Environmental Policy Stringency and Technological Innovation: Evidence from Patent Counts by Ivan Hascic ‡ , Nick Johnstone ‡ and Christian Michel † May 7 th , 2008 Abstract This paper examines the impact of public environmental policy, as reflected in expenditures on pollution abatement and control, on innovations in environment-related technology. The analysis is conducted using patent data for a panel of 16 countries between 1985 and 2004. It is found that there are important differences in innovation effects of resources spent in the public vs. private sector and resources spent on pollution control activities vs. directly on R&D. These results are broadly confirmed with a subsequent analysis on a broader cross-section of 33 countries over the period 2001-2006, using an alternative measure of environmental policy stringency. JEL codes: O31; O38; Q55; Q58 Keywords: Environmental Policy; Technological Innovation; Patents ‡ [email protected], [email protected]Empirical Policy Analysis Unit, OECD Environment Directorate, 2 rue André Pascale, 75775 Paris Cedex 16, France † Department of Economics, Oxford University, Manor Road, Oxford, United Kingdom
28
Embed
Environmental Policy Stringency and Technological ... · Environmental Policy Stringency and Technological Innovation: Evidence from Patent Counts by Ivan Hascic ‡, Nick Johnstone
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Environmental Policy Stringency and Technological Innovation:
Evidence from Patent Counts
by
Ivan Hascic‡, Nick Johnstone
‡ and Christian Michel
†
May 7th, 2008
Abstract
This paper examines the impact of public environmental policy, as reflected in expenditures on
pollution abatement and control, on innovations in environment-related technology. The analysis is
conducted using patent data for a panel of 16 countries between 1985 and 2004. It is found that
there are important differences in innovation effects of resources spent in the public vs. private
sector and resources spent on pollution control activities vs. directly on R&D. These results are
broadly confirmed with a subsequent analysis on a broader cross-section of 33 countries over the
period 2001-2006, using an alternative measure of environmental policy stringency.
There is currently much interest in the role of public policy in inducing innovations in technologies
which help reduce environmental impacts of economic activity. In many industrialized countries,
significant progress has been achieved during the past several decades on this front. For example,
emissions of pollutants into air and water have been greatly reduced1 and some advances have been
achieved in waste management.2 Most likely, this has been achieved due to structural changes in
economic activity (e.g., less emission-intensive production such as coal fired power plants), input
substitution (e.g., using coal with lower sulphur content), as well as via technological
improvements (incl. end-of-pipe solutions such as scrubbers, or production process innovations
such as fluidized bed combustion).
Understanding the factors that have determined this process is important for several
reasons. First, despite significant progress achieved to date, air and water pollution remains an
important public policy issue due its negative impacts on human health (see e.g., Cohen et al. 2005)
and ecosystem functions (see e.g., Islam and Tanaka 2004 and Lorenz 1995). Moreover, further
emissions reductions will require action on the part of more diffuse sources of pollution and may
therefore be more difficult to achieve, as their identification and measurement are complicated.
Finally, while emissions of many “traditional” pollutants are currently more-or-less controlled, new
“emerging” pollutants may become relatively more important in the future. In this context,
technological innovation is important because it allows society to further reduce environmental
impacts or to achieve a given environmental goal at lesser cost (see e.g., Kneese and Schultze
1977).
1 Between 1990 and 2005, emissions of SOx and NOx have fallen by 72% and 33% respectively in the
European Union (EU15) and 37% and 26% in the US. In some OECD countries emissions have actually
increased, notably in Australia and New Zealand with 25%-58% increase in emissions. Emissions causing
increased levels of water pollution have also been reduced in many countries. For example, the proportion of
population connected to public wastewater treatment plants has increased from 46% to 68% in OECD
countries during the last 25 years. However, enormous differences remain across countries – while as much
as 98% of population is connected in the Netherlands and the UK, the share is only 35% in Mexico and
Turkey (OECD 2007a). 2 Between 1990 and 2005, the volume of municipal waste generated per capita has remained stable in the US
(750 kg), had dropped slightly in Japan (from 410 to 400 kg), and has increased sharply in the European
Union (EU15) (from 430 to 570 kg) (OECD 2007a).
2
In the last several decades, OECD countries have introduced a number of policy measures
with the objectives to reduce environmental impacts of economic activity. However, it is difficult
to predict the effect of such policies on the pattern of technological innovation. While private (firm-
level) incentives to environment-friendly innovations may play some role3, it is public policy that
often plays the pivotal role in creating demand for technological innovation in environment-related
technologies, although its impact may vary across countries, pollutants, and over time.
In 1932, John Hicks observed that a change in the relative prices of factors of production
will motivate firms to invent new production methods in order to economise the use of a factor
which has become relatively expensive. This idea, originally developed in the context of labour
economics, came to be known as the “induced innovation hypothesis”. Applied to the public policy
framework, this implies that if governments could affect relative input prices, or otherwise change
the opportunity costs associated with the use of environmental resources, firms‟ incentives to seek
improvements in production technology would be increased. Indeed, since markets often fail to put
a price on environmental resources, the price of many environmental assets is to a large extent
formed by government regulation. Depending on the stringency of a regulation, the change in
opportunity costs of pollution then translates into increased cost of some factors of production.
Since this effect is unobservable to a researcher, pollution abatement and control expenditure
(PACE) can serve as an imperfect proxy for the changes in opportunity costs involved.
PACE has been used to examine the links between environmental regulation and
innovation in two distinct manners. In one case, the focus has been on the effects of PACE
expenditures on (plant-, sector-, or country-level) differences in productivity growth by examining
whether a given level of PACE has more or less greater impact on productivity (e.g., Gray and
Shadbegian 2003; Morgenstern, Pizer and Shih 2001; Jorgenson and Wilcoxen 1990; Gollop and
Roberts 1983; see also Jaffe et al. 1995 and papers cited therein). The basic question is whether, the
impact of a given level of PACE on productivity is more or less than unity. For instance, some
investments targeted at reducing environmental impacts may increase (or decrease) the efficiency
3 For instance, recycling of secondary materials to reduce input costs, consumer demand for „defensive‟
measures, etc.
3
associated with the use of other factors of production in production more generally (see Labonne
and Johnstone 2007 for discussion on this issue; see also Morgenstern, Pizer and Shih 2001).
In the second case, the focus has been on the effects of PACE on one aspect of productivity
– notably technological innovation, using patent data (e.g., Popp 2003; Brunnermeier and Cohen
2003; Jaffe and Palmer 1997; Lanjouw and Mody 1996). However, empirical evidence on the
effect of stringency of environmental policy on innovative behaviour remains limited, both with
respect to the overall effects of environmental policy on technological innovation, as well as the
more specific question of the extent to which this is reflected in patent activity. Nevertheless, there
is now increasing empirical evidence to support the contention that environmental policies do lead
to technological innovation. For recent reviews of the empirical literature on this theme see
Vollebergh (2007) and Jaffe, Newell and Stavins (2002).
This paper continues in the tradition of the latter approach and studies the effects of public
environmental policy (as proxied by PACE) and other factors on innovation in environmental
technologies, using patent data for an unbalanced panel of 16 countries for the period 1985-2004.
Unlike previous studies which used PACE at the sectoral level, this is the first econometric study
using PACE data at the cross-country level.4 The key hypothesis to be explored is the effect of
public environmental policy on innovation; and in particular, the possibly differential effect across
the alternative economic sectors undertaking PAC activities (i.e. public sector, private sector,
specialized producers). The role of government environmental R&D is also examined.
2. Data construction and interpretation
2.1. Patent counts as a measure of environment-related innovation
Patent data have been used as a measure of technological innovation because they focus on outputs
of the inventive process (Griliches 1990). This is in contrast to many other potential candidates
(e.g. research and development expenditures, number of scientific personnel, etc.) which are at best
imperfect indicators of the innovative performance of an economy since they focus on inputs.
4
Moreover, patent data provide a wealth of information on the nature of the invention and the
applicant, the data is readily available (if not always in a convenient format), discrete (and thus
easily subject to statistical analysis), and can be disaggregated to specific technological areas.
Significantly, there are very few examples of economically significant inventions which have not
been patented (Dernis and Guellec 2001).
However, patents are an imperfect measure of technological innovation for several reasons.
First, there is variation in the propensity for inventors to patent across countries and sectors. This is
due in part to the level of protection afforded by the patent, but also to the possibility of
appropriating rents from innovation by other means depending upon market conditions (e.g.,
industrial secrecy). In the empirical section of this paper, this concern over differences in the
propensity to patent is addressed by including a variable reflecting the overall patenting activity to
control for these effects across countries and over time.
Second, it is difficult to distinguish between the „value‟ of different patents on the basis of
patent applications. Most clearly, the use of unweighted patent counts would attribute the same
importance to patents for which there were no successful commercial applications with those which
are highly profitable. In this paper, this concern is addressed by using data on patent applications to
the European Patent Office (EPO), rather than individual patent offices.5 Through the EPO, the
applicant designates as many of the EPO member states for protection as it desires, rather than
applying to individual European patent offices among the 32 contributing countries. If the
application is successful, the patent is transferred to the individual national patent offices
designated for protection in the application. Given that EPO applications are more expensive than
applications to national patent offices, inventors typically first file a patent application in their
home country, and then apply to the EPO if they desire protection in multiple European countries
due to perceived market opportunities. As such, patent applications to the EPO are likely to be of
greater commercial value than the mean value of patent applications at national patent offices.
4 It is important to study these effects in a cross-country manner because large differences in per capita
emissions across countries exist (e.g., Australia‟s SOx emissions are almost 5-times higher than the OECD
average (OECD 2007a)).
5
While the use of EPO applications introduces a „quality‟ threshold to ensure that only
relatively valuable patents are included in the analysis, it introduces a potential source of bias.
While the European market is significant, it is still expected that there will be some bias toward
applications from European inventors (see Dernis and Guellec 2001). For a given invention, a
German inventor will be more likely to patent at the EPO than an American inventor. In the
empirical analysis undertaken in this study the bias associated with the use of EPO applications is
addressed through the inclusion of both country fixed effects and a control variable reflecting data
on total EPO applications by inventor country for all technology areas.
And finally, it can be difficult to identify relevant patent applications. Drawing upon
existing efforts to define „environmental‟ activity in sectoral terms, some previous studies have
related patent classes to industrial sectors using concordances (e.g., Jaffe and Palmer 1996). The
weaknesses of such approach are twofold. First, if the industry of origin of a patent differs from
industry of use of the patent, then it is not clear to which industrial sector a patent should be
attributed in the analysis. This is important when studying specifically “environmental” technology
because in this case the demand (users of technology) and supply (inventors of technology) of
environmental innovation may involve different entities. Often, “environmental” innovations
originate in industries which are not specifically environmental in their focus. For example,
technologies aimed at reducing wastewater effluents from the pulp & paper industry are often
invented by the manufacturing or chemicals industry (see e.g., Popp et al. 2007). On the other hand,
some “environmental” industries invent technologies which are widely applicable in non-
environmental sectors (e.g., processes for separation of packaging waste; separation of vapours and
gases).
More fundamentally, sectoral classifications are, by definition, based on commercial
outputs. As such there will be a bias toward the inclusion of patent applications from sectors that
produce environmental goods and services. The application-based nature of the patent
classification systems allows for a richer characterisation of relevant technologies. Consequently,
5 Using data on application, rather than granted patents, is more useful for international comparisons because
granting frequency varies across countries and over time (see e.g., Griliches 1990).
6
in this study patent classifications are used, rather than those of industrial or sectoral
classifications.6 Specifically, relevant patents were identified using the International Patent
Classification (IPC) system. However, IPC classes may be too broad for many areas of
“environmental” technology, leading to two possible types of error when searching for relevant
patents – inclusion of irrelevant patents and exclusion of relevant patents from the selected
classifications. Therefore keyword searches were used to filter only the relevant patents.
Patent data were extracted from the OECD Patent Database (OECD 2007b) using a search
algorithm based on a selection of IPC classes combined with keyword searches to target specific
areas of environment-related technology (Annex 1 gives the list of classes included; for the
complete search strategy see Schmoch 2003).7 The patent data are used to construct counts of
patent applications to the EPO in selected areas of environmental technology (air pollution, water
pollution, waste disposal, noise protection, and environmental monitoring), classified by inventor
country (country of residence of the inventor) and priority date (the earliest application date within
a given patent family). A panel of patent counts for a cross-section of all countries and over a time
period of 27 years (1978-2004) was obtained. Figure 1 shows the total number of EPO patent
applications by OECD countries in the five environmental domains. It shows that while water and
air pollution innovations have been increasing rapidly, the growth has been slower in the fields of
noise protection and environmental monitoring. Innovations related to waste disposal reached a
peak in 1991 and have declined since.
(Insert Figure 1 about here.)
Figure 2 gives patent counts in environmental technology for selected countries which have
exhibited significant levels of innovation. Germany has the highest number of patents, but relative
to the US and Japan, this partly reflects the „home bias‟ in EPO applications. France and the UK
6 While Jaffe and Palmer (1996) used patent totals (environmental and non-environmental patents) to study
the effect of environmental regulation on innovation, Lanjouw and Mody (1995) and Brunnermeier and
Cohen (2003) focus on environmental patents only, and their approach is thus similar to ours. However,
details on the selection of patent classes they used are not provided. 7 Following the discussion above, the search strategy includes not only „environmental‟ patent classes
covering end-of-pipe innovations, but also more general patent classes covering innovations related to
changes in production processes. In the absence of inclusion and exclusion keywords, the search algorithm
could overstate the former relatively to the latter.
7
both have at least 1000 patent applications over the period. These five countries represent between
74% and 84% of patent applications in each of the five domains. Germany alone, is responsible for
the highest number of filings in air, water, waste, and noise, while environmental monitoring is
dominated by the US and Japan.
(Insert Figure 2 about here.)
While Germany, Japan, the US, France and the UK are consistently important in
environmental technologies examined, other significant innovators in specific areas have included
Sweden (air), the Netherlands (water, monitoring), Italy (waste, noise) and Switzerland (noise)
(Table 1).
(Insert Table 1 about here.)
However, a comparison of the productivity of inventive activity across countries needs to
account for relative differences in the size of countries‟ scientific capacity.8 In Table 2, the counts
are weighted by country‟s gross domestic expenditure on R&D (GERD) to yield a measure of
patent intensity. On this basis, Germany as well as a number of smaller countries such as Austria,
Denmark, Switzerland, and Finland achieve the highest innovation output per dollar of R&D
expenditure.
(Insert Table 2 about here.)
2.2. Pollution abatement and control expenditures
Public policy may induce innovation by changing relative factor prices or introducing production
constraints. However, measurement of this effect is complicated because cross-country (or cross-
sectoral) data on regulatory stringency are rarely available or are not commensurable.
Consequently, various types of proxies have been used in the literature, including PACE measured
at the macroeconomic (e.g., Lanjouw and Mody 1996) or sectoral level (e.g., Brunnermeier and
Cohen 2003), the frequency of inspection visits (e.g., Jaffe and Palmer 1997), or various types of
derived measures (e.g., Johnstone et al. 2007).
8 For example, Madsen (2007) used the ratio of patents and real R&D expenditures as an indicator of
countries‟ research productivity.
8
The use of PACE data is very common in the empirical literature because it is one of the
few sources of quantitative information on environmental “compliance costs”. PACE includes
spending on PAC activities that are defined as “purposeful activities aimed directly at the
prevention, reduction and elimination of pollution or nuisances arising as a residual of production
processes or the consumption of goods and services” (OECD 2007c).9 This definition excludes
expenditure on natural resource management and risk prevention (such as prevention of natural
disasters and hazards), on nature protection (such as the protection of endangered species, the
establishment of natural parks and green belts), and on the exploitation and mobilisation of natural
resources (such as the supply of drinking water). Also excluded is expenditure that may primarily
satisfy health and safety requirements (such as expenditure intended for workplace protection) or
expenditure on the improvement of the production process for commercial or technical reasons,
even when they have environmental benefits (OECD 2007c).
However, PACE is only an imperfect measure of regulatory stringency. Several reasons
have been identified in the literature, including (a) the difficulty of identifying expenditures on
environmental compliance compared to what they would have been in the absence of
environmental regulations. The difficulty of establishing an appropriate baseline arises because
even in the absence of government regulation firms may still invest in such projects in order to
limit their potential exposure to liability and improve their environmental image with customers
(Jaffe et al. 1995); (b) Next, there is an important distinction between end-of-pipe solutions and
production process innovations, suggesting that it may be difficult for respondents to assign
expenditures to the latter. Specifically, firms may be unable to distinguish between the different
investment motives associated with adoption of integrated technologies. For example, what
proportion of expenditure on a new production process that increases material efficiency (and thus
reduces input costs) should be assigned to PAC? (see e.g., Lanjouw and Mody 1996); (c) Further,
firms could have an incentive to “strategically” overstate their PAC expenditures in order to
9 In this paper, consistent with this definition, the PACE data include only expenditure that is incurred
directly for PAC purposes (e.g., as a consequence of government environmental policies). Expenditure that
has positive environmental effects without being directly motivated by environmental concerns (e.g., energy
efficiency) is not included here. For the complete definition see Annex 2.
9
encourage regulators to weaken the degree of regulatory stringency, a common concern with
survey data; (d) Another concern associated with the use of aggregate measures of PACE to proxy
for stringency relates to cross-country differences in industrial composition. Countries with a lot of
polluting industry will have relatively high environmental compliance costs, regardless of the
stringency of their regulations (Levinson 1999); (e) Finally, despite significant efforts undertaken
to date, collection of PACE data is not fully harmonized across countries (OECD 2007c).
Despite these shortcomings, the PACE data is a rare source of information on the
opportunity costs created by countries‟ environmental policies and as such, if handled and
interpreted carefully, can be useful to study the effects of public environmental policies on
technological innovation. In particular, changes in opportunity costs due to increased regulatory
stringency, as proxied by higher PACE, are hypothesised to increase innovative behaviour targeted
at reducing environmental impacts of economic activity.
The PACE data used in this paper have been obtained from a series of annual surveys
published in OECD (2007c; 2003; 1996). The data are disaggregated into six environmental
domains -- including air, water, waste, land, noise, and monitoring (Table 3). In addition,
expenditures are disaggregated by the nature of the costs incurred, including (a) investment
expenditure, (b) internal current expenditure, and (c) transfer payments (such as subsidies and fees)
that are directly aimed at pollution abatement and control according to the abater principle (i.e.
sector where the PAC activity occurs) (OECD 2007c).10
Hence, PAC expenditure comprises actual
outlays and is thus conceptually different from PAC cost.
(Insert Table 3 about here.)
PACE is also disaggregated by the economic sector where the PAC activity occurs,
including the public sector, the business sector, and private and public specialised producers of
PAC services. Public PAC measures, which mainly concern waste and wastewater treatment, may
either be done directly by governments (central, regional, and local) and government agencies
10
Excluded are (a) calculated cost items (e.g. depreciation of fixed capital, cost of capital) as only actual
outlays are recorded, and (b) payments of interest, fines and penalties for non-compliance with environmental
regulations or compensations to third parties etc. as they are not directly linked with a PAC activity (OECD
2007c).
10
(further referred to as public sector) or be purchased as services from publicly-owned firms (further
referred to as public specialized producers). Private PAC measures, which mostly relate to
treatment or prevention of pollution to air and water and hazardous waste disposal from firm‟s own
operating activities, may either be done directly by the business sector or be purchased as services
from private specialized producers.11
Our dataset spans 30 countries for the period from 1985 to 2004. Three alternative PACE
variables are constructed representing PACE by the public sector including public specialized
producers (PACE_Public), PACE by the private sector (PACE_Private), and a dummy variable
(D_PrivateSP) with a unit value when data on private specialized producers is available and zero
otherwise.12
The PACE variables constructed and their interpretation are summarized in Table 4.
(Insert Table 4 about here.)
While there are a number of missing observations, Figures 3 and 4 provide time-series of
expenditures by the public and private sectors for selected countries in which the data is relatively
complete. In the case of public sector PACE Germany and Denmark appear to have relatively high
expenditures. For private sector PACE the two countries with the highest average percentages
(Czech Republic and Poland) have fallen recently.
(Insert Figures 3 and 4 about here.)
2.3. Other explanatory variables
In addition to regulatory stringency, which may induce innovation indirectly, governments often
encourage innovation directly through targeted R&D spending. Since PAC expenditures do not, by
definition, include expenditures on R&D, we use data on government budget appropriations and
outlays for R&D with the objective of control and care of the environment (GBAORD_Env). The
11
Specialized producers have grown in importance over the recent years as many of these activities have
increasingly been privatised or outsourced (e.g., municipal waste collection services, water and wastewater
treatment) (OECD 2007c). 12
Data on PACE by the public sector (PACE_Public) reported in OECD (2007c, 2003, 1996) include PACE
by public specialized producers. However, data on PACE by the business sector reported in OECD (2007c,
2003) do not, by construction, include PACE by private specialized producers (PACE_PrivateSP). This data
is available separately for some countries and some years. In order to avoid losing observations, and still be
able to isolate the “average” effect of private specialized producers, a new variable is constructed as a sum of
those two, PACE_Private = PACE_Business + PACE_PrivateSP, and a dummy variable (D_PrivateSP) is
created with a unit value when the data is available and zero otherwise. However, data on PACE by the
11
data are taken from the OECD Research and Development Statistics database (OECD 2007d) and
are normalized by GDP. The sign on this variable is expected to be positive.13
Aside from public policy, there are other important determinants of patenting activity for
environment-friendly technologies. Above all, the propensity of inventors from a particular country
to patent is likely to change over time, both because different strategies may be adopted to capture
the rents from innovation (e.g., Cohen et al. 2000) and because legal conditions may change
through time (e.g., Ginarte and Park 1997). In addition, inventors from non-European countries are
less likely to patent at the EPO (home country bias). For meaningful empirical analyses it is
therefore important to control statistically for these differences in the propensity to patent. As such
a variable was included reflecting total EPO patent applications (EPO_Total) filed across the
whole spectrum of technological areas (not only environmental). This variable thus serves both as a
„scale‟ and as a „trend‟ variable in that it controls for control for differences in the effects of the
size of an economy, its research capacity, etc. on innovation as well as changes in general
propensity to patent over time and across countries. The sign on this variable is expected to be
positive. Table 5 provides basic descriptive statistics for the dependent and explanatory variables.
(Insert Table 5 about here.)
3. Empirical model and results
An empirical model is developed in order to evaluate the effects of environmental policy and other
factors on patenting activity in selected areas of environmental technology. The following reduced-
form equation is specified:
tiititititi EPOGBAORDPACEEPATENTS ,,3,2,1, [1]
where i = 1,…,16 indexes country and t = 1985,…,2004 indexes year. The dependent variable is
measured by the number of patent applications in selected areas of environmental technology (air
pollution, water pollution, waste management, noise protection, and environmental monitoring).
private sector reported in OECD (1996) include also private specialized producers. Consequently, the dummy
variable equals unity for these observations.
12
The explanatory variables include a vector of proxies for regulatory stringency (PACEi,t),
government expenditures on environmental R&D (GBAORDi,t), and total EPO filings (EPOi,t).
Fixed effects ( i ) are introduced to capture unobservable country-specific heterogeneity. All the
residual variation is captured by the error term ( ,i t ). A negative binomial model is used to
estimate equation [1] (for details on count data models see e.g., Cameron and Trivedi 1998;
Maddala 1990; Hausman, Hall and Griliches 1984).
In the first model, a narrow definition of environmental technology is applied and a patent
count in technologies related to air pollution, water pollution and waste disposal is used as a
dependent variable. This is because these are the domains that are most affected by PAC
expenditures (OECD 2007c). Second, a broader definition of environmental technology is applied
and a patent count related to air pollution, water pollution, waste disposal, noise protection and
environmental monitoring is used as a dependent variable.
Alternative specifications are estimated for a pooled model and by including country fixed
effects. Applying a likelihood ratio test, we reject the null hypothesis that the fixed effects model
and the pooled model are equivalent. Hence, further discussion concerns only the results of the
fixed effects model. Table 6 (columns 1 and 2) gives estimated coefficients of the negative
binomial model using an unbalanced panel of 16 countries14
between 1985 and 2004.15
The
presence of missing observations and all-zero outcomes of patent count for some countries reduce
the size of the sample to 150 in the models estimated.
(Insert Table 6 about here.)
Differences in countries‟ scientific capacity and propensity to patent are mostly explained
by overall EPO patenting activity (EPO_Total), which is positive and statistically significant at the
1% level and higher in both models estimated.
13
Although these expenditures are also included in total R&D (GERD), the value of government expenditure
on environmental R&D is far too small to cause any problems of correlation. 14