Managerial share ownership and operating performance: Do independent and executive directors have different incentives?

Abstract

We investigate the relationship between managerial share ownership (MSO) and earnings as a measure of operating performance in Australia. To mitigate potential earnings management, we also use discretionary accrual adjusted earnings as an alternative measure of performance. We document a negative relation between MSO and performance followed by a positive relation. We suggest that these unique results are an artefact of certain Australian institutional features and imply that the ownership–performance relation is context-specific, with the wider corporate governance systems influencing the theorised incentive effects. We also posit that executive directors and independent directors have different ownership–performance incentives. Our results are consistent with this proposition and suggest that independent directors may be immune to the theorised incentive alignment or entrenchment effects associated with share ownership.

Keywords

Entrenchment incentive alignment managerial share ownership performance

1. Introduction

This study examines the relationship between managerial share ownership (hereafter MSO) and operating performance in Australia during the period 2000–2006. Theory suggests that MSO could affect firm performance in one of two ways. Jensen and Meckling (1976) argue that increased levels of MSO help align the interests of owners and managers and, therefore, mitigate firm agency problems. An alternative argument is that managers get entrenched when there is high MSO, thereby exacerbating firm agency problems (Demsetz, 1983).

Prior research on the relation between MSO and Tobin’s Q as a measure of performance reports mixed findings. For example, Morck et al. (1988) and McConnell and Servaes (1990) document a non-linear/non-monotonic relation showing an initial positive relation between MSO and Tobin’s Q consistent with incentive alignment up to a certain level of MSO, followed by a decrease in performance consistent with an entrenchment effect. Subsequent studies address the possibility that MSO could be endogenous, and report mixed findings. For example, Cho (1998) finds that performance affects MSO (reverse causality), whereas Davies et al. (2005) observe that both MSO and performance affect each other (bidirectional relation). Additionally, there are studies that fail to document any relationship (see for example Demsetz and Villalonga, 2001; Himmelberg et al., 1999) between MSO and performance.

Only a subset of these studies use earnings as a measure of operating performance – typically as further analysis, rather than a primary measure. These findings are also mixed. For example, Morck et al. (1988) and McConnell and Servaes (1990) report a non-linear relation between MSO and earnings, whilst Demsetz and Villalonga (2001) and Welch (2003) do not find any relation.

Several factors motivate this study. First, much of the prior literature examines the relationship between MSO and performance using share ownership by all directors. It does not distinguish between share ownership by executive directors (hereinafter ESO) and by non-executive directors –in particular, independent directors (hereinafter ISO). We argue that executive directors and independent directors have different ownership–performance incentives that are likely to impact on the relationship we examine. For example, as executive directors have responsibility for day-to-day operations, it is likely that their reputation in the managerial labour market is more closely tied to the firm’s performance than that of independent directors. In contrast, the economics of the managerial labour market provide incentives for independent directors to act as effective monitors in order to enhance their reputation and the value of their human capital (Fama and Jensen, 1983). These reputation effects are likely to outweigh any issues relating to incentive alignment or entrenchment that may otherwise arise as a result of owning shares in the firm.¹ Two prior studies separately assess the influence of share ownership by the two top executive directors (Morck et al., 1988) and the CEO (Hermalin and Weisbach, 1991). Only one study (Mura, 2007) examines the ownership–performance relation for different groups of directors. Mura (2007) uses non-executive directors as a proxy for independent directors, and uses Tobin’s Q as his measure of firm performance. We extend his work by specifically identifying directors who meet regulatory criteria for independence. Hence, amongst others, non-executive directors who nevertheless have affiliations or interests that may compromise their independence, such as having recently been executives or professional advisers, are not categorized as independent directors.² Moreover, Demsetz and Villalonga (2001), in an ownership–performance context, argued that it would be more appropriate to look at an estimate of what management has attained rather than a forward-looking metric such as Tobin’s Q. They also suggest that accounting earnings, unlike Tobin’s Q, are not affected by investor psychology. Similarly, Core et al. (2006) argue that operating performance is a more appropriate measure than stock returns to examine the relationship between corporate governance and performance. Accordingly, we use accounting earnings as the main measure of firm operating performance in this study.³

Second, whilst Australia has relatively strong legal shareholder protection (La Porta et al., 1999) it is argued that features of the Australian corporate governance system are markedly different from that of the US and other countries with strong legal shareholder protection (Dignam and Galanis, 2004). For example, Australian companies have high levels of ownership concentration (La Porta et al., 1999), proxy voting by the shareholders in Australian companies is lower than in the US and Australian blockholders are relatively passive in monitoring of management (see Dignam and Galanis, 2004 for a more detailed discussion). Furthermore, the Australian market is characterised by relatively larger private benefits of control (Nenova, 2003) and it has been argued that a shareholder does not need a particularly large shareholding to derive private benefits and maintain ‘practical control’ (Lamba and Stapledon, 2001). Thus managerial ownership is likely to play a greater role in the governance of Australian firms, and interest alignment might require significantly higher managerial ownership.⁴ In view of the above, we cannot assume that evidence reported overseas on the relationship between MSO and performance will hold in Australia.

There is very limited evidence regarding the relationship between managerial ownership and performance for Australian companies. Craswell et al. (1997) report weak evidence supporting a curvilinear relation between MSO and firm value for a sub-sample of their observations, and Welch (2003) found no evidence of a relationship between MSO and firm value once she controlled for potential endogeneity for a linear specification of MSO. These two prior studies are characterised by relatively small samples: specifically, Craswell et al. (1997) used 349 firm-year observations and Welch (2003) 113 firms for 1991–1992. Moreover, in common with much of the research at the time, Craswell et al., (1997) did not address the issue of endogeneity and reverse-causality and Welch (2003) did not do so when testing non-linear specifications of MSO.⁵ In contrast, our sample comprises 1154 firm-year observations during the period 2000–2006 and our results are robust to a range of alternative measures of performance (including accrual adjusted earnings), as well as rigorous methods to control for endogeneity and reverse-causality.

Third, managers have numerous market and/or contract-driven incentives to manage earnings. Discretionary accruals are a commonly used proxy for earnings management (see for example Dechow et al., 1995; Healy and Wahlen, 1999). Warfield et al. (1995) argue that contractual constraints designed to align interests and/or reduce the potential for opportunistic behaviour are likely to be systematically associated with the level of MSO, and find an inverse relation between the level of MSO and the level of discretionary accruals. Accordingly, in contrast to prior work examining the MSO–performance relationship, we also use adjusted earnings as a measure of operating performance.

We examine the relation between MSO and firm performance for an unbalanced panel data set of the 300 largest Australian firms between 2000 and 2006. We find evidence of a curvilinear relation between ownership and firm performance. Performance initially declines as MSO increases to an approximate range between 20% and 30%, depending on the measure. After this, increases in ownership are associated with an increase in firm performance. We also find that performance affects MSO, but only when we use adjusted earnings. Our finding of such a bidirectional relation when we use adjusted earnings supports our argument for the need to recognise the possibility of earnings management in the context of the ownership–performance relationship. We also find that results are driven by executive director ownership; ownership by grey and independent non-executive directors does not influence firm performance. Our results are robust to a range of alternative firm performance measures (accounting earnings, adjusted accounting earnings, and Tobin’s Q) and several alternative empirical methods to control for endogeneity and reverse-causality in the ownership-performance relation. Our results are also robust to alternative specification of managerial ownership, alternative measures of earnings (including operating cash flows) and alternative estimates of discretionary accruals, as well as autocorrelation, heteroskedasticity and multicollinearity.

The findings of our study contribute to the literature in a number of ways. First, this study presents some robust results which are argued to be consistent with the features of the Australian corporate governance environment; specifically, that managers have the potential to derive private benefits and maintain ‘practical control’ at relatively low levels of ownership. Second, whilst prior work focuses on MSO as a whole, this study examines the relation between ESO and performance and ISO and performance separately, and shows that executive and independent directors have different incentives. Third, it uses a much larger dataset and addresses methodological limitations associated with the two previous Australian studies (Craswell et al., 1997; Welch, 2003).

The remainder of the paper is structured as follows. Section 2 provides the theoretical development and research propositions. This is followed by an outline of the research design. The results are discussed in section 4, and section 5 outlines further analyses and robustness tests. Section 6 presents concluding remarks.

2. Theoretical framework and research propositions

A manager who owns a fraction of a firm’s shares bears the consequences of managerial actions thus aligning their incentives with other shareholders (Jensen and Meckling, 1976). However, an increase in MSO can result in managers becoming entrenched (Demsetz, 1983). The argument is that the extra voting power enables them to secure their position in the firm, thereby insulating them from certain disciplining mechanisms (for example, the managerial labour market and the market for corporate control), which is likely to have an adverse effect on firm performance. Hence the initial theory developed in this area would suggest a non-monotonic relation – more specifically, a positive relation between MSO and performance consistent with incentive alignment up to some turning point, followed by a negative relation when the costs associated with entrenchment exceed the incentive benefits of managerial ownership (see for example McConnell and Servaes, 1990; Morck et al., 1988). There are numerous other studies that fail to document any relation between MSO and firm value or performance (see for example Demsetz and Villalonga, 2001; Himmelberg et al., 1998). In short, notwithstanding the elegant theories put forward, the empirical conclusions are quite unsettled.

It is argued that features of the Australian legal system, the market for corporate control, ownership characteristics and other corporate governance features mean Australian corporate governance is markedly different from that of the US (Dignam and Galanis, 2004). Thus even in the context of an environment of relatively strong legal shareholder protection, it has been argued that a shareholder does not need a particularly large shareholding to derive private benefits and maintain ‘practical control’ in Australia (Lamba and Stapledon, 2001). Prior studies that identify an entrenchment effect document it commencing at varying levels – for example, 5% in the US (Morck et al., 1988) and 7% in the UK (Davies et al., 2005). In view of the above, it is possible that Australian managers may not need a particularly large shareholding to get entrenched and derive private benefits, hence any effects on performance may take place at relatively lower levels of ownership.

When managers’ ownership is low, they may maximise their personal wealth through increasing perquisites and guaranteeing their employment at the expense of firm performance. Although some control mechanisms are likely to be in place, the effect of alignment of interests may not be strong enough and could be dominated by entrenchment effect. Furthermore, an absence or low level of hostile takeover activity may indicate that market discipline is also lacking in Australia (Dignam and Galanis, 2004). The lack of disciplinary control over poorly performing management strengthens managers’ ability to pursue sub-optimal corporate policies, as does the relative passivity of blockholders in Australia. Beyond a certain ownership level, however, the incentive benefits of managerial ownership exceed the costs associated with entrenchment and may lead to managerial behaviour to improve firm performance. Prior studies that identify an entrenchment effect turning and being dominated by an incentive effect report that it commences around 25% in the US (Morck et al., 1988) and 26% in the UK (Davies et al., 2005).

We posit that entrenchment effects may take place at relatively lower levels of ownership and that, based on prior studies, incentive effects are likely to start dominating at ownership levels in excess of 20%–25%. On balance, theory suggests some combination of entrenchment and incentive alignment effects and, therefore, a non-monotonic relation between MSO and operating performance with a turning point. Hence, to examine such a non-monotonic relation, we use a quadratic specification of the MSO variable. The mixed evidence reported in prior research also suggests that this is ultimately an empirical issue.

As previously indicated, our measure of operating performance is accounting earnings, which may be susceptible to earnings management (see for example Healy and Wahlen, 1999). Moreover, Warfield et al. (1995) report a systematic relationship between the level of MSO and the level of discretionary accrual. Accordingly, we use accrual adjusted earnings as an additional measure of operating performance.

Much of the prior literature examining the relationship between MSO and performance does not differentiate between the roles of the managers who own the shares. This may not be appropriate. We argue that executive directors and non-executive directors (particularly independent directors) are likely to have different incentives. Executive directors are more closely involved in the operations of the business and it is likely that their reputational capital is more closely tied to firm performance, as is their ability to influence performance.

On the other hand, it is argued that the economics of the managerial labour market provide incentives for non-executive directors, more specifically independent directors, to act as effective monitors in order to enhance their reputation and the value of their human capital (Fama and Jensen, 1983). Future directorships may be a function of the reputation they develop as effective monitors. Once again, there is empirical support for this proposition. For example, Cotter et al. (1997) report that shareholders of target firms, with outside directors who have multiple directorships, receive larger premiums in tender offers. The one prior study that separately examines the influence of share ownership by executive and non-executive directors reports results consistent with different incentives, but uses non-executive directors as a proxy for independent directors and Tobin’s Q rather than any accounting measures of performance (Mura, 2007). Hence it is argued that, in the case of independent directors, concern for their reputations as effective monitors is likely to outweigh any issues relating to incentive alignment or entrenchment that may otherwise arise as a result of owning shares in the firm. On the other hand, for any given level of share ownership, executive directors are likely to be more sensitive to the effects of incentive alignment and entrenchment than are independent directors. Accordingly, we expect the relation between executive directors and operating performance to drive the MSO–performance relation, but we expect no relation between independent director share ownership and operating performance.

3. Research design

3.1. Data

We initially identified the top 300 Australian companies by market capitalisation at 30 June 1999. Consistent with the prior literature, we exclude banks, financial institutions, trusts and utility firms (49 firms) which have different disclosure requirements and/or different corporate governance structures. We exclude another 63 firms due to insufficient information for our study. Of these 63 firms, 17 were excluded since they use foreign currencies for reporting and the rest were excluded due to missing corporate governance or financial information, or both.⁶ Sample selection procedure is described in panel A of Table 1. The final sample comprises the remaining firms, with a total of 1154 firm-year observations over the seven-year period 2000–2006.⁷ As evident in panel B of Table 1, the sample firms belong to eight Global Industrial Classification Standard (GICS) sectors –material (19%), industrial (16%), health care (12%), information technology (7%), consumer discretionary (27%), consumer staples (11%), energy (5%) and telecommunication (2%). Panel C of the same table details the breakdown of observations across different years.

Table 1.

Sample description.

Panel A: Sample selection
Number of firms		300
Less:
Financial and utility companies		49
Companies without insufficient information		63
Total		188

Panel B: Analysis of sample by GICS sectors and industries
GICS sector	GICS industry group
Material	Chemicals	3
	Construction material	5
	Metal & mining	22
	Paper & forest products	6
Industrial	Capital goods	16
	Commercial service & supplies	9
	Transportation	5
Health care	Health care equipment & supplies	10
	Health care providers & services	6
	Pharmaceutical, biotechnology & life science	8
Telecommunication	Diversified telecommunication	4
Consumer staples	Food & staple retailing	5
Consumer staples	Food, beverage & tobacco	15
Consumer discretionary	Automobiles & components	7
	Consumer durables & apparels	6
	Consumer services	11
	Media	17
	Retailing	10
Information technology	Software & services	7
Information technology	Technology hardware & equipment	6
Energy	Oil and gas	10
Total		188

Panel C
Year		No of observations
2000		165
2001		169
2002		170
2003		166
2004		163
2005		159
2006		162

The required accounting information was collected from the Aspect Fin Analysis and Connect 4 databases. The ownership and other corporate governance data were hand-collected from the corporate governance disclosures, shareholding information and directors’ reports contained in annual reports.

3.2. Model specification

A number of recent studies suggest the potential for endogeneity between MSO and firm value or performance (see for example Davies et al., 2005). This would be consistent with, for example, better performing firms having higher MSO, suggesting that managers of these firms are more willing to accept shares as part of their compensation package and/or to buy shares in their own firms. To address this issue, we examine the relationship between MSO and performance using an instrumental variable regression technique. In other words, instrumental variable regression is our main method of regression analysis. This method has been commonly used in previous studies (see for example Hermalin and Weisbach, 1991). The instrumental variable needs to be correlated with the endogenous variable (MSO) and independent from the error terms of the ownership–performance regressions (Greene, 2003). Following previous studies (see for example Anderson and Reeb, 2003; Palia, 2001), this study uses the average age of the board of directors and monthly stock return volatility as the instruments. We use a two-way fixed effects model for our regression analysis. The fixed effects are dummy variables for each year of the sample and dummy variables for each GICS sector. We apply White’s (1980) heteroskedasticity-consistent standard errors for all regression analyses performed in this study. Furthermore, we apply the firm clustering technique for all the analyses, because multiple observations from the same firm (but from different years) are included in our dataset. We use three different types of managerial ownership variables (MSO, ESO and ISO) in our study.⁸ For all managerial ownership variables we use quadratic specifications (see for example McConnell and Servaeas, 1990).⁹

We use the following equation to examine the relation between MSO and performance using an instrumental variable regression.

\begin{matrix} Performance = β_{0} + β_{1} (MSO) + β_{2} {(MSO)}^{2} + β_{3} (Leverage) + β_{4} (Investment) \\ + β_{5} (Unaffiliated shareholdings) + β_{6} (Board independence) + β_{7} (Firm age) \\ + β_{8} (Size) + β_{9 to 15} (GICS Sectoral dummies) + β_{16 to 21} (Year dummies) + ε \end{matrix}

(1)

Performance is measured by earnings and adjusted earnings. Our definition of earnings is: Earnings before interest, tax depreciation and amortisation, scaled by the book value of total assets (EBITDA). We exclude discretionary accruals from the aforementioned earnings measure and obtain adjusted net earnings after tax and before abnormal items (AEBITDA). Details regarding the estimation of discretionary accruals are discussed in section 3.3. MSO, ESO, and ISO are calculated by taking the percentage of ordinary shares owned by the directors, executive directors and independent directors, respectively. Independent directors are those who meet the criteria for independence as set out in the Investment and Financial Services Association definition that was subsequently adopted by the ASX Corporate Governance Council (2003), Principles of Good Corporate Governance and Best Practice Recommendations.¹⁰

The control variables introduced in the above equation are leverage, investment, unaffiliated substantial shareholdings, board independence, firm age and size. Leverage is calculated as the ratio of book value of debt and book value of assets. It may mitigate certain agency problems through debtholder monitoring, and may also impact on MSO (Himmelberg et al., 1999). Investment is calculated as capital expenditure scaled by book value of assets (Cho, 1998; Davies et al., 2005). Cho (1998) argues that investment positively affects performance since the market reacts positively to the announcement of planned capital expenditure. Unaffiliated substantial shareholdings/blockholdings are measured by taking the percentage of ordinary shares held by substantial shareholders other than the directors (Dahya et al., 2008).¹¹ They are included to address the possibility of monitoring by blockholders (Dahya et al., 2008). Board independence is calculated as the number of independent directors scaled by the size of the board. A greater proportion of independent directors on the board may result in better performance because of improved monitoring (Anderson and Reeb, 2003). Firm age is calculated by taking the natural log of number of years since the firm was listed on the ASX (Anderson and Reeb, 2003). It is controlled since mature firms are likely to perform better than start-up firms (Fahlenbrach and Stulz, 2009). Size is proxied by the natural log of book value of assets. A large firm may generate more resources, which may result in better performance through larger projects and attaining economies of scale (Demsetz and Villalonga, 2001).

Motivated by the findings of Cho (1998) and Davies et al. (2005), we also use a simultaneous equations system (three-stage least squares). In other words, the simultaneous equations system is our secondary method of testing the relationship between MSO and performance. We use a system of three equations and introduce two additional equations (one for MSO and the other one for investment) with our performance equation. Once again, we use average age and monthly stock return volatility as the instruments for MSO. We also use firm age and board independence as the instruments for performance and investment, respectively. First we introduce the following equation, along with equation (1), for all the managerial ownership variables (MSO, ESO, and ISO).

\begin{matrix} MSO = α_{0} + α_{1} (Performance) + α_{2} (Leverage) + α_{3} (Investment) \\ + α_{4} (Volatility) + α_{5} (Liquidity) + α_{6} (Market value of equity) \\ + α_{7 to 13} (GICS Sectoral dummies) + α_{14 to 19} (Year dummies) + ε \end{matrix}

(2)

The definitions of managerial share ownership and performance variables are identical to those used in equation (1). The control variables used in this equation are leverage, investment, volatility, liquidity and market value of equity. Leverage has already been defined in equation (1). The use of leverage may lessen the need for external financing, thereby resulting in an increase in MSO (Cho, 1998), or leverage and MSO may be perceived as alternative monitoring mechanisms (Warfieldet al., 1995). Investment is calculated as the ratio of capital expenditure and book value of assets. Cho (1998) argues that investment positively affects performance, since the market reacts positively to the announcement of planned capital expenditure. Volatility is calculated as a standard deviation of earnings of the preceding five years scaled by book value of assets (Davies et al., 2005). It is included to examine the possibility that high firm-specific uncertainty affects the level of MSO (Cho, 1998). Liquidity is calculated as the ratio of net operating cash flows and book value of assets (Cho, 1998; Davies et al., 2005). Cho (1998) argues that managers may prefer to have a higher stake in highly liquid firms due to the ease of discretionary spending. Market value of equity is calculated by taking the natural log of market value of common equity (Cho, 1998). Holderness et al. (1999) report that the market value of equity is negatively associated with MSO because of managerial wealth constraints.

Prior studies (Cho, 1998; Davies et al., 2005) argue that investment is also endogenous. Following these studies, we introduce equation (3) to address the possibility that investment is endogenous when we run the simultaneous equations system (three-stage least squares).

\begin{matrix} Investment = δ_{0} + δ_{1} (MSO) + δ_{2} {(MSO)}^{2} + δ_{3} (Performance) + δ_{4} (Volatility) \\ + δ_{5} (Liquidity) + δ_{6} (Property, plant and equipment) \\ + δ_{7 to 12} (GICS Sectoral dummies) + α_{13 to 18} (Year dummies) + ε \end{matrix}

(3)

The definitions of managerial share ownership, performance and investment variables are identical to the definitions used in equation (1) and (2). The control variables used in this equation are volatility, liquidity and property, plant and equipment are consistent with (Cho, 1998). Volatility and liquidity are already defined in equation (2). Cho (1998) argues that volatility may adversely affect investments due to the uncertainty of the expected relationship between current and future profitability. He also argues that highly liquid firms will make more investments. Property, plant and equipment is calculated by taking the ratio of value of property, plant and equipment and book value of assets. Cho (1998) and Mura (2007) argue that this proxy for capital expenditure is associated with firm value.

3.3. Estimation of discretionary accruals

Given that managers have numerous market and/or contract-driven incentives to manage earnings, we argue that examination of the relationship between MSO and earnings as a performance measure could be biased. As discretionary accruals are a commonly used proxy for earnings management, we exclude discretionary accruals and use accrual adjusted earnings as an additional measure of performance. We use two models to estimate discretionary accruals. First, we use a parsimonious model used by Chan et al. (2006) to estimate discretionary accruals. The model is:

E_{t} (T A C C_{i t}) = \frac{\sum_{k = 1}^{5} T A C C_{i t - k}}{\sum_{k = 1}^{5} S a l e s_{i t - k}} S a l e s_{\begin{array}{l} i t \end{array}}

(4)

where: E_t(TACC_it) = Expected total accruals of firm i in year t; TACC_it-k = Total accruals of firm i in year t−k; Sales_it-k = Sales revenue of firm i in year t−k.

Total accruals (TACC) is estimated as

TACC = Δ C A - Δ C L - D E P

(5)

where ΔCA is the change in non-cash current assets (change in current assets less change in cash), ΔCL is the change in current liabilities excluding short-term debt (change in current liabilities less the change in debt included in current liabilities and minus the changes in income tax payable) and DEP is depreciation and amortisation.

Discretionary accrual is then given by

D A C C_{i t} = T A C C_{i t} - E_{t} (T A C C_{i t})

(6)

where: DACC_it = Discretionary accruals of firm i in year t; TACC_it = Total accruals of firm i in year t; E_t(TACC_it)= Expected total accruals of firm i in year t.

The level of total accruals has been related to current sales. To smooth any kind of transitory fluctuation, the proportion as the ratio of a moving average of the past five years’ total accruals to a moving average of sales has been estimated. The discretionary component is estimated by taking the difference between actual and estimated total accruals, as calculated in equation (6).

As an alternative measure, we also use the more data-intensive time-series version of the modified Jones model (Dechow et al., 1995). Under this model, the level of discretionary accruals for a particular firm is calculated as the difference between the firm’s total accruals and its non-discretionary accruals (NDAC), as estimated with the following equation:

N D A C_{t} = α_{0} \frac{1}{T A_{t - 1}} + α_{1} \frac{Δ R E V_{t} - Δ A R_{t}}{T A_{t - 1}} + α_{2} \frac{P P E_{t}}{T A_{t - 1}}

(7)

where

NDAC_t = Non-discretionary accruals in year t; TA_t-₁ = Total assets in year t−1; ΔREV_t = Change in revenue of firm i in year t; ΔAR_t = Change of accounts receivable of firm i in year t; PPE_t = Property plant and equipment of firm i in year t.

4. Results

4.1. Descriptive statistics

Table 2 reports the descriptive statistics.¹² It shows that the average EBITDA is 0.132 and the average adjusted EBITDA (AEBITDA) is 0.116. The average MSO is 12.54%, which is similar to the average MSO of 12.4% in the US (Cho, 1998) and 13.02% in the UK (Davies et al., 2005). The average ESO and ISO are 6.34% and 1.99%, respectively.

Table 2.

Descriptive statistics.

The following table reports descriptive statistics. Different notation used in the table are defined as follows: MSO = Percentage of ordinary shares owned by the directors of the board; ESO = Percentage of ordinary shares owned by the executive directors of the board; ISO = Percentage of ordinary shares owned by the independent outside directors of the board; USUBSP = Percentage of ordinary shares owned by the unaffiliated (excluding the directors) substantial shareholders; INV = Investment, calculated as the ratio of capital expenditure and book value of assets; BIND = Board independence calculated as the number of independent directors scaled by the size of the board; Age = Age of the board of directors; Tobin’s Q = Sum of book value of debt, preference shares and market value of equity to net book value of assets; Stock return volatility = Monthly stock return volatility; LIQ = Liquidity, calculated as the ratio of net operating cash flows and book value of assets; VOL= Volatility of earnings calculated as standard deviation of earnings of preceding five years scaled by book value of assets; .LEV = Leverage, calculated as the ratio of book value of debt and book value of total assets; MVEQ = Market value of common equity; ASST = Book value of assets, EBITDA = Earnings before interest, tax depreciation and amortisation scaled by the book value of total assets; DACC (C) = Discretionary accruals, calculated as the discretionary accruals as per Chan et al. model scaled by the book value of assets; DACC (J) = Discretionary accruals, calculated as the discretionary accruals as per modified Jones model scaled by the book value of assets; AEBITDA(C) = EBITDA − DACC(C); AEBITDA(J) = EBITDA − DACC(J).

	Mean	Median	Max	Min	Stdev
MSO (%)	12.535	2.398	87.903	0.000128	18.373
ESO (%)	6.341	0.241	75.624	0	13.164
ISO (%)	1.987	0.117	3.8	0	7.293
USUBSP (%)	37.106	34.61	94.391	0	22.547
INV	0.073	0.048	0.722	0	0.073
BIND	0.571	0.60	1	0	0.202
Tobin’s Q	1.892	1.224	8.588	0.629	1.298
Age	59	57.5	79	33	6.081
Stock return volatility	0.604	0.535	4.225	−2.016	0.662
LIQ	0.093	0.089	0.639	−0.543	0.099
VOL	0.029	0.016	0.630	0.001	0.047
LEV	0.244	0.235	4.961	0	0.247
MVEQ (in million)	2096.232	404.206	174000	3.384	8653.048
ASST (in million)	2238.082	604.332	65271.088	4.434	5863.207
EBITDA	0.132	0.128	0.604	−0.295	0.109
DACC(C)	0.016	0.009	0.626	−0.489	0.099
DACC(J)	0.012	0.007	0.727	−0.532	0.198
AEBITDA(C)	0.116	0.117	0.717	−0.592	0.148
AEBITDA(J)	0.120	0.118	0.721	−0.706	0.251

4.2. Relation between managerial ownership and performance measured by earnings

Table 3 reports the results of regression analysis for MSO and performance and ESO and performance. To address the issue of endogeneity we use an instrumental variable regression.¹³ Performance is measured by earnings (EBITDA). We test our model for alternative specifications of managerial ownership variable along with our original quadratic specification. In models 1 to 4 we explore the relations between MSO and EBITDA. In model 1 we use a linear specification of MSO. We document a negative and insignificant coefficient of MSO after controlling for other potential determinants of EBITDA. This implies an insignificant inverse relation between MSO and EBITDA. In the context of the UK, whilst examining the relation between board ownership and fraction of non-executive directors, Peasnell et al. (2003) find that a natural log of board ownership describes their data better than a quadratic specification. Therefore, motivated by the findings of the same study in model 2, we take logged value of MSO to investigate the MSO–performance relationship. However, we document a negative and insignificant coefficient of logged MSO. In model 3 we use a quadratic specification of MSO. We document a negative and significant coefficient of MSO (β= −0.158, P<0.01). We also document a positive and significant coefficient of MSO² (β = 0.352, P<0.05). The signs of MSO and MSO² imply a broadly convex relation between MSO and performance measured by EBITDA. The estimated point at which the negative MSO and EBITDA relation turns to a positive relation is when MSO is 22.4%. We find that 206 firm-year observations have MSO in excess of the estimated turning point, which corresponds to around 17.8% of the overall observations. This suggests that the estimated turning point is driven by a non-trivial number of observations. The fact that the coefficients of some other control variables are statistically significant suggests that performance is also influenced by other factors. In particular, the positive significant coefficient for investment implies that higher capital expenditure may result in better performance, since the market reacts positively to planned investment (Cho, 1998). The significant coefficient for firm size implies potential benefits through economies of scale (Anderson and Reeb, 2003). Contrary to expectations, but consistent with the findings of Anderson and Reeb (2003), we find that leverage is negatively associated with firm performance. In model 4 we use a cubic specification of MSO consistent with Short and Keasey (1999). We find a negative and significant coefficient of MSO (β= −0.775, P<0.10). However, the coefficients of MSO² and MSO³ are insignificant. This implies that cubic specification is not suitable for our sample companies. As posited, a quadratic specification in model 3 is most suitable to examine the MSO–performance relation.

Table 3.

Relations between managerial ownership and performance.

The following table reports the regression results regarding the relation between different types of managerial ownership variables and performance. Performance has been measured by earnings (EBITDA). Different notations used in the table are defined as follows: MSO = Percentage of ordinary shares owned by the directors of the board; ESO = Percentage of ordinary shares owned by the executive directors of the board; ISO = Percentage of ordinary shares owned by the independent directors of the board; NESO = Percentage of ordinary shares owned by the non-executive directors of the board; LEV = Leverage, calculated as the ratio of book value of debt to book value of total assets; INV = Investment, calculated as the ratio of capital expenditure to book value of assets; USUBSP = Percentage of ordinary shares owned by unaffiliated (excluding directors) substantial shareholders; BIND = Board independence, calculated as the number of independent directors scaled by the size of the board; FIRM AGE = Age of the firm, calculated by taking the natural log of number of years since the firm was listed on the ASX; ASST = Natural log of book value of assets; EBITDA = Earnings before interest, tax depreciation and amortisation, scaled by the book value of total assets; Year and industry dummies are not reported. P values are reported in parentheses.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
MSO	−0.043		−0.158	−0.775
MSO	(0.136)		(0.001)	(0.090)
MSO²			0.352	2.578
MSO²			(0.011)	(0.231)
MSO³				−2.161
MSO³				(0.438)
LOG(MSO)		−0.004
LOG(MSO)		(0.142)
ESO					−0.222
ESO					(0.000)
ESO²					0.536
ESO²					(0.000)
ISO						0.122
ISO						(0.713)
ISO²						−0.387
ISO²						(0.427)
NESO							0.363
NESO							(0.295)
NESO²							−0.663
NESO²							(0.173)
LEV	−0.019	−0.017	−0.022	−0.015	−0.022	−0.015	−0.019
LEV	(0.124)	(0.151)	(0.085)	(0.136)	(0.077)	(0.217)	(0.071)
INV	0.152	0.154	0.184	0.160	0.191	0.154	0.159
INV	(0.000)	(0.000)	(0.000)	(0.015)	(0.004)	(0.001)	(0.031)
USUBSP	0.021	0.019	0.002	0.015	0.002	0.022	0.027
USUBSP	(0.121)	(0.149)	(0.199)	(0.297)	(0.009)	(0.089)	(0.081)
BIND	0.010	0.011	0.009	0.006	0.017	0.021	0.027
BIND	(0.522)	(0.479)	(0.549)	(0.704)	(0.001)	(0.109)	(0.219)
FIRMAGE	0.001	0.001	0.003	0.002	0.002	0.001	0.001
FIRMAGE	(0.539)	(0.567)	(0.230)	(0.188)	(0.001)	(0.057)	(0.502)
ASST	0.005	0.001	0.012	−0.004	0.015	0.007	0.011
ASST	(0.286)	(0.863	(0.009)	(0.604)	(0.000)	(0.218)	(0.052)
Intercept	0.084	0.099	0.024	0.180	0.003	0.059	0.012
Intercept	(0.051)	(0.034)	(0.616)	(0.016)	(0.952)	(0.196)	(0.890)
Adj. R²	0.021	0.034	0.043	0.034	0.044	0.031	0.029
F stat	4.253	4.336	4.876	3.586	4.792	4.434	4.292
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	1154	1154	1154	1154	1154	1154	1154
Turning point			22.4%		20.7%
Observations above turning point			206		114

In model 5 we examine the relation between ESO and performance. We use a quadratic specification of the ESO variable. We document a negative and significant coefficient of ESO (β= −0.222, P<0.01). We also document a positive and significant coefficient of ESO² (β= 0.536, P<0.01). The signs of ESO and ESO² suggest a convex relation between ESO and EBITDA after controlling for other determinants of performance. The estimated point at which the negative ESO and EBITDA relation turns to a positive relation is when ESO is 20.7%. We find that 114 firm-year observations have ESO in excess of the estimated turning point, which corresponds to around 9.9% of the overall observations. Once again, this suggests the estimated turning point is driven by a non-trivial number of observations.

We previously argued that independent directors are less likely to be influenced by the effects of incentive alignment or entrenchment. In model 6 we replicate the analysis conducted in respect of ESO and find no significant relation between ESO and EBITDA after controlling for other determinants of performance. In model 7 we examine the impact of ownership by all non-executive directors – that is, independent directors and affiliated (grey) directors – on EBITDA. Once again, the insignificant coefficients of NESO and NESO² suggest that there is no relation between ownership by non-executive directors and performance.

Overall, these results provide an interesting contrast to those reported in the US and UK. Whilst the point at which an entrenchment effect turns and starts being dominated by an incentive effect is broadly similar to that found in Morck et al. (1988) and Davies et al. (2005), it is clear that entrenchment effects dominate at lower levels of ownership. A possible explanation of the difference in our results could be Australian institutional features that are different from the US and UK. We previously argued that managers do not need a particularly large shareholding to maintain practical control in Australia, which may allow them to derive private benefits at relatively low levels of ownership. A negative relation is likely to continue until they reach a level of ownership at which they own enough shares to have their interests aligned with the shareholders. Our results suggest that this is usually reached at about 20% ownership level. When managers own less than 20%, they may maximise their personal wealth at the expense of firm performance. Because of some other external mechanisms, incentive effects could be in operation, but that may not be strong enough to offset the entrenchment effect.

4.3. Relation between managerial ownership and performance measured by adjusted earnings

We previously argued that examination of the relationship between MSO and earnings could be biased in the event of earnings management. Accordingly, we replicate the earlier analyses using adjusted earnings/performance measured by AEBITDA, and report the results in panels A and B of Table 4. In panel A discretionary accruals are estimated using Chan et al. (2006). In model 1 we examine the relationship between MSO and AEBITDA and document a significant relation. The coefficients of MSO (β= −0.276, P<0.01) and MSO² (β=0.503, P<0.01) are significant, and their signs are negative and positive respectively. One notable feature is that the absolute size of the coefficients of the MSO variables has increased (MSO decreases from −0.158 to −0.276 and MSO² increases from 0.352 to 0.503).¹⁴ Our results suggest a convex relation between MSO and performance measured by AEBITDA.

Table 4.

Relations between MSO and performances and ESO and performance measured by adjusted earnings.

The following table reports the regression results regarding MSO and performance and ESO and performance. Performance has been measured by adjusted earnings (AEBITDA). Panel A estimates discretionary accruals as per Chan et al. model and Panel B estimates discretionary accruals as per modified Jones model. Different notations used in the table are defined as follows: MSO = Percentage of ordinary shares owned by the directors of the board; ESO = Percentage of ordinary shares owned by the executive directors of the board; LEV = Leverage, calculated as the ratio of book value of debt to book value of total assets; INV = Investment, calculated as the ratio of capital expenditure to book value of assets; USUBSP = Percentage of ordinary shares owned by unaffiliated (excluding directors) substantial shareholders; BIND = Board independence, calculated as the number of independent directors scaled by the size of the board; FIRM AGE = Age of the firm, calculated by taking the natural log of number of years since the firm was listed on the ASX; ASST = Natural log of book value of assets; EBITDA = Earnings before interest, tax depreciation and amortisation, scaled by the book value of total assets; AEBITDA = EBITDA − DACC; DACC = Discretionary accruals (measured by Chan et al. and modified Jones models) scaled by the book value of assets. Year and industry dummies are not reported. P values are reported in parentheses.

	Panel A		Panel B
	(1)	(2)	(3)	(4)
MSO	−0.276		−0.527
MSO	(0.000)		(0.007)
MSO²	0.503		1.193
MSO²	(0.002)		(0.000)
ESO		−0.327		−0.904
ESO		(0.000)		(0.075)
ESO²		0.585		1.694
ESO²		(0.001)		(0.025)
LEV	−0.047	−0.049	−0.054	−0.037
LEV	(0.007)	(0.006)	(0.109)	(0.082)
INV	0.099	0.280	1.321	1.742
INV	(0.095)	(0.124)	(0.037)	(0.036)
USUBSP	0.003	0.003	0.164	0.154
USUBSP	(0.211)	(0.496)	(0.327)	(0.292)
BIND	0.039	0.027	0.076	0.216
BIND	(0.081)	(0.389)	(0.061)	(0.537)
FIRMAGE	−0.003	−0.001	0.025	−0.008
FIRMAGE	(0.863)	(0.549)	(0.427)	(0.641)
ASST	0.008	0.010	0.007	0.009
ASST	(0.281)	(0.092)	(0.087)	(0.077)
Intercept	0.113	0.076	−0.024	−0.304
Intercept	(0.099)	(0.388)	(0.539)	(0.562)
Adj. R²	0.072	0.057	0.039	0.032
F stat	6.073	3.689	3.535	3.395
Significance	P<0.01	P<0.01	P<0.01	P<0.01
N	1154	1154	844	844
Turning point	27.5%	27.9%	22.1%	26.7%
N above turning point	173	91	141	63

In model 2 we investigate the relation between ESO and AEBITDA. We document a significant relation between ESO and performance measured by AEBITDA. The coefficients of ESO (β= −0.327, P<0.01) and ESO² (β=0.585, P<0.01) are significant and their signs are negative and positive respectively. Once again, this implies a non-monotonic convex relation between ESO and AEBITDA. The coefficient of ESO, which was −0.222 in panel A of previous table, decreases to −0.327. Similarly, the coefficient of ESO² increases from 0.536 in panel A of the previous table to 0.58, suggesting that these variables have a greater economic impact on adjusted earnings. Once again, the estimated turning point appears to be driven by a non-trivial number of observations.

In Panel B we use a modified Jones model to estimate discretionary accruals and measure adjusted earnings, replicating our analysis. The results are consistent with the findings of the first panel.

Once again we document a non-monotonic convex relation between managerial ownership and performance measured by adjusted earnings. Interestingly, the size of the coefficients of managerial ownership variables has increased. This implies that managerial ownership has greater economic impact on adjusted performance of the firm.

4.4. Relation between MSO and performance: simultaneous equations

We report the results of our simultaneous equations system in Table 5. We introduce two additional equations (MSO and INV) in our system along with the earnings equation. In panel A we measure performance by earnings (EBITDA). The results of the EBITDA regression show significant coefficients of MSO (β= −0.178, P<0.01) and MSO² (β=0.383, P<0.05). Once again, this suggests a non-monotonic convex relation between MSO and performance measured by EBITDA. The results of other variables qualitatively remain unchanged when compared with the results reported in Table 3. In the MSO regression, the coefficient of EBITDA shows a positive but insignificant value. In other words, there is no evidence to suggest that managers of better performing firms are more willing to accept shares as part of their compensation package and/or to buy shares in their own firms. This is inconsistent with the substance of the findings of Davies et al. (2005), who report that firm value (Tobin’s Q) affects MSO. We also find that MSO is affected by leverage, market value of equity and volatility. The negative significant coefficient suggests that leverage and MSO may be perceived as alternative monitoring mechanisms (Warfield et al., 1995). A negative significant coefficient of market value of equity suggests that managers are subject to wealth constraint whilst making their investments (Holderness et al., 1999). A positive significant coefficient of volatility of earnings is consistent with the findings of Demsetz and Villalonga (2001). In the investment (INV) regression we document that MSO also affects investment, which is consistent with Cho (1998).

Table 5.

Relation between MSO and performance: Simultaneous equations.

The following table reports the regression results regarding MSO and earnings and adjusted earnings. Panel B estimates discretionary accruals as per Chan et al. model and Panel C estimates discretionary accruals as per modified Jones model. Different notations used in the table are defined as follows: MSO = Percentage of ordinary shares owned by the directors of the board; LEV = Leverage, calculated as the ratio of book value of debt to book value of total assets; INV = Investment, calculated as the ratio of capital expenditure to book value of assets; USUBSP = Percentage of ordinary shares owned by unaffiliated (excluding directors) substantial shareholders; BIND = Board independence, calculated as the number of independent directors scaled by the size of the board; FIRM AGE = Age of the firm, calculated by taking the natural log of number of years since the firm was listed on the ASX; ASST = Natural log of book value of assets; MVEQ = Natural log of market value of common equity; VOL= Volatility of earnings calculated as a standard deviation of earnings of preceding five years scaled by book value of assets; LIQ = Liquidity, calculated as the ratio of net operating cash flows to book value of assets; EBITDA = Earnings before interest, tax depreciation and amortisation scaled by the book value of total assets; AEBITDA = EBITDA − DACC; DACC = Discretionary accruals (measured by Chan et al. and modified Jones models) scaled by the book value of assets. Year and industry dummies are not reported. P values are reported in parentheses.

	Panel A			Panel B			Panel C
	EBITDA	MSO	INV	AEBITDA	MSO	INV	AEBITDA	MSO	INV
EBITDA		0.240	0.136
EBITDA		(0.278)	(0.111)
AEBITDA					1.070	−0.013		0.039	0.021
AEBITDA					(0.000)	(0.352)		(0.059)	(0.026)
MSO	−0.178		0.122	−0.422		0.130	−0.513		0.179
MSO	(0.000)		(0.006)	(0.000)		(0.011)	(0.000)		(0.004)
MSO²	0.383		−0.233	0.870		−0.236	0.972		−0.287
MSO²	(0.003)		(0.003)	(0.001)		(0.006)	(0.000)		(0.007)
LEV	−0.025	−0.051		−0.045	−0.053		−0.059	−0.029
LEV	(0.043)	(0.019)		(0.008)	(0.028)		(0.376)	(0.053)
INV	0.409	−0.136		0.309	−0.269		3.238	−0.284
INV	(0.000)	(0.491)		(0.041)	(0.231)		(0.005)	(0.355)
USUBSP	0.002			0.003			0.135
USUBSP	(0.137)			(0.252)			(0.426)
BIND	0.015			0.002			0.207
BIND	(0.547)			(0.894)			(0.132)
FIRM AGE	0.003			−0.004			0.026
FIRM AGE	(0.171)			(0.905)			(0.212)
ASST	0.007			0.003			0.0291
ASST	(0.097)			(0.075)			(0.294)
VOL		0.638	0.117		0.603	0.135		0.621	0.109
VOL		(0.000)	(0.022)		(0.000)	(0.007)		(0.000)	(0.043)
LIQ		−0.061	0.039		0.734	0.070		−0.170	0.082
LIQ		(0.743)	(0.587)		(0.004)	(0.459)		(0.031)	(0.007)
MVEQ×100		−0.013			−0.013			−0.012
MVEQ×100		(0.038)			(0.052)			(0.062)
PPE			0.057			0.021			0.026
PPE			(0.117)			(0.147)			(0.107)
Intercept	0.056	0.638	0.097	0.165	0.208	0.099	−0.513		0.139
Intercept	(0.336)	(0.000)	(0.000)	(0.045)	(0.001)	(0.000)	(0.324)		(0.000)
Adj. R²	0.055	0.091	0.112	0.061	0.042	0.098	0.046	0.073	0.102
F stat	4.176	6.214	8.027	4.398	4.374	6.706	4.283	5.896	7.91
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	1154	1154	1154		1154	1154	844	844	844
Turning point	23.2%			24.2%			26.3%
N above	194			187			123

In panel B we measure performance by adjusted earnings (AEBITDA) and estimate discretionary accruals using the Chan et al. (2006) model. The result of the AEBITDA regression shows significant coefficients of MSO (β= −0.422, P<0.01) and MSO² (β=0.870, P<0.01). Once again, the signs of the coefficients support a convex relation between MSO and AEBITDA. The coefficients of other variables do not show any substantive difference to those reported in panel A. In the MSO regression, the coefficient of AEBITDA is positive and significant (β=1.070, P<0.01). Hence, in marked contrast to the earlier analysis of EBITDA, we document that adjusted earnings (AEBITDA) affect MSO, suggesting that managers of better performing firms are more willing to accept shares as part of their compensation package and/or to buy shares in their own firms. The investment (INV) regression once again shows that MSO also affects investment, which is consistent with Cho (1998).

In Panel C we use a modified Jones model to estimate discretionary accruals and measure adjusted earnings (AEBITDA). We then replicate our analysis and report results consistent with the findings reported in panel B.

4.5. Relation between ESO and performance: simultaneous equations

We examine the relation between ESO and performance using the simultaneous equations method, and report the results in Table 6. In panel A we measure our performance using EBITDA. The result of the EBITDA regression shows significant coefficients of ESO (β= −0.240, P<0.01) and ESO² (β=0.509, P<0.01) with negative and positive signs, respectively. This suggests a non-monotonic relation between ESO and EBITDA. The coefficients of the other variables remain qualitatively unchanged. The result of the ESO regression shows a positive insignificant coefficient for EBITDA, suggesting that ESO is not affected by EBITDA. Similarly, the results of investment (INV) regression suggest that ESO does not affect level of investment.

Table 6.

Relation between ESO and performance: Simultaneous equations.

The following table reports the regression results regarding ESO and earnings and adjusted earnings. Panel B estimates discretionary accruals as per Chan et al. model and Panel C estimates discretionary accruals as per modified Jones model. Different notations used in the table are defined as follows: ESO = Percentage of ordinary shares owned by the executive directors of the board; LEV = Leverage, calculated as the ratio of book value of debt to book value of total assets; INV = Investment, calculated as the ratio of capital expenditure to book value of assets; USUBSP = Percentage of ordinary shares owned by unaffiliated (excluding directors) substantial shareholders; BIND = Board independence, calculated as the number of independent directors scaled by the size of the board; FIRM AGE = Age of the firm, calculated by taking the natural log of number of years since the firm was listed on the ASX; ASST = Natural log of book value of assets; MVEQ = Natural log of market value of common equity; VOL= Volatility of earnings, calculated as a standard deviation of earnings of preceding five years scaled by book value of assets; LIQ = Liquidity, calculated as the ratio of net operating cash flows to book value of assets; EBITDA = Earnings before interest, tax depreciation and amortisation scaled by the book value of total assets; DACC (C) = Discretionary accruals, calculated as the discretionary accruals as per Chan et al. model scaled by the book value of assets; DACC (J) = Discretionary accruals, calculated as the discretionary accruals as per modified Jones model scaled by the book value of assets; AEBITDA(C) = EBITDA − DACC(C); AEBITDA(J) = EBITDA − DACC(J); Year and industry dummies are not reported. P values are reported in parentheses.

	Panel A			Panel B			Panel C
	EBITDA	ESO	INV	AEBITDA(C)	ESO	INV	AEBITDA(J)	ESO	INV
EBITDA		0.018	0.129
EBITDA		(0.915)	(0.130)
AEBITDA					0.541	0.050		0.031	0.027
AEBITDA					(0.004)	(0.592)		(0.054)	(0.036)
ESO	−0.240		0.089	−0.327		0.083	−1.029		0.189
ESO	(0.005)		(0.129)	(0.008)		(0.162)	(0.007)		(0.027)
ESO²	0.509		−0.185	0.589		−0.166	1.717		−0.267
ESO²	(0.003)		(0.128)	(0.016)		(0.174)	(0.037)		(0.074)
LEV	−0.026	−0.048		−0.049	−0.048		−0.054	−0.046
LEV	(0.031)	(0.003)		(0.005)	(0.006)		(0.217)	(0.047)
INV	0.411	0.128		0.332	0.061		1.024	−0.206
INV	(0.001)	(0.384)		(0.028)	(0.707)		(0.028)	(0.329)
USUBSP	0.003			0.003			0.176
USUBSP	(0.196)			(0.293)			(0.429)
BIND	0.029			0.032			0.194
BIND	(0.003)			(0.357)			(0.506)
FIRM AGE	0.003			−0.001			0.043
FIRM AGE	(0.270)			(0.739)			(0.426)
ASST	0.009			0.001			0.061
ASST	(0.055)			(0.075)			(0.427)
VOL		−0.056	0.071		−0.058	0.092		0.694	0.129
VOL		(0.551)	(0.014)		(0.550)	(0.051)		(0.009)	(0.027)
LIQ		−0.031	0.046		0.454	0.104		−0.189	0.069
LIQ		(0.822)	(0.516)		(0.011)	(0.235)		(0.071)	(0.027)
MVEQ×100		−0.012			−0.010			−0.011
MVEQ×100		(0.017)			(0.027)			(0.109)
PPE			0.027			0.034			0.039
PPE			(0.103)			(0.067)			(0.154)
Intercept	0.020	0.063	0.106	0.076	0.114	0.106	−0.703	0.102	0.127
Intercept	(0.731)	(0.046)	(0.000)	(0.366)	(0.002)	(0.000)	(0.269)	(0.101)	(0.000)
Adj. R²	0.024	0.027	0.112	0.049	0.052	0.114	0.044	0.141	0.097
F stat	4.317	3.981	4.105	4.602	4.001	3.875	3.926	4.679	4.027
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	1154	1154	1154	1154	1154	1154	844	844	844
Turning point	23.6%			27.8%			29.8%
N above turning point	105			92			49

In panel B we examine the same relation, measuring performance by adjusted earnings (AEBITDA). We use the Chan et al. (2006) model to estimate discretionary accruals. We document significant coefficients of ESO (β= −0.327, P<0.01) and ESO² (β=0.589, P<0.01) in the AEBITDA regression. The signs of those two variables are consistent with our previous findings and provide evidence of a convex relationship between ESO and adjusted earnings. The results of ESO regression show a positive, significant coefficient of AEBITDA (β=0.541, P<0.01), implying that AEBITDA also affects ESO. That is, there is a bidirectional relation between ESO and AEBITDA.

In panel C we use a modified Jones model to estimate discretionary accruals and measure adjusted earnings, replicating our analysis. The results are consistent with the findings of the previous panel.

5. Further analysis

5.1. Alternative measure of performance

Consistent with prior studies we also use Tobin’s Q to measure firm performance and examine the same relation accordingly. We report the results in Table 7. In panels A and B we examine the relation between MSO and performance using an instrumental variable regression and simultaneous equations system, respectively. In both panels we document a non-monotonic convex relation between MSO and Tobin’s Q. We also find that MSO is affected positively by performance measured by Tobin’s Q. In other words, consistent with the findings of Davies et al. (2005), we find a bidirectional relation between MSO and Tobin’s Q.

Table 7.

Relation between MSO and performance and ESO and performance: Alternative measure of performance.

The following table reports the regression results regarding MSO and Tobin’s Q. Panels A and C reports the results using instrumental variable regression and panels B and D reports the results of simultaneous equations system. Different notations used in the table are defined as follows: MSO = Percentage of ordinary shares owned by the directors; ESO = Percentage of ordinary shares owned by the executive directors of the board; LEV = Leverage, calculated as the ratio of book value of debt to book value of total assets; INV = Investment, calculated as the ratio of capital expenditure to book value of assets; USUBSP = Percentage of ordinary shares owned by unaffiliated (excluding directors) substantial shareholders; BIND = Board independence calculated as the number of independent directors scaled by the size of the board; FIRM AGE = Age of the firm, calculated by taking the natural log of number of years since the firm was listed on the ASX; ASST = Natural log of book value of assets; MVEQ = Natural log of market value of common equity; VOL= Volatility of earnings calculated as a standard deviation of earnings of preceding five years scaled by book value of assets; LIQ = Liquidity, calculated as the ratio of net operating cash flows to book value of assets; Tobin’s Q = Sum of book value of debt, preference shares and market value of equity to net book value of assets. Year and industry dummies are not reported. P values are reported in parentheses.

	Panel A	Panel B			Panel C	Panel D
		Tobin’s Q	MSO	INV		Tobin’s Q	ESO	INV
Tobin’s Q			0.018	0.019			0.015	0.019
Tobin’s Q			(0.059)	(0.000)			(0.025)	(0.000)
MSO	−2.273	−2.518		0.147
MSO	(0.000)	(0.000)		(0.002)
MSO²	4.761	4.481		−0.289
MSO²	(0.000)	(0.000)		(0.000)
ESO					−1.515	−1.489		0.106
ESO					(0.003)	(0.086)		(0.088)
ESO²					3.205	3.716		−0.259
ESO²					(0.000)	(0.034)		(0.045)
LEV	−0.052	−0.076	−0.039		−0.069	−0.071	−0.042
LEV	(0.701)	(0.525)	(0.054)		(0.622)	(0.557)	(0.005)
INV	3.635	3.768	−0.147		1.893	3.561	0.113
INV	(0.000)	(0.000)	(0.325)		(0.000)	(0.000)	(0.304)
USUBSP	−0.116	−0.161			−0.248	−0.185
USUBSP	(0.439)	(0.387)			(0.000)	(0.325)
BIND	0.417	0.443			0.254	0.298
BIND	(0.321)	(0.071)			(0.087)	(0.215)
FIRM AGE	−0.106	−0.098			−0.131	−0.102
FIRM AGE	(0.000)	(0.026)			(0.000)	(0.026)
ASST	−0.286	−0.267			−0.326	−0.245
ASST	(0.000)	(0.000)			(0.000)	(0.000)
VOL			0.094	0.065			−0.172	0.044
VOL			(0.355)	(0.153)			(0.221)	(0.331)
LIQ			−0.067	0.051			0.058	0.049
LIQ			(0.164)	(0.001)			(0.029)	(0.002)
MVEQ×100			−0.019				−0.012
MVEQ×100			(0.002)				(0.006)
PPE				0.032				0.037
PPE				(0.187)				(0.129)
Intercept	4.501	4.859	0.064	0.082	5.414	4.577	0.035	0.086
Intercept	(0.000)	(0.000)	(0.109)	(0.000)	(0.000)	(0.000)	(0.236)	(0.000)
Adj. R²	0.147	0.151	0.068	0.119	0.184	0.141	0.042	0.113
F stat	8.833	6.018	7.069	5.614	8.447	6.917	7.324	5.961
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	1154	1154	1154	1154	1154	1154	1154	1154
Turning point	23.1%	28.1%			23.6%	20%
N above turning point	196	166			105	123

In panels C and D we investigate the relation between ESO and Tobin’s Q using an instrumental variable regression and simultaneous equations system, respectively. Consistent with our main findings, we once again document a convex relationship between ESO and Tobin’s Q in panels C and D. We also document a bidirectional relationship between ESO and Tobin’s Q in panel D.

5.2. Sub-sample analysis

We partitioned our sample into four different sub-samples based on time periods – 2000 to 2003, 2004 to 2006, 2000 to 2002 and 2003 to 2006 – and replicated the original analysis. The purpose of partitioning the sample is to test any impact of the major corporate regulatory changes (for example, the introduction of ASX corporate governance guidelines in 2003) that took place during our study period. The results are reported in Table 8. Panels A, B, C and D include the sub-samples in respect of 2000 to 2003, 2004 to 2006, 2000 to 2002 and 2003 to 2006, respectively. We use an instrumental variable regression method for this analysis. We do not report the coefficients of control variables in the interest of brevity. The results that we document for the sub-sample periods are qualitatively similar to the results in respect of the whole sample.

Table 8.

Relations between MSO and performance and ESO and performance: Sub-sample analysis.

The following table reports the regression results regarding MSO and performance and ESO and performance. Performance has been measured by earnings (EBITDA) and adjusted earnings (AEBITDA). Panels A, B, C and D include the sub-samples of 2000 to 2003, 20004 to 2006, 2000 to 2002 and 2003 to 2006 respectively. Different notations used in the table are defined as follows: MSO = Percentage of ordinary shares owned by the directors of the board; ESO = Percentage of ordinary shares owned by the executive directors of the board; EBITDA = Earnings before interest, tax depreciation and amortisation scaled by the book value of total assets; DACC (C) = Discretionary accruals, calculated as the discretionary accruals as per Chan et al. model scaled by the book value of assets; DACC (J) = Discretionary accruals, calculated as the discretionary accruals as per modified Jones model scaled by the book value of assets; AEBITDA(C) = EBITDA − DACC(C); AEBITDA(J) = EBITDA − DACC(J). Control variables and Year and industry dummies are not reported. P values are reported in parentheses.

Panel A	EBITDA		AEBITDA(C)		AEBITDA(J)
MSO	−0.139		−0.306		−0.579
MSO	(0.032)		(0.000)		(0.000)
MSO²	0.305		0.702		0.988
MSO²	(0.051)		(0.000)		(0.000)
ESO		−0.291		−0.609		−0.588
ESO		(0.002)		(0.000)		(0.000)
ESO²		0.678		1.061		1.007
ESO²		(0.000)		(0.000)		(0.003)
Intercept	0.061	0.029	0.189	0.036	0.099	0.011
Intercept	(0.301)	(0.619)	(0.028)	(0.638)	(0.249)	(0.890)
Adj. R²	0.031	0.037	0.109	0.052	0.061	0.041
F stat	4.129	3.786	4.257	3.727	4.019	3.421
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	670	670	670	670	489	489

Panel B	EBITDA		AEBITDA(C)		AEBITDA(J)
MSO	−0.151		−0.532		−0.383
MSO	(0.007)		(0.000)		(0.000)
MSO²	0.203		0.598		0.433
MSO²	(0.025)		(0.000)		(0.000)
ESO		−0.277		−0.521		−0.496
ESO		(0.001)		(0.000)		(0.000)
ESO²		0.647		0.921		0.871
ESO²		(0.000)		(0.002)		(0.000)
Intercept	0.089	0.049	0.178	0.038	0.089	0.012
Intercept	(0.077)	(0.307)	(0.015)	(0.605)	0.225	0.875
Adj. R²	0.018	0.031	0.096	0.042	0.053	0.035
F stat	4.011	3.624	4.173	3.501	3.921	3.506
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	484	484	484	484	355	355

Panel C	EBITDA		AEBITDA(C)		AEBITDA(J)
MSO	−0.207		−0.210		-0.128
MSO	(0.005)		(0.029)		(0.086)
MSO²	0.287		0.228		0.074
MSO²	(0.022)		(0.157)		(0.049)
ESO		−0.265		−0.236		−0.210
ESO		(0.006)		(0.063)		(0.097)
ESO²		0.557		0.487		0.436
ESO²		(0.005)		(0.054)		(0.088)
Intercept	0.053	0.022	0.026	0.036	−0.015	−0.069
Intercept	(0.385)	(0.709)	(0.769)	(0.643)	(0.854)	(0.381)
Adj. R²	0.031	0.040	0.026	0.029	0.034	0.029
F stat	4.005	3.674	4.326	3.587	4.217	3.409
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	504	504	504	504	372	372

Panel D	EBITDA		AEBITDA(C)		AEBITDA(J)
MSO	−0.275		−0.219		−0.179
MSO	(0.005)		(0.067)		(0.014)
MSO²	0.395		0.269		0.249
MSO²	(0.018)		(0.191)		(0.047)
ESO		−0.343		−0.262		−0.254
ESO		(0.006)		(0.089)		(0.105)
ESO²		0.647		0.517		0.492
ESO²		(0.009)		(0.062)		(0.112)
Intercept	0.061	0.042	0.032	−0.009	0.019	−0.037
Intercept	(0.446)	(0.595)	(0.749)	(0.923)	(0.856)	(0.712)
Adj. R²	0.035	0.033	0.027	0.014	0.034	0.028
F stat	4.104	3.417	4.253	3.571	3.904	3.696
Significance	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01	P<0.01
N	650	650	504	504	472	472

5.3 Other

We performed further analysis to check the robustness of our results. First, we eliminated all accruals by using cash flow from operations as an alternate measure of performance in all our models, and found qualitatively similar results.¹⁵ Second, we repeated all the analyses using a random effect model; the results were qualitatively similar to our original results. Third, we used an alternative approach to control for industry differences. As is typical in Australia, around 16% of our sample companies are resource companies. Accordingly, we also use a resource dummy in all the regressions, and we document a significantly negative coefficient for the resource dummy variable. This suggests that non-resource companies perform better than resource companies. However, our results for the managerial ownership variables (MSO, ESO and ISO) remain unchanged.

6. Conclusion

We examined the relationship between managerial share ownership and Australian firm performance using earnings as a measure of performance. Our results suggest a non-monotonic, convex relation between MSO and earnings and MSO and adjusted earnings after addressing the issue of endogeneity and reverse-causality. Our results for ESO also show a similar pattern but, as posited, we do not find any evidence that ISO affects performance. We argue that our finding of a broadly convex relationship between MSO and performance is consistent with certain Australian institutional features, suggesting that managers do not need a particularly large shareholding to derive private benefits of control. Consistent with the above, our empirical findings suggest that in Australia, a negative ownership–performance relation dominates at lower levels of ownership. At this level some incentive effects could be in operation, but are likely to be dominated by entrenchment effects. At 20%–30% of ownership level, we see a relation consistent with incentive alignment. Overall, our results suggest to regulators and investors that, unlike in the US, relatively low levels of MSO may not be an effective governance mechanism to align incentives and improve operating performance in Australia. A broader implication is that the ownership–performance relation is context-specific, with wider corporate governance conditions impacting on the theorised incentive effects.

We also contribute to the literature by arguing that executive and independent directors have different incentives that may impact upon the relation between ownership and operating performance measured by earnings. Our results support such differential incentives and imply that independent directors in Australia are immune to the theorised incentive alignment or entrenchment effects associated with share ownership. These results provide some validation of the Australian regulators’ focus on board independence, and may also influence regulators in other jurisdictions as they look to the ASX corporate governance principles as a guide to developing their own policy and codes of best practice.

We also document that performance affects MSO and ESO – that is, the relation is bidirectional, but only when adjusted earnings is the measure of performance. We suggest that our failure to see a bidirectional relation between MSO and unadjusted earnings may be due to the distortion caused by discretionary accruals. In contrast, we document that adjusted earnings, reflecting ‘true’ performance, affect MSO, suggesting that managers of better performing firms are more willing to accept shares as part of their compensation package and/or to buy shares in their own firms. Overall, this supports our use of adjusted earnings as an additional measure of firm performance in this context.

The recognition of employee stock option plans in Australian financial statements was made mandatory with the introduction of ‘AASB 2: Share-based Payment’ in 2005. A useful extension of this research would be to consider the impact of managerial stock option plans in the relationship examined.

Footnotes

Acknowledgements

We would especially like to thank an anonymous reviewer for many helpful comments. This paper has also benefited from the helpful comments and suggestions of Mark Caylor, Yuan Ding, John Hillier, Shams Pathan, Alan Ramsay, Donald Stokes, Steve Young and participants at the 2009 Prato PhD Accounting and Finance Symposium, the 2009 FINSIA-MCFS Banking and Finance Conference, the 2010 American Accounting Association Conference, the 2010 AFAANZ Conference, the 2010 Finance and Corporate Governance Conference, Melbourne as well as research seminars at Monash University and the University of Western Australia. The usual caveat applies.

Date of acceptance of final transcript: 5 September 2012.

Accepted by Associate Editor, Peter Clarkson, (Accounting).

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Anderson

Reeb

(2003) Family ownership and firm performance: Evidence from the S&P 500. Journal of Finance 58: 1301–1328.

Australian Stock Exchange Corporate Governance Council (2003) Principles of good corporate governance and best practice recommendations. Sydney: ASX.

Chan

LKC

Jegadeesh

. (2006) Earnings quality and stock return. Journal of Business 79: 1041–1082.

Cho

M-H

(1998) Ownership structure, investment and the corporate value: An empirical analysis. Journal of Financial Economics 47: 103–121.

Core

Guay

Rusticus

(2006) Does weak governance cause weak stock returns? An examination of firm operating performance and investors’ expectations. Journal of Finance 61: 655–687.

Cornett

Marcus

Tehranian

(2008) Corporate governance and pay-for-performance: The impact of earnings management. Journal of Financial Economics 87: 357–373.

Cotter

Shivdasani

Zenner

(1997) Do independent directors enhance target shareholder wealth during tender offers? Journal of Financial Economics 43: 195–218.

Craswell

Taylor

Saywell

(1997) Ownership structure and corporate performance. Pacific-Basin Finance Journal 5: 301–323.

Dahya

Dimitrov

McConell

(2008) Dominant shareholders, corporate boards, and corporate value: A cross-country analysis. Journal of Financial Economics 87: 73–100.

10.

Davidson

Mackinnon

(1993) Estimation and Inferences in Econometrics. Oxford: Oxford University Press.

11.

Davies

Hillier

McClogan

(2005) Ownership structure, managerial behaviour and corporate value. Journal of Corporate Finance 11: 645–660.

12.

Dechow

Sloan

Sweeny

(1995) Detecting earnings management. Accounting Review 70: 193–225.

13.

Demsetz

(1983) The structure of ownership and theory of the firm. Journal of Law & Economics 26: 375–390.

14.

Demsetz

Villalonga

(2001) Ownership structure and corporate performance. Journal of Corporate Finance 7: 209–233.

15.

Dignam

Galanis

(2004) Australia inside-out: The corporate governance system of the Australian listed market. Melbourne University Law Review 28: 623–653.

16.

Fahlenbrach

Stulz

(2009) Managerial ownership dynamics and firm value. Journal of Financial Economics 92: 342–361.

17.

Fama

Jensen

(1983) Separation of ownership and control. Journal of Law & Economics 26: 327–349.

18.

Greene

(2003) Econometric Analysis. 5th ed. Upper Saddle River, NJ: Prentice Hall Inc.

19.

Healy

Wahlen

(1999) A review of the earnings management literature and its implications for standard setting. Accounting Horizons 13: 365–383.

20.

Hermalin

Weisbach

(1991) The effect of board composition and direct incentives on firm performance. Financial Management 20: 101–112.

21.

Himmelberg

Hubbard

Palia

(1999) Understanding the determinants of managerial ownership and the link between ownership and value. Journal of Financial Economics 53: 353–384.

22.

Holderness

Kroszner

Sheehan

(1999) Were good old days that good? Changes in managerial stock ownership since the great depression. Journal of Finance 54: 435–469.

23.

Jensen

Meckling

(1976) Theory of firm: Managerial behaviour, firm agency costs and ownership structure. Journal of Financial Economics 3: 305–360.

24.

Lamba

Stapledon

(2001) The determinants of corporate ownership structure: Australian evidence. Working paper no. 20, Faculty of Law, University of Melbourne. Available: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=279015 (accessed 10 September 2007).

25.

La Porta

Lopez-de-Silanes

Shleifer

(1999) Corporate ownership around the world. Journal of Finance 54: 471–517.

26.

McConnell

Servaes

(1990) Additional evidence on equity ownership and corporate value. Journal of Financial Economics 27: 595–612.

27.

Morck

Shleifer

Vishny

(1988) Management ownership and market valuation: An empirical analysis. Journal of Financial Economics 20: 293–315.

28.

Mura

(2007) Do non-executive directors and institutional investors have minds of their own? Evidence on performance of UK firms. Financial Management 36: 81–112.

29.

Nenova

(2003) The value of corporate voting rights and control: A cross-country analysis. Journal of Financial Economics 68: 325–351.

30.

Palia

(2001) The endogeneity of managerial compensation in firm valuation: A solution. Review of Financial Studies 14: 735–764.

31.

Peasnell

Pope

Young

(2003) Managerial equity ownership and the demand for outside directors. European Financial Management 9: 231–250.

32.

Short

Keasey

(1999) Managerial ownership and the performance of firms: Evidence from the UK. Journal of Corporate Finance 5: 79–101.

33.

Warfield

Wild

(1995) Managerial ownership, accounting choices, and informativeness of earnings. Journal of Accounting & Economics 20: 61–91.

34.

Welch

(2003) The relationship between ownership structure and performance in listed Australian companies. Australian Journal of Management 28: 287–305.

35.

White

(1980) A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48: 817–838.