Does Innovation Policy Attract International Competition?
Evidence from Energy Storage
Kira R. Fabrizio
Questrom School of Business
Boston University
kfab@bu.edu
Sharon Poczter
Dyson School of Applied Economics and Management
Cornell University
sharon.poczter@cornell.edu
Bennet Zelner
Robert H. Smith School of Business
University of Maryland
bzelner@rhsmith.umd.edu
ABSTRACT
Though existing studies suggest that innovation-promoting public policies are associated
with increases in the number of inventions generated, these studies do not explore the
geographic origin of such inventions. Do domestic policies aimed at developing innovation
capabilities in a particular technological area result in greater domestic competitiveness, or
do they instead invite more competition from technologies developed abroad? In this
paper, we describe the impact of two categories of innovation-supporting policies: those
focused on the supply of a given set of technologies, and those focused on the demand for
products based on these technologies. We argue that demand-pull policies will result in
more technology transfer into a given country from abroad, relative to supply-push
policies. We test this prediction using panel data on the international transfer of patented
energy storage technologies between country-pairs. Our results indicate that the transfer of
such technologies into a given country increased significantly after demand-pull policies
were put into place, but the same pattern does not hold for supply-push policies.
Keywords: Innovation policy, international technology transfer, competitive advantage
1
1. Introduction
Political leaders often promote policies that seek to bolster a country’s technological or
industrial leadership by fostering national competitive advantage in an area of special
economic or technological significance. For example, U.S. President Barak Obama declared
that the development and deployment of “green” technologies in areas such as energy
production would create a foundation on which the United States could “build the clean
energy economy that is key to our competitiveness in the 21st century” (Finance Wire,
2009). Existing studies provide partial support for this assertion by demonstrating that
firms increase their R&D expenditures and patenting rates in response to public policies
intended to stimulate innovation (Hall and Van Reenen 2000, Bloom et al. 2002, Popp 2002,
Newell et al. 1999, Jaffe et al. 2002, Johnstone et al. 2010, Horbach 2008). These studies do
not investigate, however, whether such increases are attributable to domestic inventors—
which would be consistent with the objective of building a country’s technological and
industrial leadership—or alternatively, whether they represent inventions developed
outside of the country that are transferred into the domestic market. This distinction is
critical because policies that attract technologies from abroad may incite increased product
competition at the expense of local innovators, potentially undermining the goal of the
policy. In the analysis below, we illuminate this topic by examining whether national
policies intended to promote local innovation actually result in the domestic deployment of
technologies developed abroad.
Innovation-promoting public policies are commonly justified as a solution to a market
failure inherent in the innovation process. Because companies investing in research and
development are unable to fully capture the gains from knowledge creation, the incentive
to invest in innovation remains below the socially optimal level (Geroski 1995). In the
broad context that we examine below—renewable energy production—there also exists a
second market failure that impedes innovation: companies whose inventions provide an
environmental benefit (or reduce a negative environmental impact) do not fully capture
their invention’s value, because the environmental impact is an externality that is not
priced in the market. The rationale for innovation-promoting public policies in the
2
renewable energy space, therefore, is to compensate businesses for these two innovationreducing market failures.
Innovation-promoting public policies can be broadly categorized into those that directly
promote R&D investments, and those that indirectly promote innovation by increasing the
demand for environmentally favorable products. The first category, which we label
“supply-push” is intended to increase the market supply of a particular type of technology,
include measures such as direct R&D subsidies, tax credits for R&D, and financial support
for R&D personnel. The aim of the second category, which we label “demand-pull” is to
prop up demand for environmentally favorable products, and includes mandated
purchases, minimum purchase requirements, and targeted subsidies. We argue that
supply-push and demand-pull policies have different effects on the international transfer of
technology. Specifically, we contend that, relative to supply-push policies, demand-pull
policies will lead to a larger rise in the transfer of foreign technologies into the country
implementing the policy.
The well-publicized recent bankruptcy of several federally-subsidized American solar
energy firms like Solyndra provides a suggestive example. While the Obama administration
implemented a suite of policies intended to promote “green” energy innovation, including
providing low interest rate loans to suppliers and requiring electric utilities to increase the
provision of power generated using solar technologies, energy firms failed. The U.S.
Congress attributed the failure of these firms in the spite of the supportive government to
increased competition from Chinese solar panel manufacturers (Sweet and Tracy, 2011).
Data from the Earth Policy Institute are broadly consistent with this claim: during the
period 2000 – 2010, Chinese solar panel production grew by nearly 11,000 MW and U.S.
production grew by only 1,000 MW, despite the fact that installed solar generation capacity
in the U.S. increased 65% more than capacity in China during this time. 1 This is consistent
with the Chinese solar panel manufacturers gaining from the policies promoting adoption
of solar generation in the United States.
1
Installed solar generation in the U.S. increased by 856 MW and that in China grew by 520 MW.
3
Building on this example, we examine the international transfer of energy storage
technologies, which are regarded as key to ending the world’s reliance on fossil fuels (The
Economist, December 6, 2014). Advances in energy storage are considered necessary for
the deployment of large-scale renewable energy facilities (Denholm et al. 2010), and they
also represent “the most important thing we can do to make electric vehicles more
prevalent,” according to Tesla’s Chief Technology Officer (The Economist, December 6,
2014). Technological uncertainty, the need for significant financial investment, and the
market failures discussed above have led governments around the world to promulgate
both supply-push and demand-pull policies intended to stimulate innovation in this area.
To examine whether demand-pull policies do, in fact, produce a larger increase in the
transfer of technologies developed abroad than supply-push policies do, we examine 36
national policies related to energy storage passed in 11 OECD countries during the period
1974 – 2010, categorizing 16 as demand-pull, 19 as supply-push, and one as having both
supply and demand components. We then combine the policy data with panel data on
energy storage patents granted in 21 countries during the period 1970 – 2011 to estimate
the impact of policy on the transfer of patented energy storage technologies using a
difference-in-differences methodology. We found that, consistent with our expectations,
the adoption of a demand-pull policy was followed by a significant increase in the
importation of technology developed abroad. We found no such pattern for supply-push
policies, however.
From a public policy perspective, these results are valuable because they suggest that it
may be incorrect to interpret an increase in the number of patented innovations following
an innovation-promoting policy’s adoption as evidence that the policy is “working.” In the
empirical context of energy storage, the number of energy storage patents in a given
country may grow following the adoption of a demand-pull policy, but a significant
proportion of this increase is attributable to technologies invented (and initially patented)
abroad. Thus, a domestic innovation-promoting policy may encourage competition from
abroad even as it succeeds in spurring innovation overall. If the policy’s central objective is
to create domestic competitive advantage in the global market for green technologies, then
the policy’s unintended consequence may hinder achievement of this goal. Furthermore,
4
the method that we used to trace the international movement of patented technology has
not been widely exploited in innovation management research, and it holds great promise
for examining a range of research questions related to global innovation and the flow of
knowledge and technology across countries.
2. Theory and Hypotheses
Public policies to promote innovation in renewable energy technologies seek to address
two types of market failure: environmental externalities and knowledge spillovers (Popp
2006). Environmental externalities occur because the market does not accurately price the
improvements in environmental conditions that such technologies create. Namely, in
contrast to electricity generated using traditional technologies (such as coal- and natural
gas-fired), which produces a significant fraction of atmospheric greenhouse gases,
electricity generated using renewable energy technologies (such as wind- and solar-
powered) produces almost no greenhouse gases. In the absence of policy intervention,
however, the latter’s superior environmental characteristics go uncompensated (or
equivalently, the former’s negative environmental are not priced) because electricity is a
homogenous good, competing with little differentiation to the end user. Consequently, the
amount of environmentally preferable goods produced is below than the socially optimal
level.
The second market failure is more general, and involves the weak appropriability of
knowledge generated by R&D. When the knowledge generated becomes public as a result
of new products or services embodying the knowledge being made available for sale,
inventors typically fail to appropriate the full social value of such knowledge because it is
difficult to exclude others from acquiring and using it (Arrow, 1962). This results in
“spillovers” that accrue to competing firms, downstream firms, and consumers (Griliches
1979, 1992; Jaffe 1986). As a result, private firms invest less than the socially optimal level
of R&D (Geroski 1995, Martin and Scott, 2000).
5
Policies to address these market failures take multiple forms. Though there is no
universally accepted scheme for classifying such policies, 2 we draw on elemental research
in innovation theory to develop a meaningful categorization based on a policy’s operative
mechanism for promoting innovation. Such research emphasizes innovation’s
multifactorial origins, including targeted R&D, firms’ prior experience, and customer
demands (Mowery and Rosenberg, 1979; Kline and Rosenberg 1986).
Research on “demand-pull” innovation holds that firms invest more to produce inventions
that they expect to serve larger markets. Such research traces to Schmookler’s seminal
study (1966), which demonstrated that increased purchases of railroad equipment led to
greater inventive activity by railway equipment manufacturers. Subsequent theoretical and
empirical work has established that innovation investment responds positively to increases
in anticipated demand (Schmookler, 1966; Romer, 1990; Grossman and Helpman, 1991;
Aghion and Howitt, 1992; Newell, Jaffe, and Stavins 1999; Popp 2002, Acemoglu and Linn
2004, Acemoglu et al. 2006, Adner and Levinthal 2001).
Research on “supply-push” innovation supports the commonsense notion that inventions
result from the development of technological knowledge. Such knowledge may accumulate
within the firm as the result of either R&D or prior experience in design, manufacturing,
marketing, and distribution; or outside the firm, as the result of basic science and the
innovative activities of other firms. Accordingly, providing incentives directly to suppliers
to increase the technological knowledge that defines the opportunity frontier for new
discoveries and conditions the investment choices of firms across potential research areas
(Dosi, 1982; Griliches, 1979, 1995; Nelson, 1982). The conclusion of this literature is that
the development of technological knowledge, within and outside the boundary of the firm,
will lead to more innovation (Dosi, 1982; Griliches, 1979, 1995; Nelson, 1982).
Alic et al. (2003) classified innovation-supporting policies into three groups: (1) direct government funding
of research and development; (2) policies that directly or indirectly support commercialization and adoption,
and thus development; and (3) policies that support diffusion through education. The first two categories
map roughly (though not perfectly) to the supply-push and demand-pull categorization developed in this
paper. Education-related policies in third category could be demand- or supply-push in nature, but we
observe no such policies in the empirical context that we examine.
2
6
Geography moderates the impact of both demand-pull and supply-push influences on
innovation. In terms of demand-pull, local firms have an advantage in terms of both
understanding and meeting proximate demand. Demand for new products often depends
on context-specific norms, culture, and expectations, making demand hard to anticipate for
distant firms (Fabrizio and Thomas 2012). In addition, established distribution networks of
local firms provide advantage in terms of serving growing demand. The limited empirical
work that exists examining the geographic variation in the impact of anticipated demand
on innovation supports the advantage of local firms (Fabrizio and Thomas 2012). In terms
of supply-push, local firms may benefit more from knowledge spillovers because
technological knowledge may be tacit, impeding its transferability (Polanyi, 1966;
Baumaud, 1999). Empirical research has supported this contention, demonstrating both a
pattern of geographical agglomeration of innovations across multiple industries (Marshall,
1920; Krugman, 1991; Audretsch and Feldman, 1996; Feldman, 2000; Porter, 1998), as
well as geographically mediated knowledge spillovers between firms and nearby public
institutions such as universities (Jaffe 1986; Jaffe 1989; Jaffe et al. 1993; Jaffe and
Trajtenberg 1996).
Research on national innovation systems builds on these core insights, examining the
institutional structures that shape and augment underlying supply-push and demand-pull
influences on the rate and direction of innovation (Nelson, 1993; Mowery and Oxley, 1995;
Freeman, 1995). Institutional structures that condition supply-push initiatives include
government policies, funding, and regulations; the educational and research activities
housed in a nation’s universities and laboratories; and the expectations and attitudes of
workers and investors. Clusters of interconnected firms, suppliers, related industries, and
specialized entities arise to create regional or national competitive advantage in a given
industry (Porter 1990, 1998). A country’s “demand conditions” may augment these supply-
push factors to shape “national innovative capacity” (Furman, Porter, and Stern, 2002). For
example, pharmaceutical companies have been shown to produce innovations that match
their home-country demand patterns (Fabrizio and Thomas, 2012). Because the entire
network of national institutions interacts to shape the conditions for innovation, domestic
7
firms embedded in this network might be expected to be more influenced by the policies
implemented in their home countries.
Building on these theoretical foundations, we conceptualize two categories of public
policies based on their operative mechanism for promoting innovation: (1) “supply-push”
policies that provide incentives to invest in technology-related R&D, and (2) “demand-pull”
policies that stimulate innovation in a particular area by propping up demand for a given
product or category. The former include direct government funding of R&D, R&D tax
credits, and tax credits or subsidies for firms that bring particular technologies to market.
By lowering the private cost of investing in R&D, supply-push policies create incentives for
firms to increase their R&D expenditures in targeted areas.
Empirical research supports the expected positive impact of supply-push policies on
innovation. In a review piece on the effectiveness of fiscal incentives for R&D, Hall and Van
Reenen (2000) found that each tax credit dollar produced, on average, one dollar of R&D
(i.e., the elasticity of R&D with respect to tax credits is unity). Bloom et al. (2002) also
found evidence that tax credits stimulate R&D spending in a panel dataset of nine OECD
countries, and Horbach (2008) found that “investment subsidies” were associated with
increased environmental innovation in a study based on survey data from German
manufacturers. Dechezlepretre et al. (2011) document the relationship between
environmental policies and innovation, as well as the global dispersion of environmental
innovation. Though suggestive of a link between supply-push policies and innovation, these
studies do not consider the nature or location of actual innovation outcomes. Johnstone et
al. (2010) took a step in this direction emphasizing a cross-country focus in examining the
relationship between supply-push policies and the number of patented inventions in
renewable energy generation technologies in 25 countries during the period 1978–2003.
In contrast to supply-push policies, which steer innovation efforts toward a targeted area,
demand-pull policies use mechanisms such as rebates, consumer tax credits, government
procurement, and mandated purchases to boost the prices for environmentally beneficial
products, causing the externalities previously discussed to be internalized. Such policies
effectively differentiate environmentally beneficial products from rival goods that are less
8
environmentally friendly. During the 1990s, for example, many U.S. states passed
Renewable Portfolio Standard (RPS) policies requiring utilities to source a minimum
percentage of their electricity from renewable generation sources. By increasing the
demand for renewable electricity, RPS policies effectively established a differentiated
market that could support higher prices for electricity generated from such sources.
Such differentiation is critical for stimulating innovation, particularly in the early
development phases. During development and early deployment, costs of new technologies
are typically higher, making it difficult for them to compete with entrenched technologies
benefitting from economies of scale. Demand-pull policies targeting new technology allows
firms to move down the learning curve, making improvements in the technology and its
production process, and reducing costs. Ultimately, demand-pull policies enable firms
focused on new technologies to reduce costs enough to compete with established
technologies even after policy support subsides.
Empirical research on “demand-induced innovation” recognizes that R&D is a profit-
motivated investment activity, and that the direction of innovation should respond to the
pattern of relative prices (Hicks 1932). Several empirical studies in the renewable energy
space have supported this prediction (Newell et al. 1999, Jaffe et al. 2002, Popp 2002). Most
recently, Johnstone et al. (2010) found that the number of patent applications for
renewable energy generation technologies increased following the adoption of demand-
pull policies intended to increase deployment, such as purchase obligations and renewable
energy credits.
Impact on technology transfer
Prior research on the link between environmental policy and the international transfer of
technology is limited. Studies examining the relationship between policies and innovation
outcomes have not considered the geographic location of inventions (Johnston 2010,
Aghion 2011), while studies of international technology diffusion have failed to account for
policy influences (Dechezlepretre et al. 2011). The prior study that is closest in spirit to the
current one is by Lanjouw and Moody (1996), who examined the effects of increased
pollution regulation on the deployment of pollution control technologies from both
9
domestic inventors as well as those domiciled elsewhere. They found that, even in
developed countries, a large portion of pollution control technology was developed by
foreign inventors.
No prior study of which we are aware has considered how different types of policies, such
as supply-push versus demand-pull, influence the international transfer of technologies.
Without knowledge of how different policies affect technology transfer, it is not possible to
draw sound conclusions about the impact of green energy technology policies on domestic
economic development and competitiveness, a topic of substantial interest to both
policymakers and private investors.
We consider this question directly, examining the differential impact of supply-push and
demand-pull policies on the location of innovation. We anticipate that the two different
types of policies will have different effects on the amount of the targeted technology
transferred into the country adopting the policy. Relative to supply-push policies, we
expect demand-pull policies to increase the amount of inward technology transfer. As
policy-motivated demand increases in a given geographic market increases, so too does the
expected payoff of entering that market, leading foreign firms that have developed relevant
technology elsewhere to enter and compete with domestic firms. The ultimate result is
greater foreign competition in the industry supported by the demand-pull policy. The
increased importation of Chinese solar panels in response to U.S. demand-pull renewable
energy policies exemplifies this logic.
We expect supply-push policies to attract less inward technology transfer for several
reasons. First, government funding and R&D support are more likely to be restricted to
domestic firms in the country implementing the policy for the express purpose of
developing a local innovation ecosystem. Second, incentives provided through the
corporate tax system only affect firms with taxable operations in the country implementing
the policy. Third, government-sponsored investment in knowledge development may
create knowledge spillovers, favoring proximate domestic firms.
It is still possible that the favorable innovation climate created by supply-push policies will
provide incentives for foreign firms to reorganize R&D activities geographically and
10
transfer previously patented technologies into the policy-implementing country. Several
reasons exist, however, as to why we expect supply-push policies to elicit less inward
transfer than demand-pull policies. First, this effect is less direct than in the case of
demand-pull policies, requiring several steps from the policy implementation to
transferring patenting technologies into the country. Furthermore, this impact is limited
only to a subset of supply-push policies that do not target domestic producers exclusively,
Finally, as reorganizing activities globally represent a significant cost, foreign firms may be
more reluctant to respond to supply-push policies by relocating R&D activities than they
are to respond to demand-pull policies by entering new product markets, which is
relatively less costly.
3. Empirical Context: Energy Storage Policy Technology
During the past three decades, many countries have passed policies intended to promote
the development and deployment of “clean” energy technologies. These technologies
include renewable power generation, enhanced electricity generation and use energy
efficiency, and energy storage. We take a purposefully narrow lens in our analysis,
examining only those policies intended to stimulate innovation in energy storage. This
approach is beneficial for two reasons. First, with fewer policies, we can identify the time
period when each policy is in effect, avoiding the overlap that would occur when
examining, for example, all renewable energy policies. Second, many countries have
enacted multiple energy storage policies at different points in time as awareness of the
environmental and economic importance of this technology has grown. The resultant
cross-sectional and inter-temporal variation in policy environments supports our empirical
approach and bolsters the generalizability of our results.
Several OECD countries have implemented supply-push or demand-pull policies to increase
innovation in and deployment of energy storage technologies. In 2000, for example,
Australia implemented a demand-pull policy known as the “Alternative Fuels Conversion
Program,” which was intended to increase the demand for hybrid and electric vehicles. This
program provided a subsidy of up to 50% of the extra cost incurred by fleet operators to
purchase or convert their existing fleets to diesel-fueled hybrids or fully electric vehicles.
11
The U.S. implemented a supply-push policy in 2002. The “Advanced Energy Storage
Program” provided financial support from the U.S. Department of Energy [DOE] for R&D in
energy storage performed by both the DOE alone and by the DOE in cooperation with the
private sector. Similarly, Germany’s “Sixth Energy Research Program” subsidized €3.4
billion of R&D in “innovative energy technologies,” including renewables, efficiency
enhancements, and energy storage.
4. Descriptive Statistics and Empirical Methodology
To assess the impact of innovation-promoting energy storage policies on international
technology transfer, we examine whether the number of patents transferred into a given
country changed after the country adopted a supply-push or demand-pull energy storage
policy. We consider such changes relative to the amount of international technology
transferred into countries without innovation-promoting energy storage policies.
We perform the analysis at the origin-destination country pair level. “Destination”
countries include those that passed relevant policies as well as the control countries that
did not. “Origin” countries included those where the patented inventions subsequently
transferred to destination countries were initially developed. This structure allows us to
control for relevant attributes of the origin countries, the destination countries, and origindestination dyads.
We compiled policy data from the International Energy Agency’s Global Renewable Energy
Policies and Measures database (IEA) (IEA 2007), which records energy policy initiatives in
40 countries over the period 1970 – 2010. The database includes detailed information that
allowed us to classify polices as supply-push or demand-pull. For example, we categorized
R&D subsidies and tax breaks as supply-push policies, and requirements or incentives to
procure specific amounts of “clean” energy as demand-pull initiatives.
We identified a total of 11 OECD countries that passed policies intended to promote
innovation in energy storage technologies. These countries constitute the “treated” set of
observations in our analysis, though the year in which a country received a policy
treatment and the type of treatment (i.e., supply-push versus demand-pull) varied.
12
The canonical difference-in-differences approach uses a large sample from a population to
estimate the impact of a policy treatment on the treated subset of observations. In the
current context, such an approach would consist of comparing the change in the amount of
energy storage technology transferred into a country that had adopted a policy, with the
change in the amount of such technology transferred into countries that did not pass any
such policy during the same time period. Because policies were not randomly assigned,
members of the control group might differ systematically from those of the “treated” group,
however, requiring the addition of control variables among other analyses that help justify
this empirical approach.
Table 1 displays differences in means tests for salient attributes of the 11 OECD countries
that passed energy storage policies in 1973 or later, and the 144 countries that did not.
Based on nearly all of the attributes examined, the countries in the former group have
larger economies (higher GDP) and are more innovative (more patents and energy storage
patents) than those in the control group.
Between-group differences in the value of the dependent variable just before treatment
(i.e., policy implementation) are of special interest in a difference-in-differences analysis.
The identifying assumption of this empirical method is that prior to the treatment, the
change over time of the outcome variable of interest does not differ between the treatment
and control group. If this is the case, then the assumption that post-treatment changes in
the outcome variable for the control group are a good representation of how the outcome
variable would have evolved for the treated countries if the treatment had not occurred.
Here, immediately prior to implementation of the first supply-push policy, the number of
energy storage patents transferred into the countries that passed relevant policies was
decreasing at a slightly greater rate than in the control countries. The trend in the number
of transfers immediately prior to implementation of the first demand-pull policy was not
statistically distinguishable between the two sets of countries.
In order to minimize the differences between the countries that passed policies and those
in the control group, we use the Mahalanobis distance algorithm to identify the most
closely-matched control observations based on a set of country-level attributes, including
13
the number of patents generated in a country, the number of energy storage patents
generated in the country, and GDP level, all in 1973. 3 Because the Mahalanobis algorithm
selects controls with replacement, some countries were chosen as the best controls for
more than one of the treated countries. Out of the 63 possible countries for which we have
requisite data, the algorithm identified 10 matching countries. We include these countries
in the control group in the main analysis, and for robustness we also conduct analyses
using a control group comprising all 63 countries that failed to pass a relevant policy.
Table 2 displays differences in means tests for country-level attributes of the treatment
group and the matched control group. On average, the countries in the matched control
group are more similar to the policy-passing countries in the year prior to adoption than
the countries in the unrestricted control group are, though there are still notable
differences. Importantly, the pre-adoption trends in transfers of energy storage
technologies bias our analysis against finding a positive effect of policy (based on the 1973
data) or indicate that the two groups display a similar pre-trend (based on the 1998 data).
To construct the patent-based measures, we use data from the EPO Worldwide Patent
Statistical Database (PatStat), which categorizes patents generated by inventors in over 80
countries using the international patent class (IPC) scheme. Following recent studies such
as Johnstone et al. (2010), we exploit the extensive categorization of “green” patent classes
produced by the World Intellectual Property Organization (WIPO) in the green patent
inventory (http://www.wipo.int/classifications/ipc/en/est/) to identify IPC classes for
energy storage. The WIPO green patent inventory identifies the international patent classes
(IPCs) associated with “environmentally sound technologies” as delineated by the United
Nations Framework Convention on Climate Change. Specifically, we identify patents related
to energy storage as those belonging to the “storage of electrical energy” category.
The PatStat data also include information for tracking a patent’s country of origin and its
international transfer, including the country of the inventor, the initial or “priority” country
3
The Mahalanobis distance measure is the matrix product d'Xd, where d is a vector of differences (between
two countries) in the set of variables and X is the inverse of the covariance matrix of the vector of variables
selected for the match. We implemented the algorithm using Stata’s “Mahapick” command. Doing so yielded
an identical set of control countries as did propensity score matching, implemented using Stata’s “psmatch2”
command with the Mahalanobis option, which accounts for the covariance among variables.
14
of patent application, and the recipient countries of patent duplication. The dependent
variable in the analysis, #𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖,𝑗𝑗,𝑡𝑡 represents the number of patented energy storage
technologies invented in origin country 𝑗𝑗 that are granted a patent in destination country 𝑖𝑖
in year 𝑡𝑡. This method of using patent data to identify the movement of technologies across
international borders was pioneered by Eaton and Kortum (1996). Johnstone and Hascic
(2009) describe the use of patent data to trace the transfer of technologies across patent
jurisdictions; Hascic and Johnstone (2011) and Johnstone (2010) constructed similar
variables to trace the diffusion of environmentally friendly technologies among OECD
countries. We estimate the following empirical model to predict the number of energy
storage technology transfers from origin country to destination country:
#𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖,𝑗𝑗,𝑡𝑡 = 𝛼𝛼𝑖𝑖,𝑗𝑗 + 𝛽𝛽1 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖 + 𝛽𝛽2 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖
+𝛽𝛽3 𝑋𝑋𝑖𝑖 + 𝛽𝛽4 𝑌𝑌𝑗𝑗 + 𝛽𝛽5 𝑍𝑍𝑖𝑖𝑖𝑖𝑖𝑖 + � 𝛿𝛿𝑡𝑡 + � 𝜃𝜃𝑖𝑖𝑖𝑖
On the model’s right-hand side, DemandPolicy and SupplyPolicy are indicator variables
equal to one when the destination country had a demand or supply-push policy in place
during year 𝑡𝑡, and zero otherwise. Additionally, the model includes dyadic dummy
variables, 𝛼𝛼𝑖𝑖,𝑗𝑗 , to control for any unobserved conditions specific to the country-pair, such as
the proximity of the countries, whether residents speak a common language, colonial ties,
etc. The dyadic dummy variables indicators account for many of the correlates of
international trade included in so-called gravity models (Santos Silva and Silvana 2006).
We also included year dummy variables 𝛿𝛿𝑡𝑡 to account for sample-wide changes in the
amount of international technology transfer over time.
The vectors 𝑋𝑋𝑖𝑖 and 𝑌𝑌𝑗𝑗 respectively represent destination and origin country variables that
change over time. The former includes the count of energy-storage patents to control for
the origin-country stock of intellectual property. Additionally, we experiment with different
control variables for the total number of patents and the number of “green” technology
patents in the origin and destination countries. As in gravity models of international trade,
we also include GDP levels for the origin and destination countries (Santos Silva and
Silvana 2011).
15
To account for intertemporal differences in bilateral trade that are not captured by these
controls, we include the value of imports from the origin to the destination country. 4 We
also included the number of energy storage patents in the destination country to account
for any competitive effects that might influence the decision to transfer technology into the
country. Finally, we account for the overall innovation landscape in both the origin and
destination countries by including each country’s total number of patented innovations.
Because the dependent variable is a non-negative integer, we estimate the model using the
Poisson quasi-maximum likelihood estimator, which does not rely on the assumption of
mean variance equality. Recent evidence indicates that this type of model is particularly
well-suited for estimating gravity equations (Santos Silva and Silvana, 2006, 2011). A
positive coefficient on B1 or B2 would indicate an increase in technology transferred into a
country following a policy’s implementation, and a larger value of B1 than B2 would
support our hypothesis that demand-pull policies are associated with a greater increase in
international technology transfer than supply-push policies.
Table 3 displays summary statistics for the variables in the model. The average annual
number of international transfers of patented energy storage technologies per country-pair
ranges from zero to 1272, with a mean value of 0.73. Approximately 10% of the
observations reflect an active demand-pull policy in the destination country, and 14%
reflect an active supply-push policy. Table 4 provides the correlations among the variables
of interest and the controls. Unsurprisingly, the number of patents generated in an origin
country is highly correlated with the number of energy storage patents in a country, the
country’s GDP, and the total number of technology transfers from an origin country.
5. Results
Table 5 displays the main empirical results using the unrestricted control group, and Table
6 displays analogous results using the matched control group. Table 7 displays results
from an analysis including additional control variables that reduce the estimation sample
size due to data availability limitations. All models for which estimates are reported include
4
These data are publicly available from the International Monetary Fund (IMF).
16
country-pair and year dummies, with robust standard errors clustered by country pair to
account for dependence among the multiple observations for each such dyad.
The results are similar regardless of the control group used. According to both Table 5 and
Table 6, the adoption of a demand-pull policy was associated with a 67 percent increase in
energy storage technologies transferred into a country. Supply-push policies were not
associated with such an increase.
Table 7 reports results for a model that includes additional controls for origin country GDP
and trade between the origin and destination country, at the cost of excluding smaller
countries for which such data were unavailable. The results in models 1, 2, and 3 are
comparable to those reported above: demand-pull policies were associated with a
subsequent increase in technology transfer, whereas supply-push policies were not.
The final model in Table 7 includes an additional control for the total annual number of
technology transfers from the origin country to any destination country, in any
technological area. This variable, which accounts for the origin country’s general
“openness” in technology transfer, is statistically significant, but its inclusion has little
effect on the estimated coefficients for the demand- and supply-push policy indicators.
Several of the control variables merit brief discussion. The number of energy storage
patents invented in an origin country strongly predicted the number of patented inventions
transferred from that country to a given destination country, confirming ex ante
expectations that the former represents the set of patents “at risk” for transfer. Further, the
total number of patents in a destination country was also positively associated with the
number of transferred inventions, which aligns with the fact that the transferred inventions
must themselves have been approved as patented inventions to appear in the data.
However, the number of energy storage patents in the destination country was not
predictive, which may reflect competing effects of inward technology transfer due to
market growth (positive) and increased competition (negative). Finally, the negative
coefficient on destination country GDP was due to the high correlation between a country’s
GDP and the total number of patents granted in the country over time.
17
Table 8 provides additional robustness checks. First, to focus on a period of time more
congruent with the implementation of demand-pull policies, we replicate our baseline
analysis for demand-pull polices only, using data from 1990 – 2011. The first demand-pull
policies were not adopted until 1999. Because fewer unobserved differences across
country-pairs are time invariant over a longer time horizon, the country-pair dummies may
better control for unobserved differences in the shorter panel. The first column in Table 8
reports results from this model. Consistent with the results from the full panel, these
results also indicate a significant increase in technology transfer into a country following
the implementation of a demand-pull policy.
The final sets of robustness checks address the possibility of influential origin or
destination countries. Specifically, we examined whether the results reported above were
driven mainly by policies passed in the U.S., the largest country in the sample to have
adopted relevant policies; or by technology transfers from China, which has been the focus
of many of trade policies and sanctions associated with technology trade. Models 2 and 3 in
Table 8 report results for models that respectively exclude China from the origin country
group, and the U.S. from the destination country group. Despite the lower power of Model 3
in particular, the estimated coefficients on the demand-pull policy variables in both models
remain positive and significant, albeit somewhat smaller in magnitude than those reported
above.
6. Conclusion
Though political leaders often extol the competitive benefits of policies aimed at promoting
innovation, there has existed limited empirical evidence of such benefits. In the clean
energy technology space in particular, many countries have promoted innovation and
deployment using a variety of policy mechanisms, bur prior research has not examined the
patterns of international competition associated with the use of different mechanisms. We
have sought to fill this gap.
In unreported analyses, we find that the policies investigated here were met with
subsequent increases innovation, as typically measured in existing studies. Relative to the
65 countries that did not pass policies, the number of energy storage patents increased in
18
the 11 countries that passed innovation-promoting policies. 5 Other studies have
interpreted this as evidence that the policies increase innovation; The purpose of this
paper is to probe further, and assess whether these policies have an impact on
international technology transfer in particular and whether there exists a differential
impact of demand- and supply-push policies on this international technology transfer.
The evidence presented above suggests that, while both demand- and supply-push policies
may increase domestic innovation rates, their effects on domestics players’ international
competitiveness varies. By rendering the domestic market more attractive, demand-pull
policies attract competition from abroad, but supply-push policies do not have this effect.
Thus, while prior evidence that innovation-supporting policies lead to greater innovation
may be correct, the observed increase may be due to the transfer of technology developed
abroad rather than from domestic innovation in the country that adopted the policy.
Our study serves as a counterpoint to Popp’s (2006) analysis of innovation and
technological diffusion following the adoption of NOx and SO2 regulations by the United
States, Japan, and Germany in the 1970s and early 1980s. Popp found that these policies
were associated with an increase in domestic innovation, but not with increased innovation
in other countries. However, Popp’s study did not directly investigate the transfer of
technologies across countries.
The analysis reported above does not necessarily imply that demand-pull or supply-push
policies are “good” or “bad.” The fact that the former attract competition from abroad likely
indicates that one of the policy’s goals – increasing consumption of a given technology – is
being met at lower cost than it would in the absence of international competition. In terms
of social welfare maximization, the international competition associated with demand-pull
policies is thus a favorable result, as it promotes a level of consumption closer to the social
optimum at a lower cost. When one of policy’s explicit goals is to develop domestic
capabilities and competitiveness in a given technological area, however, demand-pull
5
Based on a difference-in-differences analysis of annual patenting at the country-year level, controlling for
the total number of patents granted in the country in each year, annual time effects, and country fixed effects
using a poison quasi-maximum likelihood estimation with robust standard errors clustered at the country
level.
19
policies may be counterproductive. In contrast, supply-push policies may be more
successful at directly targeting domestic innovation than their demand-pull counterparts.
20
References
Acemoglu D, Linn J. (2004). “Market size in innovation: theory and evidence from the
pharmaceutical industry.” Quarterly Journal of Economics 119(3): 1049–1090.
Acemoglu D, Cutler DM, Finkelstein A, Linn J. (2006). „Did Medicare induce pharmaceutical
innovation?” American Economic Review 96(2): 103–107.
Adner R, Levinthal D. (2001). “Demand heterogeneity and technological evolution:
implications for product and process innovation.” Management Science 47(5): 611–
628.
Aghion, P, A. Dechezlepretre, D. Hemous, R. Martin, and J. Van Reenen (2011). “Carbon
taxes, path dependence and directed technical change: Evidence from the auto
industry,” working paper.
Aghion, P and Howitt P. (1992). “A model of growth through creative destruction.”
Econometrica 60(2): 323–351.
Alic, John A., David C. Mowery, Edward S. Rubin (2003). “US Technology and Innovation
Policies: Lessons for Climate Change,” Pew Center on Global Climate Change,
Arlington, VA.
Arrow, K. (1962). “Economic Welfare and the Allocation of Resources
for Invention,” p. 609-626 in The Rate and Direction of Inventive Activity: Economic
and Social Factors, Princeton University Press.
Audretsch DB, Feldman MP. (1996). “R&D spillovers and the geography of innovation and
production.” American Economic Review 86: 630–640.
Baumaud P. (1999). Tacit Knowledge in Organizations. Sage: London, UK.
Bloom, Nick, Rachel Griffith, and John Van Reenen (2002). “Do R&D Tax Credits Work?
Evidence From a Panel of Countries 1979-1997,” Journal of Public Economics, 85: 131.
Dechezlepretre, Antoine, Mathiew Glachant, Ivan Hasic, Nick Johnstone, and Yann Meniere.
(2011). “Invention and Transfer of Climate Change-Mitigation Technologies: A
Global Analysis,” Review of Environmental Economics and Policy, 5(1): 109-130.
Denholm, Paul, Erik Ela, Brendan Kirby, and Michael Milligan (2010). “The Role of Energy
Storage with Renewable Electricity,” NREL Technical Report, NREL/TP-6A2-47187.
21
Dosi G. (1982). “Technological paradigms and technological trajectories.” Research Policy
11(3): 147–162.
Eaton, J. and S. Kortum (1996). “Trade in ideas Patenting and productivity in the OECD,”
Journal of International Economics, 40(3–4): 251–278.
"President Obama Speaks from Electric Vehicle Technical Center; House Passes 90 Percent
Tax on Bailout Bonuses; Broadway Honors." Finance Wire Mar 19 2009. ProQuest.
Web. 9 Dec. 2014.
Fabrizio, Kira and LG Thomas (2012). “The Impact of Local Demand on Product Innovation
in a Global Industry,” Strategic Management Journal, 33(1): 42-64.
Feldman MP. 2000. Location and innovation: the new economic geography of innovation,
spillovers, and agglomeration. In Oxford Handbook of Economic Geography, Clark G,
Feldman M, Gertler M (eds). Oxford University Press: Oxford, UK; 373–394.
Freeman C. (1995). “The national system of innovation in historical perspective.”
Cambridge Journal of Economics 19: 5–24.
Furman JL, Porter ME, Stern S. (2002). “The determinants of national innovative capacity.”
Research Policy 31: 899–933.
Geroski, P. A. (1995). 'Do spillovers undermine the incentive to innovate?' In S. Dowlick, ed.,
Economic Approaches to Innovation, p. 76-97, Cheltenham: Edward Elgar.
Griliches, Zvi. (1979). "Issues in assessing the contribution of research and development to
productivity growth." The Bell Journal of Economics, 92-116.
Griliches, Zvi. (1992). “The search for R&D spillovers,” No. w3768. National Bureau of
Economic Research, 1992.
Griliches Z. (1995). R&D and productivity: econometric results and measurement issues. In
Handbook of the Economics of Innovation and Technological Change, Stoneman P
(ed). Blackwell: Oxford, UK; 52–89.
Grossman G, Helpman E. (1991). Innovation and Growth in the Global Economy. MIT Press:
Cambridge, MA.
Hascic, I. and N. Johnstone (2011). “The Clean Development Mechanism and International
Technology Transfer: Empirical Evidence on Wind Power,” Climate Policy 11(6):
1303-1314.
22
Hall, Bronwyn and John Van Reenen (2000). “How Effective are Fiscal Incentives for R&D?
A Review of the Evidence,” Research Policy, 29, p. 449-469.
Horbach, Jens (2008). “Determinants of environmental innovation—New evidence from
German panel data sources,” Research Policy, 37(1), p. 163-173.
International Energy Agency (IEA) (2007). Global renewable energy policies and measures
database. IEA, Paris, France (http://www.iea.org/textbase/pm/index.html)
Jaffe, Adam B. (1986). "Technological opportunity and spillovers of R&D: Evidence from
firms' patents, profits and market value," American Economic Review, 76(5): 9841001.
Jaffe AB. (1989). “Real effect of academic research.” American Economic Review 79(5): 957–
970.
Jaffe, A.B, R.G. Newell, and R.N. Stavins (2002). “Environmental Policy and Technological
Change,” Environmental and Resource Economics, 22 (1-2), 41-70.
Jaffe AB, Trajtenberg M, Henderson, R. (1993). „Geographic localization of knowledge
spillovers as evidenced by patent citations.” Quarterly Journal of Economics 108:
577–598.
Jaffe AB, Trajtenberg M. (1996). “Flows of knowledge from universities and federal
laboratories: modeling the flow of patent citations over time and across institutional
land geographic boundaries.” Proceedings of the National Academy of Sciences 93:
12671–12677.
Johnstone, N. and I. Hascic (2009). “Indicators of Innovation and Transfer in
Environmentally Sound Technologies: Methodological Issues,” OECD working paper
ENV/EPOC/WPNEP(2009)1, Working Party on National Environmental Policies.
Johnstone, N. (2010). “Climate Policy and Technological Innovation and Transfer: An
Overview of Trends and Recent Empirical Results,” OECD Working Party on Global
and Structural Policies, ENV/EPOC/GSP (2010)10.
Johnstone, N, I Hascic, and D Popp (2010). “Renewable energy policies and technological
innovation: Evidence based on patent counts,” Environmental Resource Economics,
45, 133-155.
Kline, Stephen J. and Nathan Rosenberg (1986). “An Overview of Innovation,” p, 275 – 305
in The Positive Sum Strategy: Harnessing Technology for Economic Growth, Ralph
Landau and Nathan Rosenberg, eds., National Academy Press, Washington, DC.
23
Krugman P. (1991). “Increasing returns and economic geography.” Journal of Political
Economy 99(3): 483–499.
Lanjouw, Jean Olson and Ashoka Mody. (1996). “Innovation and the international diffusion
of environmentally responsive technology,” Research Policy, 25: 549-571.
Marshall A. (1920). Principles of Economics. MacMillan: New York.
Martin, S. and J.T. Scott. (2000). “The nature of innovation market failure and the design of
public support for private innovation.” Research Policy, 29(4–5), 437-447.
Mowery DC, Oxley JE. (1995). “Inward technology transfer and competitiveness: the role of
national innovation systems.” Cambridge Journal of Economics 19(1): 67–93.
Mowery DC, Rosenberg N. (1979). “The influence of market demand upon innovation: a
critical review of some recent empirical studies.” Research Policy 9:102–153.
Nelson RR. (1982). “The role of knowledge in R&D efficiency.” Quarterly Journal of
Economics 97(3): 453–470.
Nelson RR. (1993). National Innovation Systems: A Comparative Analysis. Oxford University
Press: New York.
Newell, Richard G., Adam B. Jaffe, and Robert N. Stavins. (1999). “The Induced Innovation
Hypothesis and Energy-Saving Technological Change,” The Quarterly Journal of
Economics, 114(3): 941-975.
Polanyi M. (1966.) The Tacit Dimension. Doubleday: Garden City, NY.
Popp, David (2002). “Induced innovation and energy prices,” American Economic Review
92(1), 160–180.
Popp, David (2006). “International innovation and diffusion of air pollution control
technologies: The effect of NOx and SO2 regulation in the US, Japan, and Germany,”
Journal of Environmental Economics and Management, 51: 446-71.
Porter ME. (1990.) The Competitive Advantage of Nations. Free Press: New York.
Porter ME. (1998.) “Clusters and the new economics of competition.” Harvard Business
Review 76(6): 77–91.
Romer PM. (1990). “Endogenous technological change.” Journal of Political Economy 98:
S71–S102.
Santos Silva, J.M.C. and Tenreyro, Silvana (2006). “The Log of Gravity,” The Review of
Economics and Statistics, 88(4), pp. 641–658.
24
Santos Silva, J.M.C. and Tenreyro, Silvana (2011). “Further simulation evidence on the
performance of the Poisson pseudo-maximum likelihood estimator,” Economics
Letters, 112(2): 220–222.
Schmookler J. (1966). Invention and Economic Growth Development. Harvard University
Press: Cambridge, MA.
Sweet, C and R Tracy (2011). “Solar Firms Seek Duties in China,” Wall Street Journal,
October 20, 2011.
The Economist (2014), “Tesla’s Electric Man,” December 6, 2014.
25
Table 1: Difference of Means Tests: Policy-passing countries versus all other countries, for years
1973 (before first Supply-push policy) and 1998 (before first Demand-pull policy)
Variable
Mean
Non-policy
countries
(1)
Mean
Policy
countries
(2)
Difference
(3)=(1)-(2)
1973
%Growth in ES transfersa
-0.00
-0.02
0.02**
# ES transfersa
0.01
0.20
-0.19**
# ES patentsb
0.35
56.45
-56.11**
b
# Patents
448.76
23365.55
-22916.79**
GDP (million $US) b
21,849.97
271,538.10
-249,688.10**
1998
%Growth in ES transfersc
-0.00
-0.00
-0.00
# ES transfersc
0.04
0.56
-0.52**
# ES patentsd
3.89
468.91
-465.02**
d
# Patents
1211.74
75641
-74429.26**
GDP (million $US)d
102,018.60
1,799,894.00
-1,697,876**
acountry-pair level observations, N=29494 for non-policy countries and 2232 for policy countries.
bcountry-level observation, N=144 for non-policy countries and 11 for policy countries
ccountry-pair level observations, N=29520 for non-policy countries and 2255 for policy countries.
dcountry-level observation, N=144 for non-policy countries and 11 for policy countries
*difference significant at 5% level **difference significant at 1% level
Table 2: Difference of Means Tests: Policy-passing countries versus control countries, for years
1973 (before first Supply-push policy) and 1998 (before first Demand-pull policy)
Variable
Mean
Non-policy
countries
(1)
Mean
Policy
countries
(2)
Difference
(3)=(1)-(2)
1973
%Growth in ES transfersa
-0.01
-0.02
0.01*
# ES transfersa
0.07
0.20
-0.13**
# ES patentsb
2.10
56.45
-54.35*
# Patentsb
2907.70
23,365.55
-20,457.85
GDP (million $US) b
52,391.88
271,538.10
-219,146.20
1998
%Growth in ES transfersc
-0.01
-0.00
-0.00
c
# ES transfers
0.13
0.56
-0.43*
# ES patentsd
21.50
468.91
-447.41
# Patentsd
6806
75641
-68835
d
GDP (million $US)
358,629.40
1,799,894.00
-1,441,265
acountry-pair level observations, N=2038 for non-policy countries and 2232 for policy countries.
bcountry-level observation, N=10 for non-policy countries and 11 for policy countries
ccountry-pair level observations, N=2050 for non-policy countries and 2255 for policy countries.
dcountry-level observation, N=10 for non-policy countries and 11 for policy countries
*difference significant at 5% level **difference significant at 1% level
26
Table 3: Summary Statistics (N=66,664 country-pair observations)
Variable
Mean
Std. Dev.
Min
Max
#ES transfers
0.73
13.57
0
1272
DemandPolicy
0.09
0.28
0
1
SupplyPolicy
0.14
0.35
0
1
#ES Pats_origin
54.13
46.10
0
8536
#Pats_origin
8094.16
39639.22
1
466637
#ES Pats_ destination
271.64
1007.55
0
8536
#Pats_ destination
36838.30
84499.34
1
466637
GDP_ origin
251925.20
973675.50
18.43
15000000
GDP_destination
1001873.00
2094610
4317.04
15000000
Imports
1.46e+09
7.84e+09
0
3.40e+11
# Transfers_origin
4086.685
16334.450
0
170173
NOTE that in the variable names, “destination” refers to the destination country of the transfer and
“origin” refers to the origin of the transferred technology.
Table 4: Correlations (N=66,664 country-pair observations)
1
2
3
4
5
6
7
8
9
10
1
DemandPolicy
2
SupplyPolicy
0.34
3
#ES transfers
0.08 0.07
4
#ES Pats_origin
0.02 0.02 0.31
5
#Pats_origin
0.00 0.00 0.28 0.86
6
#ES Pats_ destination 0.46 0.11 0.06 -0.00 -0.01
7
#Pats_ destination
0.38 0.23 0.08 -0.01 -0.01 0.86
8
# Transfers_origin
0.01 0.00 0.26 0.66 0.81 -0.00 -0.00
9
GDP_ origin
0.04 0.02 0.20 0.54 0.74 0.01 -0.00 0.78
10 GDP_destination
0.48 0.53 0.12 0.01 -0.01 0.52 0.72 0.00 0.01
11 Imports
0.12 0.16 0.40 0.19 0.26 0.10 0.15 0.33 0.35 0.24
NOTE that in the variable names, “destination” refers to the destination country of the transfer and
“origin” refers to the origin of the transferred technology.
27
Table 5: Energy Storage International Technology Transfer as a Function of Destination Country
Policies, using all available control countries, 1970-2011
(3)
0.663
(0.133)**
SupplyPolicy
-0.057
0.020
(0.173)
(0.145)
ln_#ES Pats _origin
1.114
1.147
1.115
(0.063)**
(0.085)**
(0.063)**
ln_#Pats _origin
-0.180
-0.180
-0.181
(0.087)*
(0.119)
(0.087)*
ln_#ES Pats_ country
0.065
0.082
0.064
(0.066)
(0.067)
(0.067)
ln_#Pats _country
0.905
0.891
0.909
(0.091)**
(0.095)**
(0.090)**
ln_GDP_country
-0.528
-0.715
-0.528
(0.153)**
(0.163)**
(0.152)**
# Country Pairs
1018
1018
1018
N
40,649
40,649
40,649
* p<0.05; ** p<0.01 Robust standard errors, clustered by country pair, in parentheses
NOTE that in the variable names, “destination” refers to the destination country of the transfer and
“origin” refers to the origin of the transferred technology. All regressions include country pair fixed
effects and year fixed effects.
DemandPolicy
(1)
0.661
(0.134)**
(2)
28
Table 6: Energy Storage Technology Transfer as a Function of Destination Country Policies, using
selected comparable control countries, 1970-2011
DemandPolicy
SupplyPolicy
ln_#ES Pats_ origin
ln_#Pats _origin
ln_#ES Pats _country
ln_#Pats_ country
ln_GDP_country
# Country Pairs
N
(1)
0.659
(0.132)**
1.093
(0.066)**
-0.146
(0.094)
-0.001
(0.081)
0.979
(0.144)**
-0.532
(0.214)*
591
24,822
(2)
-0.068
(0.178)
1.117
(0.085)**
-0.141
(0.122)
-0.044
(0.100)
1.092
(0.175)**
-0.814
(0.242)**
591
24,822
(3)
0.664
(0.132)**
0.037
(0.153)
1.094
(0.066)**
-0.147
(0.095)
-0.002
(0.081)
0.987
(0.137)**
-0.536
(0.209)*
591
24,822
* p<0.05; ** p<0.01 Robust standard errors, clustered by country pair, in parentheses
NOTE that in the variable names, “destination” refers to the destination country of the transfer and
“origin” refers to the origin of the transferred technology.
All regressions include country pair fixed effects and year fixed effects.
29
Table 7: Energy Storage Technology Transfer as a Function of Destination Country Policies, using
selected comparable control countries, with additional controls, 1970-2011
(1)
(2)
(3)
(4)
DemandPolicy
0.709
0.718
0.670
(0.146)**
(0.147)**
(0.126)**
SupplyPolicy
-0.066
0.063
0.055
(0.174)
(0.150)
(0.159)
ln_#ESPats _origin
1.113
1.141
1.114
1.019
(0.082)**
(0.100)**
(0.081)**
(0.062)**
ln_#Pats _origin
-0.256
-0.248
-0.258
-0.814
(0.124)*
(0.139)
(0.123)*
(0.165)**
ln_#ESPats _country
-0.067
-0.127
-0.070
-0.036
(0.093)
(0.114)
(0.092)
(0.072)
ln_#Pats_ country
1.180
1.423
1.199
1.102
(0.235)**
(0.287)**
(0.228)**
(0.203)**
ln_GDP_country
-0.944
-1.270
-0.946
-0.924
(0.223)**
(0.218)**
(0.224)**
(0.227)**
ln_GDP_origin
-0.016
-0.067
-0.017
0.018
(0.256)
(0.290)
(0.260)
(0.262)
ln_imports
0.148
0.185
0.141
0.229
(0.211)
(0.245)
(0.205)
(0.208)
ln_#Transfers_origin
0.955
(0.196)**
# Country Pairs
402
402
402
402
N
15,489
15,489
15,489
15,489
* p<0.05; ** p<0.01 Robust standard errors, clustered by country pair, in parentheses
NOTE that in the variable names, “destination” refers to the destination country of the transfer and
“origin” refers to the origin of the transferred technology.
All regressions include country pair fixed effects and year fixed effects.
30
Table 8: Energy Storage Technology Transfer as a Function of Destination Country Policies,
Robustness tests
1990-2011
(1)
0.439
(0.092)**
Excluding China
Excluding U.S.
(2)
(3)
DemandPolicy
0.622
0.241
(0.122)**
(0.105)*
SupplyPolicy
0.042
0.012
(0.158)
(0.137)
ln_#ESPats _origin
0.967
1.008
1.042
(0.066)**
(0.055)**
(0.059)**
ln_#Pats _origin
-0.545
-0.658
-0.487
(0.204)**
(0.138)**
(0.138)**
ln_#ESPats _country
0.036
0.028
0.182
(0.085)
(0.063)
(0.066)**
ln_#Pats_ country
0.775
0.908
0.814
(0.250)**
(0.134)**
(0.153)**
ln_#Transfers_origin
0.659
0.851
0.609
(0.153)**
(0.159)**
(0.138)**
ln_GDP_country
-0.340
-0.468
-0.428
(0.226)
(0.185)*
(0.212)*
Country-pairs
454
583
529
N
9,988
24,486
22,218
* p<0.05; ** p<0.01 Robust standard errors, clustered by country pair, in parentheses
NOTE that in the variable names, “destination” refers to the destination country of the transfer and
“origin” refers to the origin of the transferred technology.
All regressions include country pair fixed effects and year fixed effects.
31