The influence of institutional investors on risk and performance in the renewable energy sector

The influence of institutional investors on risk and

performance in the renewable energy sector

Thesis for the Master thesis, MSc. Business and Management, specialization International Financial Management

University of Groningen, Faculty of Economics and Business

June 30, 2012 ALEXANDER GLADISCH Studentnumber: 1942166 Meeuwenlaan 83b 1021HW Amsterdam phone: +31 (0)621199964 e-mail: a.gladisch@student.rug.nl ´ Supervisor/ university Dr. H. Gonenc

Abstract ... 4

1 Introduction ... 5

2 Literature review ... 7

2.1 Problem indication, originality and relevance ... 7

2.1.1 Risk and return in the renewable energy industry ... 8

2.1.2 Institutional investor and company risk ... 10

2.1.3 Institutional investor and company value ... 11

2.2 Investor characteristics ... 14

2.2.1 Institutional investor actions and behaviour ... 15

2.2.2 Different institutional types of investors ... 16

2.3 Research question and model ... 19

2.4 Hypotheses and expectations ... 21

3 Data and variable description ... 24

3.1 Data ... 24 3.2 Response variables ... 26 3.3 Explanatory variables ... 27 4 Methodology ... 31 4.1 Theory ... 31 4.2 Model creation ... 33 5 Results ... 36

5.1 Relationship between institutional ownership and firm performance ... 36

5.2 Relationship between institutional ownership and risk ... 39

5.3 Country of origin effects ... 41

5.4 Time effects ... 44

5.5 Robustness checks ... 47

7 References ... 52

Appendix ... 56

Sample descriptives and industry information ... 56

Results ... 57

A

BSTRACT

1 I

NTRODUCTION

Some investors emphasize that “...the biggest investment factor of our lifetime” will be the climate change (Lomax 2012). The importance of renewable energy is growing around the globe because it promises a more secure supply of energy, independence from fossil resources and the limitation of carbon dioxide emissions. The future growth of this industry, which has grown at rates between 30% and 40% in recent years, requires an investment of 10 trillion US dollars according to the International Energy Agency (IEA) (Birol 2009). Till today the main part of the investments needed to change the energy supply from fossil sources towards renewable energy sources are funded by governments and subsidiaries. It is expected that private investors will play a bigger role in the future capital supply for the renewable energy industry (Organisation for Economic Co-operation and Development 2009).

Climate change has a direct impact on the investments of pension funds, foundations and endowments, due to their long investment horizon. Investments in the renewable energy sector become attractive because they can act as a natural hedge against climate change risks (Miller 2011). In 2009 Deutsche Bank Group conducted a survey of institutional investors representing more than 1 trillion US dollars. Results show that 75% of institutional investors want to invest more in renewable energy by 2012 (Deutsche Bank Group 2009). Furthermore several annual international conferences like the Renewable Energy Finance Forums draw the attention of the investment community to the renewable energy industry (Euromoney Energy Events 2010). A study of the United Nations in cooperation with Bloomberg and the Frankfurt School of Finance found private equity investments into the renewable energy industry rose 60% from 2009 to 2010. Still investments of both private equity and investment funds into renewable energy are below the record levels of 2008 but are recovering fast from the financial crisis (United Nations Environment programme 2011).

performing CEOs (Aggarwal, Erel, Ferreira, Matos 2011; Cronqvist, Demiralp, D'Mello, Schlingemann, Subramaniam 2011; Fahlenbrach 2009; Nielsen 2008). The impact on the firm seems to differ depending on the country of origin of the investors. Also higher commitment of the investor to the company is found to result in better corporate governance and higher company value (Baik, Jun-Koo, Jin-Mo 2010; Ferreira, Matos 2008).

On the other hand investors with a short investment horizon tend to foster misreporting and force the company to cut back research expenses and thereby decrease firm value in the long run (Burns, Kedia, Lipson 2010; Guthrie, Sokolowsky 2010). Other authors looking into the role of hedge funds find a negative relationship between firm value and the investors’ engagement (Brophy, Ouimet, Sialm, 2009).

The main impact of institutional investors appears to lie in corporate governance problems that arise through separation of control and ownership. This is known as the principle-agent problem. This paper analyses the influence of institutional ownership on financial performance and risk in the renewable energy industry. I expect institutional ownership to lessen agency costs through better monitoring. The engagement of institutional investors is consequently expected to increase firm value. Regarding the relationship between operational risk in companies and the investment of institutional investors, a clear tendency cannot be given. This research is going to examine the question: How does institutional investor involvement influence the financial performance and risk of companies in the renewable energy industry? If the future expansion of renewable energy supply can only be financed by private investors, it is of relevance for practitioners, politicians and scientists to know how this engagement affects the industry. It is of interest how much the renewable energy industry might differ from other investments and whether institutional investors have the same impact on the renewable energy industry as was found by previous researches in other industries. Also companies based in the US and the EU are compared in order to find out if differing market conditions and political agendas are beneficial for renewable energy investments. In general not much is known about the characteristics of the renewable energy industry so this research has an explorative character.

minority of companies. Outliers are also a problem. In my analyses I found that the involvement of institutional investors in renewable energy companies have a positive impact on the firm value as measured by the Tobin´s q ratio. The impact on risk differs between types of institutional investors. Private equity funds reduce risk while they improve financial performance. The other investor types increase risk.

This paper continues with a summary of the related literature, which is the background for the research framework and the hypotheses. After laying out this framework, the data and methodology used, is described. After this, the results are described and the paper ends with the conclusion and discussion.

2 L

ITERATURE REVIEW

Before addressing in depth empirical findings and the expected relationships in the proposed theoretical model, it is essential to provide a description of institutional investors and the renewable energy industry. The basic concept of risk and return is used throughout this research to analyse the relationship between investors and industry.

2.1 P

,

Political and social interest groups and economic conditions foster changes in energy supply. This requires huge investment funds. According to estimations of the IEA the investment volume will increase for an extra 10 trillion US dollars during the time span till 2030 in order to reach the two degree goals.1 An IEA scenario anticipates that the major part of these additional funds has to be provided by private investors. The role of governments is expected to be limited to setting the right incentive systems (Organisation for Economic Co-operation and Development 2009). Even though more than 70 countries around the world have introduced goals and support mechanisms to foster renewable energy production, the political motives and emphasis around the globe differ. The question is whether this national and international activism provides a sufficient base to encourage investors to

1

supply the necessary funds (Wooldridge 2010). Thus for the purpose of this research the basic risk-return relations of investments in the renewable energy industry illustrate the background for the role institutional investors play in the renewable energy industry. The nature of the renewable energy sector and the impact of institutional investors will be researched along these two characteristics.

It is difficult to give a clear definition of institutional investors. Generally institutional investors are understood as commercial banks, insurance companies, funds (pension funds, mutual fund, private equity, endowment funds and hedge funds) and other financial institutions that possess huge amounts of money to invest and face special regulative treatment. These organisations gained importance on the international capital markets in recent years. The percentage of financial assets over GDP hold by institutional investors in the OECD countries increased from 110.2% (1995) to 162.6% (2005) to a total of 40.3 trillion US dollars (Monks, Minow 2004). These numbers should be seen in relation to the current size of investments in the renewable energy projects. The investment size in renewable energy projects reached 1.2 trillion US dollars in 2010 which is around six times the level of 2004 but still just a sixth of all investments undertaken in the field of energy or 1.5% of all world investments (United Nations Environment programme 2011).

2.1.1 R

Besides the production of energy from renewable sources, the sector contains a wide range of companies that are engaged in supply and development of technologies. These technologies are innovative and have no or a short market history which increases the risk associated with the product. A lack of economy of scale effects, insurances, guarantees and higher planning expenses increases the initial investment costs. On the other hand due to the resource efficiency of the employed technique, operating costs are usually lower. With expected high future market potential, once the technology is mature and a high industry risk (an average beta of 2) the renewable energy industry is comparable with the high tech industry, rather than the traditional utility industry (Henriques, Sadorsky 2008).

Another challenge to the attractiveness of renewable energy projects is investment grading. The small size of projects and their ownership by small companies or special entity vehicles (SPVs) makes it difficult to obtain an investment grade by a rating agency. Especially considering the risk connected to the technology, the newness of the industry and the changing regulative environment investors are asked by regulation to look for rated investment opportunities (Justice 2009).

Besides the risk of the investments, also the return of investments in the renewable energy sector can be a challenge. The attraction of private money for renewable energy projects faces a major problem. Renewable energy projects compete with traditional energy projects concerning cost competitiveness. The technology is new and the enhanced investment risk is difficult to estimate due to lack of experience. The early stage on the learning curve and the degree of maturity of the used technology causes high costs and a lack of cost competitiveness on most markets. Therefore politicians have installed financial incentive systems in order to develop innovative technology and to reach a market size at which production costs decline to a competitive level (Donovan, Nuñez 2012). Understanding the political impact on risk and return in the renewable energy industry is a crucial difference to investment decisions in other industries (Justice 2009; Teppo, Wustenhagen 2006; United Nations Environment programme 2011).

industry. The discrepancy between political, societal and economic cost-benefit analyses is frequently mentioned and discussed (Bergmann, Hanley, Wright, 2006; Jacobsson, Johnson 2000; Valentina 2006; Wiser, Bolinger 2006). Researchers criticise common risk-return calculations for not including external factors into the estimations. It is expected that inclusion of these external factors would make renewable energy favourable over traditional energy source even without governmental subsidies (Bergmann et al. 2006). Jacobsson and Johnson (2000) follow this critique in describing the current risk assessment systems as locked in on a path dependency of the established technologies. Bhattacharya and Kojima (2012) present a scenario analysis to show the case that risk in traditional energy investments often is underestimated while it is overestimated for renewable energy investments. Further the study underlines a risk reduction potential of renewable energy when it is part of the energy portfolio of the whole economy.

The traditional energy production faces increasing risks. Up till 80% of the world electricity is produced from fossil fuels. These are limited and often imported from delicate security regions. The reduction of import dependency and greenhouse gas emissions present a challenge that increases the importance of renewable energy sources. The increasing support and enforcement of more sustainable energy sources around the world mitigates risks of renewable energy investments (United Nations Environment programme 2011). On the other hand also investments in the traditional energy industry face risks. If climate change aspects have to be taken into consideration in the future these risks will increase (Teppo, Wustenhagen 2006).

2.1.2 I

After the description of risk and return in the renewable energy industry this section mentions findings about the behaviour of institutional investors towards risk.

positive skewed portfolios was already found by Kraus and Litzenberger (1976). The given conclusion is that most investors unequally weigh the possibility of high losses and high returns and therefore often prefer a highly positive skewed portfolio over a diversified one with low return variance and less extreme losses (Hueng, Yau 2006).

It appears like institutional investors do not follow the prudent man principle considering risk but also prefer high volatile stocks (Ferreira, Matos 2008; Gompers, Metrick 2001). A positive relation between the presence of institutional investors and the volatility of return is mentioned by Sias (1996). The mentioned relationship leads to his question whether institutional investors prefer riskier assets to invest in or that the presence of institutional investors increases the risk of the company. Sias (1996) finds significant results for the latter, meaning that the presence of institutional investors’ increases risk. Following these research results, institutional investors seem to prefer portfolios that are not well diversified and therefore risky. But more important: institutional investors increase the risk of companies they invest in.

2.1.3 I

The expected additional investment volume into the renewable energy industry of 11 trillion US dollars demonstrates huge investment opportunities for institutional investors (Birol 2009). The question stays how the involvement of institutional investors influences the renewable energy industry. Generally there are chances and risks related to the involvement of institutional investors. Some authors see the principle-agent problem as too simple to explain the impact of institutional investors and mention mixed impacts of the influence of institutional investment activity (Connelly, Hoskisson, Tihanyi, Certo 2010; Faccio, Lasfer 2000; Miller 2011).

improves the corporate governance by reducing agency costs and thereby is able to ensure a satisfying result and to hinder harmful actions of the management.

Baik et al. (2010) show that local institutional ownership concentration is a good predictor for future positive stock returns. The argument in his article is that local institutional investors have an informational advantage over non- institutional and non-local owners. Due to the better information employed by local institutional investors they are capable to predict positive future stock returns. The results are confirmed by the impact of local investors in firms with high information asymmetry like small and young firms, firms with high R&D or high return volatility (Chiang, Qian, Sherman 2010; Yan, Zhang 2009a).

The second major impact of institutional investors considers the principal-agent problem. The principal-agent problem arises through the separation of ownership and control. This separation often is necessary since risk bearing and managerial abilities are not in the hands of the same person (Fama, Jensen 1983). Both owners (principals) and managers (agents) follow their own agendas. Problems arise when the managers´ interest is not in line with the owners´ interest. The principal takes actions in order to ensure that the management works in his or her best interest. The principal’s efforts result in agency costs consisting of monitoring and bounding costs. Even after these investments a gap between the interest of owners and management remains, which causes a residual loss that adds to the agency costs (Jensen, Meckling 1976). Agency costs diminish shareholder value measured as the market value of the company. Institutional ownership reduces the agency costs through monitoring and controlling efforts. In order to protect the investment an institutional investor has an incentive to control and monitor the company´s management, thereby decreasing the residual loss. Thus institutional ownership leads to an increase in market value of companies if they are able to lower agency costs. Additionally to improve corporate governance, investors also control the company from outside to ensure the profitability of their investment. Such free control through outsiders increases firm value since controlling costs are accounted for by the investor outside the company (Hamberg 2004).

investing into shareholder value. Hence cash that is available in the company and not used for profitable investments should be paid out to the shareholders in order to avoid misuse. Institutional investor’s involvement often results in lower cash levels (Achleitner, Betzer, Gider 2010). A high, stable dividend policy relates to institutional investors as well. Such a dividend policy forces the management to generate funds by profitable investments and can be seen as a means by which institutional investors reduce the agency conflict. Similar arguments are connected to high leverage. Since debt financing forces the management to earn money to pay back the loans it reduces possibilities of inefficient investments. Further the lender, usually a bank, takes over a part of the monitoring costs in order to ensure that the company is able to pay back the loan. Hence higher leverages as well as higher dividend payouts are ways for institutional investors to reduce agency costs and increase firm value. On the other hand, higher leverage can lead to financial distress which increases bankruptcy risk (Cronqvist, Fahlenbrach 2009). Demiralp et al. (2011) and Guthrie and Sokolowsky (2010) researched the effects of institutional investors around seasoned equity offerings (SEO). Since companies possess substantial amounts of free cash after a SEO, there is a higher potential for misallocation. Hence in such a situation the enhanced monitoring by institutional investors should be easily measurable. In fact Demiralp et al. (2011) found that a firm with active monitoring investors show positive performance after SEOs.

strategies of institutional investors. They find advantages of monitoring for the institutional investor. These advantages increase with the stake and the time horizon of the investment as well as independency of the investor. They stress that while all shareholders benefit from the monitoring efforts institutional investors still gain an information advantage. Boehmer and Kelley (2009) find that stocks of companies with institutional ownership are more efficiently priced, meaning that they follow their fundamental values more closely.

Leuz, Lins and Warnock (2010) researched the international investment behaviour of institutional investors and found that investors hesitate to engage in countries with weak shareholder protection and known corporate governance problems because of higher information asymmetry and consequently higher monitoring costs. Missing disclosure and corporate governance regulations increase the information asymmetry between corporate insiders and outside investors. The information asymmetry and the costs to overcome these are found to be especially high for foreign investors, explaining the reluctance of foreigners to invest in such countries (Leuz et al. 2010). Guthrie and Sokolowsky (2010) on the other hand find evidence that firms with an outside blockholder tend to inflate earnings. In this case current shareholders use their influence on the company at the expense of future shareholders.

2.2 I

Institutional investors differ in size, time and purpose of their investment in companies. These characteristics influence the investors’ approach to monitoring as well as divestment and can be expected to influence risk and financial performance of the target company.. Investors influence a range of firm decisions like investment decisions, growth, leverage and executive pay because of their contact with the board and management (Cronqvist, Fahlenbrach 2009).

2.2.1 I

corporate governance especially in countries with weak shareholder protection. Corporate governance on the firm level can be seen as a substitute for shareholder protection on the country level. Aggarwal et al. (2011) stress that institutional investors enhance better corporate governance while no support is found for the opposite relationship. For example, institutional ownership of companies results more often in the replacement of poor performing CEOs. Also the firm performance measured by Tobin´s q increases due to higher institutional ownership while the relationship again does not work in the other direction. Institutional investors irrespectively from their country of origin prefer large, widely hold companies with good governance (Ferreira, Matos 2008).

Gompers and Metrick (2001) add to this list of investment preferences. They find a comparable low stock return in the previous year important and report that institutional investors predict higher future returns due to a demand shock. Yan and Zhang (2009b) also claim that short term institutional investors predict future returns especially from small and growth firms. They state that short term oriented institutional investors are better informed than long term oriented and use this advantage through active trading.

2.2.2 D

After the presentation of the general findings on institutional investors and the main characteristics, like active and passive or dependent and independent, some investor specific characteristics are presented in the following part.

V

categories, namely: market adoption risk, technology risk, people risk, regulatory risk, and exit risk (Teppo, Wustenhagen 2006). Besides the provision of money, knowledge and expertise that is provided, the investor tends to take an active role in the company. Typical sectors for venture capital investments are the IT industry and the biotech industry that account for more or less two thirds of the venture capital investments. The sources of venture capital funds are pension funds, insurances, endowments or wealthy private investors (Moore, Wustenhagen 2004). The authors conclude that no industry specific reason can be found to explain the low investment of VC in renewable energy. Since VC invest in companies mainly before they are listed on the stock market, these investor types are of less importance for this research due to data collection as will be explained later on.

P

–

H

(2010), Kein and Zur (2009) find evidence that hedge funds reduce the agency conflict arising from free cash flows by increasing dividend pay-outs, interest rates paid to creditors and increase the debt to equity ratio.

M

P

F

Like other institutional investors, mutual funds vary in their actions as investors. Overall mutual funds act as monitors, for example through voting on shareholder proposals (Morgan, Poulsen, Wolf, Yang 2011). The majority of mutual funds invest in a broad benchmark with a tendency to invest in growth stocks (Chan, Chen, Lakonishok 2002). Especially mutual and pension funds often show herding behaviour (Hill Griffiths, Lim 2008).

Research on the impact of pension funds on corporate governance in the UK does not reach clear results. First, the research disagrees on the level of influence pension funds have on the management of the firms they invest in. Second, even though the governance system in the UK and US are both market based, major differences in the behaviour of pension funds can be observed between both markets (Faccio, Lasfer 2000). The existence of differences between pension funds is also supported by a study undertaken by Hao (2010). This study reports that pension funds successfully introduce changes in the corporate governance and monitoring of the target firm. An impact of pension funds activism on firm value and performance can however not be concluded from the results. Pension funds also trade on intangible information and increase the stock price, leading to a market premium over the intrinsic value (Hao 2010).

B

I

for offshore wind parks and hedges against fluctuating energy productions). The ways how energy efficiency and renewable energy can moderate or reduce insurance losses or become a competitive advantage for insurance and risk management are not being fully explored (Evan 2003).

2.3 R

The presented literature discusses the consequences of institutional investors’ involvement. The results showed positive and negative influences. This section sets and describes the scope, model and questions of the proposed research based on the presented literature.

The research focus lies on the renewable energy industry because the industry is expected to become more important in the future. The amount of research concerning the renewable energy industry is limited. In the field of renewable energy the main research question, apart from technological issues, consider the marketability of renewable energies. Herein the political aspects of support and protection mechanisms are central. As mentioned before the industry is growing and is considered to continue growing in the future at high rates (Birol 2009). A necessity for the expected growth is the availability of sufficient financial funds. Governments, currently supplying a main share of these funds through subsidies and other support mechanism, are less willing and capable to provide further financing. A recent example can be found in the cutbacks in subsidies due to austerity measures in Europe. As can be seen in Spain, which cutback subsidies and was downgraded in the renewable energy attractiveness index, an index summarizing the investment friendliness of the renewable energy industry of countries (Ernst & Young Project Finance International 2012). Due to the uncertainties, complexity and size of the renewable energy projects, new investments are expected from attracting private funds professionally managed by institutional investors (Organisation for Economic Co-operation and Development 2010).

follows these two points of interest in researching the renewable energy industry and the institutional investors´ impact. Hence the research question is stated as:

Q: How does institutional investor involvement influence the financial performance and risk of companies in the renewable energy industry?

The Orbis database offers data on a broad range of different institutional owners that allows for detailed analyses in most cases. In order to reach most accurate results I research the effect of each investor type separately. A grouping of investors into active and passive, dependent and independent or similar classifications as used by other author will be used to interpret the results if this is appropriate (Ferreira, Matos 2008; Bushee 2001; Demiralp et al. 2011; Miguel, Pindado, La Torre 2004).

Further aspects of the proposed research are country differences. The sample used in this research combines companies from countries all over the world which allows looking into country specific effects on financial performance and risk. Previous research found differences between developed and developing countries in their attractiveness to institutional investors. Also institutional investors were found to employ different investment strategies in these two groups of countries (Aggarwal et al.; 2011Baik et al. 2010; Chen et al. 2009; Leuz et al. 2010).

The focus of this research lies in the difference between companies from the US and the EU. Both areas are comparable according to general industrial development but differ in areas like political and financial support for renewable energies (Heal 2010; Delbeke Klaassen, van Ierland, Zapfel 2010). There are structural differences when comparing the capital markets, like the civil law versus the common law tradition. Since institutional investors are a major player on the capital markets, different capital market traditions appear to be an interesting aspect of potential influence (Aggarwal et al. 2005).

Environment program 2011). The crisis as well could have an impact on financial performance and risk in the renewable energy industry and is included in the research model.

2.4 H

Following the broad research framework proposed before, this section will describe the developed hypotheses and expectations marking the foundation of this research.

In line with the presented literature I assume that institutional investors have a positive impact on the company value of renewable energy companies. Due to improvements in corporate governance and the reduction of agency conflicts through institutional investors, the financial performance is expected to increase (Achleitner et al. 2010; Baik et al. 2010; Cronqvist, Fahlenbrach 2009; Demiralp et al. 2011). I use seven different ownership types. Six are institutional investors. These different investor types are banks, insurances, financial institutions, mutual and pension funds, private equity funds and foundations. Additionally public authorities are included to detect a potential political influence. The influence on the companies is expected to differ between the investor types. I do not combine investor types into two groups as performed by other authors because the information of the specific influence of each investor type would be lost. Nevertheless the grouping characteristics can be used to formulate expected results. Active and independent investors are expected to have a bigger positive influence on financial performance. This group includes private equity funds, venture capital and financial institutions. Rather passive and dependent investors are banks, insurances and foundations which are expected to have less influence on the firm value (Bushee 2001; Demiralp et al. 2011; Ferreira, Matos 2008; Miguel, Pindado, La Torre 2004). Mutual and pension funds are combined as one investor type by the Orbis database They are hard to place in one of these groups due to the ambiguous research findings (Chan, Chen, Lakonishok 2002; Hao 2010).

Especially long term oriented investors should be able to decrease financial distress and improve corporate governance which can be expected to reduce the risk a company faces. On the other hand the combination of lower risk and higher financial performance conflicts with the general theory of efficient markets that excludes ‘free lunches’. Investment opportunities that combine higher returns with lower risk should vanish in an efficient market (Fama 1970). A possible explanation why such situations occur anyhow could be an informational advantage of institutional investors as mentioned by Baik (2010). Another reason is a short investment horizon like in the research of Achleitner (2010). The combination of short term investments and informational advantages could result in the case of higher financial performance and lower risk. Other researchers found that institutional investors increase the risk of companies (Sias, 1996). Therefore the hypothesis assumes the impact on the risk of companies to differ depending on the type of investor. Overall no specific expectations can be given about the impact of institutional investor involvement on the risk of a company.

The first hypothesis therefore is proposed as:

H1a: Companies owned partly by an institutional investor have higher financial performance than companies without institutional investors.

H1b: Companies owned partly by an institutional investor face different risks than companies without institutional investors.

2009; Ferreira, Matos 2008; Leuz et al. 2010). The country of origin effect considered in this research gives attention to the difference between the US and the EU. The home country of the company is expected to have an influence on the ownership structure so differences in the capital markets become interesting. Also political support for companies can be expected to depend on the country of origin. Companies in the EU are expected to have advantages over companies in the US due to higher subsidiaries and proactive government intervention. The capital markets differ between countries with a common law tradition like the US and countries with a civil law tradition like most parts of the EU (Aggarwal et al. 2005). The US capital market is characterized through a high number of small shareholders in comparison with the EU where traditionally large shareholders like families or, especially in Germany, banks and insurances play an important role. Due to these differences institutional investors play a more important role reducing the principal-agent problem in the US than in the EU where this problem can be expected to be smaller, because of the presence of large shareholders. The ambitious renewable energy agenda of the EU in combination with subsidiaries in a range of EU countries lead to growth in the renewable energy industry. Also environmental goals and regulations are advanced and similar throughout the EU, even though single countries tailor made regulations and support systems due to specific national circumstances and preferences. The US, as another big area of comparable economic strength and development, chose for a more market oriented support system which offers companies’ tax cuts for renewable energy investments. Here the energy independency motive ranks higher than the greenhouse gas reduction goals that are the political incentive behind regulations and supports in the EU. These differences should allow the EU based industry to have a higher financial performance. This economical advantage depends to a degree on the political situation. Like in the earlier mentioned case of Spain the political situation runs the risk of sudden changes. If EU based companies rely partly on the favourable political circumstances, they might be less competitive compared with the US based companies. Hence the risk of the companies is expected to be higher in the EU compared to companies in the US.

H2a: Alternative energy companies in the EU have higher firm values than companies in the US.

Justice S. (2009) discusses the impact of the crisis on the renewable energy industry in detail. Referring to the introduction of this paper I mentioned that investments from institutions into the renewable energy sector plumbed in the end of 2008 as a result of the financial crisis of 2008 (United Nations Environment programme 2011). This reduction in invested funds should have had a measurable effect on the renewable energy sector. The financial crisis cannot be expected to change the fundamental risk that companies in the renewable energy industry have to manage. This risk comes from new technologies with unsure market chances and stays independent of economic developments. Still the financial crisis can be expected to increase the risk for the industry on the financial side. While it should be easier to find investors into risky technologies during good economic times the recent financial crisis leads to a scarcity of financial funds for the whole economy. Hence problems to attract sufficient financing add risk to the renewable energy sector after the crisis 2008. As the problems of the financial crisis where followed by government debt crisis in parts of the EU, I expect the risk in the alternative energy sector to be higher for the years after 2008. The hypotheses are expressed as follows:

H3a: The period 2007-2008 shows an increase in financial performance in the renewable energy industry.

H3b: The period 2009-2010 shows an increase of risk in the renewable energy industry.

3 D

ATA AND VARIABLE DESCRIPTION

3.1 D

and 99% level to handle outliers. Companies that do not have data for two consecutive years for the dependent variables are deleted from the sample so that the final sample size consists of 355 companies.

The data for ownership is provided by Orbis and used as follows. The database offers 16 ownership classifications2. The database reports controlling ownership, therefore voting rights are mentioned where available. The ownership is reported as a percentage of voting shares. There are two types of ownership reported. “Direct ownership” describes a direct link between two companies like firm A owns a certain percentage of firm B. The second case “Total Ownership” describes situations where firm A owns a stake in firm B but it is unclear in which way the ownership is acquired. For example firm A could own firm B via a company C. Note that ownership can exceed 100 % (Orbis - User Guide). I used the values for “direct ownership” where available. Gaps in direct ownership and cases where no “direct ownership” was mentioned are filled with the figures for “total ownership”. This means in cases where values for both direct and total ownership are available I ignored the values for total ownership in order to avoid double counting.

The values for the three ownership types “Public”, “Unnamed private shareholders, aggregated” and “Other unnamed shareholders, aggregated” are deleted from the ownership data base. Since it is mentioned in the ORBIS descriptions that these groups are having no control over the company. Further, the data about the nationality of the company and the investors are taken from the Orbis database.

The financial data obtained from DataStream includes market value of the company, total assets, cash and cash equivalence, net income, total debt, capital expenditures, research and development expenditures and total sales all for the time frame from 2005 to 2011. All variables and their modifications will be described in more detail in the following part.

2

3.2 R

F

Institutional investors are expected to decrease the principle-agent problem through improved monitoring and controlling of the company and a positive effect is assumed on firm value and performance. In order to measure this impact, different variables for firm value are suitable as dependent variable. One option is the usage of the change in stock returns or return on assets as a dependent variable. Another often used measure for firm performance to test effects of corporate governance is the Tobin´s q ratio (Doidge, Karolyi, Stulz. 2004; Fahlenbrach, Stulz 2009; Konijn et al. 2011). Mc Connell, Servaes and Lins (2008) use stock returns as a proxy for Tobin´s q in a study about insider ownership and its impact on firm value. Cronqvist and Fahlenbrach (2009) use ROA and Tobin’s q as variables for firm performance in a research about the impact of block holders on firm policy. The variable is calculated following Doidge et al. (2004) and Ferreira and Matos (2008) as the sum of total assets (WorldScope: 02999) plus market value of equity (DataStream: MV) minus book value of equity (WorldScope: 03551). Tobin’s q is a ratio of the market value of a firm to the replacement costs of its assets.

One problem pointed out in a paper by Doidge et al. (2004) who used the same database and formula for estimating Tobin´s q is that accounting practises differ between countries. Since DataStream values are based on local accounting principles the obtained numbers might not reflect comparable values. In cases where accounting numbers report historical instead of replacement values for assets the Tobin’s ratio becomes inaccurate. Reserves and R&D expenses are also mentioned examples for such different principles. Even though various measurement errors and problems of the Tobin´s q ratio are known, discussed and researched it is still widely used (Gompers, Ishii, Metrick 2010; Perfect, Wiles 1994). Also because other studies used the same measure which makes comparison of results easier I chose the Tobin´s q as the measure for financial performance.

RISK

companies, e.g. the z-score for banks and financial intuitions (Valentina 2006). For this research the standard deviation of Returns on Assets (ROA) is chosen as risk measure. This measure is an easy to use, standard measure of operational risk (Agusman 2008). The variable is constructed as the three year standard deviation of the company’s ROA (WorldScope: 08326)(Teddlie, Tashakkori 2008). ROA is a profitability measure. The standard deviation describes the variation of the ROA from the mean over three years. A high value of this measure stands for large changes of ROA from year to year, hence the operational risk of the company is high. While Tobin´s q measures the financial performance, the risk variable describes how stable the company´s operations are. General the standard deviation is calculated over five or even ten years. Due to the short research period and data availability a three year calculation was chosen. This gives single years more impact and can therefore inflate the risk measure.

3.3 E

O

equity funds since the few observations do not allow to analyse them separate and private equity funds appear to resemble investor types the most.

Table 1:

Descriptives: ownership of different investor types in the renewable energy industry in percent (FIN: Financial Institutions; FOU: Foundations; HF: Hedge Funds; INS: Insurance companies; MPF: Mutual and Pension Funds; PEF:

Private Equity Funds; PUB: Public agencies, governments, states; VC: Venture Capital Funds)

The geographic distribution of 355 companies in the sample is shown in Figure 1. The companies are spread over 33 countries. As can be seen, the majority of the companies are from the United States (US) with Germany on the second place. Almost two thirds of the sample, 230 companies, have their origin in the EU or the US.

Figure 1

Country of origin of companies from the renewable energy industry in the sample.

Details about the origin and types of the investors into renewable energy companies are summarized in Table8 in the appendix. Overall 2463 investors from 56 different countries

BANK FIN FOU HF INS MPF PEF PUB VC

invested in the companies represented by the sample. 115 out of the 355 companies do not have an institutional investor. Most investors come from the US (989), followed by Great Britain (362) and Germany (151). The additional data given in Table 8 in the appendix consider non institutional investors like industrial firms and the governments show that industrial companies play a more important role as investors in the EU (192) than in the US (92). Especially Germany with 78 investors from this category sticks out. Governments on the other hand account for smaller amounts of investments, namely: US: 3; EU: 9.

CONTROL VARIABLES

working capital (Burns et al. 2010; Guthrie, Sokolowsky 2010). The expenses for research and development (WorldScope: 01201) dived by the amount of total assets is used as a measure of how much the company invests into the development of future products (R&D). These investments are certainly necessary to stay competitive and increase market share. Hence the R&D variable should have an impact on the financial performance. On the other hand the variable also indicates how much money the company puts into projects with unsure payoffs. Since the outcomes and profitability of research investments are difficult to estimate beforehand, the variable is also related to the risk a company faces. Capital expenditures (WorldScope: 04601) divided by total assets as used by Guthrie and Sokolowsky (2010) and Konijn et al.(2011) is ratio that put the cost for acquiring and maintaining assets in relation to the overall asset base (Capex). Expecting that investors see these investments as necessary and financially sound, an increase in the capital expenditure ratio should increase the firm value. Net income (WorldScope: 01551) dived by total sales (WorldScope item 01001) shows how much the company gains from its business. An increase in this profitability ratio should lead to an increase in firm value (NPM) (Aggarwal et al. 2011). The variables Capex and NPM are found not to have an impact in the variables and are therefore dropped from the analyses. Table 2 presents descriptive data from the control variables. Some outstanding points are the debt ratio and cash ratio with a maximum of 1 as well as the negative average net profit margin. A company with research and development cost more than six times the value of all assets, can only be explained by a huge one time investment and cannot be seen as typical. This summary shows the huge differences between the companies even after winsorizing the variables; some values seem unrealistic for the long run. The values are often caused by companies at the edge of bankruptcy.

Table 2

Descriptives of the control variables. All variables are winsorized at 1% and 99%

4 M

ETHODOLOGY

An econometric model that is correctly specified, in the sense that it includes all relevant explanatory variables, is essential in order to have unbiased estimators (Hill et al. 2008). The following section discusses the specification of the model and the operationalisation of the introduced variables.

4.1 T

For the drawing of the sample from the population a non-probability purposive sampling selection is used (Teddlie, Tashakkori 2008; Thomas 2004). Still, the cross-section observations can be seen as independent, identically distributed and draws from the population since the number of observations is much larger compared to the number of periods (Wooldridge 2010). The combination of time series and cross sectional data in a panel of data provides several advantages. Panel data analyses have higher degrees of freedom and therefore are more powerful tests. Also it allows mitigating possible problems of multicollinearity (Brooks 2008). Probably the main advantage of panel data is that it provides the option to control for heterogeneity, i.e. variables that are not measurable or directly observable (Baltagi 2008). In this case, such variables are company specific characteristics which affect the firm performance and cannot be found through accounting data. Furthermore, it is possible to account for time-variant factors, which change in time, but stay constant across entities, such as market conditions and regulatory policies (Baltagi 2008).

A range of models are available to analyse panel data. Some of these are more sophisticated than others and have different advantages and disadvantages. The following part gives an inside in the model chosen and the reason why.

The method ignores any common structure in the data as well as any common variation in the data over time (Brooks 2008).

Next to the POLS regression model a range of more sophisticated models are available for panel data analysis. The easiest of these analyses is the use of a SUR (seemingly unrelated regression) model. The use of the SUR model is limited to cases where time series observations are equal or above the number of cross-sectional observations. Hence SUR is not a visible option for the analysis of the data in this research. For a sample that is “short and wide” as the case in this research, the two models that allow more sophisticated analysis are the random effects model and the fixed effects model. The fixed effect model allows the intercept to differ cross sectional but is stable over time while the slope is fixed. The model divides the error term into an individual specific part and one that captures the remaining variation over time and entity. The assumptions are that errors are independent, with a zero mean and a constant variance of time and entities (Baltagi 2008, Hill et al. 2008). A disadvantage of the fixed effects model is that the influence of variables that do not change over time but still influence the dependent variable is lost (Brooks 2008). A variation of the fixed effect model is the time fixed model. In this model the dependent variable changes over time and not cross-sectional. This model is appropriate to use if it can be expected that changes in the data happen over time in a comparable way to all individuals in the sample. A similar model as the fixed effects models can be created by the use of dummy variables either for each individual or for each time period for the case of time fixed effects. Obviously this strategy is only feasible if the number of observations is small (Hill et al. 2008).

either justifies the use of the random effect model which is generally believed to be more efficient or it indicates the better suitability of the fixed effect model (Baltagi 2008; Brooks 2008).

4.2 M

As mentioned the advantage of panel data are the opportunities to deal with heterogeneity and dynamic effects. Differences between the companies in the sample that cannot be captured by variables as well as factors that impact the sample over time, can be captured in the panel data models. The descriptive analyses of the variables in the previous chapter already showed the variety of companies included in the sample. Further, as the number of data gaps shows, the industry changes fast which results in companies entering the market and leaving. The nature of the industry, including high degrees of research and new and risky technology, adds to the dynamic. Hence it can be assumed that factors within companies that are not captured by the variables influence their riskiness and financial performance. The model therefore should account for cross-sectional heteroskedasticity. The sample includes companies that actually produce renewable energy, companies that deliver technology and other that contribute in a broader sense to the alternative energy sector, so it would be desirable to control for group differences. Unfortunately no data concerning detailed business models is available so that this difference cannot be controlled for. It has to be assumed that all companies belonging to the alternative energy sector do not differ group wise. Further the period under investigation matches with the recent financial crisis. Since dramatic economic changes affect all companies in an industry in a similar way, like for example decrease in demand and financial support, it is plausible to investigate a time effect on risk and financial performance.

Building up on the previous discussed methodology the model to test hypothesis 1 is developed. The model consists of the dependent variable Tobin´s q (TQ) for hypothesis 1a and risk for hypothesis 1b. The percentage of institutional ownership of one investor type and the mentioned company characteristics as control variable are part of the equation. Dummy variables for the years 2007 till 2010 are included to control for time effects. The model for hypothesis 1a,b is specified as:

(1a,b) TQit or Riskit= α + β1*cashit + β2*levit + β3*R&Dit + β4*sizeit + β5*d2007+ β6*d2008 + β7*d2009 + β8*d2010 + β9*Inv Xit +δi + εit

FORMULA 1

with αas the intercept parameter, δi the fixed effect containing the omitted variable for every entity i, β1..i the estimated parameters, Inv Xit as the variable describing the ownership percentage of a certain type of investor for every entity i at time t, cash, lev, R&D and size as control variables for every entity i at time t, d2007 till d2010 as year dummy variables and εit the error term capturing the remaining heteroskedasticity for every entity i at time t.

As mentioned before previous research found differing impacts of institutional investors depending on the type of institution. The separate analyses of the X investor types as proposed in the model by the variable Inv Xi allows to grasp these investor type specific impacts.

The statistical model for hypothesis 2 tests the country of origin effect. Therefore the model includes dummy variables for the companies based in the US and in the EU. These two areas capture a large part of the sample as well as differing political agendas and capital market traditions. Since not all countries are included, the dummy variable trap is avoided. Unfortunately the combination of the fixed effects model with dummy variables for the US and EU leads to a near singular matrix. Hence the model to test hypothesis 2a,b is a simple panel least square model without fixed effects. The year dummies to capture time effects are still included in the model. The resulting model to test hypothesis 2a, b is specified as:

(2a/b) TQit or Riskit= α + β1*cashit + β2*levit + β3*R&Dit + β4*sizeit + β5*d2007+ β6*d2008 + β7*d2009 + β8*d2010 + β9*Inv Xit + β10*dumUS + β11*dumEU + ε

with α the intercept parameter, β1..i the estimated parameters, Inv Xit as the variable describing the ownership percentage of a certain type of investor for every entity i at time t. Cash, lev, R&D and size are control variables for every entity i at time t, dumUS and dumEU are the dummy variable that capture the country of origin of companies from the US and the EU respectively, d2007 till d2010 are the year dummy variables and ε is the error.

The third hypothesis suggests an effect of the financial crisis in 2008 on the financial performance and risk in the renewable energy industry. In order to find such an effect two dummy variables, one for the years 2005 till 2008 and one for the years 2009 till 2010 are included into the model. Also the analysis is split and performed once for the time frame 2005 till 2008 and once for the time frame 2009 till 2010. Again a normal panel least square model is used for this analysis since data quality does not allow fixed effects analysis. The expectations are a higher financial performance for the years 2007 till 2008 before the crisis and higher risk for the time period 2009-2010 after the crisis.

(3a/b) TQit or Riskit= α + β1*cashit + β2*levit + β3*R&Dit + β4*sizeit + β5*d2005-2008 + β6*d2009-2010 + β7*Inv Xit + β8*dumUS + β9*dumEU + ε

FORMULA 3

with α the intercept parameter, β1..i the estimated parameters, Inv Xit as the variable describing the ownership percentage of the seven different types of investors for every entity i at time t. Cash, lev, R&D and size are the control variables for every entity i at time t, dumUS and dumEU are the dummy variables that present the country of origin of companies from the US and the EU respectively, dummy variable d2005 -2008 captured the time effect before the crisis and d2009-2010 after the crisis and is ε the error term.

impact that is measurable in the dependent variables. Therefore the analyses are run also with lagged values for the ownership variables.

Further the linearity of the relationship between institutional ownership and financial performance and risk is controlled for. As Guthrie and Sokolowsky (2010) and Konijn et al. (2011) mention in their research, institutional investors do not always need to improve control and monitoring in a company but can act in the opposite way when they become a dominant investor. In order to control if the relationship of institutional ownership on the dependent variables changes with the concentration of ownership, a quadratic ownership variable is added.

As an effort to improve the results of the proposed models, cross-sectional White is used as the coefficient covariance method. So that heteroskedasticity – robust standard errors are achieved in all models.

5 R

ESULTS

The results presented in this section are based on the described data by performing the mentioned tests in the statistic program EViews 7.2. The presented tables consist of the essential information needed to report on the developed hypothesis. Additional tables are placed in the appendix. The analysis uses 1071 observations for the dependent variable Tobin’s q and 779 observations for the analysis of company´s risk.

5.1 R

The result reported in Table 3 show the impact of the different types of institutional ownerships on the financial performance in the sample of companies from the alternative energy industry. The fixed effects model with year dummies as described in the methodology section has an adjusted R2 of 0.81, indicating a good fit of the model to the used data. The model tests the relationship between institutional ownership and financial performance as described in hypothesis 1a. , which assumed a positive relationship between the variables.

the highest coefficients but the results for foundations are only significant at a 10% level. Only the results for banks are significant at the 1% level. Banks are also the institutional investor with the smallest impact on the company value.

The control variables leverage (lev), research and development expenditures (R&D) and size are also found to be significant at a 1% level. The coefficients for leverage and research and development expenditures are positive with coefficients of around 2.22 and 0.33 respectively for all investor types. This indicates a strong positive effect of higher leverage on the financial performance. Research and development expenditures have smaller impacts. Both observations match the expectations and previous research findings (Chiang, Qian, Sherman 2010; Yan, Zhang 2009a). The positive impact of higher leverage can be interpreted as a better use of the available funds. The use of more debt allows for more investments that increase the financial performance. Expenditures for research and development are important to allow future growth. As mentioned before the renewable energy industry is described as a research intensive industry, therefore the positive coefficient for this variable hardly surprises either. The variable size has a negative coefficient of around -0.52 for all types of institutional ownership. Larger companies have a lower financial performance than smaller ones which was expected since larger firms usually grow at lower rates.

Year dummies are included in the analyses to capture possible time effects. Only the year dummies for the years 2010 shows significant results. Also the United Nations Environment programme (2011) mentions an increase of investments into the renewable energy industry for 2010. The other year dummies do not show significant results so that no change in the financial performance over the years can be observed.

Table 3

This table reports estimates of coefficients of the annual fixed effects panel analysis on Tobin´s q (TQ).

Eq Name: Bank Financial

institution Foundation Insurance

The regressors include total cash divided by total assets (cash), total debt divided by total assets (lev), research and development expenses divided by total assets (R&D), log of total assets (size), dummy variables for the years 2007 till 2010 and one of the following institutional ownership variables: Banks (BANK), financial institutions (FIN), foundations (FOU), Insurance companies (INS), mutual and pension funds (MPF), private equity funds (PEF), public authorities (PUB).

Standard errors are given in brackets. Significant levels indicated by asterisks, 1% significance level ***, 5% significance level **, 10% significant level *.

The check for the right relationship between the variables of institutional ownership and financial performance is reported in Table 9 in the appendix. A quadratic relationship between institutional ownership and financial performance does not show significant results for most investor types. Only insurances and private equity funds have significant negative results for the quadratic term at a 1% significance level. Small investments by these investor types still have a positive influence on the firm’s performance but there are indications that the effect changes to a negative influence for very large stakes.

5.2 R

The fixed effects model analyses the relationship between the different types of institutional ownership and the risk in companies of the alternative energy industry as mentioned in hypothesis 1b. The results show significant results for all investor types except mutual and pension funds. Furthermore all results are positive except the value for private equity funds. The fit of the model to the data is high with an adjusted R² of around 0.74. Banks, financial institutions, foundations, insurance companies, and public authorities increase the risk in the invested companies as can be seen by the positive coefficients. Only private equity funds decrease risk in their investments. Foundations and insurance companies both with significant results at the 1% level show the highest coefficients 2.5 and 1.4 respectively. Public authorities also report a high coefficient but the results only have a low significance (10% level) and the standard error is with 1.5 much higher than any other standard error. The results for public authorities therefore appear less reliable. The impact of banks and financial institutions is way smaller. Only private equity funds have a negative coefficient. Private equity funds are the only investor type who improve the financial performance and at the same time reduces risk in the companies they invest in. Unfortunately the significance of this result is low.

Table 4

Eq Name: Bank Financial

institution Foundation Insurance

Mutual and Pension funds Private equity funds Public authorities

Dep. Var: Risk Risk Risk Risk Risk Risk Risk

The regressors include total cash divided by total assets (cash), total debt divided by total assets (lev), research and development expenses divided by total assets (R&D), log of total assets (size), dummy variables for the years 2007 till 2010 and one of the following institutional ownership variables: Banks (BANK), financial institutions (FIN), foundations (FOU), Insurance companies (INS), mutual and pension funds (MPF), private equity funds (PEF), public authorities (PUB).

Standard errors are given in brackets. Significant levels indicated by asterisks, 1% significance level ***, 5% significance level **, 10% significant level *.

Considering the control variables again, leverage and size are significant at a 5% and 1% significance level respectively. As is expected, higher leverage increases the risk as can be seen by the positive sign of the coefficients. This result fits to the explanation suggested in the previous part, that higher leverage increases financial performance but increases financial distress and thereby risk at the same time. The use of more debt increases the risk of bankruptcy. The risk reducing impact of size can be interpreted in a similar fashion. Larger companies have a lower financial performance but are more stable or less risky.

The year dummies show significant negative results for all years and all investor types. The coefficients for the year 2009 and 2010 are slightly smaller than the ones for the years 2007 and 2008. This would mean that the risk was lower for the years after the financial crisis in 2008. The effect of the financial crisis will be further analysed in section 5.4.

The test for a quadratic relationship between the ownership variable are presented in the appendix in Table 10. This analysis leads to no significant results for the quadratic term. As far as risk is concerned a linear relationship between the dependent variable and the institutional ownership variables appear appropriate. The size of the investment of institutional investors does not change the effect on the risk in the company.

5.3 C

Table 5

Country of origin effects on Tobin´s q (TQ).

This table reports estimates of coefficients of the annual panel least square analysis of Tobin´s q (TQ). The regressors include total cash divided by total assets (cash), total debt divided by total assets (lev), research and development expenses divided by total assets (R&D), log of total assets (size), dummy variables for the years 2007 till 2010, dummy

variable for country of origin from the EU and US and one of the following institutional ownership variables: Banks (BANK), financial institutions (FIN), foundations (FOU), Insurance companies (INS), mutual and pension funds (MPF), private equity funds (PEF), public authorities (PUB). Standard errors in brackets. Significant levels indicated by asterisks.

Eq Nam e: Bank Financial

ins titution Foundation Ins urance

Table 6

Country of origin effects of Risk.

This table reports estimates of coefficients of the annual panel least square analysis of Risk. The regressors include total cash divided by total assets (cash), total debt divided by total assets (lev), research and development expenses divided by total assets (R&D), log of total assets (size), dummy variables for the years 2007 till 2010, dummy variable for country

of origin from the EU and US and one of the following institutional ownership variables: Banks (BANK), financial institutions (FIN), foundations (FOU), Insurance companies (INS), mutual and pension funds (MPF), private equity funds (PEF), public authorities (PUB). Standard errors in brackets. Significant levels indicated by asterisks. 1% significance level

***, 5% significance level **, 10% significant level *. Eq Nam e: Bank Financial

ins titution Foundation Ins urance

Mutual and Pens ion funds Private equity funds Public authorities