The relation between board size and funding ratio of Dutch pension funds
January, 2015
Elwin van Rooijen (10002294) Supervisor: dr. I. J. Naaborg
BSc. Economics & Business (Finance & Organization) Research Topic: Financial Governance
______________________________________________________________________________ This paper provides empirical evidence that there is a relation between board size and
funding ratio based on a unique panel dataset of the largest 50 Dutch pension funds between 2009 and 2013. This result is partly consistent with Yermack (1996), but is different in the sense that the relation between board size and funding ratio of Dutch pension funds is
nonlinear. The empirical results suggest that neither a very small nor a large board is optimal, regarding the funding ratio of Dutch pension funds. Pension funds with a board size of seven directors have on average the highest funding ratio.
Table of Contents:
1. Introduction ... 3
2. Literature Review ... 4
2.1 Theory ... 4
2.2 Empirical Results: Firms... 5
2.3 Empirical Results: Mutual Funds ... 6
2.4 Empirical Results: Banks ... 6
2.5 Empirical Results: Non-Profit Organizations ... 7
2.6 Pension Funds ... 8
3. Hypothesis, Methodology and Data... 9
3.1 Hypotheses and Methodology... 9
3.1.1 Hypotheses ... 9 3.1.2 Methodology ... 10 3.2 Data ... 13 4. Empirical Analysis ... 16 4.1 Empirical Results ... 16 4.2 Robustness Check ... 19
5. Discussion & Conclusion ... 22
References ... 25
Appendix ... 30
This document is written by Student Elwin van Rooijen who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents
1. Introduction
The Dutch pension system is ranked as one of the best pension systems in the world, based on indicators for adequacy, sustainability and integrity (Mercer, 2014). Nevertheless, the recent financial crisis had a negative effect on the financial position of Dutch pension funds. A substantial number of Dutch pension funds were forced to increase premiums, freeze pensions for inflation or even cut benefits (De Nederlandsche Bank, 2015). In many cases, the employee foots the bill when pension funds fail to achieve a sufficient funding ratio. While institutional investors are interested in corporate governance of the firms they invest in (Faccio & Lasfer, 2000), there is a growing public interest in improving governance of Dutch pension funds due to underfunded pension funds.
As a reaction to the public debate about pension funds governance, a parliamentary committee, called Commissie Frijns, was appointed to design a bill to improve pension funds governance. The law, called empowerment of pension fund governance recently passed (Commissie Frijns, 2013). However, this law does not give any suggestions about optimal board size for Dutch pension funds, although this could be an important factor in optimizing pension fund governance. Board size has been studied extensively in economic literature. Research on the impact of board size on different organizations show different results. However, the effect of board size on pension funds has not been explored yet. The similarities and differences with the previously researched organizations and the social impact of potential pension cuts in situations of underfunding makes the research question scientifically as well as socially relevant.
This study presents the results of empirical research on the effect of board size on funding ratio of Dutch pension funds, using a unique database of 50 pension funds between 2009 and 2013. It intends to answer the following question:
What is the relation between board size and funding ratio of Dutch pension funds?
This paper is structured as follows: the second section will provide a review of literature on board size and will link the literature to pension funds. The third section will explain the methodology that is used to answer the research question and will describe the dataset. The fourth section will show empirical results and a robustness check. In the fifth section, a
discussion will be conducted about the interpretation of the empirical results, recommendations for further research will be made and the paper will come to a conclusion.
2. Literature review
The purpose of this section is to give a review of the theory and empirical results regarding the effect of board size on the value and performance of firms, mutual funds and non-profit organizations. Moreover, a link between literature on board size and the type of organization probed in this study, pension funds, will be made.
2.1 Theory
Although this study has an economic perspective, the fundamentals of the theory about the influence of board size on value and performance of firms, mutual funds and non-profit organization can be found in social sciences. For instance, psychologists Latane, Williams & Harkins (1979) observes that in simple tasks such as clapping, individual effort reduces with increasing group size, but that total output diminishingly increases. Latane et al. suggest that this loss of efforts is caused by ‘free riding’ of group members and call this phenomenon social loafing.
Later on, these findings were applied in economic literature. Lipton and Lorsch (1992) describe the lack of a critical attitude towards top managers in U.S. boardrooms. Lipton and Lorsch believe that this problem increases when boards become larger. Directors are more familiar with each other when working in smaller boards. A smaller board creates more room for discussion and makes it easier to reach general agreement. The suggestion of Lipton and Lorsch is to limit board size to ten people. A larger board size leads to situations in which directors have a lack of time to determine the strategy of the firm. Furthermore, Lipton and Lorsch mention decreasing cohesiveness with increasing board size. A board size of eight or nine directors is assumed to be optimal, following the theory of Lipton and Lorsch. Jensen (1993) stresses the problem of oversized boards as well, and proposes an optimal board size of seven or eight directors. According to Jensen, boards that are beyond this size will lose effectiveness, because monitoring the CEO and other board members becomes harder. To conclude, in their comprehensive overview of economic literature on board size, Hermalin and Weibach (2001) describe a consensus among economists that there is a negative relation between board size and corporate value and performance.
However, there are exceptions on this consensus. According to Dalton, Daily, Johnson and Ellestrand (1999) a disadvantage of a small board might be that they lack experience and knowledge. Large boards might suit better for complex firms that require abundant knowledge and experience.
2.2 Empirical Results: Firms
Probably the most cited paper1 on board size, and the key paper to this study as well is Yermack (1996). Yermack empirically examines the theories regarding board size, and observes an inverse, convex-shaped relation between board size and Tobin’s Q2, studying 452 large U.S. corporations over the period 1984-1991. Firm value decreases mostly when board size increases from small to medium. Yermack proposes that this correlation could be caused by a loss of effectiveness within large boards. Eisenberg, Sundgren and Wells (1998) also observe an inverse relation, in their research on the influence of board size on return on assets. However, their panel differs considerably from the panel used by Yermack (1996). Instead of large U.S. corporations, Eisenberg et al. perform research on 879 small Finnish firms over the period 1992-1994.
Research of Coles, Naveen and Naveen (2008) shows somewhat different results. The relation between firm value and board size appears to be u-shaped. Coles et al. uses a sample of 8,165 year-firm observations over the period 1992-2001 in the United States. According to Coles et al., either a very large or a very small board size is optimal. Large boards are optimal for multi-segment, complex firms, while for less complex, single-segment firms small boards are optimal. Furthermore, Coles et al. show that for complex firms mainly outside directors contribute to an increasing firm value, when board size grows.
Where the research discussed so far mainly focuses on firm value, Cheng (2008) studies the effect of board size on the variability of firm performance. Cheng finds a negative relation between board size and variability of firm performance, using a sample of 1,252 U.S. firms over the period 1996-2004. Cheng suggests that this relation is caused by a higher level of difficulty to coordinate within large boards. Board members have to make more concessions to reach consensus in a larger board. For that reason, strategies are relatively more moderate. This generally leads to constant performances.
1 Google Scholar
2 Market Value Assets/Replacement Value Assets, (Tobin, 1969)
5
2.3 Empirical Results: Mutual Funds
Tufano and Sevick (1997) study the effect of board structure on fee-setting among 1,587 American open-end mutual funds in the year 1992. Tufano and Sevick find that mutual funds with smaller boards are more active negotiators achieving lower shareholder fees. Shareholder fees are paid by shareholders to cover the costs of running the mutual fund. Del Guercio, Dann and Partch (2003) discover that among 476 closed-end U.S. mutual fund in the year 1996, smaller boards have lower expense ratios. Moreover, Del Guercio et al. observe a relation between board size and restructures, such as share repurchases or restructures of the portfolio that add value to the mutual fund. These results suggest that smaller boards are able to manage a fund more effective.
Khorana, Servaes and Wedge (2007) do not find any significant relation between board size of 614 U.S. mutual funds over the period 1999-2001 and their performance after a merger with another mutual fund. Performance seems to be related to ownership of board members. Cremers, Driessen, Maenhout, and Weinbaum (2009) state that board effectiveness is related to fund performance, but cannot find a significant relation between board size and fund performance using a sample of 134 U.S. mutual funds over the period 1996-2004 either. Nevertheless, Cremers et al. do find a relation between ownership stakes of the board members in the mutual fund and performance. Mutual funds with boards that have low ownership stakes underperform compared to other mutual funds.
2.4 Empirical Results: Banks
Adams and Mehran (2008) study the effect of board size on Tobin’s Q of 35 large U.S. banks over the period 1986-1999. Opposed to what they expect, they find a positive relation between board size and Tobin’s Q of banks. However, Adams and Mehran find that merger & acquisition activity seems to increase with board size as well, because some directors of the target bank are added to the board of the acquiring bank. Furthermore, Adams and Mehran find a positive relation between M&A activity and Tobin’s Q because banks that are involved with M&A activity are likely to have a higher Tobin’s Q. Their suggestion is that the relation between M&A activity and board size, and the relation between M&A activity and Tobin’s Q causes endogeneity and partly explains the counterintuitive positive relation between board size and bank performance.
Andres and Vallelado (2008) find an inverted u-shaped relation between board size and bank performance, using a dataset of 69 large commercial banks in U.S, Canada and Europe. A growing board size leads to increasing bank performance, with a maximum of 19 directors according to Andres and Vallelado. Opposed to Adam and Mehran, Andres and Vallelado do not take M&A activity into account. Agoraki, Delis and Staikouras (2010) observe a negative, linear relation between board size and cost and profit efficiency, using a sample of large European banks over the period 2002-2006. Moreover, Agoraki et al. find a nonlinear relation between share of outside directors and profit efficiency. An increase in share of outside directors leads to an increase in profit efficiency, but this effect is diminishing.
2.5 Empirical Results: Non-Profit Organizations
In contradiction to previous research, Aggarwal, Evans and Nanda (2012) observe a positive effect of board size on the performance of 35,040 US nonprofit organizations over 1998-2003. Performance is measured by indicators such as the ability to raise funds and to complete projects successfully. Aggarwal et al. state that every board member brings knowledge and effort in the organization. This theory is more or less consistent with the findings of Latane, Williams and Harkins (1979) that the total output of a group, despite loss of individual effort, still increases with board size. Garner and Harrison (2013), perform research on the influence of board size on nonprofit firm effectiveness. Garner and Harrison study 7,148 U.S. nonprofit firms and find a positive relation between board size and nonprofit firm performance. Effectiveness is measured by expenses and direct benefits. The relation is nonlinear, suggesting that very large boards are optimal for nonprofit firms. Garner and Harrison propose that larger boards are better in allocating workload.
To encapsulate the literature review, Hermalin and Weisbach (2001) describe in their literature overview a consensus about a negative relation between board size and firm performance and value. However, this relation is not applicable to any organization. Positive, negative as well as no relationship between board size and different aspects of different types of organizations are observed throughout time.
2.6 Pension Funds
Research on asset management of institutional investors is comprehensive, but a full review would be beyond the scope of this study. Still, the following papers are notable for the remainder of this study. A fundamental paper is written by Sharpe and Treynor (1977), who explain that pension funds are social as well as financial institutions. The responsibility of pension fund directors is to make decisions regarding asset management (investing decisions and hedging market and currency risks), and to hedge interest rate risks that can influence the value of the liabilities. Furthermore, pension fund directors are responsible for the determination of premiums and benefits which affects the liquidity of the funds.
Agnew, Balduzzi and Sunden (2003) perform research to the asset allocation of 7.000 U.S. retirement accounts over 1994-1998. According to Agnew et al., age is an important predictor for the allocation to equity. The higher the age of the participant, the lower the allocation to equity. Agnew et al. suggest that older people are more cautious with their pension savings. Research from both Alestalo and Puttonen (2006) and Gerber and Weber (2007) shows empirical evidence for respectively 44 Finnish pension funds in 2002 and around 1,000 Swiss pension funds in 2000 and 2002 that the composition of participants of pension funds influences the asset allocation of European pension funds as well. Similar to U.S. retirement accounts, the average age of the participants is an important determinant of the allocation to equity for European pension funds. Equity is regarded as a more risky asset class. This cut back of risk during the life of a participant is called lifecycle investing. The older the participant the more conservative and cautious the investments, because there is less time left to correct for losses. Bikker, Broeders and Hollander (2012) observe that lifecycle investing is being practiced among Dutch pension funds, using a dataset of 378 Dutch pension funds in 2007.
Even though economic literature provides a broad perspective on the impact of board size on performance and value of different types of organizations, it is still unclear which result makes the best prediction for the relation between board size and funding ratio for Dutch pension funds. Basically, pension funds are non-profit organizations, since the goal is not to be as profitable as possible, but to provide their participants a proper retirement. However, a pension fund makes an obligation to their participant in the future. The feasibility to meet this obligation depends on the funding ratio of the pension fund. Pension funds are, in contrast to most non-profit organizations, large players on financial markets. Dutch pension funds have
an accumulated assets under management of over one trillion euros (Centraal Bureau voor de Statistiek, 2014).
Similar to banks, pension funds are strong forces on financial markets. Pension funds are large financial institutions as well. Nonetheless, pension funds are both social and financial institutions. The goal of the pension fund is to be revolving, while the banks researched in the literature review are mainly commercial financial institutions.
Investment and hedging decisions of the board are essential in keeping the funding ratio sufficient. This is in in accordance with the role of the board of a mutual fund. Despite this similarity, pension funds generally have a more long-term view than mutual funds. The main objective is not to make high short-term profits, but to provide a good funding ratio to meet future obligations (Del Guercio & Tkac, 2002).
If the obligation cannot be met, either the firm, the government or the employee has to do a concession. For corporate pension plans, in case of underfunding the firm has to complement cash. This makes the corporate pension plan part of the firm (Sharpe & Treynor, 1977). Contrary to firms, the main activity of a pension fund is investing and hedging.
3. Hypothesis, Methodology and Data
In this section the hypothesis will be formulated. The methodology that will be used to test the hypothesis will be explained and the panel dataset will be described.
3.1 Hypothesis and Methodology
3.1.1 Hypotheses
To test which findings of previous research are most applicable for Dutch pension funds the following hypotheses will be tested:
H0 = There is no relation between board size and funding ratio of Dutch pension funds H1 = There is a relation between board size and funding ratio of Dutch pension funds
Additionally, to test whether the thesis of the key paper for this study, Yermack (1996), also hold for Dutch pension funds, the following hypothesis will be tested:
H2 = Smaller boards manage Dutch pension funds more effectively
3.1.2 Methodology
H0 and H1 will be tested by the following, fixed effects regressions:
𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑅𝑅𝑅𝑅𝑅𝑅𝐹𝐹𝑅𝑅 = 𝛼𝛼 + 𝛽𝛽1𝑙𝑙𝑅𝑅𝐹𝐹 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽3𝑃𝑃𝑆𝑆𝐵𝐵𝑃𝑃𝑅𝑅𝐵𝐵𝐷𝐷𝑅𝑅𝐹𝐹𝐹𝐹𝑆𝑆 + 𝛽𝛽4 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽5 % 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 + 𝛽𝛽6−9 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑅𝑅𝑅𝑅𝑅𝑅𝐹𝐹𝑅𝑅 = 𝛼𝛼 + 𝛽𝛽1𝑙𝑙𝑅𝑅𝐹𝐹 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽3 𝑃𝑃𝑆𝑆𝐵𝐵𝑃𝑃𝑅𝑅𝐵𝐵𝐷𝐷𝑅𝑅𝐹𝐹𝐹𝐹𝑆𝑆 + 𝛽𝛽4 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽5% 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 + 𝛽𝛽6𝑙𝑙𝑅𝑅𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽7−10 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀
The estimators are measured by a panel of 50 Dutch pension funds. The regression is an interpretation of the models that are introduced by Yermack (1996) and Coles, Naveen and Naveen (2008), applied to a panel of Dutch pension funds. The dependent variable, funding ratio, differs from the commonly used Tobin’s Q (e.g. Yermack (1996)). Tobin’s Q is a ratio of market value of the assets to replacement value of the assets (Tobin, 1969). However, the financial structure of pension funds is different from firms, because all assets are linked to claims of participants of the pension funds in the future (Sharpe, 1977). Therefore, Tobin’s Q cannot be applied to pension funds. Funding ratio is a widely accepted measure for the welfare of a pension fund, and is defined as (De Nederlandsche Bank, 2014):
𝑀𝑀𝑅𝑅𝐵𝐵𝑀𝑀𝑆𝑆𝑅𝑅 𝑉𝑉𝑅𝑅𝑙𝑙𝐹𝐹𝑆𝑆 𝑅𝑅𝑃𝑃𝑜𝑜𝑃𝑃𝑃𝑃𝑆𝑆𝑅𝑅𝑃𝑃
𝐿𝐿𝐹𝐹𝑅𝑅𝑉𝑉𝐹𝐹𝑙𝑙𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝑃𝑃(𝐷𝐷𝐹𝐹𝑃𝑃𝐹𝐹𝑅𝑅𝐹𝐹𝐹𝐹𝑅𝑅𝑆𝑆𝐹𝐹 𝑉𝑉𝐷𝐷 𝐶𝐶𝑆𝑆𝐹𝐹𝑅𝑅𝐵𝐵𝑅𝑅𝑙𝑙 𝐵𝐵𝑅𝑅𝐹𝐹𝑀𝑀 𝐼𝐼𝐹𝐹𝑅𝑅𝑆𝑆𝐵𝐵𝑆𝑆𝑃𝑃𝑅𝑅 𝑅𝑅𝑅𝑅𝑅𝑅𝑆𝑆) 𝑥𝑥 100
Funding ratio is chosen as dependent variable, because the main objective of a pension fund is to be revolving for the participant. The ability to be revolving depends on the financial position of the fund. Pension fund directors can affect both sides of the ratio: on one hand their asset management influence the value of the assets, on the other hand their hedging
(1A & 2A)
(1B & 2B)
decisions regarding interest rate risk determines the value of the liabilities (Sharpe & Treynor, 1977). Nevertheless, funding ratio is only an indicator for the financial position of a pension fund. Macroeconomic events such as booms and busts can influence the market value of the assets, while interest rate adjustments directly affect the value of the liabilities. These events are directly reflected in the funding ratio, but the effect of these events on the funding ratio can be beyond the influence of pension fund directors. To control for these time fixed events, control variables for each year are incorporated in the regression (Stock & Watson, 2012).
Board size is the core variable of interest in finding an answer to the research question. A log-transformation is used to cope with skewness and kurtosis and to normalize the distribution of the observations (Stock & Watson, 2012). To test if the board size effect is a nonlinear relation, an additional regression with the nonlinear function of Log Board Size will be performed:
𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑅𝑅𝑅𝑅𝑅𝑅𝐹𝐹𝑅𝑅 = 𝛼𝛼 + 𝛽𝛽1𝑙𝑙𝑅𝑅𝐹𝐹 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2𝑙𝑙𝑅𝑅𝐹𝐹 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆2
+ 𝛽𝛽3𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽4 𝑃𝑃𝑆𝑆𝐵𝐵𝑃𝑃𝑅𝑅𝐵𝐵𝐷𝐷𝑅𝑅𝐹𝐹𝐹𝐹𝑆𝑆 + 𝛽𝛽5 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽6 % 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 + 𝛽𝛽7−10 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀
Fiduciary management is a governance structure for pension funds to outsource the responsibility of picking asset managers for asset classes such as fixed income, equity or real estate to a third party. This third party is commonly an asset manager as well and is integrated in the pension fund board, to provide them profound advice (van Nunen, 2007). Because fiduciary managers are integrated in the board, they can be considered as important partners for the pension fund board. Fiduciary managers share characteristics with the concept of an outside director. Research shows different results of outside directors on variables such as performance or Tobin’s Q: Weisbach (1988), Byrd & Hickman (1992) and Baysinger & Butler (1985). Furthermore, a fiduciary manager is an important source of investment advice for pension funds. Coles et al (2008) included advice as control variable in their regression, and find a positive interaction effect between board size and advice on Tobin’s Q. Because of these contradicting results, fiduciary management is included as a control variable in this regression.
(1C & 2C)
Performance is measured by the annual return on assets of the pension fund. It is included as a control variable, because return on assets directly influences the funding ratio, since assets are part of this ratio. Return on assets can be volatile, and might be subjected to coincidence. Since this paper is interested in the long-term financial position of the pension fund, funding ratio, the regression controls for performance as pension fund-specific variable. Unfortunately, this dataset is not extensive enough to provide information over past performance. However, even if the dataset was extensive enough, this could lead to misleading results because crisis year 2008, with outstanding poor performance for pension funds, would be included in the model.
Other pension fund-specific control variables that are included are: liquidity and share of retired participants. Liquidity is measured by a ratio of premiums and benefits, which are the incoming and outgoing cash flow of the fund. A similar control variable is used by Coles et al. (2008) as well.
As observed by Agnew et al. (2003), Alestalo and Puttonen (2006) and Gerber and Weber (2007) the demographic composition of a pension fund influences the asset allocation. The older the average participant and the larger the share of retired participants, the more risk averse the investment behavior of a fund. This phenomenon is called lifecycle-investing, and is observed for Dutch pension funds as well by Bikkers, Broeders, Hollanders and Ponds (2012). Share of retirees is used as an indicator to control for the effects of lifecycle investing on the funding ratio.
Fund size is included as a control variable for several reasons. First, as mentioned by Coles et al. (2008) optimal board size seems to increase with complexity of firm activities. Assuming that complexity increases with size of the fund, it makes sense that fund size and board size are positively correlated. Furthermore, it is unclear what the effect of fund size is on performance. Chen, Hong and Huang (2004) observe a negative relation to size of a mutual fund and their performance, studying 165 U.S. mutual funds over 1981-1999. Contrary, Dyck and Pomorski (2011), show that pension funds benefit from economies of scale and that larger pension funds outperform smaller pension funds, studying a database of 842 world-wide pension funds over 1990-2008. Moreover, fund size is used as a control variable by Yermack (1996) and Coles et al (2008) as well.
In the dataset, two outliers regarding fund size can be found: ABP (governmental pension fund, 300 billion euros Assets under Management3) and Zorg en Welzijn (pension
3 Annual Report 2013 ABP
12
fund for healthcare workers, 140 billion euros Assets Under Management4). Fund size of other pension funds in the dataset ranges from 2.5 to 45 billion euros. To control for potential influence of these outliers, regression 2A, 2B and 2C are performed. These regressions exclude ABP and Zorg en Welzijn. Another potential problem of including fund size as a control variable is the correlation between fund size and board size. This could lead to multicollinearity. Therefore, both regression 1A and 2A are performed, in which fund size is not included as control variable. However, omitting fund size might be problematic as well, because this could lead to endogeinity (Stock & Watson, 2012). For that reason, regression 1B and 2B are performed. In 1C and 2C fund size is excluded, because the correlation between fund size, board size (log) and board size (log)2 would lead to strong multicollinearity. Similarly to board size, the log-transformation supports coping with skewness and kurtosis.
As mentioned by Petersen (2009), a researcher should be careful with standard error when a panel dataset is used. In case of a panel dataset, heteroskedastic robust standard errors, as introduced by White (1980), might lead to biased standard errors because these standard errors are not corrected for autocorrelation. Autocorrelation could be problematic for the data used in this paper. A funding ratio is based upon assets and liabilities that are accumulated over a long period, and are probably correlated over time. Board size can be correlated over time as well, since board compositions might be constant over time. The solution used in this paper to cope with biased standard errors is to use clustered standard errors. These more conservative standard errors correct for both heteroskedasticity and autocorrelation within clusters. The standard errors are clustered by pension funds.
3.2 Data
This paper uses the database of the Dutch consultancy firm, DutchInvestor. From this database, which is based on information from annual reports of pension funds, a panel of the largest 50 pension funds is selected, based on assets under management. The pension fund size in this dataset ranges from 2.5 to 300 billion euros, with two positive outliers: ABP and Zorg en Welzijn. Without these two pension funds, the dataset would range from 2.5 to 45 billion euros. The dataset consists of 218 observations. A complete list of involved pension
4 Annual Report 2013 Zorg en Welzijn
13
funds can be found in the appendix. The panel data set is unbalanced due to incomplete information. Financial key figures, such as funding ratio and performance, could be obtained from the database. Unfortunately, this database, as well as other databases such as De Nederlandsche Bank or Pensioen Federatie lacks information about board size of Dutch pension funds. Therefore, this information is hand-picked from publicly available annual reports of the Dutch pension funds, and results in a unique database. A time-frame of 2009-2013 is chosen to exclude crisis year 2008. Including 2008 would lead to large outliers in performance and funding ratio, which could make the estimators less precise.
Table 1 provides descriptive statistics about the board size of Dutch pension funds throughout the whole period, while table 2 provides descriptive statistics about the dependent variable, funding ratio.
Table 1
Board Size Dutch Pension Funds 2009-2013
2009 2010 2011 2012 2013 Observations 42 48 49 48 50 Min 5 6 5 5 6 Max 16 16 16 13 14 Mean 9.64 9.71 9.14 9.21 9.12 Median 9 9 9 9 8 Std. Dev. 2.90 2.52 2.47 2.17 2.28
Source: Annual Reports Dutch Pension Funds
Table 2
Funding Ratio5 Dutch Pension Funds 2009-2013
2009 2010 2011 2012 2013 Observations 49 50 53 51 50 Min 98 95 88 95.8 101.9 Max 154 147 132 133.1 139.3 Mean 114.02 111.52 103.72 109.07 114.25 Std. Dev. 10.69 10.11 8.79 9.24 9.13 Source: DutchInvestor
5 Funding Ratio is measured by Market Value of Assets/Liabilities (Discounted By Central Bank Interest Rate),
(De Nederlandsche Bank, 2014)
14
Table 3 provides the correlations between the different variables. The most notable correlations are the significant positive correlation between Log Board Size and Log Fund Size and the significant negative correlation between Funding Ratio and both Log Board Size and Log Fund Size. It is important to keep the strong correlation between fund size and board size in mind during the continuation of the study, since it could lead to multicollinearity between the variable of interest and the control variable. This can cause misleading standard errors, and very conservative t-values (Stock & Watson, 2012).
Table 3
Correlations between variables of Dutch pension funds 2009-2013
Funding Ratio Log Board Size Log Fund Size Perfor mance Fidu ciary Liquidity % Retired Funding Ratio 1 Log Board Size -0.1578** 1
Log Fund Size -0.1193* 0.4519*** 1
Performance 0.0824 0.1052 .0550 1
Fiduciary 0.0808 0.0539 .1336** 0.1101* 1
Liquidity 0.0073 -0.0625 -.1398** 0.0055 .0122 1
% Retirees 0.2339*** 0.1211* 0.1192* 0.0061 -.0163 -.5763*** 1
* = 10% significance, ** = 5% significance, *** = 1% significance
Figure 1
Histogram Board Size Dutch Pension Funds 2009-2013
Source: Annual Reports Dutch Pension Funds
Figure 1 shows a histogram with information about the frequency of every board size between 2009 and 2013. This histogram confirms that a log-transformation for board size is essential to normalize the distribution. The histogram also shows that the mode board size of Dutch pension funds between 2009 and 2013 is eight directors.
4. Empirical Analysis
4.1. Empirical Results
This section provides the empirical result of the panel data research. Table 4 shows the empirical results of the analysis among Dutch pension funds between 2009 and 2013, and tests H0 and H1. Furthermore, in this section an economic interpretation of the results will be given. Table 5 and 6 shows a robustness check, and tests H2.
The first notable result is that, within this panel, there is a significant negative relation between the core variable of interest, board size, and the dependent variable funding ratio in both regressions 1A and 2A. As mentioned in the data section, adding or dropping fund size as control variable, influences the coefficient and standard error of the board size estimator due to a strong and significant correlation between both variables, which is shown in regressions 1B and 2B. However, both 1A and 2A and 1B and 2B show that board size is negatively related to funding ratio.
Regressions 1C and 2C suggest that the relation between board size and funding ratio might be nonlinear. Funding ratio increases, when board size grows from very small to small. When board size goes from relatively small to medium/large, funding ratio decreases. Both board size estimators in regression three are imprecise, have large standard errors and are therefore insignificant. This is caused by multicollinearity, since Board Size (Log)2 is a function of the variable Board Size (Log). Nevertheless, an F-test proves that both estimators deviate significantly from zero.
The pension fund-specific control variables are quite constant in all regressions, and are almost all significant. The composition of the participants is a significant predictor for funding ratio. The larger the share of retired participants, the higher the funding ratio. Liquidity has a significant positive effect on the funding ratio. This research does not provide any evidence that fiduciary management leads to a higher or lower funding ratio. Moreover, it does not provide any evidence that fund size influences funding ratio.
Table 4 Empirical Results
All regressions are time fixed effects regressions, using data from an unbalanced panel of 50 Dutch pension funds. 218 observations are made, between 2009 and 2013. Values of t-statistics based on clustered robust standard errors are reported between brackets, which are corrected for heteroskedasticity and autocorrelation. Funding ratio is measured by assets divided by liabilities. Board Size (Log) is equal to the natural log of the total amount of directors of the pension fund. Board Size (Log)2 is the quadratic function of Board Size (Log).
Fiduciary management is a governance style for pension funds in which asset managers are picked by one, fiduciary asset manager. Performance is equal to annual return on assets. Liquidity is measured by annual premiums divided by benefits. % Retirees is a measure of the demographic composition of the participants to control for the effects of lifecycle investing. Fund size is measured by the log of assets under management. Dummy variables for each year are included to control for time-fixed events such as interest-rate adjustments and booms or busts on financial markets.
* = 10% significance, ** = 5% significance, *** = 1% significance
With ABP and Zorg en Welzijn Without ABP and Zorg en Welzijn
(1A) (1B) (1C) (2A) (2B) (2C) Independent Variables Funding Ratio Funding Ratio Funding Ratio Funding Ratio Funding Ratio Funding Ratio Board Size (Log)
(ŋ1) -9.544** (-2.43) -7.812* (-1.68) 8.231 (.20) -8.724** (-2.05) -8.306* (-1.76) 9.395 (.22) Board Size (Log)2
(ŋ2) -4.024 (-.45) -4.127 (-.44) Fiduciary 1.279 (.61) 1.479 (.72) 1.368 (.65) 1.629 (.75) 1.676 (.80) 1.734 (.80) Performance .335** (2.32) .374** (2.57) .333** (2.32) .368** (2.470 .376** (2.32) .365** (2.45) Liquidity 1.619*** (2.74) 1.492** (2.60) 1.596*** (2.71) 1.546** (2.64) 1.515** (2.61) 1.528** 2.62) % Retirees .271*** (3.50) .264*** (3.38) .269*** (3.47) .264*** (3.26) .263*** (3.26) .263*** (3.25)
Fund Size (Log) -.903
(-.95) -0.327 (-.22) N 218 218 218 208 208 208 F-Statistic 43.27 35.35 38.51 38.42 33.23 34.35 R2 .3086 .3155 .3097 .2966 .2972 .2976 F-Statistic (ŋ1+ŋ2) = 0 3.89** 2.61* 17
The results provide significant evidence to reject H0 in favor of H1. There is a relation between board size and funding ratio. The economic interpretation of the empirical result is that Dutch pension funds with very small or very large boards have, generally spoken, a lower funding ratio. Previous research suggests that this relation is caused by ineffectiveness of large boards and a lack of mutual monitoring. However, very small boards may lack enough knowledge to govern a pension fund. Pension funds operate in complex financial systems, and this lack of effectiveness, monitoring, experience or knowledge might lead to poor investing and/or hedging decisions. Additionally, hectic financial markets require fast decision-making. As mentioned by Cheng (2008), in larger boards it might be harder to find consensus. This makes it more complex to adapt the strategy of the fund to market circumstances. On the other hand, it requires knowledge and experience to steer a billion euro pension fund through capriciously moving financial markets. Furthermore, very large or very small boards may lack the capability of determining a robust policy regarding premiums and benefits. To keep the funding ratio sufficient in the long run, it might be necessary to increase premiums and/or decrease benefits.
The following economic interpretations of the results regarding the control variables can be made. First, the share of retired participants positively influences the funding ratio of Dutch pension funds. This might be explained by a lifecycle investing style, in which a more conservative investment style leads to a higher funding ratio. Liquidity is a positive, significant predictor of funding ratio. Higher cash inflows are reflected in funding ratio. Furthermore, this study does not provide any evidence that Dutch pension funds benefit from economies of scale regarding funding ratio, or that fund size erodes funding ratio. Fiduciary management is not a significant predictor as well, but this might be caused by the fact that fiduciary management is quite a young management style among Dutch pension funds. Potential effects might be observed in the future.
Figure 2 provides a graphical presentation of the relation between board size and average funding ratio between 2009 and 2013. This graph seems to support the thesis from regression 2 that the relation between board size and funding ratio of Dutch pension funds is nonlinear. Moreover, this graph suggests that pension funds with a board size of seven directors have on average the highest funding ratio. Funding ratio decreases most when board size moves from seven to eight or more board members, or moves from seven to six or five board members. However, these results should be interpreted carefully. An optimal board size might depend
on pension fund specific characteristics, such as fund size, as well. This graph is only a rough, visual description of the board size effect. Furthermore, the graph may exaggerate the true relation due to positive outliers within the group of pension funds with a board size of seven. However, even when the fund with the highest average funding ratio (Pensioenfonds voor Huisartsen) is excluded, the remaining pension funds with a board size of seven directors have an average funding ratio of 118%. This is distinctively higher than 110%, which is the average funding ratio of all pension funds involved in this research6.
Figure 2
Average Funding Ratio Dutch Pension Funds by Board Size 2009-2013
4.2 Robustness check
The robustness check tests whether the suggestion from the empirical results that neither very small nor very large boards are optimal holds for Dutch pension funds. Furthermore, this robustness check intends to test H2 that smaller boards manage Dutch pension funds more effective. The two alternative dependent variables that are used to measure the effectiveness of a pension fund board are return on assets and total costs (total expenditures as a percentage of total assets). This results in the following regressions.
6 DutchInvestor
19
𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝐹𝐹 𝑅𝑅𝐹𝐹 𝑜𝑜𝑃𝑃𝑃𝑃𝑆𝑆𝑅𝑅𝑃𝑃 = 𝛼𝛼 + 𝛽𝛽1𝑙𝑙𝑅𝑅𝐹𝐹 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2𝑙𝑙𝑅𝑅𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆
+ 𝛽𝛽3 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽4 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽5 % 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 +𝛽𝛽6−9 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀
𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝐹𝐹 𝑅𝑅𝐹𝐹 𝑜𝑜𝑃𝑃𝑃𝑃𝑆𝑆𝑅𝑅𝑃𝑃 = 𝛼𝛼 + 𝛽𝛽1log 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2log 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆2+ 𝛽𝛽3log 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + + 𝛽𝛽4 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽5 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽6 % 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 + 𝛽𝛽7−10 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀
𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑙𝑙 𝐶𝐶𝑅𝑅𝑃𝑃𝑅𝑅𝑃𝑃 = 𝛼𝛼 + 𝛽𝛽1𝑙𝑙𝑅𝑅𝐹𝐹 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2𝑙𝑙𝑅𝑅𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆
+ 𝛽𝛽3 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽4 𝑃𝑃𝑆𝑆𝐵𝐵𝑃𝑃𝑅𝑅𝐵𝐵𝐷𝐷𝑅𝑅𝐹𝐹𝐹𝐹𝑆𝑆 + 𝛽𝛽5 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽6 % 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 + 𝛽𝛽7−10 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀
𝑇𝑇𝑅𝑅𝑅𝑅𝑅𝑅𝑙𝑙 𝐶𝐶𝑅𝑅𝑃𝑃𝑅𝑅𝑃𝑃 = 𝛼𝛼 + 𝛽𝛽1log 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + 𝛽𝛽2log 𝐵𝐵𝑅𝑅𝑅𝑅𝐵𝐵𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆2+ 𝛽𝛽3log 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑆𝑆𝐹𝐹𝑆𝑆𝑆𝑆 + + 𝛽𝛽4 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐵𝐵𝐷𝐷 𝑀𝑀𝑅𝑅𝐹𝐹𝑅𝑅𝐹𝐹𝑆𝑆𝐷𝐷𝑆𝑆𝐹𝐹𝑅𝑅 + 𝛽𝛽5 𝑃𝑃𝑆𝑆𝐵𝐵𝑃𝑃𝑅𝑅𝐵𝐵𝐷𝐷𝑅𝑅𝐹𝐹𝐹𝐹𝑆𝑆 + 𝛽𝛽6 𝐿𝐿𝐹𝐹𝐿𝐿𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝑅𝑅𝐷𝐷 + 𝛽𝛽7 % 𝑅𝑅𝑆𝑆𝑅𝑅𝐹𝐹𝐵𝐵𝑆𝑆𝐹𝐹 𝑃𝑃𝑅𝑅𝐵𝐵𝑅𝑅𝐹𝐹𝐹𝐹𝐹𝐹𝑃𝑃𝑅𝑅𝐹𝐹𝑅𝑅𝑃𝑃 + 𝛽𝛽8−11 𝐷𝐷𝐹𝐹𝐷𝐷𝐷𝐷𝐷𝐷 𝑉𝑉𝑅𝑅𝐵𝐵𝐹𝐹𝑅𝑅𝑉𝑉𝑙𝑙𝑆𝑆𝑃𝑃 𝑌𝑌𝑆𝑆𝑅𝑅𝐵𝐵𝑃𝑃 + 𝜀𝜀
The estimators of interest, Board Size (Log) and Board Size (Log)2, are insignificant in all regressions. There is not enough evidence to accept H2. However, the results give a rough indication that the relation between board size and both performance and total costs have a same pattern as empirical results in table 4. Return on assets first increases and then decreases when moving from a small to a larger board size. However, return on assets is not the main objective of a pension fund. This might explain why the estimators are insignificant.
Total costs move the other way around: it first decreases and then increases when moving from a small to a larger board size. This makes sense, because high performance and low cost can be seen as characteristics of an effective board. However, this effect disappears when ABP and Zorg en Welzijn are omitted from the regression. The insignificance and the considerable difference between regressions 5B and 6B might be caused by the small sample. This leads to less precise estimators, larger standard errors and lower t-values. Reporting costs of asset management became mandatory for pension funds only in the last years (Autoriteit Financiële Markten, 2011), so information about costs of pension funds of previous years is not accessible via public information.
(3A & 4A)
(3B & 4B)
(5A & 6A)
(5B & 6B)
Table 5
Robustness Check Table 5 & 6
All regressions are time fixed effects regressions, using data from an unbalanced panel of 50 Dutch pension funds. Respectively 218 and 43 observations are made, between 2009 and 2013. Values of t-statistics, based on clustered robust standard errors, are reported between brackets. The standard errors are corrected for heteroskedasticity and autocorrelation.
Performance is equal to annual return on assets. Total costs are equal to expenditures as a percentage to assets under management. Board Size (Log) is equal to the natural log of the total amount of directors of the pension fund. Board Size (Log)2 is the quadratic function of Board Size (Log). Fiduciary management is a governance style for pension funds in which asset managers are picked by one, fiduciary asset manager. Liquidity is measured by annual premiums divided by benefits. % Retirees is a measure of the demographic composition of the participants to control for the effects of lifecycle investing. Fund size is measured by the log of assets under management. Dummy variables for each year are included to control for time-fixed events such as interest-rate adjustments and booms or busts on financial markets.
With ABP and Zorg en Welzijn Without ABP and Zorg en Welzijn
(3A) (3B) (4A) (4B)
Independent Variables
Return on Assets Return on Assets Return on Assets Return on Assets Board Size (Log)
(ŋ1) -.534 (-.56) 2.269 (.16) . -.605 (-.61) 5.79 (.40) Board Size (Log)2
(ŋ2) -.635 (-.21) . -1.45 (-.45) Fund Size .627*** (3.29) .628** (3.27) .548* (1.77) .533* (1.74) Fiduciary -.029 (-.05) -.015 (-.03) -..090 (-.14) -.051 (-0.08) Liquidity .488* (1.88) .484* (1.86) .499* (1.92) .491* (1.89) % Retirees .067** (2.15) .067** (2.12) .070** (2.22) .070** (2.20) N 218 218 208 208 F-Statistic 114.33 105.73 106.71 96.41 R2 .6354 .6355 .6292 .6296 F-Statistic (ŋ1+ŋ2) = 0 .21 .35
* = 10% significance, ** = 5% significance, *** = 1% significance.
Table 6
With ABP and Zorg en Welzijn Without ABP and Zorg en Welzijn
(5A) (5B) (6A) (6B)
Independent Variables
Total Costs Total Costs Total Costs Total Costs Board Size (Log)
(ŋ1) .058 (.62) -.087 (-.07) .074 (.81) .861 (.68) Board Size (Log)2
(ŋ2) .034 (.12) -.185 (-.64) Fund Size .011 (.37) .010 (.33) .-.036 (-1.14) .036 (-1.11) Fiduciary .046 (.59) .046 (.57) .061 (.84) .066 (.90) Performance .031*** (2.93) .031*** (2.87) .033*** (3.04) .031*** (2.87) Liquidity -.031* (-1.75) -.032* (-1.69) -.031* (-1.75) -.036** (-2.02) % Retirees -.002 (-1.26) -.002 (-1.24) -.002 (-1.30) -.002 (-1.32) N 48 48 46 46 F-Statistic 3.13 2.59 2.82 2.60 R2 .2058 .2060 .2174 .2242 F-Statistic (ŋ1+ŋ2) = 0 .20 .42
* = 10% significance, ** = 5% significance, *** = 1% significance.
5. Discussion & Conclusion
The empirical results provide significant evidence to reject the hypothesis that there is no relation between board size and funding ratio of Dutch pension funds. The relation between board size and funding ratio partly seems to be consistent with the predictions of, among others: Lipton & Lorsch (1992), Jensen (1993), the literature review of Hermalin & Weisbach (2001) and the empirical results of Yermack (1996) that large boards are not optimal. Moreover, these empirical results suggest that small boards are not optimal as well, which is consistent with the less cited theory of Dalton, Daily, Johnson and Ellestrand (1999) and Coles, Naveen and Naveen (2008). Neither a large nor a very small board seems to be
optimal for Dutch pension funds. Empirical results give an indication that a board of seven directors gives on average the highest funding ratio for Dutch pension funds. Theory of Jensen (1993) proposes a board size of seven directors as well. However, there are five suggestions and points to take into account when interpreting these results.
First, one should be careful when drawing conclusions about causality based on the correlation that this study shows. Although previous research provides suggestions how to interpret the results, further research on the origins of this nonlinear relation is necessary to draw complete conclusions about causality. To gain more insight in eventual causality, it could be helpful to study the effects of a pension fund that shrinks or enlarges its board size. However, to observe potential effects, this research should take a long time frame because of the long term objectives of the pension funds. The characteristics of a capable board will probably only be reflected in funding ratio in the long run.
Second, research on the influence of the composition of the board and the characteristics of board members on funding ratio could be helpful in understanding pension fund governance. It is likely that optimal governance is not only about quantity, but about quality as well. Attributes of pension fund directors that could be interesting to investigate are: sex, educational background, age, independency or stakes of the board member in the pension fund. Independency is especially interesting because the law about empowerment of pension fund governance introduces a new governance structure with independent directors. Empirical findings of, among others, Coles et al. (2008) suggest that independent directors contribute to firm value.
Likewise, it would be interesting to investigate the influence of stakes of board members in the pension fund. As suggested by Khorana et al. (2007) and Cremers et al (2009), ownership and stakeholdership are important predictors for mutual fund performance. This effect might also occur if board members are participants of the pension fund. Moreover, it would be interesting if the age of the board member with stakes in the fund have any influence in the light of lifecycle investing and intergenerational solidarity. It could be the case that older board members act more in the interest of older participants.
Third, a qualitative research on the behavior of pension fund directors within small or large boards could contribute to the understanding of pension fund governance. An in-depth analysis of the decision-making process among pension fund boards could be a method to gain more knowledge about pension fund governance. Because economics might be considered as a behavioral science, qualitative research could contribute to the understanding of the origins of the behavior of directors among small or large boards. Furthermore, it is
recommended to perform further research on the monitoring behavior of pension fund directors among different board sizes. As mentioned by Yermack (1996) and Lipton and Lorsch (1992), one of the explanations of the negative relation between board size and funding ratio might be the lack of monitoring within very large boards.
Fourth, a larger panel would strengthen the analysis and could make the estimators more precise. A transnational research, like research on the influence of board size on performance of European firms performed by Conyon and Peck (1998), would help to provide pension funds worldwide with more knowledge about optimal governance of their pension fund. This would increase the external validity of this research.
Fifth, while empirical results suggest that larger boards lead to a lower funding ratio, this cannot be used as an absolute directive for every individual pension fund. Although this regression controls for pension fund specific factors, the optimal board size for a specific pension fund might still deviate from the results of this study. Especially since this study is based on a panel of only the largest 50 pension funds, one should be cautious.
To summarize, board size is an extensively researched topic in economic literature. This paper attempts to make a contribution to this literature, by using a new methodology for a unique new database of Dutch pension funds. The research intends to be an extension of previous research by, among others Yermack (1996) and Coles, Naveen and Naveen (2008). Additionally, this paper attempts to contribute to the search for improvement in Dutch pension fund governance.
Empirical results from this study provide enough evidence to reject the null hypothesis that board size does not influence the funding ratio of Dutch pension funds. However, contrary to the findings of Yermack (1996), this relation seems to be nonlinear. Decreasing board size does not linearly increase funding ratio for Dutch pension funds. Economic literature suggests that on the one hand this nonlinear relation is caused by lack of knowledge and experience within very small boards. On the other hand, large boards suffer from a lack of effectiveness and monitoring power. The empirical results do not provide enough evidence to conclude that there is a significant, robust nonlinear relation between board size and effectiveness of the pension fund board. Nevertheless, these insignificances could be caused by the fact that return on assets is not the main priority of a pension fund and that the sample for regression five and six is quite small. Further research might give a decisive answer on the question if there is a relation between board size and board effectiveness for Dutch pension funds.
Two recommendations can be made based on the outcomes of this study. First, further research is desirable to gather more knowledge about the origins of the nonlinear relation between board size and funding ratio of Dutch pension funds. Further investigation on causality, characteristics of board members and transnational effects would help to gain more knowledge on pension fund governance. Furthermore, it could give more clarity about an eventual relation between board size and effectiveness. A larger sample would strengthen the analysis. Second, to improve pension fund governance, it is recommended for Dutch pension funds with a low funding ratio to review their board size. It might be a first step towards a better funding ratio. While this research cannot give an unambiguous advice about the perfect board size, this research and economic literature in general strongly suggests that there is a relation between board size and funding ratio.
Special gratitude to DutchInvestor for sharing its data for this research.
References
Adams, R. B., & Mehran, H. (2008). Corporate performance, board structure and its determinants in the banking industry. Federal Reserve Bank of New York Staff
Reports, No. 330
Aggarwal, R. K., Evans, M. E., & Nanda, D. (2012). Nonprofit boards: Size, performance and managerial incentives. Journal of Accounting and Economics, 53 (1), 466-487.
Agnew, J., Balduzzi, P., & Sunden, A. (2003). Portfolio choice and trading in a large 401 (k) plan. American Economic Review, 93 (1), 193-215.
Agoraki, M. E. K., Delis, M. D., & Staikouras, P. K. (2010). The effect of board size and composition on bank efficiency. International Journal of Banking, Accounting and
Finance, 2 (4), 357-386.
Alestalo, N., & Puttonen, V. (2006). Asset allocation in Finnish pension funds. Journal of
Pension Economics and Finance, 5 (1), 27-44.
Andres, P. D., & Vallelado, E. (2008). Corporate governance in banking: The role of the board of directors. Journal of Banking & Finance, 32 (12), 2570-2580.
Autoriteit FinanciëleMarkten (2011), Kosten pensioenfondsen verdienen meer aandacht,
retrieved on January 26 2015, from:
http://www.afm.nl/~/media/files/rapport/2011/onderzoeksrapport-kosten-pensioenfondsen.ashx
Baysinger, B. D., & Butler, H. N. (1985). Corporate governance and the board of directors: Performance effects of changes in board composition. Journal of Law, Economics, &
Organization, 1 (1), 101-124.
Bikker, J. A., Broeders, D. W., Hollanders, D. A., & Ponds, E. H. (2012). Pension Funds’
Asset Allocation and Participant Age: A Test of the Life‐Cycle Model. Journal of
Risk and Insurance, 79 (3), 595-618.
Byrd, J. W., & Hickman, K. A. (1992). Do outside directors monitor managers? Evidence from tender offer bids. Journal of Financial Economics, 32 (2), 195-221.
Centraal Bureau voor de Statistiek (2015), Beleggingen institutionele beleggers 1.8 biljoen
euro, retrieved on January 29 2015 from:
http://www.cbs.nl/NR/rdonlyres/5C14F2D2-71E5-403F-AC67-7DF87D06722E/0/2014Beleggingeninstitutionelebeleggersart.pdf
Chen, J., Hong, H., Huang, M., & Kubik, J. D. (2004). Does fund size erode mutual fund performance? The role of liquidity and organization. American Economic Review, 94 (5), 1276-1302.
Cheng, S. (2008). Board size and the variability of corporate performance. Journal of
Financial Economics, 87 (1), 157-176.
Coles, J. L., Daniel, N. D., & Naveen, L. (2008). Boards: Does one size fit all? Journal of
Financial Economics, 87 (2), 329-356.
Conyon, M. J., & Peck, S. I. (1998). Board size and corporate performance: evidence from European countries. The European Journal of Finance, 4 (3), 291-304.
Commissie Frijns (2013), Wet versterking pensioenfonds bestuur, Retrieved on January 19 2015 from:
https://www.eerstekamer.nl/wetsvoorstel/33182_wet_versterking_bestuur, http://wetten.overheid.nl/BWBR0033702/geldigheidsdatum_19-01-2015
Cremers, M., Driessen, J., Maenhout, P., & Weinbaum, D. (2009). Does skin in the game matter? Director incentives and governance in the mutual fund industry. Journal of
Financial and Quantitative Analysis, 44 (6), 1345.
De Nederlandsche Bank (2014), Wat is de dekkingsgraad?, retrieved on January 8 2015 from: http://info.dnb.nl/vragen/wat_is_de_dekkingsgraad
De Nederlandsche Bank (2015), Zijn pensioenfondsen geraakt door de crisis?, retrieved on January 17 2015 from:
http://info.dnb.nl/vragen/Zijn_pensioenfondsen_zwaar_geraakt_door_de_crisis
Dalton, D. R., Daily, C. M., Johnson, J. L., & Ellstrand, A. E. (1999). Number of directors and financial performance: A meta-analysis. Academy of Management Journal, 42 (6), 674-686.
Dyck, A., & Pomorski, L. (2011). Is bigger better? Size and performance in pension plan
management. Rotman School of Management Working Paper (1690724)
Eisenberg, T., Sundgren, S., & Wells, M. T. (1998). Larger board size and decreasing firm value in small firms. Journal of Financial Economics, 48 (1), 35-54.
Faccio, M., & Lasfer, M. A. (2000). Do occupational pension funds monitor companies in which they hold large stakes? Journal of Corporate Finance, 6 (1), 71-110.
Gerber, D. S., & Weber, R. (2007). Demography and investment behavior of pension funds: evidence for Switzerland. Journal of Pension Economics and Finance, 6 (3), 313-337.
del Guercio, D., Dann, L. Y., & Partch, M. M. (2003). Governance and boards of directors in closed-end investment companies. Journal of Financial Economics, 69 (1), 111-152.
del Guercio, D., & Tkac, P. A. (2002). The determinants of the flow of funds of managed portfolios: Mutual funds vs. pension funds. Journal of Financial and Quantitative
Analysis, 37 (4), 523-557.
Garner, J. L., & Harrison, T. D. (2013). Boards, executive excess compensation, and shared power: evidence from nonprofit firms. Financial Review, 48 (4), 617-643.
Hermalin, B. E., & Weisbach, M. S. (1988). The determinants of board composition. The
RAND Journal of Economics, 19 (4), 589-606.
Hermalin, B. E., & Weisbach, M. S. (2001). Boards of directors as an endogenously
determined institution: A survey of the economic literature, National Bureau of
Economic Research.
Jensen, M. C. (1993). The modern industrial revolution, exit, and the failure of internal control systems. The Journal of Finance, 48 (3), 831-880.
Khorana, A., Tufano, P., & Wedge, L. (2007). Board structure, mergers, and shareholder wealth: A study of the mutual fund industry. Journal of Financial Economics, 85 (2), 571-598.
Latane, B., Williams, K., & Harkins, S. (1979). Many hands make light the work: The causes and consequences of social loafing. Journal of Personality and Social Psychology, 37 (6), 822.
Lipton, M., & Lorsch, J. W. (1992). A modest proposal for improved corporate governance.
The Business Lawyer, 48 (1), 59-77.
Mercer (2014), Melbourne Mercer Global Pension Index, retrieved on January 17 2015 from: http://www.mercer.com/insights/focus/melbourne-mercer-global-pension-index.html
van Nunen, A. (2007). Fiduciary management: Blueprint for pension fund excellence (Vol. 415). John Wiley & Sons.
Petersen, M. A. (2009). Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22 (1), 435-480.
Sharpe, W. F., & Treynor, J. L. (1977). The principles of corporate pension finance. The
Journal of Finance, 32 (2), 627-638.
Stock, J. H., & Watson, M. W. (2012). Introduction to econometrics. Essex: Pearson Education Unlimited.
Tobin, J. (1969). A general equilibrium approach to monetary theory. Journal of Money,
Credit and Banking, 1 (1), 15-29.
Tufano, P., & Sevick, M. (1997). Board structure and fee-setting in the US mutual fund industry. Journal of Financial Economics, 46 (3), 321-355.
Weisbach, M. S. (1988). Outside directors and CEO turnover. Journal of Financial
Economics, (20), 431-460.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica: Journal of the Econometric Society, 48 (4), 817-838.
Yermack, D. (1996). Higher market valuation of companies with a small board of directors.
Journal of Financial Economics, 40 (2), 185-211.
Appendix: List of involved pension funds
ABN AMRO Bank N.V., St. Psf. van de ABP, St. Psf. Achmea, St. Psf. Ahold, St. Psf. APF, St. Psf. (AkzoNobel) Apotheken, St. Psf. Medewerkers Architectenbureaus, St. Psf. voor de Bakkersbedrijf, St. Bpf. voor het Bouw, St. Bpf. voor de
Delta Lloyd, St. Psf.
Detailhandel, St. Bpf. voor de DSM Nederland, St. Psf. Fysiotherapeuten, St. Psf. voor
Grafische Bedrijfsfondsen, St. Psf. voor Heineken Psf., St.
Hoogovens, St. Psf.
Horeca en Catering, St. Bpf. Huisartsen, St. Psf. voor
IBM Nederland, St. Psf. (SPIN) ING Groep, St. Psf. KLM Algemeen, St. Pensioenfonds KLM Vliegend personeel, St. Psf. KLM-Cabinepersoneel, St. Psf. Koopvaardij, St. Bpf. voor de KPN, St. Psf.
Landbouw (Colland), St. Bpf. voor de Levensmiddelenbedrijf, St. Bpf. Medische Specialisten, St. Psf.
Meubelindustrie en Meubileringsbedrijven, St. Bpf. voor d Openbaar Vervoer, St. Psf.
Philips Psf., St.
PME, St. Bedrijfs- en vroegpensioenfonds PMT, St. Bpf. Metaal en Techniek
PNO Media, St. Bpf. voor het PostNL, St. psf. Rabobankorganisatie, St. Pensioenfonds Robeco, St. Psf. Schilders, St. Bpf. Schoonmaak- en Glazenwassersbedrijf, St. Bpf. Shell Psf., St. SNS Reaal Groep, St. Psf. Spoorwegpensioenfonds, St. TNO, St. Psf. Unilever Psf. 'Progress', St. UWV, St. Psf Vervoer, Pensioenfonds Werk en (re)integratie, St. Psf. Wonen, St. Bpf. Woningcorporaties, St. Psf. voor de Zorg en Welzijn, St. Psf. Zorgverzekeraars, St. Bpf. (SBZ) 30
