D come from?
Nonetheless, there has been little empirical evidence asking whether or not new energy R&D comes at the expense of other R&D. In Popp and Newell (2009), we examine this question at both the industry and firm level. We begin at the industry level, asking whether investment dollars flowing to energy-related research are coming at the expense of overall R&D in specific sectors. We find no evidence that increased economy-wide energy R&D efforts draw resources away from sectors not actively performing energy R&D. However, the data are not strong enough to allow definitive conclusions within the energy R&D sector – that is, whether energy R&D represents new R&D spending, or shifts R&D efforts away from other projects performed by these firms.
To address this second question, we turn to firm-level data, linking financial data with patent data for publicly-traded firms in the automotive and alternative energy industries. Because firm-level R&D data are not broken down by field, we use patents as a proxy for R&D effort, asking whether increases in clean energy patents lead to decreases in patents for other technologies. We focus on large firms with diverse research portfolios, as the question of firm-level crowding out is not relevant for companies specialising in a specific energy technology (e.g. International Fuel Cells Corp).1
Figure 1 shows alternative energy and non-energy patents for the firms in our sample. Note that alternative energy patents peak during the energy crises of the 1970s, while other patents are at their lowest. Afterwards, energy patents decline, and non-energy patents increase, suggesting possible crowding out. Our regression results confirm this crowding out in the alternative energy industry, where one new alternative energy patent results in one less non-alternative energy patent. Moreover, research efforts by firms in this industry appear to be financially-constrained, as increased sales revenue also leads to more patenting activity. In contrast, crowding out does not appear to be a problem within the automotive industry.2 Energy patents have no statistically significant effect on other patents within this industry, and the R&D efforts of these firms do not appear to suffer from financial constraints.
Figure 1. Total alternative energy patents from all energy firms, by application year
The figure shows the total number of alternative energy patents and non-energy patents assigned to firms in our regression analysis. The number of energy patents is on the left axis, and number of other and related patents is on the right axis. Patents are sorted by the year of application.
Substituting clean for dirty
Finding evidence of crowding out among firms working on alternative energy technologies, we focus on a subset of these firms to see what types of innovations are being crowded out. We categorise all of the patents from 12 energy refinery companies into one of 7 technology groups: alternative energy, coal-based fuels, environmental, refining, chemistry, drilling wells, and other technologies.3 Alternative energy patents include traditional renewable sources such as wind, solar, and geothermal. Coal-based fuels include coal gasification and coal liquefaction patents. Environmental patents pertain to pollution control and include techniques for cleaning up oil spills, as well as isomerisation refining techniques developed as part of the phase-out of lead from gasoline and reformulated fuels. Examining correlations among these seven technology groups, we find alternative energy patents to have strong negative correlations with both patents for refining and drilling wells, both of which would increase production of fossil fuels. This suggests that one effect of policies designed to increase alternative energy R&D is to lower incentives for R&D on traditional fossil fuel energy sources, suggesting that changing R&D portfolios may have positive effects on environmental quality. In essence, these changes are simply a reaction of profit-maximising firms to the changing value of clean versus dirty energy R&D.
The social value of energy R&D
Finally, we use patent citation data to compare the social value of new energy innovations to other innovations done by these firms. If the social returns to new alternative energy R&D are no different from the social returns to the R&D that is crowded out, then simply measuring the magnitude of crowding out is sufficient. However, if these social returns differ, the estimated economic effect of crowding out must account for these differences.
There are two reasons to expect that alternative energy research may have a greater social value than other research. First, comparatively less research has been done on alternative energy than other fields, particularly at the beginning of our sample. As a result, alternative energy starts from a lower knowledge base, leading to greater opportunities for big breakthroughs and positive spillovers than more mature technologies. Second, energy technologies may have influence in many sectors, raising the possibility that innovations will have the characteristics of general purpose technologies (e.g. Helpman 1998). We find alternative energy patents are 15% more likely to be cited, and automotive energy patents are 53% more likely to be cited, than other patents from these firms. In addition, both alternative energy patents and automotive energy patents are more general than other patents from these firms, meaning that they are cited by patents in a broader range of technological fields. However, confining our analysis to refineries, we find no difference in the social value of alternative energy patents compared to the refining and wells patents most likely to be crowded out by alternative energy research.
Our results have implications both for policy and for integrated assessment modellers.
For policy, a key result is that any crowding out that may occur appears to affect dirty technologies. Thus, policies enhancing research incentives for green technologies have the additional desired effect of reducing incentives for research on dirty technologies such as fossil fuels. Indeed, our results suggest that concerns about crowding out from new energy R&D are minimal. One important caveat is that any expansion of energy R&D should be done in ways consistent with economic theory. For instance, Goolsbee (1998) provides evidence that one beneficiary of increased government R&D support are scientists and engineers, who experience short-run increases in earnings due to the inelastic supply of researchers. This suggests that consistent, gradual increases in R&D funding are more likely to be successful than periodic funding surges, as the former is more likely to provide incentives for training new researchers.
For integrated assessment modellers of climate policy, our results highlight the importance of incorporating the costs of R&D in the model. Many integrated assessment models include detailed descriptions of alternative energy research, but treat other forms of technological change as exogenous. As this exogenous technological change includes research that may decline when alternative energy research increases, not adjusting this exogenous rate will double-count gains from induced technological change, even if the decrease in research results from profit-maximising decisions of how to allocate research inputs, rather than crowding out from financial constraints.
1 Note that our patent data show that the alternative energy industry consists of many small firms for which crowding out is not a concern. Using patent data from 1971-2002, we identify 18,107 alternative energy patents. There are 3,059 unique patent assignees within this field, of which 1,935 have just one alternative energy patent. Only 17% of alternative energy patents are assigned to the top 20 assignees.
2 Within this industry, we classify patents on energy efficiency, fuel cells, and hybrid vehicles as clean energy technologies.
3 These firms are Amoco, Atlantic Richfield, Chevron, Conoco, ExxonMobil, Gulf, Kerr-McGee, Mobil, Occidental, Standard Oil, Texaco, and Tosco.
Buonanno, P., Carraro, C., & Galeotti, M. (2003). “Endogenous induced technical change and the costs of Kyoto”. Resource and Energy Economics , 25 (1), 11-34.
Goolsbee, A. (1998). "Does government R&D policy mainly benefit scientists and engineers?," American Economic Review, 88(2), 298-302.
Helpman, E. (ed.) (1998) General Purpose Technologies and Economic Growth. Cambridge, MA: MIT Press.
Kammen, D. M. & Nemet, G. F. (2005).“Reversing the Incredible Shrinking Energy R&D Budget”. Issues in Science and Technology, 22 (1), 84-88.
Nordhaus, W. D. (2002). “Modeling induced innovation in climate-change policy”. In A. Grübler, N. Nakicenovic, & W. D. Nordhaus (Eds.), Technological Change and the Environment (pp. 182-209). Washington, DC: Resources for the Future.
Popp, D. (2006). “R&D Subsidies and Climate Policy: Is there a "Free Lunch"?” Climatic Change, 77 (3-4), 311-341.
Popp, D. (2004). “ENTICE: Endogenous Technological Change in the DICE Model of Global Warming”. Journal of Environmental Economics and Management , 48 (1), 742-768.
Popp D. & Newell R.G. (2009). “Where does energy R&D come from? Examining Crowding Out from Environmentally-Friendly R&D”. NBER Working Paper 15423.
Schneider, S. H., & Goulder, L. H. (1997). “Achieving low-cost emission targets”. Nature , 389, 13-14.
Yang, C.-J., & Oppenheimer, M. (2007). “A 'Manhattan Project' for climate change?" Climatic Change , 80 (3-4), 199-204.