Journal of Business Ethics

, Volume 117, Issue 1, pp 137–151

The Drivers of Responsible Investment: The Case of European Pension Funds

Authors

  • Riikka Sievänen
    • Department of Economics & Management, Faculty of Agriculture & ForestryUniversity of Helsinki
  • Hannu Rita
    • Department of Forest Sciences, Faculty of Agriculture & ForestryUniversity of Helsinki
    • Department of Economics, Econometrics & Finance, Faculty of Economics and BusinessUniversity of Groningen
Article

DOI: 10.1007/s10551-012-1514-0

Cite this article as:
Sievänen, R., Rita, H. & Scholtens, B. J Bus Ethics (2013) 117: 137. doi:10.1007/s10551-012-1514-0

Abstract

We investigate what drives responsible investment of European pension funds. Pension funds are institutional investors who assure the income of part of the population for a long period of time. Increasingly, stakeholders hold pension funds accountable for the non-financial consequences of their investments and many funds have engaged in responsible investing. However, it appears that there is a wide difference between pension funds in this respect. We investigate what determines pension funds’ responsible investments on the basis of a survey of more than 250 pension funds in 15 European countries in 2010. We use multinomial logistic regression and find that especially legal origin of the country, ownership of the pension fund and fund size-related variables are to be associated with pension funds′ responsible investment. For fund size, we establish a curvilinear relationship; especially the smallest and largest pension funds in the sample tend to engage with responsible investing.

Keywords

Pension fundsSocially responsible investmentSurveyCorporate social responsibilityResponsible investingEuropeInvestmentsMultinomial logistic regression analysis

Introduction

Responsible investing has become mainstream (Sparkes and Cowton 2004; Eurosif 2010; UN PRI 2012). Especially pension funds, who are the main shareholders in many OECD countries, have amended their traditional investment strategies. Increasingly, they decide to include environmental, social, corporate governance and ethical criteria in their investment decisions (Eurosif 2010). Apart from a general call in society that wants to hold companies accountable for what they do, many funds are convinced themselves too that responsible investing needs to be included in their business models (UN PRI 2011a; 2012). Furthermore, there is increasing evidence that it does matter what pension funds do. For example, Neubaum and Zahra (2006) find that long-term institutional ownership is positively associated with corporate social responsibility (CSR). They also find that activism interacts with long-term institutional holdings and that this positively affects corporate social performance. In contrast, Barnea and Rubin (2010) find that institutional ownership is not significantly related to CSR. The reason may be that the institutional investors’ duration is in fact much smaller than generally suggested, in particular because of the ageing population in industrialized economies (OECD 2009b). Aguilera et al. (2006) reveal that differences in corporate governance arrangements in the UK and the US translate into differences in the importance ascribed to a company’s CSR policy. David et al. (2007) find that shareholder activism actually reduces corporate social performance: activism engenders diversion of resources away from CSR to political activities by managers to resist external pressures and to retain discretion. Prior et al. (2008) find that part of the investments in CSR is used for managerial entrenchment to gain support from stakeholders after having employed practices that damage shareholders’ interests like earnings management. Dam and Scholtens (2012) show that ownership by institutional investors is neutral regarding CSR performance of the investment object.

A major problem regarding CSR is that we lack a scientifically as well as a practically accepted definition of what socially responsible behaviour actually is. CSR tends to vary across industries, cultures, firms—but also among individuals. Although CSR is often viewed as an aggregate construct, social responsiveness is fundamentally multidimensional (McWilliams et al. 2006). It embodies a large and varied range of corporate behaviour in relation to its resources, processes and outputs, and it impacts both on the costs and the revenues of the firm (see Waddock and Graves 1997; Wood 2010). Furthermore, CSR generally includes much more than only social responsibility in a strict sense; there is a general understanding that economic and environmental responsibility is included too (Derwall 2007). Showing a high degree of CSR may therefore require a very diverse range of activities (such as engagement in community activities, diversity of management and workforce, philanthropy, reduction of environmental impact, employee empowerment, etc.). Each of these activities may have a separately identifiable impact on firm performance and reputation.

Responsible investing has grown rapidly in the last decade (Eurosif 2010). Responsible investment (RI), which is also called socially responsible investment (SRI) or sustainable investment (SI) in the academic literature, is done via screening investments, engaging with companies, shareholder activism, community investing, and social venture capital funding. In this article, we use ′responsible investment′ when addressing the different names that describe investment that takes into account non-financial criteria. Most responsible investment is undertaken by large investors; retail investors make up only a small fraction of total responsible investment (Eurosif 2010). In responsible investment, investors try to account for environmental, social, governance (ESG) and ethical issues in the investment process. Thus, it encompasses different stakeholder interests, ranging from economic (such as institutional investors, banks, venture capitalists), organizational (such as labour unions), and societal (such as international organizations, governments, non-governmental organizations, academics). The responsible investment market has become more diverse over time, which has been accompanied by a change in the qualitative nature of these investments. The early responsible investment predominantly was based on negative screens, but current practices are much more based on pro-active positive screening and shareholder engagement. Initially, responsible investment mainly was undertaken by retail investors. Since the 1980s, however, the overwhelming majority of responsible investment is via institutional investors. The global financial crisis of 2007/2009 has led consumers and investors to pay more attention to democracy and responsibility in the markets (Banerjee 2010). This is accompanied by increased attention for transparency and accountability of market participants. As such, responsible investing is becoming mainstream.

We investigate what determines responsible investment with pension funds. More specifically, we try to complement the literature by investigating what determines pension funds′ responsible investment. To this extent, we will rely on information about the institutional context of the pension fund as well as on fund-specific information. By focusing on the key business of the funds, we investigate what determines their (lack of) responsible investment. We investigate the legal system, the ownership of the funds, the organization of the pension plan and several size-related variables. We have a survey of more than 250 pension funds from 15 European countries in 2010. We use multinomial logistic regression to determine whether particular variables do play a significant role in pension funds′ responsible investment. We find that especially size, having Scandinavian or English legal origin, being a public pension fund, having a defined-benefit system, and offering a statutory pension plan appear to have a significant positive impact on responsible investment by European pension funds.

The structure of this article is as follows: Section two provides the background about CSR, responsible investment, and pension funds. The data and methodology is introduced in section three. We present the results of our analysis in section four. The last section concludes.

Background

Elkington (1998) suggests that sustainability translates into CSR at the level of firms and industries. In his view, firms need to manage their business and keep a balance between people, the planet and profits (Triple P). The literature provides several insights into the determinants of CSR (Ioannou and Serafeim 2010; Wood 2010; Gjølberg 2009a; Waddock 2008; Campbell 2007). Bénabou and Tirole (2010) find that corporate social behaviour is linked to motives both at the individual and at the company level. Hockerts and Moir (2004) and Harjoto and Jo (2011) argue that CSR focuses on issues which impact stakeholders, whereas responsible investment is a way to evaluate a company’s response to several stakeholders.

Why might pension funds care about CSR? An important reason is the connection with financial performance. Margolis et al. (2007) perform a meta-analysis on 167 studies of how CSR relates to financial performance. In 58 % of the studies there is a non-significant relationship between CSR and financial performance. There is a positive relationship in 27 % of the studies, and a negative relationship in 2 %. The other 13 % did not report sample size or other key characteristics that made it impossible to test for significance. These meta-analytic results suggest that companies do not appear to suffer financially for behaving responsibly. The recent empirical evidence supports this conjecture for the investment community (see Bauer et al. 2005; Galema et al. 2008). Clark and Hebb (2005) also find that investors seeking to protect the value of their investment are very attentive to the sensitivity of the share price regarding the reputation of the firm they invest in. The non-negative association may be a reason for some shareholders to advance CSR. Then, management may act differently regarding social activities in the presence of a particular type of shareholder than they would without this shareholder.

From the perspective of the firm, instrumental stakeholder theory helps explain why firms might account for owners when investing in CSR (see Jones 1995). This theory suggests that firms can reduce transaction and agency costs by taking social initiatives that affect stakeholder relations. The initiatives may directly affect the stakeholder if the initiatives have a positive contribution regarding the goals and aims of that stakeholder. For example, a firm which is partially owned by unions could improve health and safety conditions far beyond what is legally required. There can also be an indirect effect namely when the initiatives of the firm are viewed as a signal of acting in a responsible manner. This can improve the bond between the firm and its stakeholders. Both channels will mitigate agency problems. Then, CSR is used as a means of conflict resolution between stakeholders and the firm (see also Harjoto and Jo 2011). Thus, on the basis of the instrumental stakeholder approach, CSR can be regarded as a means to ‘neutralize’ agency problems. However, when additional (non-financial) motives enter the investment decision, there is a potential conflict of interest with CSR that may not be fully reflected in market prices (McWilliams and Siegel, 2001).

Bengtsson (2008a, b), Juravle and Lewis (2009) and Sandberg et al. (2009) argue that institutional characteristics can help explain the emergence of responsible investment. Scholtens and Sievänen (2012) find that economic openness, the size of the pension industry, and cultural values such as masculinity/femininity and uncertainty avoidance can be associated with the differences in responsible investment in the Nordic countries. Thus, the national context is likely to be linked to the way phenomena like responsible investment and its closely linked affiliate, corporate social responsibility, take place. Campbell (2007) argues that the relationship between basic economic conditions and corporate behaviour is mediated by several institutional conditions, such as public and private regulation. Gond et al. (2011) point out that the current CSR literature has not reserved for government politics and, as a consequence, for regulation and public policy in relation to CSR. They argue that it is the institutional embeddedness of market mechanisms within broader systems of governance that reflect social relations as well as the national legal and political governance systems (see also Haigh and Hazelton 2004). Gjølberg (2009a, b) finds that CSR has many forms depending on the social, economic, cultural, legal and political contexts. Argandoña and Von Weltzien (2009) suggest that the content of CSR has evolved over time, depending on historical, cultural, political, and socio-economic drivers and particular conditions in different countries and also at different points in time. Cheah et al. (2011) show a connection between responsible investment’s demographic characteristics and views related to CSR. Matten and Moon (2008) document a link between CSR and the national context. Furthermore, international portfolio investments by institutional investors seem to promote good corporate governance practices around the world (Aggarwal et al. 2009). Galbreath (2010) suggests that formal strategic planning is linked to CSR as well as culture. Furthermore, the motives and values of corporate leaders appear to be linked with the drivers of socially responsible decisions (Waldman and Siegel 2008). Harwood et al. (2011) find that relational and moral motives, as well as compliance to regulation, account for the emergence of CSR. Woods and Urwin (2010) suggest that the investment beliefs of pension funds affect their behaviour.

Boatright (1999) defines responsible investment as investing that takes account of ‘people’ and the ‘planet’. Responsible investment seems to provide investors with a framework to include moral considerations, where CSR is a framework to investigate how the investment targets actually operate in ESG arenas. Responsible investment can relate to both investment and credit practices (Scholtens 2006). Thus, it may relate to loans, bonds, stocks, commodities, and other financial instruments, including financial derivatives. Usually, it is done by negative screening (leaving out controversial firms and industries, like those involved tobacco, weapons, gambling), positive screening (concentrating on particular favourable firms and industries), best-in-class (focusing on the top-30 %/50 % of firms with respect to particular social or environmental performance criteria), activism and engagement (discussing with firm boards and directors), and combinations (Cox and Schneider 2010; Renneboog et al. 2008; Scholtens 2006; Hockerts and Moir 2004). In addition, integration, i.e. incorporating ESG-issues into the traditional financial analysis, has become a popular means of responsible investment (UN PRI 2012). By engaging in responsible investment, investors account for environmental, social, governance and ethical issues in the investment process (Renneboog et al. 2008). As such, investors try to affect the social responsibility of firms and the sustainability of countries, while at the same time trying to optimize their financial risk-return trade-off. A large number of studies has investigated responsible investment (for overviews, see Hoepner and McMillan 2009; Capelle-Blancard and Monjon 2012). This literature almost exclusively focuses on the financial consequences of accounting for responsibility and the drivers of responsible investment is not an issue at all.

Pension funds are institutions that are set up by a public or by private entity (OECD 2009a). Their aim is to secure the pensions of the funds’ participants. This is usually ensured by investing the contributions paid by the participants and their employers in a profitable and secure way (Eurostat 2009; OECD 2009a, b). Pension funds receive funds (premiums) from employees and employers mainly in the form of statutory pension payments and try to manage these assets profitably to ensure the pay-out of pensions during retirement. It is the fiduciary duty of pension funds to maximise the financial return of the invested contributions (Sethi 2005; OECD 2006; Woods and Urwin 2010; Hoepner et al. 2011). As such, pension funds have become large investors in several industrialized countries, and their assets represent on average 34 % of GDP in OECD countries (OECD 2009b). As the stock market (Levine 1998) and the credit market (Scholtens 2006) are means to promote economic and social development, the way the pension assets are invested can impact the financial markets and, thereby, the economy at large.

Given the fact that pension funds are large investors in several European and OECD countries (Eurosif 2010, OECD 2009b), it comes as no surprise that they dominate the responsible investment landscape in many countries too.1 For example, Eurosif (2010) mentions that, in most European countries, more than 90 % of the responsible investments are done by pension funds. Several public sector pension funds have supported actions to improve corporate governance and to promote social responsibility (Blackburn 2006), and have considered factors such as sustainability and its impact on companies’ long-term performance (Sethi 2005). Banerjee (2010) suggests developing more democratic forms of global governance of corporations and institutional investors. Aggarwal et al. (2009) suggest that institutional investors may want to improve companies’ corporate governance practices, or invest in companies which already have satisfactory corporate governance practices. As such, pension funds’ responsible investment is a means to advance corporate social responsibility (Solomon et al. 2004). Especially engagement is regarded as an efficient way to promote CSR (Clark and Hebb 2005). Heinkel et al. (2001) find that responsible investors would need to have at least 20 % of the stock to have a material impact on firms’ cost of capital. As such, corporate social responsibility and responsible investments are closely linked.

There appears to be a difference in the management style of conventional pension funds and those which consider responsible investment. Ferruz et al. (2010) show that conventional pension fund managers use superior information to follow stock picking strategies, whereas pension fund managers who consider responsible investment use superior information to follow market timing strategies. Furthermore, there can be differences in their beliefs of the importance of ethical issues, their investment decision-making style, and their perception of moral intensity (McLachlan and Gardner 2004). Eurosif (2010) and the United Nations Principles for Responsible Investment (UN PRI 2011a) suggest that funds also expect to be financially better off in the long-run by investing in a responsible manner. Woods and Urwin (2010) mention that investment beliefs, especially conjectures and assumptions about the investment world, are important in explaining pension fund behaviour as well.

The issue of whether responsible investment delivers superior or inferior financial return when compared to conventional investments has been studied at great length. Renneboog et al. (2008) suggest that the existing literature hints at investors not being ready for suboptimal financial return. Wen (2009) finds that financial expectations are among the central drivers of institutional investors’ responsible investment. Barnett and Salomon (2006) and Lee et al. (2010) examine the impact of screening intensity on the financial return in mutual funds that practice responsible investment. The former find a U-shaped relationship between risk-adjusted performance and screening intensity: the funds that screen little and those that screen most appear to deliver higher returns. Lee et al. (2010) find that the screening intensity does not have an impact on unsystematic risk, but it does seem to reduce the fund’s total risk. Bauer et al. (2005) and Galema et al. (2008) establish that there is no significant difference between returns of responsible investments compared to those investments that do not account for responsibility.

The literature suggests that there will be both external (environmental) and internal (firm-specific) drivers of responsible investment. For example, Sandberg et al. (2009) and Bengtsson (2008a, b) argue that especially country-specific regulations and institutional settings explain the emergence of responsible investment in a Scandinavian context. Juravle and Lewis (2009) find that government initiatives, such as the US pension act, boosted the development of responsible investment in the US (see also Gond et al. 2011). In the Netherlands, the favourable tax treatment of financing green projects appears to have advanced Dutch responsible investments (Scholtens 2005). Chiou et al. (2010) find wide variation in stock performance across legal origins. This corresponds to the results of La Porta et al. (1998, 2000) who find that the protection provided to shareholders and creditors varies from one legal origin to another, and that the legal origin is a crucial means to understand corporate governance and its reform. Cox and Schneider (2010) suggest that the legal and regulatory environment may affect pension funds regarding responsible investing, as well as the perceived legitimacy of corporate social performance among corporations. Kho et al. (2009) suggest that portfolio investors exhibit a large home bias against countries with poor governance. Especially weak accounting standards, weak shareholder rights and a poor legal framework appear to scare US investors (Cox and Schneider 2010). Aggarwal et al. (2009) suggest that international portfolio investment by institutional investors promotes good corporate governance practices around the world. Institutional investors may want to improve companies’ corporate governance practices, or invest in companies which already have satisfactory corporate governance practices (Aggarwal et al. 2009). Banerjee (2010) argues that governance has to become more democratic and requires new forms of multi-actor and multi-level governance arrangements in an attempt to create forms of power that are more compatible with the basic principles of economic democracy. Cumming and Johan (2007) find that responsible investment is more common among institutional investors with an international orientation. Johnson and Greening (1999) find that the proportion of shares owned by US pension funds in US companies is positively related to the corporate social performance dimensions that the pension funds try to promote.

Public pension funds appear to be the pioneers and the most activist in pension fund responsible investment (Cox and Schneider 2010). Their actions include divestment from tobacco companies (Wander and Malone 2006) and engagement (Alm 2007). In this respect, the Norwegian Government Pension Fund Global, which actually is a sovereign wealth fund, has become a role model for many other investors worldwide (Eurosif 2010). Sethi (2005) suggests that responsible investment is the best choice for public pension funds as they can play a critical role in improving the overall quality of corporate conduct in society. Furthermore, Sethi (2005) observes public pension funds expand their investment strategy by taking into account companies’ long-term risks such as environmental protection and sustainability. This may be due to public pension funds being more exposed to the public eye and less affected by conflicts of interest, e.g. between the sponsor company of the pension fund, and ethical viewpoints (Juravle and Lewis 2009). Other potential drivers of responsible investment may include pension plan funding type and pension plan status type.

Last but not least, the size of pension funds is related to pension funds’ responsible investments (Hawley and Williams 2000, 2007; Sethi 2005; Cox and Schneider 2010). Langley (2008), Monks (2001) and Useem (1996) study pension fund capitalism which means pension funds have the power to impact and direct the investments in the market. Hawley and Williams (2000, 2007) elaborate on the concept of the universal owners, i.e. large pension funds being entitled to impact corporate behaviour and in that way increase long-term absolute returns (see also Thamotheram and Wildsmith 2007).

Data and Methodology

We use a survey to assess the drivers of pension funds’ responsible investment. The survey data consist of 15 countries: Austria, Belgium, Denmark, Finland, France, Germany, Iceland, Italy, Luxemburg, Norway, Spain, Sweden, Switzerland, the Netherlands and the United Kingdom. The data collection took place in spring and summer 2010. The sampling targeted registered pension funds, the financial authorities of each country, and the directory of “Pension funds and their advisers 2008” (Wilmington 2008) as contact information sources. Due to significant differences in the number and size of pension funds as well as the availability of information and research resources, we were unable to include all registered pension funds in each country. From the following countries we targeted all registered pension funds, but due to missing or incorrect contact information the number of contacted pension funds may differ from that of the registers provided by the authorities: Austria (26), Belgium (161), Denmark (57), Finland (67), Iceland (30), Luxemburg (10), Norway (119) and Sweden (113). For Italy (93), France (37) and Spain (30), the sample consisted of the funds listed in the directory only. With respect to Germany (177), Switzerland (168) and the Netherlands (147), we applied the above-mentioned inclusion criteria in addition to a portfolio size based inclusion. We included a pension fund if its portfolio size was > € 10 million, > € 100 million, or > € 150 million in these countries, respectively. As such, we investigate above the median size pension funds. The differences in these limits relate to the number and size of pension funds. For the UK (97 funds), we drew the sample from the directory, and it included 100 largest pension funds in terms of portfolio size. This results in an unbalanced sample. The data targeted pension funds regardless of their type or whether they considered, or were familiar with, responsible investment. To ensure a diversity of pension funds in the sample, we clearly communicated this intention to the key financial decision makers of the pension funds.

The data collection took place in three phases. The first phase was preparation, which consisted of the preparation of a list of pension funds included. The sampling source was, first, the financial authorities of each country, and second the ′Pension funds and their advisers 2008′ book. We formed preliminary country-specific lists first. We completed them by making internet searches and by calling the pension funds, if needed, in order to obtain the contact details of the key financial decision maker. The availability and ′reachability′ of this information varied widely, and in some cases the survey targeted the ′person responsible for the investments′. For some countries, we excluded funds based on their (small) portfolio size, which we related to the total number of pension funds in that country. We excluded pension funds for which we could not obtain contact information (address, e-mail address and phone number). The second phase was data collection, which consisted of contacting the (named) key financial decision-maker(s) by mail in spring and summer 2010. In total, we sent 1,332 letters. The letter explained the study and encouraged participation regardless of whether the pension fund was familiar with responsible investment. In approximately two weeks’ time after sending the letter, we sent an e-mail to the (named) key financial decision-makers with a link to the online questionnaire. The e-mail encouraged recipients to participate and to forward the link to colleagues who work with investments. The third phase consisted of three reminder rounds in total, sent by e-mail, in spring and summer 2010. Altogether 281 respondents participated in our survey. We classified pension funds on the basis of whether they claimed to engage with responsible investment. As with all surveys, there can be a self-reporting bias in the funds’ answers. This suggests that the results have to be dealt with great care.

Our dependent variable is whether a fund engages with responsible investment. We measured this variable on a Likert scale, where 1 = totally disagree, 7 = totally agree and 8 = I do not know. We excluded the last response class, 8, (0.4 % of respondents) from the analysis, as well as from the characteristics we relate to pension funds’ responsible investment. Table 1 shows the distribution of the funds along all the response classes. To ensure a sound analysis, we classified the respondents into three classes: pension funds which have no responsible investment [classes 1–3; 37.0 % of respondents; ′conventional pension funds (PF)′], neutral ones [class 4; 10.9 %; “neutral pension funds (PF)”], and pension funds with responsible investment [classes 5–7; 51.8 %; ′pension fund (PF) with responsible investment (RI)′]. Based on this classification, we used a multinomial logistic regression (see Lemeshow and Hosmer 1984, Hosmer and Lemeshow 1989). Examples of studies in finance that use multinomial logistic regression include e.g. Chatterjee et al. (1996), Chi (2009) and Ongena and İlkay (2011).
Table 1

Distribution of the observations of the dependent variable (′Our pension fund engages with responsible investment′) sample along the response classes

Response class

Number of respondents

% of respondents

1 = Totally disagree

51

18.5

2

25

9.1

3

26

9.4

4

30

10.9

5

33

12.0

6

22

8.0

7 = Totally agree

88

31.9

8 = I do not know

1

0.4

Total

276

100.0

To find out about the determinants of responsible investment, we investigate external and internal characteristics. More specifically, we look into the following eight variables we took from the economics and finance literature and report the descriptives in Table 2:
Table 2

Distribution of the sample along the categories (% of total is in brackets)

1.

Legal origin

No. of pension funds

English

22 (8)

German

60 (22)

French

82 (30)

Scandinavian

106 (38)

Country not indicated

5 (2)

 

Total

275

2.

Ownership of the pension fund

No. of pension funds

Public

63 (23)

Corporate

166 (61)

Other

42 (15)

Corporate and other

4 (1)

  

Total

275

3.

Pension plan funding type

No. of pension funds

Defined benefit

144 (53)

Defined contribution

74 (28)

Hybrid

41 (15)

Other

10 (4)

  

Total

269

4.

Pension plan status type

No. of pension funds

Statutory

80 (30)

Complementary

151 (56)

Both

29 (11)

Other

7 (3)

  

Total

267

5.

Number of employees the pension scheme covers

No. of pension funds

0–199

23 (9)

200–999

34 (14)

1,000–4,999

60 (26)

5,000–24,999

49 (20)

250,00–99,999

42 (17)

>100,000

35 (14)

Total

243

6.

Number of staff the pension fund employs

No. of pension funds

0–3

77 (30)

4–9

56 (21)

10–49

65 (250)

50–199

25 (10)

>200

38 (14)

 

Total

261

7.

Number of persons currently receiving a pension

No. of pension funds

0–99

26 (11)

100–999

49 (21)

1,000–4,999

58 (24)

5,000–14,999

34 (14)

15,000–49,999

37 (16)

>5,0000

34 (14)

Total

238

8.

Portfolio financial size (as of 31.12.2009), millions €a

No. of pension funds

€ 0–99

50 (18)

€ 100–499

62 (23)

€ 500–999

30 (11)

€ 1,000–4,999

70 (26)

€ 5,000–9,999

26 (9)

€ 10,000–49,999

29 (11)

Total

271

a In the classes of 50 billion to 100 billion and above 100 billion there are 2 funds (1 %) each

  1. 1.

    Legal origin. We constructed this variable ′Please specify your country′, which consisted of 15 alternatives, corresponding to the participating countries. The variable legal origin of the country consisted of four classes, based on La Porta et al’s (1998) classification, namely: German-origin (Austria, Germany, Luxemburg and Switzerland), French-origin (Belgium, France, Italy, Spain and the Netherlands), Scandinavian-origin (Denmark, Finland, Iceland, Norway and Sweden), English-origin (or common law origin; the United Kingdom) and the fifth class was ′Country not specified′. Luxemburg and Iceland were not included in La Porta et al.’s (1998) classification. For Luxemburg, we took the example of Chiou et al. (2010) and included it in the German-origin. Based on established geographical terminology, ′the Nordic countries′, we assign Scandinavian-origin to Iceland. Table 2 shows that most funds in the sample (38 %) have Scandinavian-origin, French-origin ranks second (30 %) and German third (22 %).

     
  2. 2.

    Ownership of the pension fund. The respondents had to choose the closest alternative on the following scale: Public pension fund, corporate pension fund or other pension fund (see UN PRI 2011b). As the respondents could choose multiple responses, we further formed a class of ′corporate and other pension fund′. Corporate funds make up the majority of the sample (61 %) and public funds constitute 23 % of the total.

     
  3. 3.

    Pension plan funding type. The respondents had to choose the closest alternative on the following scale [see UN PRI (2011b) for the scale, OECD 2009a for definitions]: Defined benefit (the level of pension benefits promised to participating employees is guaranteed), defined contribution (the plan sponsor pays fixed contributions and has no legal or constructive obligation to pay further contributions to an ongoing plan in the event of unfavourable plan experience), hybrid (a combination of the two previous) or other. Most funds (53 %) in the sample are defined benefit, and 28 % is defined contribution.

     
  4. 4.

    Pension plan status type. The respondents had to choose the closest alternative on the following scale [see UN PRI (2011b) for the scale, OECD 2009a for definitions]: Statutory (mandatory social security), complementary (voluntary supplementary pensions), both (a combination of the two) and other. Table 2 shows that most funds (56 %) have a complementary status (56 %) and 30 % is of the statutory type.

     
  5. 5.

    Number of employees the pension plans covers [number of participants/members—see UN PRI (2011b)]. The respondents inserted their answer as free text. Again, we formed the following classes based on the distribution of the responses: 0–199, 200–999, 1000–4999, 5000–24999, 25000–99 999 and >100000. Table 2 shows that 26 % of the funds cover between one and five thousand employees; one-fifth of the funds cover between 5 and 25 thousand employees.

     
  6. 6.

    Number of staff with the pension funds (UN PRI 2011b). The respondents inserted their answer as free text. For our analysis, we formed the following classes, which are motivated by the distribution of the responses: 0–3, 4–9, 10–49, 50–199 and >200 persons. Table 2 shows that 30 % of the funds has at most a staff of three persons, and 14 % have more than 200 staff.

     
  7. 7.

    Number of persons currently receiving a pension (see UN PRI 2011b). The respondents inserted their answer as free text. Again, we formed the following classes based on the distribution of the responses: 0–99, 100–999, 1000–4999, 5000–14999, 15000–49999 and >50000. Table 2 shows that 24 % of the funds in our sample grant a pension to one to five thousand persons; the other classes are of about the same size.

     
  8. 8.

    Portfolio size (as of 31.12.2009) (see UN PRI 2011b). The respondents inserted their answer as free text in boxes and they chose the currency in question. We converted the currencies into euros using the currency rates of the Finnish National Bank, as per 31st December 2009. We classified the responses based on the distribution: € 0–99 million, € 100–499 million, € 500–999 million, € 1–<5 billion, € 5–<10 billion, € 10–<50 billion, € 50–<100 billion and >€ 100 billion. Table 2 shows that most funds (26 %) have a fund size that ranges between € 1 and € 5 billion. The size class of € 100–500 million ranks second.

     

We characterise the relationship of these eight variables with respect to the existence of responsible investment with a pension fund by calculating the proportions of the existence within each category of the eight variables. The existing literature shows the connection between the legal origin and protection provided by shareholders and creditors (La Porta et al. 1998), variation in stock performance (Chiou et al. 2010), and legal and regulatory environment (e.g. Campbell 2007; Gond et al. 2011). Authors such as Sethi (2005) and Blackburn (2006) find a connection between the ownership of the pension fund and their responsible investment. Pension plan funding type and pension plan status type fit well to a pension fund investigation as the former is a central topic in many pension debates (OECD 2009b), and the latter may relate to obligations to consider ethical issues. Lastly, the literature also relates the size of the pension funds to responsible investing (e.g. Hawley and Williams 2000, 2007; Sethi 2005; Cox and Schneider 2010).

We compare the proportions of ′conventional PF′, ′neutral PF′ and ′PF with RI′ by the odds ratio (abbreviated to OR). The odds ratio is a measure of the discrepancy between two proportions, and it carries information on both the direction and the size of the effect, and it is a common measure in (multinomial) logistic regression (Hosmer and Lemeshow 1989; Lemeshow and Hosmer 1984; Collet 2003; Rita and Komonen 2008; see Champagne and Kryzanowski 2008; Gaganis et al. 2010 and Laitinen 2011, for examples of the application of logistic regression in business and economics). An odds ratio above 1 suggests that the probability is larger in a specific class compared to that of the reference class; values below 1 indicate the opposite. From below, odds ratios are limited by 0, whereas from above, there is no limit (Rita and Komonen 2008). We use ′neutral PF′ as the reference group to calculate the odds ratios. We calculate the odds ratios by first transforming the two proportions under comparison into odds. The odds corresponding to a proportion p is defined as p/(1 − p), meaning the ratio of the probability that an incident takes place to the probability that it does not (Rita and Komonen 2008). Then, for each of the eight variables, we compare the odds’ of the proportions of ′conventional PF′ and ′PF with RI′, to the odds of ′neutral PF′ by calculating their ratio, i.e. the odds ratio. For the reference category (′neutral PF′), the odds ratio is always 1, and this makes it easy to interpret the results of the two other groups versus the reference group. This methodology enables us to compare the divergence of conventional pension funds and pension funds with responsible investment in relation to the neutral category. We use PASW Statistics 18.0 for the analysis.

Results

This section reports the results of the analysis. Table 3 highlights the first four variables, namely legal origin, ownership of the pension fund, pension plan funding type and pension plan status type. Table 4 reports about the pension fund size-related characteristics.
Table 3

The effects of the legal origin, the ownership of the pension fund, the pension plan funding type and the pension plan status type on the probability of having responsible investment (RI)

Variable

% out of respondents per response class (N)

OR

Conventional PF

Neutral PF

PF with RI

Total frequency

Conventional PF

Neutral PF

PF with RI

Panel A

Legal origin (p = 0.022, df = 8)

 German-origin

52 % (31)

15 % (9)

33 % (20)

60

6.1

1

2.8

 English-origin

27 % (6)

5 % (1)

68 % (15)

22

7.9

1

45.0

 Scandinavian-origin

31 % (33)

8 % (8)

61 % (65)

106

5.5

1

19.4

 French-origin

35 % (29)

13 % (11)

51 % (42)

82

3.4

1

6.8

 Country not indicated

60 % (3)

20 % (1)

20 % (1)

5

6.0

1

1.0

 Total

37 % (102)

11 % (30)

52 % (143)

275

4.8

1

8.8

Panel B

Ownership of pension fund (p = 0.006, df = 6)

 Public PF

29 % (18)

6 % (4)

65 % (41)

63

5.9

1

27.5

 Corp. PF

44 % (73)

13 % (22)

43 % (71)

166

5.1

1

4.9

 Other type of PF

26 % (11)

10 % (4)

64 % (27)

42

3.4

1

17.1

 Corp. and Other PF

0 % (0)

0 % (0)

100 % (4)

4

1

(∞)

 Total

37 % (102)

11 % (30)

52 % (143)

275

4.8

1

8.8

Panel C

Pension plan funding type (p = 0.132, df = 6)

 DB

39 % (56)

9 % (13)

52 % (75)

144

6.4

1

11.0

 DC

36 % (27)

14 % (10)

50 % (37)

74

3.7

1

6.0

 Hybrid

37 % (15)

17 % (7)

46 % (19)

41

2.8

1

4.2

 Other

10 % (1)

0 % (0)

90 % (9)

10

(∞)

1

(∞)

 Total

37 % (99)

11 % (30)

52 % (140)

269

4.6

1

8.6

Panel D

Pension plan status type (p = 0.014, df = 6)

 Statutory

40 % (32)

6 % (5)

54 % (43)

80

10.0

1

17.4

 Complementary

40 % (61)

11 % (17)

48 % (73)

151

5.3

1

7.4

 Both

31 % (9)

24 % (7)

45 % (13)

29

1.4

1

2.6

 Other

0 % (0)

0 % (0)

100 % (7)

7

1

(∞)

 Total

38 % (102)

11 % (29)

51 % (136)

267

5.1

1

8.5

Represented as frequencies, and as odds ratios (OR)

Table 4

The effects of pension fund (PF) size as number of employees covered by the plan, the number of staff working in the pension fund, the number of current pensioners, and the size of portfolio on the probability of having responsible investment (RI)

Variable

% out of respondents per response class (N)

OR

Conventional PF

Neutral PF

PF with RI

Total frequency

Conventional PF

Neutral PF

PF with RI

Panel A

No. of empl. covered by plan (p = 0.008, df = 2)

 0–199

52 % (12)

4 % (1)

43 % (10)

23

24.0

1

16.9

 200–999

44 % (15)

12 % (4)

44 % (15)

34

5.9

1

5.9

 1,000–4,999

48 % (29)

12 % (7)

40 % (24)

60

7.1

1

5.0

 5,000–24,999

39 % (19)

16 % (8)

45 % (22)

49

3.2

1

4.2

 25,000–99,999

29 % (12)

10 % (4)

62 % (26)

42

3.8

1

15.4

 100,000>

23 % (8)

6 % (2)

71 % (25)

35

4.9

1

41.3

 Total

39 % (95)

11 % (26)

50 % (122)

243

5.4

1

8.4

Panel B

No. of staff working in pension fund (p = 0.000, df = 2)

 0–3

52 % (40)

9 % (7)

39 % (30)

77

10.8

1

6.4

 4–9

41 % (23)

20 % (11)

39 % (22)

56

2.9

1

2.7

 10–49

32 % (21)

11 % (7)

57 % (37)

65

4.0

1

11.0

 50–199

28 % (7)

8 % (2)

64 % (16)

25

4.5

1

20.4

 200>

16 % (6)

5 % (2)

79 % (30)

38

3.4

1

67.5

 Total

37 % (97)

11 % (29)

52 % (135)

261

4.7

1

8.6

Panel C

No. of persons receiving a pension (p = 0.141, df = 2)

 0–99

35 % (9)

12 % (3)

54 % (14)

26

4.1

1

8.9

 100–999

43 % (21)

14 % (7)

43 % (21)

49

4.5

1

4.5

 1,000–4,999

45 % (26)

12 % (7)

43 % (25)

58

5.9

1

5.5

 5,000–14,999

41 % (14)

15 % (5)

44 % (15)

34

4.1

1

4.6

 15,000–49,999

35 % (13)

3 % (1)

62 % (23)

37

19.5

1

59.1

 50,000>

26 % (9)

9 % (3)

65 % (22)

34

3.7

1

18.9

 Total

39 % (92)

11 % (26)

50 % (120)

238

5.1

1

8.3

Panel D

Portfolio size (€ million) (p = 0.000, df = 2)

 € 0–99 million

40 % (20)

8 % (4)

52 % (26)

50

7.7

1

12.5

 € 100–499 million

55 % (34)

15 % (9)

31 % (19)

62

7.2

1

2.6

 € 500–999 million

50 % (15)

27 % (8)

23 % (7)

30

2.8

1

0.8

 € 1,000–4,999 million

29 % (20)

9 % (6)

63 % (44)

70

4.3

1

18.1

 € 5,000–9,999 million

19 % (5)

8 % (2)

73 % (19)

26

2.9

1

32.6

 € 10,000–49,999 million

21 % (6)

0 % (0)

79 % (23)

29

(∞)

1

(∞)

 € 50,000–99,999 million

0 % (0)

0 % (0)

100 % (2)

2

1

(∞)

 € 100,000 million>

0 % (0)

0 % (0)

100 % (2)

2

1

(∞)

 Total

37 % (100)

11 % (29)

52 % (142)

271

4.9

1

9.2

Represented as frequencies, and as odds ratios (OR)

Table 3 consists of four panels, one per variable, and of two main columns. The rows in panel A of Table 3 list the percentages and frequencies for the response classes of legal origin, and the first column, ′% out of respondents per response class (N)′, shows the distribution of the observations by the three pension fund classes (conventional PF, neutral PF and PF with RI) as well as their sum. The percentages add up to 100 % per row. The second column, ′OR′, reports the odds ratios for each pension fund category (conventional PF, neutral PF, PF with RI). The neutral pension fund group is the reference group and therefore its value is always 1. We report statistical significance at the variable level, but not at the response class level as the data sample usually is too small for reliable calculations for several variables. The odds ratios give the size of the effects, whereas the p value reveals the significance of the variable in discriminating between categories. The odds ratios show how the variables relate to the funds’ responsible investment. They characterise the effect of a variable irrespective of the magnitude of the probabilities. This shows that the odds ratio cannot be reduced to probabilities alone; it can be used to compare probabilities, not to indicate their individual values.

As for legal origin (Panel A), the results indicate that having responsible investment is the dominant form (68 %) in the group of English-origin. For French- and Scandinavian-origin, funds, responsible investment is the dominant form too (51 and 61 %, respectively). For the German-origin, as well as for the respondents who have not indicated their country of origin, the result of not having responsible investment is more likely (52 and 60 % respectively). The 52 % for funds with German-origin are almost double that of those with English-origin (27 %). We also observe that for ′conventional PF′, the odds ratio is highest in the English-origin group (7.9), followed by the German-origin (6.1), whereas for the French-origin the odds ratio is lowest (3.5).2 For the response group ′PF with RI′, the English-origin has the highest odds ratio (45), followed by the Scandinavian-origin (19). If we do not account for the respondents that have not indicated their country, the odds ratio is lowest (2.8) in the German-origin group. In other words, the odds ratio of the German-origin group is 183 % larger when compared to the odds of the respective neutral group, whereas for the English-origin group it is 4400 % larger.3 This suggests that the observations in the German-origin group are in the conventional pension fund group, whereas in the English-origin, pension funds with responsible investment dominate.

At the variable level, legal origin significantly explains whether a pension fund does have responsible investment or not (p = 0.022). This p value indicates that the distributions of the observations in the different response classes significantly differ from each other. We find similar results in an additional analysis (not reported but available upon request) with six indicators of La Porta et al. (1998): efficiency of judicial system, rule of law, corruption, risk of expropriation, risk of contract repudiation and rating on accounting standards. These results provide additional support for the significant role of legal origin as a driver of pension funds’ responsible investment. The inclusion of these indicators into the analysis made the odds ratios much larger, and in the multivariate model the legal origin became more significant (p = 0.007) when explaining pension fund responsible investment. These results are in line with the findings of several authors who suggest that the legal origin matters in various contexts (see for example Chiou et al. (2010), who connect legal origin to stock risks and performance).

La Porta et al. (1998, 2000) argue that the protection provided to shareholders and creditors varies from one legal origin to another, and that corporate governance and its reform can be understood well by legal origin (Bengtsson 2008a, b, Juravle and Lewis 2009, Sandberg et al. 2009). Gjølberg (2009a, b) suggests that the social, economic, cultural, legal and political contexts impact corporate social responsibility. In addition, Matten and Moon (2008) find a correspondence between corporate social responsibility and the national context as well. Gond et al. (2011) relate CSR to government and politics (see also Haigh and Hazelton 2004). Furthermore, our findings are congruent with those of Cox and Schneider (2010) who suggest that legal and regulatory environments may affect pension funds regarding responsible investing, and legal and regulatory environments can affect the perceived legitimacy of corporate social performance among corporations (see also Gond et al. 2011). We arrive at additional evidence that legal origin drives pension funds’ responsible investment too. This is in line with the findings of Aggarwal et al. (2009) and Kho et al. (2009) about selective institutional investor investment across nations and along the level of corporate governance.

As for pension fund ownership (Panel B in Table 3), our results show that 65 % of the public pension fund respondents indicate their pension fund engages with responsible investment. This result is closely followed by ‘other’ type of pension fund, 64 %. It is mostly the Scandinavian pension funds which do belong to this latter category. They have characteristics of both public and corporate pension funds (OECD 2009a, b). As for corporate pension funds, the percentages are almost the same for ‘conventional PF’ and for ‘PF with RI’. Corporate pension funds stand out with the highest percentage of conventional funds when compared to the other pension funds. We do not find that corporate pension funds closely follow the ‘pioneer’ group of public pension funds: the odds ratios of pension fund type appear to be much higher in two of the classes in the group of ‘PF with RI’, than in the respective classes of ‘conventional PF’: the odds ratio is the largest with public pension funds (27.5), followed by other type of pension fund (17.1). We observe the highest odds ratio for public pension funds (5.9), and the lowest for other type of pension fund (3.4). In all, the ownership of the pension fund significantly describes whether the pension fund engages with responsible investment (p = 0.006). These results are in line with the existing literature (e.g. Sethi 2005; Blackburn 2006; Wander and Malone 2006; Alm 2007; Juravle and Lewis 2009; Cox and Schneider 2010).

Funding type of the pension funds is addressed in panel C of Table 3. For pension funds with a defined-benefit (DB) scheme, the odds ratio of ‘PF with RI’ group is as high as 11.0. For the ‘conventional PF’, the odds ratio is 6.4. Thus, the percentage of pension plans with responsible investment is almost twice that of conventional funds. It seems that both DB and defined-contribution (DC) funding schemes differ more from neutral than the hybrid pension types do. However, the p value (0.132) indicates that the distributions between the response classes do not significantly differ from each other. Thus, we conclude that the funding type of the fund is not a discriminating factor with respect to responsible investment.

As for pension plan status type (panel D in Table 3), the results suggest that this variable significantly explains whether the pension fund has responsible investment or not (p = 0.014). Statutory pension plans relate more to responsible ways of investing than not doing so (17.4 vs. 10.0). Thus, the discrepancy between responsible investing and remaining neutral is higher than that of conventional investing and neutral with respect to responsible investing. For the other classes of pension plan status type, we observe clearly lower discrepancies. The lowest one is in the class of ‘both’, in which the odds for conventional pension funds is only slightly larger when compared to the corresponding neutral class, but substantially larger than the corresponding neutral class for pension funds which have responsible investment.

The results of the pension fund size variables are in Table 4. It appears that responsible investing is much more common with large funds than with smaller ones. This happens to be the case with all the size-related variables. This finding is in line with most of the literature (see Monks 2001, Hawley and Williams 2000, 2007). Panel A in Table 4 relates to the number of employees covered by the pension plan. The two largest plan sizes with respect to the number of employees covered, 62 and 71 % of the respondents report they have responsible investment. In the smallest plan size, only 43 % of the respondents list that they have responsible investment. In the group of ‘PF with RI’, the odds ratio is largest (41) in the plan size of >100,000, whereas for ‘conventional PF’, the largest odds ratio (24) is in the smallest plan size. However, there is no linearity in the odds ratios’ along the plan sizes, but rather a U-shaped relationship. In the smallest and largest plan sizes, the odds ratios diverge the most from the reference group, for pension funds that have responsible investment. We are not familiar with studies that document a curvilinear relationship between the size of pension funds and their responsible investment. Udayasankar (2008) finds a U-shaped relation between the firm size and CSR. We conclude that the number of employees covered by the scheme is a significant factor in explaining responsible investment.

With the number of staff in the pension fund (panel B), the results also suggest that this variable plays a significant role in discriminating between conventional funds and funds with responsible investment (p = 0.000). The smallest pension funds are the least likely to have responsible investment (39 %), whereas with larger funds the percentage can be as high as 79 %. This is supported by the odds ratio: in the group of ‘PF with RI’, the odds ratio is highest in the largest staff size group. For ‘conventional PF’, the odds ratio is the highest in the smallest group. Thus the odds of the group ‘PF with RI’, for the pension funds which employ 200 persons or more, is much larger than that of the corresponding neutral group. For smaller pension funds, the differences are much less pronounced. Again, these results indicate responsible investment is much more common in large pension funds than in small ones.

The number of current pensioners (panel C) does not significantly discriminate the different groups (p = 0.141). However, here too we observe that responsible investment becomes more common when the size of the pension funds increases.

As for portfolio size (panel D in Table 4), we find a curvilinear trend in the ‘PF with RI’ group. In this group, it is the category of € >5–10 billion which most clearly relates to responsible investment (33): the odds ratio diverges most from ‘neutral PF’ group. At the same time, in ‘conventional PF’ group, the odds ratio is among the two lowest ones for the same portfolio size group (2.9). As such, it seems that the odds of ‘PF with RI’, in the group of portfolio size between € 5,000–9,999 million, is very much larger than that of the respective neutral group (and for ‘conventional PF’ only 186 %). If we compare the odds of ‘conventional PF’ to the respective neutral class in an ascending order, starting from the smallest portfolio size, the differences become smaller but are still substantial. This suggests that the odds’ of not having responsible investment, when compared to the neutral group, are larger among small portfolio sizes than among large portfolio sizes. This suggests that the divergence from the respective neutral group is large in the smallest portfolio size group. Then, it then diminishes strongly in the two following ones, and becomes again larger in the larger portfolio sizes. This too is congruent with Udayasankar’s (2008) results for corporates. On the variable level, portfolio size significantly explains whether the pension fund has responsible investment or not (p = 0.000).

Thus, it appears that for the number of employees covered, funds’ portfolio size as well as their staff numbers, the size variable significantly explains whether a pension fund is likely to engage with responsible investment. Only with the number of people receiving a defined benefit from the fund, we do not establish a statistically significant relationship. However, here too the same pattern appears to hold, namely that larger funds are more likely to engage with responsible investing. Useem (1996), Monks (2001), Hawley and Williams (2007) and Langley (2008) argue that due to their large size, pension funds are in the position to impact on corporate behaviour. In addition, we can relate our findings to a curvilinear trend (Lee et al. 2010; Udayasankar 2008; Barnett and Salomon 2006), as the odds ratios diverge most in small and large pension funds, whereas the divergence is at its smallest in the middle classes. This U-shaped relationship is especially visible in the group of ‘PF with RI’.

Conclusion

The literature suggests that institutional and economic settings impact on corporate social responsibility and responsible investment (e.g. Bengtsson 2008a, b; Gjølberg 2009a, b; Cox and Schneider 2010; Gond et al. 2011; Scholtens 2012). However, these studies do not investigate pension fund characteristics in relation to responsible investment. This is the challenge in our study. To this extent, we investigate more than 250 pension funds in 15 European countries. We analyse eight different variables and can link them to pension funds’ responsible investment: legal origin, ownership of the pension fund, pension plan funding type, pension plan status type, number of staff in the pension fund, number of employees covered by the plan, number of current pensioners and funds’ financial assets size. We find that legal origin significantly relates to pension funds’ responsible investment. Especially with Scandinavian- and English-origin, the pension funds are likely to engage with responsible investment. As for the ownership of the pension fund, public pension funds are the most likely ones to engage in responsible investment. This also is the case with defined-benefit and statutory pension plans. Furthermore, it predominantly is larger funds (measured as assets, staff, participants) which are most likely to have responsible investment than smaller ones. The latter observation is congruent with the existing literature (Langley 2008; Hawley and Williams 2007; Monks 2001; Useem 1996). We show that this is the case for different proxies for fund size. In addition, we establish that there is a curvilinear relationship between fund size and responsible investment. More specifically, we find that it in particular is the smallest and the largest pension funds in the sample that engage with responsible investment. Median sized funds engage the least. Apparently, it is all these characteristics that offer most scope for accounting for ESG- issues in pension fund management. Small size funds have a close relationship with their stakeholders and the very large ones attract most attention and, thus, public pressure. Public pension funds and defined-benefit schemes offer more scope for stakeholder commitment regarding non-financial aspects of the pension. The legal system sets the stage for the interaction between financial institutions and shareholders and this seems most favourable in the Scandinavian and English and least so in the German system.

This exploratory analysis is the first to relate a large number of pension fund characteristics to responsible investment. Although its restrictions include the fact that what responsible investment means for a pension fund is based on self-reporting, the choice of the key characteristics, as well as the somewhat unbalanced sampling in different countries, we find that it considerably extends the understanding of the drivers of pension funds’ responsible investment in several respects. First, our findings complement the literature about the drivers of responsible investment at the pension fund level, and contribute to the existing findings on a macro level (e.g. La Porta et al. 1998; Hawley and Williams 2000, 2007; Sethi 2005; Cox and Schneider 2010). Second, we both qualify and quantify the connection between the legal origin, the ownership of the pension fund, the size-related variables and responsible investment. Third, our multinomial logistic regression approach and the odds ratios enable a clear comparison of the impact of each the eight different pension fund characteristics. Our findings help to lay the groundwork for further investigation of responsible investment with pension funds. In future research, we want to investigate the dynamics of pension funds and responsible investment. Our research also will be directed at further investigating the transmission mechanisms between the characteristics detected in this article and the pension funds’ responsible investment. Furthermore, it seems relevant to find out whether the characteristics also hold outside the group of countries studied and whether alternative dimensions might play a role there.

Footnotes
1

Of the 15 countries we investigate, pension fund assets to GDP are above 50 % in Denmark, Finland, Iceland, the Netherlands, Norway, Sweden, Switzerland, and the United Kingdom. In the other countries investigated, they are less than 10 % (OECD 2009b).

 
2

Although the odds ratio is a common way to report results in (multinomial) logistic regression, the concept might be unfamiliar for some readers. Odds ratio is the ratio of two odds. The odds are retrieved by transforming the proportions under comparison (for example the English-origin and its respective neutral class) into odds (Rita and Komonen 2008). Thus, the odds of the English-origin group to be conventional pension funds are 7.9 times larger than for the respective neutral group, whereas for the French-origin it is 3.5 times larger.

 
3

This percentage is calculated as follows for e.g. the German-origin: the odds for the proportion of ′PF with RI′ is (20/60)/[1 − (20/60)] = 0.50000; the odds for the respective ′neutral′ proportion is (9/60)/[1 − (9/60)] = 0.17647. Odds ratio is obtained as the ratio 0.50000/0.17647 = 2.833; 0.50000 is 100 × (2.833 − 1) = 183 % larger than 0.17647.

 

Acknowledgments

We want to thank the editor and the anonymous referees for their useful comments. Riikka Sievänen acknowledges the financial support from the Foundation of Economic Education and Jenny and Antti Wihuri Foundation. We are grateful for the comments and suggestions of the participants of the UN’s PRI and MISTRA Conference in Sigtuna, Sweden, September 26–28, 2011. The usual disclaimer applies.

Copyright information

© Springer Science+Business Media Dordrecht 2012