The Evolution of Persistency in Innovation Activities in the Renewable Energy Sector

(1)

The Evolution of Persistency in Innovation Activities in the Renewable Energy Sector

1

UNIVERSITY OF GRONINGEN – FACULTY OF ECONOMICS AND BUSINESS

The Evolution of Persistency in Innovation

Activities in the Renewable Energy Sector

AN EMPIRICAL ANALYSIS OF EUROPEAN COUNTRY-LEVEL PATENT APPLICATION DATA

Master’s Thesis

By

Stefan Wiersema (S2807254)

Word count: 9704

MSc. Strategic Innovation and Management Faculty of Economics and Business

University of Groningen

(2)

2

ABSTRACT

In recent years, differences among European countries have arisen in the development of renewable energy technology (RET) for the upcoming transition to a more sustainable energy supply. However, limited research can be found explaining the differences between European countries over time. Taking innovation activity as an important stimulant for the development of RET, this research examines whether there exists persistency in national innovation activities in the European renewable energy sector and the effect policy intervention measures have on this persistency over time. Persistency of innovation highlights the influence of past innovation activities of a country on current and future innovation behaviour and success in this area. Drawing upon existing literature, two hypotheses were developed, first, seeking persistency in countries’ innovation activities in RET by focussing on a country’s past innovation activities in this area, and second regarding the effect policy intervention measures have on the persistency of a country’s innovation activities over time. These hypotheses were tested on a dataset of 28 European countries, using fixed-effects panel regression. This research finds persistency in a country’s innovation activities in RET when a country’s past innovation activities in this area are considered. Furthermore, the results indicate that although more European countries are developing policy intervention measures to stimulate the development of RET, the persistency of innovation activities of countries with past experience in the development of renewable energy technologies became stronger over time. The results of this research contribute to the persistency of innovation literature, assists policymakers in reaching their renewable energy targets and enhance the energy transition for a more sustainable energy supply.

Keywords: Energy transition, Europe, Patents, Persistency of innovation, Policy intervention measures,

(3)

3

1. INTRODUCTION

‘There’s one issue that will define the contours of this century more dramatically than any other, and that is the urgent and growing threat of a changing climate.’

- Barack H. Obama (September, 2014)

In the scientific community, strong consensus exists on the conclusion that climate-warming trends over the past century are the result from human activities (NASA, 2017). In an analyses of the peer- reviewed scientific literature on global warming and global climate change, Cook et al. (2013) have shown that 97% of climate scientists agree on this matter. Climate change will alter the earth’s climate system in a negative way (NCSE, 2017). The human activity that primarily effects the amount and rate of climate change is greenhouse gas emissions from the burning of fossil fuels (EPA, 2017). In the coming decades, preventing and reducing carbon emissions (the primary greenhouse gas) is necessary to reduce the negative effects of climate change on the earth’s climate system (IEA, 2017). Making the energy transition from high-carbon fossil energy to low-carbon renewable energy is considered to be an important step in achieving this goal (Stern, 2007).

To enable the large-scale adoption of renewable energy, innovation in renewable energy technology (RET) is critical. In this context, innovation is seen as an important stimulus for the energy transition to renewable energy sources and for reducing carbon emissions (Veugelers, 2012). Although RET can help countries in reducing carbon emissions, these technologies are also more costly to use than fossil fuel technology (Popp, Hascic & Medhi, 2011). Innovation is needed on a large scale to lower the costs of energy produced by renewable energy sources and make it possible to compete with energy produced by fossil fuels (IEA, 2017).

(5)

5

Innovation is a requirement of the energy transition and differences in the levels of innovation activity among European countries can be noticed, specifically in the sector of electricity generation. This sector is of special interest, because the generation of electricity is the largest source of carbon emissions in Europe and will thus be at the centre of the attempt to reduce carbon emissions (EC, 2016). Germany and Denmark have been leading European countries in the development of RET in this sector. In 2010, these countries exceeded their European renewable energy target for the percentage of renewable energy sources in electricity generation, while other European countries, such as the United Kingdom and The Netherlands, did not reach their renewable energy targets (Klessmann, 2012). In the 2017 progress report published by the European commission, the commission urges countries such as The Netherlands, Luxembourg and France to substantially increase their efforts to reach their binding renewable energy targets for 2020 and beyond (EC, 2017).

Limited research can be found exploring the reasons for the differences among European countries in innovation activities related to RET and the effect of policy intervention measures on these innovation activities in RET over time. The available literature (Johnstone, Hascic & Popp, 2010; Lund, 2009) has focussed mainly on the effect policy intervention measures have on the development of RET within a single country and over short span of time.

This research considers the differences among European countries in innovation activities in RET by exploring the importance of past innovation activity in this area (a phenomenon called ‘persistency of innovation’) for innovation activity in future periods and the effect policy intervention measures have on this phenomenon over time. The phenomenon of persistency of innovation highlights the influence of the past innovation activities of a country on current and future innovation behaviour and success (Ganter & Hecker, 2013). In recent innovation literature, persistency of innovation has been used to find and clarify differences among firms and regions in innovation activity over time (Antonelli, Crespi & Scellato, 2012; Cefis & Orsenigo, 2001; Karlsson & Tavassolia, 2016; Tavassolia & Karlsson, 2015).

Therefore this research, designed at the country level, analyses whether there is persistency in innovation activities in the RET of European countries and the effect of policy intervention measures on this persistency over time, in the European electricity sector.

(6)

6

understanding is important, because more and more countries are becoming involved in the transition to a more sustainable energy supply.

To test the hypotheses of this research, a fixed-effects panel regression model was used. This fixed-effects panel regression model was used on a sample of 28 European countries,1 including seven technologies2 in a timeframe of 22 years (1990–2011). Data for this research was obtained from the patent database of the Organization for Economic Co-operation and Development (OECD) and the World Bank.

This paper is structured as follows. Section 2 presents the theoretical background and the hypotheses. After this discussion, Section 3 describes and justifies the methodological choices made during this research. Then, Section 4 presents the results of the fixed-effects panel regression analyses. In Section 5, the results are linked to the hypotheses and the theoretical and the practical implications of the research findings are discussed. Finally, section 5 outlines the limitations of this study, future research directions and some closing comments.

2. THEORETICAL BACKGROUND

This section discusses the theoretical background of renewable energy innovation and persistency of innovation. Section 2.1 provides a general discussion of the development of RET and gives an overview of the literature that has focussed on innovation activity in RET. Next, Section 2.2 briefly introduces the phenomenon of persistency of innovation and reviews the existing literature on persistency of innovation activities. In Section 2.3, hypotheses are formulated that focus on the effect of past innovations and the influence of policy intervention measures on a country’s persistency of innovation activities in RET over time.

2.1 Renewable energy innovation

2.1.1 Renewable energy introduced

In this research, renewable energy is defined as energy derived from natural processes (e.g. sunlight and wind) and is replenished at a higher rate than it is consumed (IEA, 2017). According to the International Energy Agency (2007), three generations of RET can be distinguished. The first generation of renewables were developed over 100 years ago during the industrial revolution and includes technologies such as hydropower, biomass combustion and geothermal power and heat. The second

1_{Austria, Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany,} Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, The Netherlands, Poland, Portugal, Romania, Slovak Republic, Slovenia, Spain, Sweden and the United Kingdom.

(7)

7

generation of RET includes solar heating and cooling, wind power, modern forms of bioenergy and solar PV. This generation of RET is now considered to consist of mature technologies, because of R&D investments made after the oil crisis in the 1970s and the consequent energy security concerns in the 1980s. Third-generation RETs are relatively new technologies that are not ready for the mass market and still depend on sufficient R&D funding to be further developed. These technologies centre on solar power, ocean energy, enhanced geothermal systems, and integrated bioenergy systems. Of these RETs, technologies based on solar and wind energy have had the most success in recent decades. According to the OECD/IEA (2014), these technologies will contribute the vast majority of non-hydro renewable energy generation in the short term and long term, especially because these technologies can nowadays compete on costs with their fossil fuel-based counterparts.

This research concerns itself with the electricity generation sector, because this is the sector with the highest carbon emissions in the world, twice that of the second sector: transportation (IEA, 2015). In the electricity generation sector, coal remains the predominant fuel used to generate electricity, with a worldwide market share of 72% in 2013 (IEA, 2015). Furthermore, the carbon emissions in this sector grew worldwide by 50% between 2000 and 2013, mainly fuelled by coal as a source to generate electricity. Carbon emissions in this sector are expected to increase sharply (by more than 40%) due to an increase in electricity demand from third-world countries (Noailly & Smeets, 2012; OECD/IEA, 2015). This forecast shows that technological developments in the electricity sector can have a potentially massive impact on lowering carbon emissions. Although solar- and wind-energy-based technologies have been relatively successful in past decades, much has yet to be accomplished to make RETs cost competitive with fossil-fuel-based technologies. In 1990, 19.5% of global electricity was produced by renewable energy sources; this share has increased only slightly to 22% in 2013 (IEA, 2015), which clearly signals that innovation in RET is necessary in the coming decades, given the expected increase in electricity demand from third-world countries, in particular.

(8)

8

2.1.2 Innovation activity in renewable energy technology

The literature on renewable energy innovation has focussed on the factors needed to enhance innovation in RET to make the technical change from fossil-fuel-based energy to renewable energy. Acemoglu, Aghion, Bursztyn and Hemous (2012) highlight that the direction of technical change, as for example the change from fossil-fuel-based technologies to RETs, is effected by three main factors:

• the price effect, which encourages innovation in the sector with higher prices;

• the market size effect, which encourages research in markets with a large (possible) market; and • the direct productivity effect, which encourages innovation in technologies of which a country has

an existing stock of knowledge or with greater productivity potential, thereby generating path-dependencies in knowledge creation by building on previous innovations.

Noailly and Smeets (2012) confirm these three main factors in a study of the determinants of directed technical change in the European electricity sector. They point out that for specialised firms in RET, the main drivers of innovation are fossil fuel prices, market size and firms’ past knowledge (Noailly & Smeets, 2012).

2.2 Persistency of innovation

Following Malerba, Orsenigo and Peretto (1997), this research defines ‘persistency’ as the conditional probability that countries innovating at time t will also innovate at time t + 1. Persistency thus addresses the lack of change in a phenomenon or its distribution over time (Andersson & Koster, 2010). As such, persistency in the innovation activities of firms has been mainly associated with the Schumpeterian hypotheses of ‘creative accumulation’ (or Schumpeter Mark II model) (Schumpeter, 1942). The Schumpeterian view of creative accumulation emphasises that technical knowledge has a strong tacit component and is highly specific to individual firms and applications (Malerba, Orsenlgo & Peretto, 1997). This view posits that a country’s past innovation behaviour and success make it more likely that the country will continue to innovate (Ganter & Hecker, 2013).

(9)

9

may seem too affect future innovation because it shows the effect of the persistent unobservable characteristics of a firm or region (Pohjola, 2013).

Studies of persistency in innovation can be divided into two groups: (i) those building upon the analysis of large samples of patents and (ii) those using data from innovation surveys repeated over time. Studies using patent data generally find low levels of persistency. However, they also find that both great innovators and non-innovators have a high probability of remaining in their current state, and in this sense persistency of innovative activities among firms is quite high (Cefis & Orsenigo, 2001; Cefis, 2003; Geroski, van Reenen & Walters, 1997). Studies that use survey data generally find strong evidence for persistency of innovation activity (Antonelli, Crespi & Scellato, 2012; Clausen et al., 2011; Flaig & Stadler, 1994; Ganter & Hecker, 2013; Karlsson & Tavassoli 2016; Peters, 2009; Raymond et al., 2010; Roper & Hewitt-Dundas, 2008; Tavassoli & Karlsson, 2015; Triguero & Córcoles, 2013). Most studies related to persistency focus on the firm level and within one country. In this field, Cefis and Orsenigo (2001) have performed one of the few cross-country studies into a firms’ persistency in innovation activities. More recently, Tavassoli and Karlsson (2015, 2016) have begun exploring the impact of regions within a country on a firm’s persistency in innovation activities.

2.3 Hypotheses

2.3.1 Past innovation activity

Malerba and Orsenigo (2000) argue that persistency in innovation activities over time can be explained by past innovation activities in this area; new innovation activities build upon current knowledge generated from past innovation activity in a country (Breschi, Malerba & Orsenigo, 2000; Malerba & Orsenigo, 2000). In the literature, three perspectives illuminate why past innovation activity determines persistency in a country’s innovation activities over time (Ganter & Hecker, 2013; Peters, 2009; Pohjola, 2013; Raymond et al., 2010).

(10)

10

innovation activities of a country nurture that country’s knowledge base, which becomes a foundation for future innovation activities and creates persistency (Raymond et al., 2010). The third perspective focusses on resource constraints or ‘success breeds success’ (Alfranca, Rama & Von Tunzelmann, 2002; Nelson & Winter, 1982; Flaig & Stadler, 1994; Raymond et al., 2010). From this perspective, successful innovation activities in the past create profits that can be invested into future innovation activities. Innovation success thus breeds innovation success by facilitating access to financial resources (Flaig & Stadler, 1994).

Karlsson and Tavassoli (2016) argue that the three perspectives are built on a combination of learning effects and economies of scale. Learning effects are generated from the innovation process and dynamic scale economies are generated from positive feedback mechanisms between the accumulation of knowledge and innovation processes (Geroski, Van Reenen & Walter, 1997). Thus, one can argue from these perspectives that persistency in a country’s innovation activities in RET is the result of cumulative knowledge patterns and learning dynamics from past innovation activity in the relevant area. Based on the above arguments, the following hypothesis was developed:

H1: A country’s past innovation activity in renewable energy technologies (RETs) explains a country’s

innovation activity in this area in the following periods. 2.3.2 Persistency of innovation activities over time

Hypothesis 1 focusses on determining whether persistency exists in a country’s innovation activity in RETs over time. Hypothesis 2 shifts focus by arguing that increased governmental policy intervention measures for RET have a negative impact on persistency in a country’s innovation activity over time. Policy intervention measures for RET are seen in literature as the main driver for firms to innovate these technologies, so have a large impact on the development of persistency in a country’s innovation activities in RET over time (Horbach, Rammer & Rennings, 2012; Johnstone, Hascic & Popp, 2010; Rennings & Rammer, 2011).

(11)

11

Other European countries, with the adoption of a white paper in 1997, committed themselves to incorporating renewable energy into their electricity markets (EC, 2007). More specifically, the Directive 2001/77/EC of the European Parliament in 2001 (EUR-Lex, 2001) and the ratification of the Kyoto protocol in 2002 gave European countries clear compulsory renewable energy targets for their electricity markets. To reach the compulsory renewable energy targets, policymakers in European countries have increasingly introduced policies to reduce costs and accelerate market penetration for RET (Schmidt & Marschinski, 2009). Groba and Breitschopf (2013) have shown that these policies effectively stimulated firms to increase innovation activity in RET. This findings is in line with Acemoglu et al. (2012) and Noailly and Smeets (2012), who have found that innovation activity in RET is directed towards markets with a great potential.

For firms located in lead markets, this relationship between policy and innovation makes expanding into foreign markets attractive, especially because lead markets are becoming saturated (Lewis & Wiser, 2007). To succeed in foreign markets, firms will need to use their domestic knowledge base and knowledge of the foreign market to innovate. This adaptive process means that the knowledge of RET that is necessary to keep innovating will disseminate over time into new markets, because firms are relocating their innovation facilities into these new markets. Thus, it can be argued, increasing governmental support policies for RET over time will lead to a decrease in the persistency of a country’s innovation activities over time, leading to the following hypothesis:

H2: Persistency of a country’s innovation activities in renewable energy technologies (RETs) will

decrease over time.

3. METHODOLOGY

This section first explains the data collection. Second, a justification for the statistical test used to evaluate the hypotheses is given. Third, the dynamic panel model used for this research is introduced and explained.

3.1 Data collection

3.1.1 Patent data

(12)

12

Table 1: Advantages and disadvantages for using patent data (Source: OECD, 2009).

Advantages Disadvantages

Patents provide information on a broad range of technologies that are not always covered by other sources of data.

Because of high costs or for strategic reasons (for instance, secrecy), not all inventions are patented.

Patents have a close link to invention, because most significant inventions are patented.

The value distribution of patents is highly

skewed. Many patents have no industrial application, and some have a really high value. A patent application is a rich ‘free’ source of

information containing, for instance, the names and addresses of inventors.

Patent offices around the world use different patent laws and practices. Using a homogenous patent dataset is preferable,

With more than a million patents applied for worldwide each year, the information available for researchers is vast.

Patent laws change over years, which necessitates caution when analysing trends over time.

Patent data are quite readily available from national and regional patent offices.

Patent data are complex. When compiling and interpreting patent data, this complexity needs to be taken into consideration.

According to Johnstone, Hascic and Popp (2010) patents are, despite their shortcomings, the best available source of data on innovation activity that is readily available and comparable among countries. Patents have already been used to study persistency of innovation activities of firms by Cefis and Orsenigo (2001), for example. This research used patent data on the country level. Section 3.1.3 deals to a certain extent with the disadvantages of using patents by choosing the appropriate methods to minimise bias.

3.1.2 Control variables

(13)

13

3.1.3 Collection of patent data

Under the International Patent Classification (IPC), patent applications are divided among several classes, ranging from A to H. Patent applications may contain several technical objects and therefore be assigned several IPC classes (OECD, 2009). The problem in relation to renewable energy patents is that they are divided between several technical areas, making the analysis of patent applications in this area difficult. To enhance research into energy-related patents, the OECD has developed algorithms to classify various technology fields in which patent applications of a certain technology are clustered: for instance, the technology field used in this research is ‘environment-related technologies’. With the data provided by the OECD, it is possible to do statistical analysis only on a country level.

The OECD patent manual was used to ensure that the appropriate patent sets were selected for this research. The relevant patent set was selected by excluding the Japan Patent Office and the United States Patent and Trademark Office and to include the European Patent Office (EPO). The EPO database was used primarily because all patent applications are filed under the same conditions and secondly because the EPO covers all European countries in a comparable and timely manner. Triadic patent families and patent applications filed under the Patent Cooperation Treaty (PCT) are not included. The patent offices use different time-paths for the application of patents, combining patent counts is therefore not an option for this research. In the case of the Euro-PCT, the information on the effective transfer to the EPO is available 36 months after priority (first filing) (OECD, 2009). This late availability of patent information influences the timeliness of patent indicators. This research focusses on one patent office to minimise the amount of bias in the patent application data.

To divide patents across countries or regions, the OECD database uses information provided on the front page of the patent applications, with the place of residence of the inventors, among other things (OECD, 2009). In this research, the inventor’s country of residence was used, because this country is in most cases the one in which the invention was developed. Fractional counts are used by the OECD when an invention is patented by inventors with different countries of residence. This use of fractions guarantees that the correct share of a patent is added to the correct country.

(14)

14

than the actual invention, which would bias the results. Therefore, priority date was used in this research. Priority date represents the first date of filling of a patent application anywhere in the world to protect an invention. Priority date records the closest date to the actual invention and thus the most suitable for this research.

3.1.4 Collection of control variables data

The World Bank’s database was used to collect data for the control variables. Data on GDP per capita (in 2017 U.S. dollars) and electricity consumption per capita (in kilowatt-hours) were extracted from the publically available database of the World Bank.

3.1.5 Country and timeframe selection

In this research, 28 countries that are part of the European Union were selected. European countries were of interest for the following reasons. First, the countries within the European Union share similar policy goals for the development of renewable energy. Second, the economies of European countries are similar.

A timeframe of 22 years (1990–2011) was selected for the following reasons. First, before 1990, innovation activities in RET were limited. In the chosen timeframe, innovation activities started to rise in Europe. Extending the timeframe backwards was therefore considered not useful for this research. Second, a considerable number of harmonisation of patent rules across countries took place in the 1990s, especially with the creation of the Agreement on Trade-Related Aspects of Intellectual Property Rights in 1994 (OECD, 2009). This agreement enhanced the value of cross-country patent data and made it possible to interpret and analyse the data. The timeframe of this research was limited to 2011. The decision to limit the timeframe and corresponding dataset to 2011 was made after consulting an expert in the field of renewable energy research.3 The expert argued that patents are generally published 18 months after the first date of filling (priority date). No patent data is available until 18 months after the priority date, because this data is not published until then. The environment-related technologies database of the OECD was updated with PATSTAT data for autumn 2015.4 To prevent data reliability concerns, the expert advised a cut-off date of at least three years prior to the most recent dataset. Therefore the timeframe of this research was limited up to and including 2011. The method of ‘nowcasting’ (predicting patent applications) could be deployed to overcome this problem and provide patent data after 2011, but in this research only factual data was used, to increase the reliability of the final results (OECD, 2009).

3

(15)

15

3.1.6 Descriptive statics

This research focusses on the current 28 countries that are full member states of the European Union.5 The selected timeframe was from 1990–2011. In the timeframe, there are a total of 28513.13 patent applications, divided among seven technologies: wind energy, with 9155.75; solar thermal, with 6819.20; solar PV, with 7602.25; solar thermal PV hybrids, with 382.83; geothermal, with 593.4; marine, with 1342.4; and hydro, with 2617.3.

3.2 Statistical analysis

To choose the appropriate analytical analysis, Cameron and Trivedi’s (2013) study was used. To test the relationship between the dependent variable and lagged dependent variables as regressors, regression analysis was chosen. Regression analysis was appropriate because it helped to characterise the effect lagged dependent variables (past innovation activity) have on a country’s innovation activities in future periods and the effect persistency in innovation activities have over time.

The appropriate regression model for this research depended heavily upon the dependent variable. Data on patent applications was used as the dependent variable in this research. Patent application data can be regarded as count data, which refers to the number of times an event occurs. Cameron and Trivedi (2013) describe count data as the realisation of a non-negative integer-valued random variables. The patent application data in the dataset is highly positively skewed (see Appendix A). Therefore, this research used a log transformation of the dependent variable for the statistical analysis.

The cross-sectional ordinary least squares (OLS) method was deemed inappropriate for this research because this method does not take the panel data structure of the dataset into account. This research therefore used panel regression techniques. For the panel regression techniques, a choice had to be made between fixed-effects and random-effects. The difference between the two is that a fixed-effects technique assumes that the individual-specific effects (in this research, countries) are correlated to the independent variable (in this research, lagged time lengths of innovation activity in RET); a random-effects technique assumes that they are not correlated. To test this assumption, a Hausman specification test was performed on the patent database (Cameron & Trivedi, 2013). The Hausman specification test suggests to use fixed effects, because the results shows that there is correlation (Prob > chi2 = 0.0000) between the individual specific effect and the independent variable. In sum, this research used fixed-effects panel regression techniques to test the hypotheses.

5

(16)

16 3.3 Dynamic panel model

To estimate the effect of a country’s past innovation activities in RET on innovation activities in future periods, a dynamic panel model was employed:

Log (𝑅𝐸

it

) = 𝛾

n-1

log (𝑅𝐸

i,t-n + 1

) + x’

it

β +

ai

+

εit

(1)

Equation 1 is based on Cameron and Trivedi (2013) and Andersson and Koster (2011). In the model, REit is the renewable energy activity in country i in year t, and t-n denotes the lag lengths; x’it

allows for heterogeneity among countries captured with the control variables discussed in Section 3.1.2; ai denotes unobserved time-invariant heterogeneity captured by country-specific effects. By using fixed effects, unobserved time-invariant heterogeneity captured by country-specific effects is controlled for. Finally, εit is an error term.

4. RESULTS

In this section the results of the fixed-effects regressions are discussed. Section 4.1 discusses the results of the effects panel regression for Hypothesis 1. Section 4.2 presents the results of the fixed-effects panel regression for Hypothesis 2.

4.1 Past innovation activity

(17)

17

Table 2: Estimation of lagged variables for the logarithm of innovation activity in RET over time (including control variables)

GDP pc 1.46e-05*** 1.90e-05*** 2.13e-05*** 2.39e-05*** 2.58e-05*** 2.70e-05*** 2.70e-05*** 2.80e-05*** 2.95e-05*** 2.95e-05*** 3.10e-05*** 3.50e-05*** 4.02e-05*** 4.00e-05*** 3.77e-05*** 2.66e-05*** 3.92e-06 (1.10e-06) (1.19e-06) (1.24e-06) (1.27e-06) (1.30e-06) (1.33e-06) (1.35e-06) (1.37e-06) (1.40e-06) (1.43e-06) (1.56e-06) (1.84e-06) (2.32e-06) (2.90e-06) (3.50e-06) (4.54e-06) (6.24e-06) Electricity consumption pc 5.19e-05*** 5.90e-05*** 8.84e-05*** 0.000100*** 9.58e-05*** 9.19e-05*** 9.59e-05*** 7.66e-05*** 4.14e-05 1.69e-05 -1.64e-05 -3.43e-05 -0.000102** -0.000155*** -0.000196*** -0.000157*** -6.02e-05 (1.49e-05) (1.68e-05) (1.85e-05) (2.02e-05) (2.19e-05) (2.41e-05) (2.64e-05) (2.85e-05) (3.07e-05) (3.34e-05) (3.66e-05) (4.06e-05) (4.40e-05) (4.65e-05) (4.85e-05) (5.02e-05) (5.50e-05) Innovation activity (t -1) 0.594*** (0.0135) Innovation activity (t -2) 0.512*** (0.0152) Innovation activity (t -3) 0.453*** (0.0170) Innovation activity (t -4) 0.397*** (0.0189) Innovation activity (t -5) 0.338*** (0.0208) Innovation activity (t -6) 0.283*** (0.0226) Innovation activity (t -7) 0.266*** (0.0242) Innovation activity (t -8) 0.236*** (0.0253) Innovation activity (t -9) 0.183*** (0.0263) Innovation activity (t -10) 0.207*** (0.0277) Innovation activity (t -11) 0.210*** (0.0296) Innovation activity (t -12) 0.145*** (0.0320) Innovation activity (t -13) 0.0757** (0.0345) Innovation activity (t -14) 0.0899** (0.0375) Innovation activity (t -15) -0.0318 (0.0402) Innovation activity (t -16) -0.0866** (0.0427) Innovation activity (t -17) -0.0646 (0.0452) Constant -0.268*** -0.314*** -0.472*** -0.538*** -0.491*** -0.440*** -0.437*** -0.307* -0.0763 0.0830 0.267 0.305 0.629** 0.997*** 1.418*** 1.589*** 1.749*** (0.0804) (0.0912) (0.102) (0.113) (0.124) (0.139) (0.155) (0.170) (0.187) (0.207) (0.231) (0.262) (0.292) (0.317) (0.336) (0.365) (0.374) Observations 3953 3792 3631 3470 3304 3108 2912 2716 2520 2324 2128 1932 1736 1540 1344 1148 952 R -squared 0.518 0.443 0.390 0.349 0.308 0.282 0.269 0.257 0.241 0.234 0.219 0.197 0.173 0.135 0.101 0.046 0.004 *** p < 0.01, ** p < 0.05, * p < 0.1

Standard errors in parentheses

(18)

18

In Table 2, each of the 17 fixed-effects panel regressions has a different lag length as the explanatory variable. The lag lengths varies from 1–17 years. The results in Table 2 show with a lag length of t − 1, an increase of 1% in a country’s innovation activity in the previous year, the innovation activity in the next year is expected to increase by 0.594%. The coefficient only slowly drops when the lag lengths increase. For example, with a lag length of t – 10, an increase of 1% in a country’s innovation activity in the past, a country’s innovation activity after 10 years is expected to increase with 0.207%. After t – 16 the effect of a country’s past innovation activity on a country’s future innovation activity is no longer significant.

Regarding the control variables, the results in Table 2 show that GDP per capita has a significant effect on future innovation activities for the lag lengths until and including a lag length of t − 16 (p < 0.01). For example, with lag length t − 1, an increase of 1 U.S. dollar for GDP per capita, it is expected that the innovation activity in the following year will increase by 0.000146%. Electricity consumption per capita has a significant effect on future innovation activity for the lag lengths up to and including t − 8 (p < 0.01). For example, with lag length of t − 1, an increase of 1 kWh of electricity consumption per capita, it is expected that the innovation activity in the following year will increase by 0.000519%. Only between the lag lengths t − 14 and t − 16 does electricity consumption per capita again have a significant effect on future innovation activities (p < 0.01). After lag length t − 16, Electricity consumption per capita is no longer significant.

(19)

19

Figure 1: R-squared with increased lag lengths

4.2 Persistency in innovation activities over time

Hypothesis 2 examined the effect of increased governmental support policies for RET on a country’s persistency in innovation activities over time. Hypothesis 2 posited that RET knowledge in lead countries will disseminate to new markets, lowering the persistency in a country’s innovation activities in RET over time.

To answer Hypothesis 2, this research first focussed on the correlation between a country’s innovation activity in RET in one year and in previous years. For this reason, 231 separate correlation analyses were performed. The results of these correlation analyses can be found in a correlation matrix in Appendix C. The correlation matrix gives a better idea of how the correlation between a country’s innovation activity in RET in one year and previous years has developed over time. For example, the correlation of 2011 with 2010 (t – 1; 0.9341) is stronger than the correlation of 1991 with 1990 (t – 1; 0.8256).

(20)

20

Table 3: Fixed-effects panel regression results with interaction terms and control variables

Innovation activity Model 1 Model 2

(21)

21

2011 0.302*** 0.284***

(0.0441) (0.0442)

GDP pc 1.52e-06

(1.67e-06)

Electricity consumption pc 1.22e-05

(2.13e-05)

Constant 0.419*** 0.317***

(0.0353) (0.117)

Observations 4676 4521

R-squared 0.605 0.612

Standard errors in parentheses *** p < 0.01, ** p < 0.05, * p < 0.1

(22)

22

Figure 2: Average marginal effect over time

Figure 2 shows that over time having innovation activities in the previous year has a strong and increasing positive effect on a country’s innovation activities in the following year. This effect is insignificant until 1998, but from 1998 onwards the overall year effect on a country’s innovation activities becomes very strong and persistent.

5. DISCUSSION & CONCLUSION

5.1 Discussion

In recent years, European countries have increasingly showed interest in the development of RET to make the transition to electricity produced by renewable energy sources. This increased interest has spurred innovation activity in RET and currently the European Union is a world leader in the development of RET (Kerebel & Stoerring, 2016). Differences can be noticed in the development of RET among European countries over time (EC, 2017). Limited literature could be found explaining the reasons for the differences between European countries. This research aimed to analyse these differences by searching for persistency in a country’s innovation activities in RET and by analysing the effect of policy intervention measures on this persistency over time.

To analyse whether persistency exists in a country’s innovation activities in RET, patent application data from seven technologies and 28 European countries over a timeframe of 22 years (1990–

.2 .4 .6 .8 Ef fe ct s o n L in e a r Pre d ict io n 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 Year

(23)

23

2011) was collected. Fixed-effects panel regression techniques were performed on the dataset to analyse whether there was persistency in a country’s innovation activities and the effect of policy intervention measures on this persistency over time.

The first outcome of the analytical analysis shows that there is persistency in a country’s innovation activities in RET and supports Hypothesis 1 of this research. This research argued that a country’s past innovation activity in RETs explains a country’s innovation activity in this area in the following periods, because of a combination of cumulative knowledge patterns and learning dynamics. The results show that persistency in a country’s innovation activities is strong, with a lag length of t – 1, and only slowly decreases with increasing lag lengths. After a lag length of t – 16, no signification relation remains. The results of this first outcome align with the conclusions of Karlsson and Tavassoli (2016), who have argued that persistency in innovation activities is built on a combination of learning effects and economics of scale.

Contrary to the expectation of this research, the second outcome of the analytical analysis does not support Hypothesis 2, which conjectured that a country’s persistency in innovation activities in RET will decrease over time. This research argued that persistency in innovation would decrease over time, because firms located in lead markets would over time relocate their firms innovation activities to new foreign markets. These new markets have opened up for RET after more European countries became committed to incorporating renewable energy sources into their electricity generation sector. The results of the statistical analysis show that over time the effect of a country’s past innovation activity on a country’s innovation activity in the following year became stronger. Figure 2 shows that after 1998, persistency in a country’s innovation activities in RET is very strong and became increasingly persistent. Although more European countries are introducing policies to reduce costs and accelerate market penetration for RET, firms are not relocating their innovation facilities to these countries. Instead, the results suggests that firms increase their innovation efforts in the countries in which they are currently located. A possible explanation for this result could be that firms in the RET sector form geographical clusters to benefit from economies of localisation; in other words, firms form geographical clusters with other firms to benefit from specialised skills, growth of trade and services, and the availability of specialised machinery (Keeble & Wilkinson, 1999).

(24)

24

5.1.1 Theoretical implications

This research provides some important implications for the existing literature. First of all, this study represents one of the first to research the phenomenon of persistency of innovation at the country level. Cefis and Orsenigo (2001) were the first to conclude that countries can play an important role in the persistency of innovation activities of firms. This research underlines their conclusion by revealing the existence of persistency in a country’s innovation activities and that this persistency became stronger over time. Further, the results of this research underline previous persistency of innovation literature stating that both great innovators and non-innovators have a high probability of remaining in their current countries (Cefis & Orsenigo, 2001; Cefis, 2003; Geroski, van Reenen & Walters, 1997). As well, this research advances the understanding of the differences between European countries in the development of RET in the electricity generation sector, showing that the phenomenon of persistency contributes to the differences noticed among European countries in the renewable energy sector.

5.1.2 Practical implications

(25)

25 5.2 Limitations and future research

This research has several limitations, providing avenues for future research. First of all, this research used patent application data. Patent data has some well-known drawbacks. Future research could use alternative measures to assess a country’s innovation activities in the renewable energy sector. For example, product announcements or innovation activity input data, such as a country’s R&D expenditures, could be deployed to enrich the dataset with inventions not patented by firms in a country. Second, this research focusses on innovation activity in RET in European countries. For the generalisability of the results of this research, it would be interesting to replicate this research on different continents around the world. For example, it would be interesting to research whether there is persistency in innovation activities in RET in developing continents, such as Asia. This extension would considerably expand the understanding of the phenomenon of persistency in developing countries. Third, by using patent data at a country level, heterogeneity among regions in a country is lost. Future research could investigate persistency of innovation activities in RET among regions in a large European country, such as Germany, to incorporate this heterogeneity.

5.3 Conclusion

(26)

26

REFERENCES

Acemoglu, D., Aghion, P., Bursztyn, L., & Hemous, D. (2012). The environment and directed technical change. The American Economic Review, 102(1), 131-166.

Alfranca, O., Rama, R., & von Tunzelmann, N. (2002). A patent analysis of global food and beverage firms: the persistence of innovation. Agribusiness, 18(3), 349-368.

Andersson, M., & Koster, S. (2010). Sources of persistence in regional start-up rates—evidence from Sweden. Journal of Economic Geofigurey, lbp069.

Antonelli, C., Crespi, F., & Scellato, G. (2012). Inside innovation persistence: New evidence from Italian micro-data. Structural Change and Economic Dynamics, 23(4), 341-353.

Arrow, K. (1962). Economic welfare and the allocation of resources for invention. In The rate and

direction of inventive activity: Economic and social factors (pp. 609-626). Princeton University

Press.

Breschi, S., Malerba, F., & Orsenigo, L. (2000). Technological regimes and Schumpeterian patterns of innovation. The economic journal, 110(463), 388-410.

Cameron, A. C., & Trivedi, P. K. (2013). Regression analysis of count data (Vol. 53). Cambridge university press.

Cefis, E., & Orsenigo, L. (2001). The persistence of innovative activities: A cross-countries and cross-` sectors comparative analysis. Research Policy, 30(7), 1139-1158.

Cefis, E. (2003). Is there persistence in innovative activities?. International Journal of Industrial

Organization, 21(4), 489-515.

Clausen, T., Pohjola, M., Sapprasert, K., & Verspagen, B. (2011). Innovation strategies as a source of persistent innovation. Industrial and Corporate Change, dtr051.

Cook, J., Nuccitelli, D., Green, S. A., Richardson, M., Winkler, B., Painting, R., ... & Skuce, A. (2013). Quantifying the consensus on anthropogenic global warming in the scientific literature.

Environmental research letters, 8(2), 024024.

Dosi, G. (1988). Sources, procedures, and microeconomic effects of innovation. Journal of economic

(27)

27

EC (1997). Renewable Energy Roadmap Renewable energies in the 21th century: Building a more

sustainable future. Retrieved 9 March 2017, from

http://www.europarl.europa.eu/RegData/docs_autres_institutions/commission_europeenne/co m/2006/0848/COM_COM(2006)0848_EN.pdf

EC (2007). Report from the commission to the European parliament, the council, the European Economic

and social committee and the committee of the region. Renewable Energy Progress Report.

Retrieved 22 March 2017 from

http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:52017DC0057&qid=1488449105433&from=EN

EC (2016). File:Greenhouse gas emissions by economic activity, EU-28, 2009 and 2014 (% of total

emissions in CO2 equivalents) YB17.png - Statistics Explained. Ec.europa.eu. Retrieved 10 March

2017, from http://ec.europa.eu/eurostat/statistics-

explained/index.php/File:Greenhouse_gas_emissions_by_economic_activity,_EU28,_2009_and_ 2014_(%25_of_total_emissions_in_CO2_equivalents)_YB17.png

("File:Greenhouse gas emissions by economic activity, EU-28, 2009 and 2014 (% of total emissions in CO2 equivalents) YB17.png - Statistics Explained", 2016)

EPA (2017). Causes of Climate Change. Retrieved 15 May 2017 from

https://www.epa.gov/climate-change-science/causes-climate-change

EUR-Lex. (2011). Renewable energy: the promotion of electricity from renewable energy sources.

Retrieved 17 May 2017, from

http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=LEGISSUM:l27035

Flaig, G., & Stadler, M. (1994). Success breeds success. The dynamics of the innovation process.

Empirical Economics, 19(1), 55-68.

Fuchs, G., & Wassermann, S. (2012). From niche to mass markets in high technology: The case of photovoltaics in Germany. Innovation policy and governance in high-tech industries (pp. 219-244). Springer Berlin Heidelberg.

Ganter, A., & Hecker, A. (2013). Persistence of innovation: discriminating between types of innovation and sources of state dependence. Research Policy, 42(8), 1431-1445.

Geroski, P. A., Van Reenen, J., & Walters, C. F. (1997). How persistently do firms innovate?. Research

(28)

28

Groba, F., & Breitschopf, B. (2013). Impact of renewable energy policy and use on innovation: A literature review.

Horbach, J., Rammer, C., & Rennings, K. (2012). Determinants of eco-innovations by type of environmental impact—The role of regulatory push/pull, technology push and market pull.

Ecological economics, 78, 112-122.

IEA (2007). Renewables in Global Energy Supply: An IEA Fact Sheet. IEA Publications.

IEA (2015). Key CO2 Emissions Trends. IEA Publications.

IEA (2017). Renewables. Retrieved 24 March 2017, from https://www.iea.org/topics/renewables/

Jacobsson, S., & Lauber, V. (2006). The politics and policy of energy system transformation—explaining the German diffusion of renewable energy technology. Energy policy, 34(3), 256-276.

Jensen, N. (2017). Denmark invest 25m in China, 22m in Vietnam to reduce CO2 – ScandAsia. Retrieved 11 June 2017, from http://scandasia.com/denmark-invest-25m-in-china-22m-in-vietnam-to-reduce-co2/

Johnstone, N., Hascic, I., & Popp, D. (2010). Renewable energy policies and technological innovation: evidence based on patent counts. Environmental and resource economics, 45(1), 133-155.

Karlsson, C., & Tavassoli, S. (2016). The role of firms? location on persistency of various types of innovation. In ERSA conference papers (No. ersa16p154). European Regional Science Association.

Keeble, D., & Wilkinson, F. (1999). Collective learning and knowledge development in the evolution of regional clusters of high technology SMEs in Europe. Regional studies, 33(4), 295-303.

Kerebel, C. & Stoerring, D. (2016). Renewable energy | EU fact sheets | European Parliament.

Europarl.europa.eu. Retrieved 9 March 2017, from

http://www.europarl.europa.eu/atyourservice/en/displayFtu.html?ftuId=FTU_5.7.4.html

Klessmann, C. (2012). EU missed target of 21% RES in the electricity mix in 2010 - Ecofys. Ecofys. Retrieved 9 March 2017, from http://www.ecofys.com/en/news/eu-missed-target-of-21-res-in-the-electricity-mix-in-2010/

(29)

29

Lewis, J. I., & Wiser, R. H. (2007). Fostering a renewable energy technology industry: An international comparison of wind industry policy support mechanisms. Energy policy, 35(3), 1844-1857.

Licht, G., & Peters, B. (2013). The Impact of Green Innovation on Employment Growth in Europe (No. 50). WWWforEurope.

Lund, P. D. (2009). Effects of energy policies on industry expansion in renewable energy. Renewable

energy, 34(1), 53-64.

Malerba, F., Orsenigo, L., & Peretto, P. (1997). Persistence of innovative activities, sectoral patterns of innovation and international technological specialization. International Journal of Industrial

Organization, 15(6), 801-826.

Malerba, F., & Orsenigo, L. (2000). Knowledge, innovative activities and industrial evolution. Industrial

and corporate change, 9(2), 289-314.

Mathew, S. (2006). Wind energy: fundamentals, resource analysis and economics (Vol. 1). Heidelberg: Springer.

Meisler, L. (2017). This Is Where to Expect the Biggest Job Growth in the Future. Bloomberg.com. Retrieved 11 June 2017, from https://www.bloomberg.com/graphics/2017-job-projections-2024/

NASA (2017). Scientific consensus: Earth's climate is warming. Retrieved 17 April 2017, from http://climate.nasa.gov/scientific-consensus/

NCSE (2017). How Will Climate Change Affect the World and Society?. Retrieved 17 April 2017, from https://ncse.com/library-resource/how-will-climate-change-affect-world-society

Nelson, R. R., & Winter, S. G. (1982). The Schumpeterian tradeoff revisited. The American Economic

Review, 72(1), 114-132.

Noailly, J., & Smeets, R. (2012). Directing Technical Change from Fossil-Fuel to Renewable Energy Innovation: An Empirical Application Using Firm-Level Patent Data. CPB Discussion Paper,

237.

Obama, B. H. (2014). Remarks by the President at U.N. Climate Change Summit. whitehouse.gov. Retrieved 19 April, 2017, from https://obamawhitehouse.archives.gov/the-press-office/2014/09/23/remarks-president-un-climate-change-summit

(30)

30

OECD/IEA (2014). The Power of Transformation: Wind, Sun and the Economics of Flexible Power Systems. IEA Publications.

OECD/IEA (2015). World Energy Outlook Special Briefing for COP21. IEA Publications.

Peters, B. (2009). Persistence of innovation: stylised facts and panel data evidence. The Journal of

Technology Transfer, 34(2), 226-243.

Pohjola, M. U. P. (2013). A Compilation of Studies on Innovation in Firms: Persistence, Strategies, and

Capabilities (Doctoral dissertation, Utrecht University School of Economics).

Popp, D., Hascic, I., & Medhi, N. (2011). Technology and the diffusion of renewable energy. Energy

Economics, 33(4), 648-662.

Raymond, W., Mohnen, P., Palm, F., & Van Der Loeff, S. S. (2010). Persistence of innovation in Dutch manufacturing: Is it spurious?. The Review of Economics and Statistics, 92(3), 495-504.

Rennings, K., & Rammer, C. (2011). The impact of regulation-driven environmental innovation on innovation success and firm performance. Industry and Innovation, 18(03), 255-283.

Roper, S., & Hewitt-Dundas, N. (2008). Innovation persistence: Survey and case-study evidence.

Research Policy, 37(1), 149-162.

Schmidt, R. C., & Marschinski, R. (2009). A model of technological breakthrough in the renewable energy sector. Ecological Economics, 69(2), 435-444.

Schumpeter, J. (1942). Creative destruction. Capitalism, socialism and democracy, 825.

Stern, N. H. (2007). The economics of climate change: the Stern review. Cambridge University press.

Sutton, J. (1991). Sunk costs and market structure: Price competition, advertising, and the evolution of

concentration. MIT press.

Tavassoli, S., & Karlsson, C. (2015). Persistence of various types of innovation analyzed and explained. Research Policy, 44(10), 1887-1901.

Triguero, A., & Córcoles, D. (2013). Understanding innovation: An analysis of persistence for Spanish manufacturing firms. Research Policy, 42(2), 340-352.

(31)

31

Appendix A: Distribution of patent applications in the dataset

(32)

32

Appendix B: Estimation of lagged variables for the logarithm of innovation activity in RET over time Innovation activity (t -1) 0.712*** (0.0121) Innovation activity (t -2) 0.656*** (0.0142) Innovation activity (t -3) 0.618*** (0.0165) Innovation activity (t -4) 0.578*** (0.0189) Innovation activity (t -5) 0.523*** (0.0214) Innovation activity (t -6) 0.471*** (0.0236) Innovation activity (t -7) 0.449*** (0.0254) Innovation activity (t -8) 0.408*** (0.0267) Innovation activity (t -9) 0.339*** (0.0281) Innovation activity (t -10) 0.342*** (0.0297) Innovation activity (t -11) 0.322*** (0.0319) Innovation activity (t -12) 0.240*** (0.0347) Innovation activity (t -13) 0.148*** (0.0373) Innovation activity (t -14) 0.131*** (0.0400) Innovation activity (t -15) 0.00754 (0.0422) Innovation activity (t -16) -0.0601 (0.0434) Innovation activity (t -17) -0.0638 (0.0451) Constant 0.281*** 0.371*** 0.448*** 0.527*** 0.611*** 0.686*** 0.739*** 0.803*** 0.881*** 0.922*** 0.977*** 1.062*** 1.157*** 1.223*** 1.336*** 1.430*** 1.488*** (0.0123) (0.0137) (0.0150) (0.0160) (0.0170) (0.0178) (0.0184) (0.0188) (0.0192) (0.0197) (0.0204) (0.0213) (0.0221) (0.0228) (0.0234) (0.0233) (0.0236) Observations 4088 3892 3696 3500 3304 3108 2912 2716 252 2324 2128 1932 1736 1540 1344 1148 952 R -squared 0.471 0.365 0.287 0.220 0.161 0.120 0.103 0.085 0.059 0.059 0.050 0.027 0.010 0.008 0.000 0.002 0.003

Innovation activity at time t

(33)

33

Appendix C: Correlation matrix

t - 1 t - 2 t - 3 t - 4 t - 5 t - 6 t - 7 t - 8 t - 9 t - 10 t - 11 t - 12 t - 13 t - 14 t - 15 t - 16 t - 17 t - 18 t - 19 t - 20 t - 21 t = 2011 0.9341 0.9163 0.9145 0.8862 0.8769 0.8816 0.8487 0.8563 0.8132 0.83 0.8233 0.7747 0.7046 0.7307 0.7021 0.6912 0.6697 0.613 0.6353 0.7203 0.7004 t = 2010 0.9354 0.9129 0.8919 0.8745 0.8536 0.852 0.8554 0.8219 0.8126 0.8251 0.7937 0.7309 0.7289 0.7018 0.6717 0.6476 0.5862 0.6065 0.7048 0.6946 t = 2009 _{0.9234 0.8894 0.8875 0.8627 0.8415 0.8448 0.8169 0.8071 0.8194 0.8037 0.7461 0.7371 0.7316 0.6888 0.6693 0.6076 0.629} _{0.6884 0.6781} t = 2008 0.9098 0.8961 0.8587 0.8473 0.8601 0.8336 0.8372 0.8564 0.808 0.7739 0.7625 0.7397 0.7062 0.6836 0.6132 0.6002 0.6909 0.6772 t = 2007 0.8849 0.8627 0.8639 0.8758 0.8401 0.8111 0.8216 0.8082 0.7509 0.7372 0.7585 0.7214 0.6911 0.6232 0.6368 0.7075 0.6641 t = 2006 0.8826 0.8695 0.8565 0.8089 0.822 0.8048 0.806 0.7613 0.7608 0.7483 0.7376 0.688 0.6291 0.6553 0.6942 0.6763 t = 2005 0.8915 0.8801 0.8325 0.8578 0.8409 0.8133 0.7742 0.758 0.7245 0.7377 0.6954 0.6386 0.6562 0.75 0.7244 t = 2004 0.9143 0.8626 0.8658 0.8501 0.8486 0.8086 0.7994 0.79 0.756 0.7366 0.6744 0.6803 0.7627 0.7351 t = 2003 0.8877 0.903 0.8492 0.8385 0.8136 0.7993 0.8116 0.7709 0.7716 0.6832 0.723 0.8004 0.7449 t = 2002 0.8521 0.8557 0.8364 0.8092 0.7871 0.7764 0.758 0.7096 0.6457 0.6468 0.7091 0.7299 t = 2001 0.8692 0.8238 0.7866 0.807 0.7902 0.7693 0.7856 0.702 0.7068 0.7479 0.7058 t = 2000 0.8773 0.8265 0.7968 0.7722 0.7559 0.7569 0.7035 0.6784 0.7338 0.698 t = 1999 0.8403 0.7864 0.8047 0.7713 0.7275 0.6955 0.6572 0.7216 0.7041 t = 1998 0.7851 0.7645 0.7811 0.7732 0.6964 0.6853 0.7265 0.7213 t = 1997 0.8442 0.7937 0.8001 0.7333 0.7444 0.7338 0.6784 t = 1996 0.8295 0.8156 0.7838 0.7605 0.7642 0.7208 t = 1995 0.8485 0.7922 0.7474 0.7522 0.7121 t = 1994 0.7796 0.7908 0.7646 0.7391 t = 1993 0.7871 0.7783 0.7026 t = 1992 0.8182 0.722 t = 1991 0.8256 t = 1990

(34)