3.1 Relevant literature

While there is an extensive literature on the governance of publicly listed companies,9 few studies have examined the impact of governance on the performance of, and member fees paid to, superannuation funds. This research focuses on providing evidence on self-managed superannuation funds10 and the behavioural aspects of Australian superannuation funds such as superannuation fund members' financial literacy,11 members' behaviour12 and executives' behaviour.13

Although the literature on superannuation funds is scant, there are several studies using small sample sizes that examine the effect of governance practices on superannuation fund performance and fees (Benson et al., 2011; Liu, 2014; Liu & Ooi, 2019; Nguyen et al., 2012; Nisbet, 2013; Tan & Cam, 2015). Some studies find that larger fund boards are associated with better fund performance (Benson et al., 2011) and charge higher fees (Tan & Cam, 2015). Nguyen et al. (2012) argue that as board and fund size increase, more asset consultants are hired, resulting in higher fees and agency costs. In addition to board size, Nisbet (2013) reports that independent directors and board gender diversity of industry superannuation funds are associated with superior performance, whereas independent directors on the board of retail superannuation funds are associated with poorer fund performance. Studies conducted in Poland also confirm a positive effect of independent directors on fund performance (Jackowicz & Kowalewski, 2012; Kowalewski, 2012). In contrast, some Australian studies document that board independence has no association with performance (Liu, 2014) or fees (Tan & Cam, 2015). Tan and Cam (2015) suggest that independent directors do not influence fund costs as they are nominated and elected by other directors on the board. Liu (2014) documents that directors' age, director busyness and a directors' prior superannuation fund experience are negatively associated with performance; and directors' age is positively associated with fees. A more recent study (Liu & Ooi, 2019) investigates the effect of related party outsourcing and director affiliation on investment performance. They find that retail superannuation funds with a greater proportion of affiliated directors generate lower investment performance. In their additional tests, Liu and Ooi (2019) include several governance variables such as independent chairperson, board size and directors with investment experience; however, the variables have an insignificant association with investment performance.

The Australian empirical evidence is conflicting for several possible reasons. First, prior studies use different measures of performance. There is an inherent difficulty in measuring performance due to various investment options and asset allocations. The various performance measures used are rate of return (ROR) (Nguyen et al., 2012; Nisbet, 2013), return on assets (ROA) (Nisbet, 2013), the excess return over a benchmark (Benson et al., 2011) and risk-adjusted over-benchmark return (Liu, 2014). Second, prior studies use small sample sizes, a short time period and focus on particular types of superannuation funds. For example, Benson et al. (2011) use survey data from 34 superannuation funds (industry and public sector superannuation funds) over two consecutive years (2005 and 2006). Nguyen et al. (2012) examine 52 corporate superannuation funds for one year (2010), while Nisbet (2013) uses a sample of 72 observations comprising retail and industry superannuation funds for two separate years (2009 and 2012). Similarly, Liu (2014) uses 100 observations (21 corporate, 46 industry, six public sector and 27 retail superannuation funds) for a single year (2006), and Liu and Ooi (2019) observe 101 superannuation funds between 2015 and 2016. This study builds on and significantly extends these prior studies using a benchmark-adjusted ROA and draws on a larger dataset including both industry and retail superannuation funds (in additional tests) over a longer sample period of 11 years (2010–2020).

3.2 Hypotheses development

The superannuation industry lacks external governance mechanisms. The independent reviews highlight that an effective board is critical in enhancing members' retirement benefits. Based on the recommendations in the Cooper and Murray reviews, more independent directors are viewed as beneficial in terms of increasing the quality of board oversight and lessening the conflict of interest between managers and fund members. Independent directors also bring specialist knowledge and expertise that is different to member and employer representatives.14 The specialist knowledge should allow independent directors to better perform their monitoring and advising roles. Independent directors are also less likely to be beholden to management and are more likely to question management decisions, resulting in a reduction in agency problems. Independent directors and an independent chairperson can also enhance performance as they have incentives to retain their reputation and reduce litigation risks. However, empirical evidence for both public listed firms and superannuation funds suggest that board independence does not improve firm/fund performance (Adams et al., 2010), nor is it related to lower fees charged to members (Nguyen et al., 2012; Tan & Cam, 2015).

Despite the current regulatory push, which calls for more independent boards, the lack of supporting empirical evidence questions whether director independence alone is the most important trait of superannuation fund boards.15 Rather, an emerging literature suggests that the monitoring and advising capacity of independent directors is heightened when they have both the ability and motivation to monitor and advise (Adams & Ferreira, 2007; Bedford et al., 2022; Coles et al., 2014; Crespí-Cladera & Pascual-Fuster, 2014). This literature acknowledges the heterogeneity in independent directors and provides evidence consistent with the view that some directors have greater monitoring and advising abilities, which are associated with improved firm outcomes. In our study, we focus on two characteristics of independent directors that have received significant attention in both the academic literature and in the superannuation governance reviews: financial competence and director independence from the CEO. It is expected that both independence from the CEO and financial competence are associated with the incentive and ability of independent directors to monitor and advise superannuation fund managers on complex investment decisions.

One important board characteristic that has attracted academic attention and has been identified in both the Cooper Review and the Murray Inquiry is the financial competence of independent directors. While financial expertise enhances board oversight (Güner et al., 2008), empirical evidence suggests that superannuation fund directors lack such expertise (Besley & Prat, 2003; Clark et al., 2006; Jackowicz & Kowalewski, 2012), which hampers their ability to advise fund managers. The level of competence and training of superannuation fund directors is not currently regulated. Despite the requirement for directors to be ‘fit and proper’ under the APRA Prudential Standard 520 Fit and Proper and s29D(1)(d) of the SIS Act 1993, and obtain a Registrable Superannuation Entity (RSE) licence, the level of education, qualifications, experience and training are not specified. An inadequate level of skills and knowledge in a specialised and relevant area, such as investment management, potentially leads to poor decision making and, as a result, can have an adverse effect on fund performance and fees charged to members.

Based on the 2018 APRA thematic review on the governance of Australian superannuation funds, APRA indicates that most superannuation funds do not have an optimal board composition with appropriate skills, capabilities and experience of directors.16 Financial expertise is an important skill on superannuation fund boards as their main function is to make investment decisions on behalf of members. As a result, we predict that boards with independent directors with financial expertise are better monitors and advisors and as a result make better investment decisions, which benefits fund performance and fees. Accordingly, we predict:

Another desirable characteristic of directors is that they are independent from management. Prior research indicates that independent directors who join the board after the CEO are less independent (Cohen et al., 2012; Coles et al., 2014; Core et al., 1999). This arises due to the CEO's influence over board appointments and the selection of directors who are more ‘sympathetic’, have similar views to them or with whom they share social ties (Bedford et al., 2022; Bruynseels & Cardinaels, 2014; Cohen et al., 2008). As such, directors who are appointed by the CEO (co-opted directors), are expected to be poorer monitors who are less likely to act to curtail managerial discretions due to their loyalty to the CEO. The potential conflict of interest arising from co-option is also recognised in the Cooper Review, which recommends that funds develop a conflict-of-interest policy to address potential conflicts that may arise (Recommendation 2.17). Similarly, the Productivity Commission recommends that the interest and duties of directors need to be clarified to highlight that directors act in the best interest of members (Recommendation 22).17 As such, independence as the sole criterion for improved board functions in superannuation funds is not sufficient. Rather, we propose that non-co-opted independent directors lower agency costs and improve the monitoring and effectiveness of the board as they are more likely to be willing to question the decisions of superannuation fund executives. Hence, we predict that non-co-opted independent directors improve superannuation fund outcomes for members as measured by fund performance and fees. This leads to our second hypothesis:

A possible issue with empirically examining the predictions in Hypotheses 1 and 2 is that there may be an insufficient number of independent directors with financial expertise, or that are not co-opted, to make a difference in superannuation fund board level decision making. For instance, the data presented in Panel C of Table 2 shows that when an industry superannuation fund board includes an independent director, the number of independent directors is usually one. If a single independent director is unable to exert sufficient influence over other directors in board-level discussions, then their advisory and monitoring roles may be diminished to such an extent that they have no influence on superannuation fund performance or fees. This effect would suggest that, inconsistent with Hypotheses 1 and 2, there would be no association between superannuation fund performance and fees, and the presence of non-co-opted independent directors or independent directors with financial expertise. Prior governance research in other contexts, however, has established that even a low number of directors with a particular characteristic or attribute is sufficient to influence corporate outcomes. For example, research on board gender diversity indicates that even one or a small number of female directors influences corporate outcomes such as CEO compensation (Bugeja et al., 2016), stock price informativeness (Gul et al., 2011) and takeover outcomes (Levi et al., 2014). Similarly, Gyapong et al. (2016) document that the presence of only one ethnic minority director is positively associated with Tobin's Q, while Moscariello et al. (2019) find that firm value is improved in the presence of a small number of independent directors appointed by minority shareholders. These research findings suggest that even the presence of a small number of independent directors on superannuation fund boards can impact fund performance and fees.

4.1 Regression model

EXCESS ROA is measured as the difference between a superannuation fund's ROA and the median ROA for each year.18 EXCESS TOTAL EXPENSE RATIO is calculated as the difference between a superannuation fund's TOTAL EXPENSE RATIO (calculated as total administration, operating and investment expenses divided by total assets) and the median TOTAL EXPENSE RATIO for each year. The dependent variables and all continuous variables are winsorised at the top and bottom 5 percent to reduce problems related to outliers. Models (1) and (2) are estimated using ordinary least squares (OLS) regressions.

We measure BOARD EFFECTIVENESS using the traditional measure of board independence (%IND DIR) and, to test our hypotheses, using an alternative measure of board independence highlighting whether independent directors possess financial expertise (%FINANCIAL_IND DIR) and are non-co-opted by management (%NOT CO-OPTED_IND DIR). We also examine the interaction between %FINANCIAL_IND DIR and %NOT CO-OPTED_IND DIR (%NOT CO-OPTED_FINANCIAL IND DIR). The variable %IND DIR is measured as the total number of independent directors divided by the total number of directors on the board.19 %FINANCIAL_IND DIR is the percentage of independent directors with accounting and financial qualifications on the board. Consistent with Coles et al. (2014), we define %NOT CO-OPTED_IND DIR as the ratio of the number of non-co-opted independent directors, meaning those appointed before the CEO assumes office, to board size.

To capture directors' competence and ability to advise fund managers, this study focuses on the financial expertise and superannuation fund industry experience of board members. Although neither the SIS Act 1993 nor the APRA Prudential Standards require directors to have a financial qualification, prior studies suggest that directors who have accounting and financial expertise improve firm value (Davidson et al., 2004; DeFond et al., 2005). It is conjectured that these directors use their financial skills and expertise to achieve better outcomes for members of superannuation funds. Thus, the presence of directors with financial qualifications suggests better governance practices. The measure %FINANCIAL is included as a control variable and is measured as the percentage of directors with financial expertise, including accounting and finance qualifications on the board.20 Moreover, directors do not always act in the best interest of shareholders as they may assign their loyalty and allegiance to the CEO who recruited them (Coles et al., 2014). Therefore, directors are expected to be better monitors when they make decisions that are not influenced by management. We control for %NOT CO-OPTED as the percentage of directors who were appointed before the CEO.

As many of the reviews (e.g., the Cooper Review and the Murray Inquiry) have pushed for an independent chair on the board, we include an indicator variable to control for superannuation funds with an independent chairperson (IND CHAIR). Prior studies suggest that female directors provide different perspectives and a greater level of discussion in decision-making processes (Gul et al., 2011; Srinidhi et al., 2011).21 We include %FEMALE DIR measured as the percentage of female directors on the board. Directors who hold multiple outside directorships may lack time and effort to perform their responsibilities and duties well (Ferris et al., 2003; Fich & Shivdasani, 2006; Schnake et al., 2005). In contrast, it is also possible that busy directors have expertise in decision making gained from experience in multiple directorships (Fama & Jensen, 1983; Ferris et al., 2003). We measure %BUSY DIR as the total number of outside directorships on ASX-listed companies held by directors divided by the total number of directors on the board.

Prior studies also suggest that directors with related industry experience provide better advice to the board and are associated with higher firm value (Dass et al., 2014; Faleye et al., 2018). Due to the complexity of the superannuation fund industry, directors who have prior superannuation industry experience should promote better superannuation fund outcomes. We measure %EXPERIENCE as the percentage of directors with prior superannuation fund experience on the board.22 Arguably, a long tenure may diminish a director's independence, and monitoring and advising capacity; consequently, long tenure is expected to have a detrimental effect on firm outcomes (Bhagat & Black, 2002; Vafeas, 2003) and directors with a shorter tenure are expected to have better governance practices. TENURE is measured as the total number of years directors have been employed by the fund divided by the total number of directors on the board. We also control for board size. Larger boards provide less effective monitoring and efficient decision making due to poor communication and difficulty in making decisions from various perspectives (Lipton & Lorsch, 1992; Yermack, 1996). A number of studies provide evidence that smaller boards are associated with better firm value (Yermack, 1996) and lower fund fees (Tufano & Sevick, 1997). Board size (BSIZE) is defined as the total number of directors on the board.

Following prior literature, we include five control variables to capture underlying fund characteristics (Liu, 2014; Liu & Ooi, 2016; 2019; Nguyen et al., 2012; Nisbet, 2013; Tan & Cam, 2015). First, the natural logarithm of total assets (LnTA) is included to control for fund size, as large superannuation funds may benefit from diversified investments through a wide range of asset classes and lower operating costs that are dispersed among larger assets (Cummings, 2016). Second, the natural logarithm of investment options (LnINV OPTIONS) is applied as a proxy for investment complexity. Third, preservation age is included as a control (PRS AGE) as members are more inclined to have a conservative investment portfolio to manage the liquidity issue of benefit payments when they are reaching retirement (Cummings & Ellis, 2015).23 We include %PRS AGE measured as the percentage of members who are aged 50 or over.24 We also control for the percentage of assets under management which are respectively invested in default funds (%DEFAULT FUND), and for fund asset allocation by including variables for the percentage of assets allocated to cash (%CASH) and equity (%EQUITY). Year fixed-effects are included in all models.

4.2 Sample selection

This study focuses on the period 2010–2020. Financial data is obtained from the APRA Annual Fund-level Superannuation Statistics Back Series and the APRA Superannuation Fund-level Profiles and Financial Performance statistics. All governance characteristics, except BUSY DIR (which is calculated from data available from the Connect 4 Boardroom database), are hand collected from annual reports, superannuation fund websites and relevant documents (disclosed under s29QB of the SIS Act 1993).

This search identified an initial sample of 2614 fund years. As this study examines industry superannuation funds, 609 observations of corporate and public sector funds and 701 observations of retail funds are excluded. We also exclude 66 defined benefit funds. A further 863 observations are removed from the sample due to missing financial (223 observations) and governance (640 observations) information.25 The final sample consists of 375 industry fund-year observations for 36 unique industry funds. A summary of the sample selection process is presented in Table 1.

5.1 Descriptive statistics

Table 2 provides descriptive statistics for the variables included in Models (1) and (2). Panel A of Table 2 reveals that the mean (median) ROA is 6.28 percent (7.24 percent) and EXCESS ROA is 0.54 percent (0.63 percent) for industry funds. The mean (median) TOTAL EXPENSE RATIO is 0.77 percent (0.68 percent), while the mean operating, and investment expense ratios are 0.48 percent and 0.28 percent, respectively. On average, industry fund boards comprise 10.4 percent independent directors (%IND DIR) which is similar to the sample in Tan and Cam (2015). The proportion of independent directors with financial expertise is 5.3 percent (%FINANCIAL_IND DIR), while 41.6 percent of all types of directors possess financial expertise (%FINANCIAL). Only 4.8 percent (%NOT CO-OPTED_IND DIR) of independent directors are not co-opted by the CEO. We find that 36.5 percent of the superannuation funds have an independent chair (IND CHAIR). Among the other descriptive statistics, we find that the average board size (BSIZE) is 9.26 directors, 24.7 percent of each board are female (%FEMALE DIR) and 18.8 percent of directors are defined as busy (%BUSY DIR). For industry funds the mean percentage of assets invested in cash (%CASH) and equity (%EQUITY) is 10 percent and 51.5 percent, respectively.

Panel B of Table 2 presents additional detail on the distribution of board size across industry funds. The most common board size is nine directors for 23.47 percent of funds, followed by 10 directors for 15.73 percent of funds. In Panel C of Table 2 we provide additional information on the distribution of independent, financial and non-co-opted directors at the fund level. The descriptive evidence indicates that 36.80 percent of boards have no independent directors, while 38.67 percent have one independent director and 24.53 percent have two or more independent directors. The data also indicate that 64.53 percent of boards have no independent directors with financial expertise. In contrast, 24 percent and 11.47 percent of boards have one, or two or more, independent directors with financial expertise, respectively. Turning to non-co-opted independent directors, 66.93 percent of boards have zero independent directors that are non-co-opted. In turn, 24.53 percent of boards have one non-co-opted independent director, while 8.54 percent of boards have two or more non-co-opted directors.

In Panel D of Table 2 we provide additional statistics on the sample at the director level. These statistics indicate that 11.12 percent of directors in our sample consisting of 89 unique directors are classified as independent. In terms of independent directors with financial expertise (non-co-opted), this comprises 5.34 percent (4.67 percent) of directors in the sample and 40 (46) unique directors. These data highlight that within the sample there are a sufficient number of independent directors with financial expertise and independent directors that are not co-opted to undertake formal empirical testing. Finally, there are 21 (2.62 percent) unique directors who are both independent, non-co-opted and possess financial expertise.

5.2 The effect of board effectiveness on performance and fees of industry superannuation funds

Table 3 presents the results of the effect of board effectiveness on excess return on assets (EXCESS ROA).26 In column (1) of Table 3 we find that the traditional measure of board independence (%IND DIR) is not associated with superannuation fund performance. Similarly, there is no association between fund performance and independent directors' financial expertise (column 2). The results in column (3) indicate that the appointment of non-co-opted independent directors (%NOT CO-OPTED_IND DIR) to the board is significantly associated (at the 5 percent level) with superannuation fund EXCESS ROA. This finding provides evidence to support H2a.

Among the control variables, we find a positive association between the presence of busy directors (except in column 1) and fund performance. This positive association suggests that rather than busyness distracting directors from completing their workload, busy directors draw on their experience to improve returns for members. Moreover, funds with a higher percentage of assets invested in default funds (%DEFAULT FUND) and equity (%EQUITY) also perform better.

Table 4 presents findings on the association between board effectiveness and the EXCESS TOTAL EXPENSE RATIO of industry superannuation funds. The coefficient on %IND DIR in column (1) indicates that there is an insignificant association between the traditional board independence measure and superannuation fund fees. This suggests that despite the recommendations of the Cooper and Murray reviews, industry superannuation fund board independence does not enhance fund performance or reduce fees. When we shift our focus to director co-option, we find a significant negative association (at the 5 percent level) between the percentage of non-co-opted independent directors (column 3) and total fees. The effect of independent directors with financial expertise is, however, insignificant (column 2). Similarly, the joint effect of non-co-opted directors with financial expertise (column 4) on fees is also insignificant. Overall, these results provide some support for H2b.

Among the control variables we document a negative coefficient on LnTA, indicating that industry superannuation funds with larger assets manage costs across asset-units more efficiently. The results also indicate that fees are higher for funds that are governed by a board with more experienced directors. This finding is potentially explained by more experienced directors commanding higher director fees.

6.1 Endogeneity

This study acknowledges that endogeneity concerns (simultaneity-equation bias and omitted correlated variable bias) could affect the findings (Ferris & Yan, 2007; Wintoki et al., 2012).27 One endogeneity concern arises due to reverse causality. For instance, the current performance and fees of superannuation funds may influence the choice of governance practices and consequently influences future performance and fees (Hermalin & Weisbach, 2003; Wintoki et al., 2012). In other words, superannuation funds with higher performance and lower fees tend to have boards of directors with characteristics aligned with better governance practices, and vice versa.

This study attempts to mitigate the endogeneity concern of reverse causality by estimating OLS regression models using lagged governance and control variables and employing fixed-effects models.28 The results examining the association between governance practices, and EXCESS ROA and EXCESS TOTAL EXPENSE RATIO with lagged variables and fixed-effects for industry funds, are presented in Tables 5 and 6, respectively. Consistent with the main findings, the results in Table 5 show an insignificant association between fund excess performance and the traditional measure of board independence. However, non-co-opted independent directors continue to be positively associated with excess performance. The findings in Table 6 also show that board independence is not associated with lower excess total expense ratio. We find that both financial expertise and a non-co-opted independent board is not statistically associated with the excess total expense ratio.

Another endogeneity problem arises due to omitted correlated variable bias, where some other factors may drive both the outcome and the explanatory variable. Some prior studies have used instrumental variables and two-stage least squares (2SLS) regression to address this problem of endogeneity (Larcker & Rusticus, 2010). However, Larcker and Rusticus (2010) argue that weak instrumental variables provide estimates that are more biased and are likely to provide wrong statistical inference than OLS estimates (Larcker et al., 2007; Larcker & Rusticus, 2010). This study does not perform a 2SLS regression as it is unable to identify an instrumental variable that explains governance practices that does not determine superannuation fund performance and fees.

6.2 Governance effectiveness in retail funds

Industry superannuation funds differ from other superannuation funds, especially retail funds. For instance, industry funds operate on a not-for-profit basis where all profits are returned to members, whereas retail funds operate to generate profits for the company's shareholders. Moreover, industry funds have an equal representation of member and employer representatives on the board (equal representation model) who are typically appointed by trade unions and employer associations (i.e., third parties). In contrast, the directors of retail funds are appointed by the company. In addition to differences in board composition, the age profile of industry and retail funds is different, with 43 percent (23 percent) of industry fund members being under 35 years old, and 19 percent (33 percent) of industry (retail) members being over 55 in 2021 (APRA, 2021b). This allows industry superannuation funds to invest in longer-term assets such as unlisted assets.

As an additional test we examine the effect of governance practices on the performance and fees of 701 retail superannuation funds, using hand collected data from retail fund annual reports. Descriptive statistics for retail funds are presented in Panel A of Table 2, along with a statistical comparison of the mean value between industry and retail funds. The data show that retail funds have significantly higher fees and lower performance than industry funds. In terms of director characteristics, we find that retail funds have a significantly greater proportion of independent directors than industry funds. Additionally, retail funds have significantly more independent directors that are non-co-opted and possess financial expertise. Other notable differences between retail and industry funds are that retail funds have more female and more busy directors, are smaller in size, and offer more investment options to their members. Retail funds also have less investments in default funds, and a greater (lower) proportion of assets allocated to cash (equity) compared to industry funds.

Table 7 presents the results of estimating the regression model testing the effect of board effectiveness on the excess ROA for retail funds. The results show that none of the coefficients on board effectiveness are significant, suggesting that there is no association between fund performance and independent directors, independent directors with financial expertise and non-co-opted independent directors. Among the control variables, we find that retail fund performance is positively associated with the percentage of assets allocated to equity. There is also evidence that when the directors on the board are busy, hence distracted, fund performance is lower.

The results testing the influence of board effectiveness in retail funds on excess total expenses are presented in Table 8. We do not find any association between board effectiveness and total operating expenses, irrespective of whether the independent directors possess financial expertise or are not co-opted. Among the control variables, there is a significant positive association between excess total expense ratio and board financial expertise. We also document that larger retail funds charge lower fees.

6.3 MySuper

We next examine the effect of governance practices on performance and fees of industry superannuation funds using ‘MySuper’ products. MySuper products are designed and formulated by the fund as the default investment option for members. MySuper provides an opportunity to analyse a relatively consistent set of products that are not impacted by member choice of investment option.29 MySuper was introduced on 1 January 2014 to provide a simple, cost-effective and balanced default investment product with a single diversified investment strategy.30 Although MySuper products have common characteristics (as set out in s29TC of the SIS Act 1993), the proportion of assets allocated to growth and defensive assets, and asset classes, are dissimilar.31 These differences in the proportion of asset allocations and asset classes have led to variation in the performance of MySuper funds. Panel E of Table 2 shows that the standard deviation of the MySuper EXCESS ROA (1.64 percent) is greater than that for other industry funds (1.24 percent). In contrast to the variation in asset allocations and performance, MySuper members are charged a standardised operating fee – at the same flat fee and same percentage of account balance – in accordance with the s29VA of the SIS Act 1993. This is evident in the smaller standard deviation of the excess total expense ratio of MySuper funds (0.24 percent) compared with other industry funds (0.32 percent).

We estimate our regression models for MySuper funds only; Tables 9 and 10 show the results for fund performance and fees respectively.32 Similar to the main results, Table 9 reveals that there is an insignificant association between independent directors and the performance of MySuper funds. We also find that the percentage of independent directors that are non-co-opted or that possess financial expertise is insignificantly associated with fund performance. In column (4) we find that non-co-opted independent directors with financial expertise are significantly and positively associated with MySuper fund performance. This result highlights the effectiveness of independent directors when they possess financial knowledge and when they are free from conflict of interests when making investment decisions. As this variable is only significant when the sample is restricted to MySuper funds, it is possible that the benefits of advice from directors with financial expertise is greatest when selecting the specific assets and investments to be included in default funds and those which are managed using a lifecycle approach. This result on %NOT CO-OPTED_FINANCIAL IND DIR, however, needs to be interpreted with caution due to the small number of independent directors that possess financial expertise who are non-co-opted in the sample of MySuper funds.

Table 10 shows no association between fees of MySuper funds and board independence. Similar to the findings in Table 4, we document that the percentage of directors with financial expertise is not significantly associated with fees (column 2), while having non-co-opted independent directors leads to significantly lower fees (column 3).

6.4 Other tests

To test the robustness of the main findings, alternative raw measures of performance (ROA) and fees (TOTAL EXPENSE RATIO) are used as the dependent variable. The results (untabulated) remain similar to the main findings. Specifically, we continue to find no association between measures of board independence and fund performance and fees. In addition, we continue to find that the percentage of non-co-opted directors are positively associated with fund performance and negatively associated with fees charged by superannuation funds.

Further, as Panel C of Table 2 highlights that 36.8 percent of funds do not have independent directors, we conduct additional analyses to alleviate concerns that the insignificant results on %IND DIR are driven by funds with either zero or only one independent director. First, we re-run our main analyses on funds having at least one independent director. Second, we re-run our main analyses with different measures of independent directors, using an indicator variable for funds with one or more independent directors and an indicator variable for funds with two or more independent directors. We also use a raw measure of the number of independent directors, non-co-opted independent directors and independent directors with financial expertise. The findings from these analyses (untabulated) provide a significant coefficient on the alternative measures of IND DIR, while the coefficients on independent directors who are not co-opted remain consistent with the main results.

We also examine whether a co-opted independent chair and an independent chair with financial expertise are associated with fund outcomes. We do not find that a co-opted independent chair and an independent chair with financial expertise are associated with fund performance. However, we find that a non-co-opted independent chair is negatively associated with fund fees, hence highlighting the value of an independent chair whose decisions are not influenced by management.

6.5 Limitations of the study

The results in this study suggest that director co-option is associated with worse outcomes for superannuation fund members in terms of higher fees and lower performance. In contrast, the findings do not document a statistically significant association between the traditional measure of board independence or independent director financial expertise and superannuation fund performance and fees. While we fail to document a statistically significant result on these two variables, it is acknowledged that the lack of significance on these two variables may be driven by the small sample size used in the study, which results in the tests lacking sufficient power. We leave it to future research to further examine these issues using a larger sample of superannuation funds. Finally, as with most governance research, our findings are subject to endogeneity concerns. While we have attempted to address endogeneity concerns using lagged variables in Tables 5 and 6, it is notable that the significant effect of director co-option on superannuation fund fees is not robust in this additional test.