Innovation from Broad-Based Policies

I first present evidence on innovation resulting from market forces such as higher energy prices or from technology-neutral policies. Most technological solutions to reduce climate emissions address the energy sector through one of two mechanisms: providing cleaner energy resources or improving energy efficiency. Understanding what innovation the private sector will do on these technological areas without targeted support is important for understanding when targeted support will be most effective. Three key lessons emerge from the literature.

First, higher energy prices encourage innovation on alternative energy sources and on some energy efficiency technologies. For example, Popp (Reference Popp2002) finds that a 10 percent increase in energy prices leads to a long-run 3.5 percent increase in the number of U.S. patents in alternative energy and energy efficiency technologies, with most of the effect on patenting activity observed 3.71 years, on average, after an energy price change. Verdolini and Galeotti (Reference Verdolini and Galeotti2011) use a multicountry sample over a 25-year study period and find a similar relationship between energy prices and patents. Aghion et al. (Reference Aghion, Dechezleprêtre, Hemous, Martin and Reenen2016) provide differentiated patenting responses for firms in the automotive industry facing higher fuel prices. They find that a 10 percent increase in fuel price is associated with about 10 percent more low-emission energy patents (such as electric and hybrid cars) and 7 percent fewer fossil-fuel patents (those that improve internal combustion engines).

Second, market prices alone do not encourage sufficient energy efficiency innovation. Since energy efficiency both reduces emissions and reduces costs for firms, there are incentives to develop and deploy energy efficient technologies even without climate or other energy policies in place. However, because reduced emissions are an external benefit, environmental market failures mean that individuals will not consider the social benefits of their energy efficiency investments, leading them to underinvest in energy efficient technologies relative to the social optimum. The effect of energy prices on energy efficiency innovation is also limited by their saliency. While studies on the auto industry and on renewable energy find higher energy prices spur innovation, energy prices are less effective at driving innovation and technology adoption geared toward enhancing residential energy efficiency. Prices are particularly ineffective for inducing innovation on less-visible and longer-lived technologies such as insulation that are installed by builders and are not easily modified post-construction. Instead, residential energy efficiency innovations are induced by command and control approaches like building code changes (Noailly, Reference Noailly2012).

Third, even the choice of broad-based policies focusing on overall emissions (e.g., a carbon tax) or on technology-neutral goals (e.g., renewable energy mandates) implicitly favors some technologies over others. Technology-neutral policies promote those technologies that are closest to being competitive in the market without policy support (Johnstone et al. Reference Johnstone, Hascic and Popp2010). For example, because the cost of wind energy has historically been lower than other renewables, innovation in countries with renewable energy mandates to provide alternative energy focused on wind. In contrast, technology-specific policies such as feed-in tariffs supported technologies that were less competitive with traditional energy technologies. These include solar and waste-to-energy technologies that required the guaranteed revenue from a feed-in tariff to compete with traditional energy technologies.

Together, these lessons make it challenging for policy makers to encourage long-term innovation in technologies that are not close to being competitive in the market. Technology-neutral policies that let firms choose those technologies closest to being competitive in the market lead to lower compliance costs in the short-term but may not provide enough incentive to develop technologies with longer-term research needs, such as offshore wind energy. Because no one technology will be fully able to meet all energy demands, complementary, technology-specific targeted policies to promote those technologies that are further from being competitive in the market are also needed.

When Should Policy Target Specific Technologies?

Recent theoretical work suggests other market failures may justify government support for specific technologies—particularly those furthest from market. Such market failures include learning-by-doing, path dependency, and capital market failures (Acemoglu et al. Reference Acemoglu, Akcigit, Hanley and Kerr2016, Fischer, Preonas, and Newell Reference Fischer, Preonas and Newell2017; Lehmann and Söderholm Reference Lehmann and Söderholm2018). Both learning-by-doing and path dependency justify technology-specific deployment policies such as feed-in tariffs or tax credits. Learning-by-doing occurs when early producers and users of a technology generate knowledge through the production and use of technology rather than through R&D activity. This learning results in decreasing costs for themselves and other future users/producers (e.g., Lehmann and Söderholm Reference Lehmann and Söderholm2018). While such spillover benefits justify subsidizing early adopters of clean energy technology, recent evidence indicates that the external benefits are small relative to existing levels of subsidization (e.g., Nemet Reference Nemet2012, Fischer, Preonas, and Newell Reference Fischer, Preonas and Newell2017; Tang, Reference Tang2018).

Path dependency creates a market failure only if the costs of shifting from one type of technology to profitable opportunities in another act as a hurdle, preventing adoption of an otherwise more profitable technology (Lehmann and Söderholm Reference Lehmann and Söderholm2018). Empirical evidence on path dependency is slim. There is some evidence that path dependency matters in the context of energy innovation (e.g., Aghion et al. Reference Aghion, Dechezleprêtre, Hemous, Martin and Reenen2016; Stucki and Woerter Reference Stucki and Woerter2017), but none of these studies test whether the observed path dependency results from high switching costs or is simply a response to better research opportunities.

In contrast, there is evidence both suggestive of market failures resulting from imperfect capital markets and that government investments can help. Howell (Reference Howell2017) shows that early financing from the U.S. Department of Energy Small Business Innovation Research (SBIR) program helps clean energy firms overcome capital market failures. However, this same intervention did not influence traditional energy industries such as fossil fuels and biofuels. Likewise, some evidence suggests that clean energy innovations take longer to reach commercialization than other high-technology fields (Popp, Reference Popp2017; Branstattter and Ogura Reference Branstatter and Ogura2005; Finardi Reference Finardi2011), providing another potential barrier to raising private sector financial support.

Finally, the government can help remove obstacles to commercialization through direct funding and/or performance of research and development (Mowrey, Nelson, and Martin Reference Mowrey, Nelson and Martin2010; Weyant Reference Weyant2011). The disproportionately large capital expenses required for significant energy innovations leave a role for collaboration between the public and private sectors to support both initial project development and demonstration projects. Such demonstration projects can promote further learning (Mowrey, Nelson, and Martin Reference Mowrey, Nelson and Martin2010). Palage, Lundmark, and Söderholm (Reference Palage, Lundmark and Söderholm2019) find supporting evidence, showing that advanced biofuel patenting increases after investments in demonstration projects in EU countries. While more research is needed, the evidence to date suggests a need for polices that help bridge the gap between laboratory research and commercial success.

The Effectiveness of Government R&D

Most policies addressing learning-by-doing, path dependency, and capital market failures focus on commercialization and deployment rather than the supply of innovation itself. High social returns to R&D justify government research investment to increase the supply of innovation in all industries. Thus, it is important to understand whether the identified spillovers from clean innovation are larger than in other industries. If so, a larger role for government R&D in clean energy is justified. Several recent papers use patent citation data as evidence of spillovers in clean energy industries, as patent citations are an indicator of how successful an innovation is in spurring additional innovations. These studies generally provide support for a larger role for government-funded clean energy R&D, particularly for technologies that are still emerging. Both Dechezleprêtre, Martin, and Mohnen (Reference Dechezleprêtre, Martin and Mohnen2017) and Popp and Newell (Reference Popp and Newell2012) find that clean energy R&D generates large spillovers, comparable to spillovers in other emerging fields such as IT or nanotechnology. Noailly and Shestalova (Reference Noailly and Shestalova2017) find similar results, but only for younger clean energy technologies. For emerging technologies such as energy storage, spillover occurs across technology domains, making it less likely private sector inventors can capture the full benefits of these innovations.

Popp (Reference Popp2016) studies the effectiveness of government-sponsored energy research by linking data on scientific publications to public energy R&D funding and finds the following four key results. First, while $1 million in additional government R&D funding leads to 1–2 additional publications, the lag between initial funding and publication can be as long as ten years. Second, adjustment costs associated with large increases in research funding are of little concern at current levels of public energy R&D support. These first two results together suggest that while there is room to expand public R&D budgets for renewable energy, the impacts of any expansion may be delayed for several years. Third, market factors and demand-push policies found to influence private R&D activity in other papers have little impact on publications, suggesting that current R&D funding efforts do appear to support different types of research than generated by the private sector. Finally, as the ultimate goal of government energy R&D funding is not an article but a new technology, Popp (Reference Popp2016) uses citations from patents to scientific literature to link these articles to new energy patents. Lags in both the creation of a new publication and the transfer of this knowledge to applied work mean that public R&D spending may take over a decade to go from a new article to a new patent.

In general, as evidenced by the discussion above, government R&D should focus on technologies furthest from the market. Costantini, Crespi, and Curci (Reference Costantini, Crespi and Curci2015) compare patenting in conventional first-generation biofuels to patenting in more advanced second-generation biofuels. While technology-push policies do not induce innovation for more mature technologies (e.g., first-generation biofuels), they are important for fostering development in emerging, more advanced technologies. Thus, government support for clean energy R&D should focus on emerging technological areas such as energy storage rather than more established technologies such as on-shore wind energy.

1. Coordinate targeted policy efforts

For state governments, many of the benefits from technology policy will spill over to other states. For instance, some states require that a minimum share of renewable energy come from solar technology (e.g., a carve-out for solar). These requirements promote the development of solar energy but also increase the short-run costs of generating electricity. Similar to the public goods problem for knowledge discussed earlier, the resulting technology improvements benefit consumers everywhere, not just consumers in the local community. For example, feed-in tariffs for solar energy raised electricity prices in Germany. Innovation induced by these feed-in tariffs increase the external benefits of those subsidies by 22 percent. However, most of those benefits are generated by future solar panel adoption that occurs outside Germany due to the lower solar panel prices that result (Gerarden Reference Gerarden2018). For smaller local governments, appropriating the external benefits from innovation will be even more difficult. Thus, while first-order benefits such as reduced emissions may be enough to justify the use of clean energy policies at the state and local levels, the second-order benefits of induced technological change should be of less concern to local policy makers in all but the largest states. As the work by Fu, Li, Ondrich, and Popp (Reference Fu, Li, Ondrich and Popp2018) demonstrates, there is no guarantee that clean energy innovations will occur in the communities most strongly promoting clean energy.

One way to overcome this appropriation problem is for states to coordinate renewable energy policies. The Regional Greenhouse Gas Initiative limiting emissions from the electricity sector among ten northeastern states is an example of such coordination. Since overall market size is more important than local environmental regulations for spurring clean energy innovation, coordinated policies can better create the market scale necessary for inducing private sector innovation. An example where policy coordination may be particularly helpful is in building codes. As noted in Noailly (Reference Noailly2012), changes to building codes in Europe increased energy efficiency innovation. Because building codes are set at the state and local level in the United States, coordination among local governments is more likely to create the market scope necessary to spur innovation in response to building code changes.

2. Coordinate research and development

Improving coordination among jurisdictions is particularly important for direct R&D funding. The federal government is in a better position to fund energy research and development spending, both because it is better able to diversify risk and because federal research spending avoids potential duplication of research programs across multiple states. Nonetheless, states do fund some energy R&D. In 2017, U.S. state governments spent $307 million on energy R&D. Although that is just a fraction of the $9.2 billion spent by the DOE, almost 80 percent of this funding came from just two states: California ($185.6 million) and New York ($55.9 million). Thus, there is room for other states to expand energy R&D budgets. However, enhancing state energy R&D investments faces two particular challenges.

First, despite the earlier evidence on the importance of government funding for early-stage research, much of the R&D support from states focuses on deployment rather than on early-stage research. The California Energy Commission (2015) explicitly states, “Energy Commission RD&D [research, design, and development] programs bridge the gap between the laboratory and the market.” California provides initial funding for green energy entrepreneurs through California Sustainable Energy Entrepreneur Development Initiative (CalSEED) awards. Similarly, the NY Green Bank, founded in 2014, leverages private sector financing to increase clean energy investments. NY Green Bank addresses potential capital market failures by operating in markets with limited competition. Such policies can address capital market failures noted earlier, but they do not substitute for a lack of investment in basic energy research at the federal level.

Second, although both California and New York have taken leading roles in clean energy investment, for smaller states it is difficult to invest in research at levels that would have an impact on technological improvements. For smaller states, coordinating energy R&D efforts would increase the impact of individual investments. Coordination would also help states avoid potential duplication of research effort. As an example, in 2018 staff from New York and California helped review research proposals on microgrids and energy storage to each other's state R&D programs to learn about the national state of research on each technology (Orta Reference Orta2019). Coordination would also help diversify risk by both spreading initial investments across multiple jurisdictions and providing sufficient funding to develop a diverse portfolio of projects.

The EU Strategic Energy Technology Plan (SET-Plan) provides an example of efforts to increase coordination of research efforts across jurisdictions (Dechezleprêtre and Popp Reference Dechezleprêtre, Popp, Parry, Pittel and Vollebergh2017). Implementation of the SET-Plan currently includes 17 joint programs in the European Energy Research Alliance; these programs set research priorities for various renewable technologies and encourage coordination among researchers in different countries and different sectors, including industry (European Energy Research Alliance n.d.). Since the inception of the SET-Plan in 2006, EU researchers have become more integrated, as demonstrated by increasing rates of patent citations among EU researchers in different countries (Conti et al. Reference Conti, Mancusi, Sanna-Randaccio, Sestini and Verdolini2018). A consortium of states could similarly pool funds for energy R&D, both allocating it among researchers in member states and setting aside a share of funding specifically for cross-state collaborations.

3. Provide a supportive regulatory environment

State governments can also help foster clean energy innovation by fostering a supportive regulatory environment. Tang (Reference Tang2018) finds that learning-by-doing is greatest for wind farms operating on transmission systems operated by independent system operators or regional transmission organizations rather than privately owned and vertically-integrated transmission systems owned by major utilities. Similarly, Nesta, Vona, and Nicolli (Reference Nesta, Vona and Nicolli2014) show that deregulated or liberalized energy markets in combination with supportive renewable energy policies foster more innovation than when similar policies are implemented in more traditionally regulated energy markets. Likewise, Jamasb and Pollitt (Reference Jamasb and Pollitt2011) show that patents for non-nuclear electricity and renewable technologies increased in the UK after market liberalization. Thus, increasing competition and flexibility in electricity markets through deregulation may also increase clean energy innovation.