Managerial incentives and corporate leverage: evidence from the United Kingdom
The authors would like to thank Turalay Kenc, Alexandros Kostakis, Peter N. Smith, an anonymous referee and seminar participants at the European Financial Management Association Annual Meeting (2008) for helpful comments and suggestions.
Abstract
This paper investigates the effect of managerial incentives and corporate governance on capital structure using a large sample of UK firms during the period 1999–2004. The analysis revolves around the view that managerial incentives are important in determining a firm's leverage. However, we argue that the exact impact of these incentives on leverage is likely to be determined by firm-specific governance characteristics. To conduct our investigation, we construct a simple corporate governance measure using detailed ownership and governance information. We present evidence of a significant non-monotonic relationship between executive ownership and leverage. There is also strong evidence suggesting that corporate governance practices have a significant impact on leverage. More importantly, the results reveal that the nature of the relation between executive ownership and leverage depends on the firm's corporate governance structure.
1. Introduction
The relationship between corporate governance and leverage has been extensively investigated in the literature (see e.g. Stulz, 1990; Berger et al., 1997; Litov and Kose, 2006; Kayhan, 2008). Prior research suggests that there are two important aspects of the interaction between governance and leverage. First, corporate leverage can act as a self-disciplining internal governance mechanism to mitigate the costs of the manager–shareholder agency conflict (Jensen and Meckling, 1976; Jensen, 1986). Second, better governance is associated with lower costs of debt financing (see e.g. Cremers et al., 2004; Klock et al., 2005) and, therefore, plays an important role in determining a firm's choice of capital structure. According to this view, although leverage does not necessarily play a significant governance role, strong corporate governance, all else equal, leads to higher leverage.
These two attributes of the governance–leverage relationship lead to a number of interesting predictions and raise further research questions (see e.g. Berger et al., 1997; Brailsford et al., 2002; Kayhan, 2005). For example, it is expected that firms with entrenched managers are likely to have low leverage. Moreover, in the absence of strong corporate governance, managers can voluntarily raise more debt in an attempt to increase firm value by reducing the expected costs of the manager–shareholder agency conflict. However, it is also expected that the agency benefits of leverage decrease when the overall governance quality within the firm is good. It is then less likely that firms with stronger corporate governance will use higher leverage for governance purposes.
The objective of the present paper is to provide further insights into these predictions by investigating empirically the interactions between corporate governance, managerial incentives and leverage. The empirical analysis of this paper is conducted in two stages. In the first stage, we investigate the effects of corporate governance and managerial incentives on leverage separately. We hypothesize that in the presence of information asymmetry and costly agency problems, the use of better corporate governance practices facilitates access to debt by reducing its cost, leading to a significant positive relationship between effective corporate governance and leverage (Berkovitch and Israel, 1996; Klock et al., 2005).
Additionally, we expect a strong link between executive equity ownership and corporate leverage. In line with the alignment and the entrenchment effects of managerial ownership on firm value, we expect a non-monotonic relationship between executive ownership and leverage (see Brailsford et al., 2002). Using executive ownership to proxy for managerial incentives, we predict that the relationship between executive ownership and leverage is positive at lower levels of ownership. This prediction is based on the view that greater executive ownership leads to a better alignment of the interests of managers and outside investors (alignment effect) and, hence, a lower cost of debt financing, which in turn enables firms to raise more debt financing. However, high levels of executive ownership strengthen managerial discretion, possibly leading to managerial entrenchment, and reinforcing managerial incentives to choose a lower level of leverage (entrenchment effect). This is because managers are normally expected to avoid high leverage in an attempt to minimize the discipline provided by debt and to reduce the probability of bankruptcy.
In the second stage of our empirical investigation, we test the extent to which corporate governance and managerial incentives interact in influencing corporate financing policy. It is likely that the entrenchment effect of executive ownership on leverage is less pronounced in firms with effective internal corporate governance mechanisms. We argue that even if the incentives of managers to adopt a suboptimal capital structure policy remain intact, their ability to execute these incentives would be limited in firms with strong corporate governance. Similarly, corporate governance is likely to affect the alignment effect of executive ownership on leverage because corporate governance characteristics, such as board structure and concentrated (non-managerial) ownership, may act as substitutes for executive equity ownership in influencing leverage. Put differently, the value of leverage as a self-disciplining governance mechanism diminishes in the presence of strong corporate governance.
Our paper is similar in spirit to the studies by Friend and Lang (1988), Brailsford et al. (2002) and Pindado and De La Torre (2005). Specifically, Friend and Lang (1988) show that the existence of large non-managerial investors in US corporations prevents managers from choosing leverage levels that serve their own interests. Brailsford et al. (2002) and Pindado and De La Torre (2005) also report significant interaction effects between managerial ownership, ownership concentration and leverage for a sample of Australian and Spanish firms, respectively. However, in contrast to earlier studies, our analysis focuses on UK companies. Furthermore, in addition to ownership concentration, we also consider board size and board composition as additional governance variables that are also likely to influence the relationship between executive ownership and leverage. Last but not least, we use principal component analysis (PCA) in measuring corporate governance, which helps avoid problems that may arise from the potential interdependence between corporate governance and control variables (see e.g. Agarwal and Knoeber, 1996).
We believe that our study provides important insights into at least two important related research questions that have not been addressed before. First, if there is a non-linear relationship between executive ownership and leverage, does the nature of this relationship depend on the corporate governance environment in which firms operate? Second, if there are devices that may act as substitutes to leverage in monitoring and disciplining managers, should one still expect a significant relationship between executive ownership and leverage?
Our empirical investigation is based on a large sample of non-financial UK listed firms over the period 1999–2004. The UK market bears several important characteristics that, we believe, may have significant implications with regard to the capital structure decision of firms. For example, the existing UK takeover code and the strong minority protection laws make the accumulation of controlling blocks expensive. The most significant holders in the United Kingdom are financial institutions, but there is a great deal of evidence that they do not take a very active role in corporate governance. Furthermore, UK boards of directors are generally seen as providing a weak monitoring function, mainly due to their weak power to enforce fiduciary responsibilities on directors (see e.g. Franks et al., 2001). All these arguments lead to the hypothesis that managers in the United Kingdom enjoy more freedom to pursue their own interests that may include deviating from an optimal leverage ratio.1
The analysis of the present paper reveals several important findings. First, the relationship between leverage and executive ownership is non-monotonic. We observe a positive relationship at moderate levels of executive ownership but the relationship becomes negative at higher levels. Second, the results suggest that firms with better corporate governance practices have a greater ability to issue debt financing. More importantly, there is strong evidence that the significant relationship between leverage and managerial incentives holds mainly for firms with weak corporate governance. Finally, consistent with the results reported in recent capital structure studies, our dynamic panel data regressions show that UK firms partially adjust towards a target leverage ratio.
The remainder of the present paper is organized as follows. In Section 2, we formulate our empirical hypotheses whereas in Section 3, we provide a brief description of the data and the variable construction. Section 4 presents the results and Section 5 concludes.
2. Managerial incentives, corporate governance and leverage
This section develops the hypotheses regarding the effects of managerial incentives and corporate governance on leverage.
2.1. Managerial incentives and leverage
The relationship between managerial ownership and leverage is often ambiguous. It is argued that managers usually have incentives to keep borrowing at lower levels than optimal because this reduces the probability of bankruptcy and provides managers with a greater discretion over the use of excess cash (Jensen, 1986). A lower than optimal leverage ratio, then, is expected to adversely affect firm value.
Managerial ownership serves as a mechanism that can potentially help align the interests of managers with those of shareholders (see e.g. Jensen and Meckling, 1976; McConnell and Servaes, 1990). According to this view, managers with greater ownership are less likely to engage in value-decreasing activities because they bear part of the costs of their actions. These arguments lead to a predicted positive relationship between managerial ownership and leverage (incentive-alignment effect).
However, the relationship between leverage and managerial ownership is likely to be non-monotonic. Specifically, when managers have a significant ownership in the firm, they also have greater discretion over its capital structure decision. For example, an increase in managerial ownership may lead to an increase in managerial opportunism and, hence, decreased debt levels (Brailsford et al., 2002). The above discussion implies that an entrenchment effect, which follows the initial alignment effect of managerial ownership, sets in at relatively high levels of managerial ownership, resulting in a curvilinear ownership–leverage relationship.2
2.2. Corporate governance and leverage
It is acknowledged that the corporate governance structure of firms has a significant impact on their capital structures (see e.g. Berkovitch and Israel, 1996; Klock et al., 2004). It is argued that effective corporate governance attributes, such as the presence of large shareholders among owners and an effective board of directors, signal the firm's governance quality to its prospective lenders. Consequently, firms with strong corporate governance are expected to have easier access to capital markets and, in general, are subject to lower expected agency costs of debt. This implies that these firms can afford greater leverage. Therefore, we expect a positive relationship between the quality of corporate governance and leverage.
2.3. Interaction between corporate governance and managerial incentives
The expected relationship between corporate governance and leverage mentioned above does not consider the interaction between corporate governance and managerial incentives in influencing leverage. We argue that allowing governance mechanisms and managerial incentives to interact may lead to different predictions for the overall impact of managerial incentives on leverage. For example, we do not expect the non-linear relationship between leverage and managerial ownership to hold for firms with strong corporate governance mechanisms in place. Put differently, corporate governance and executive equity ownership can act as substitutes in determining leverage. A possible explanation for this result is that the potential agency costs of the manager–shareholder conflict are lower in firms with strong corporate governance and, therefore, the disciplining and monitoring roles of leverage are weakened. This leads to a weak link between managerial ownership and leverage. Additionally, the ability of managers to use their discretion would also be limited in the presence of effective internal corporate governance mechanisms. We argue that the hypothesized non-linear relationship between executive ownership and leverage is held more strongly in firms with weak corporate governance.
3. Data and variable construction
In this section, we describe our data sources, explain how we construct the corporate governance measure using PCA, and provide a discussion as to why we use two different definitions of leverage (namely, book and market leverage).
3.1. Data
The data used in this study are obtained from two different sources. We use Thomson DataStream to collect accounting and market data for the sample period 1999–2004, whereas information on the level of executive ownership, ownership concentration and board characteristics is derived from the Hemscott Guru Academic Database.
We merge data provided from DataStream and Hemscott and exclude financial firms and utilities from the sample. Then, we drop missing firm-year observations and outliers (i.e. those observations that lie below the 1st percentile and above the 99th percentile for each variable). These criteria lead to a final sample of 956 firms for our empirical analysis. Table 1 provides the definitions of the variables used in the paper, whereas Table 2 summarizes the key descriptive statistics over the sample period. We observe that the average book leverage for UK companies is 16.79 per cent over the period 1999–2004. However, the observed market leverage during the sample period is much lower, at 9.73 per cent. The average value of the tangibility is 26.65 per cent and the market-to-book ratio is 2.12. The profitability ratio of the average firm is 0.02 (or 2 per cent per annum) during the sample period. As for the ownership and the board characteristics, executive directors hold a significant fraction of the total shares of firms, 11.60 per cent on average, whereas the average ownership concentration is 35.61 per cent. Finally, the average proportion of non-executive directors is 49.52 per cent and the number of directors for the average firm is about 7.
| Variable | Definition | Source |
|---|---|---|
| Dependent variables | ||
| LEVERAGE (BOOK) | The ratio of the book value of total debt to the book value of total assets (%) | DataStream |
| LEVERAGE (MKT) | The ratio of the book value of total debt to the sum of the book value of total assets and the market value of total equity (%) | DataStream |
| Independent variables | ||
| TANGIBILITY | The ratio of total fixed assets to the book value of total assets (%) | DataStream |
| MKTBOOK | The ratio of book value of total assets minus the book value of equity plus the market value of equity to book value of assets | DataStream |
| SIZE | Total assets (in logarithm) | DataStream |
| PROFITABILITY | The ratio of earnings before interest, taxes, depreciation and amortization (EBITDA) to the book value of total assets | DataStream |
| EXECOWNER | The percentage of equity ownership held by executive directors (%) | Hemscott |
| CONCENTR | The sum of the stakes of firm's shareholders (other than managers) with equity ownership greater than 3 per cent (%) | Hemscott |
| NON-EXEC | The ratio of the number of non-executive directors to the number of total directors on the board (%) | Hemscott |
| BOARD SIZE | The total number of directors on the board | Hemscott |
- Notes: This table provides the definitions of the main variables used in our analysis as well as some information on our data sources. DataStream database provides accounting and market data. Hemscott Guru Academic database provides financial data for the UK's top 300 000 companies and detailed data on all directors of UK listed companies.
| Mean | SD | Minimum | 25% | Median | 75% | Maximum | |
|---|---|---|---|---|---|---|---|
| LEVERAGE (BOOK) | 16.79 | 15.29 | 0.00 | 3.68 | 13.72 | 26.31 | 77.84 |
| LEVERAGE (MKT) | 9.73 | 10.02 | 0.00 | 1.50 | 7.05 | 14.70 | 61.10 |
| TANGIBILITY | 26.65 | 24.75 | 0.00 | 6.93 | 18.22 | 40.17 | 98.64 |
| MKTBOOK | 2.12 | 1.78 | 0.17 | 1.12 | 1.55 | 2.49 | 19.58 |
| SIZE | 10.89 | 2.24 | 6.22 | 9.20 | 10.65 | 12.26 | 18.87 |
| PROFITABILITY | 0.02 | 0.21 | −1.00 | −0.04 | 0.08 | 0.14 | 0.43 |
| EXECOWNER | 11.60 | 16.03 | 0.00 | 0.36 | 4.04 | 17.80 | 89.55 |
| CONCENTR | 35.61 | 26.62 | 0.00 | 21.69 | 35.68 | 49.71 | 92.88 |
| NON-EXEC | 49.52 | 13.81 | 0.00 | 40.00 | 50.00 | 58.74 | 100.00 |
| BOARDSIZE | 6.77 | 2.19 | 3.00 | 5.17 | 6.33 | 8.00 | 15.00 |
- Notes: This table provides descriptive statistics for the main variables used in our analysis. The means of the variables are measured over the period 1999–2004. Definitions for all the variables are provided in Table 1.
3.2. Construction of the governance measure
In our analysis, we use PCA to aggregate individual governance characteristics into a single governance index.3 PCA helps control for potential multicollinearity problems that may arise when several governance variables in a cross-sectional regression are included independently (also see Agrawal and Knoeber, 1996, for further discussion). Additionally, PCA automatically produces weights so that the governance measure will explain as much of the variance in the group of corporate governance attributes and, therefore, does not require the ex ante determination of the weights. Most of the earlier studies that attempt to establish corporate governance ranking variables count on the strong assumption that all the corporate governance attributes contribute equally to the corporate governance index (see e.g. Gompers et al., 2003).
The governance measure we use in this paper is based on two important guidelines provided by the Combined Code of Corporate Governance (2003) that relate to the composition of the board of directors and the role of large outside shareholders. In particular, it was argued that the board of directors should have the right balance between executive and non-executive directors so that no individual or a small group of individuals can dominate the decision taking process of the board. It was also argued that large shareholders should have a responsibility to make considered use of their votes. Accordingly, we use two aspects of corporate board structure (namely, board size (BOARDSIZE) and board composition (NON-EXEC)) and one aspect of corporate ownership structure (namely, ownership concentration (CONCENTR)) to construct our corporate governance measure (GOVERNANCE).
We expect that ownership concentration has a positive weight in the governance measure. This is based on the conjecture that shareholders with substantial equity stakes have more incentives than small shareholders to supervise management and can do so more effectively (Friend and Lang, 1988; Shleifer and Vishny, 1997). On the other hand, we expect board size to contribute negatively to the corporate governance measure. It is argued that larger boards are relatively less effective because coordination, communication and decision-making are more cumbersome in large boards (Yermack, 1996). As for the role of non-executive directors, there is no clear-cut prediction. On one hand, it is widely acknowledged that non-executive directors can contribute to better governance by limiting the exercise of managerial discretion within the firm (Byrd and Hickman, 1992). On the other hand, it is argued that non-executive directors do not add much to the governance of firms possibly because they lack information about the firm, do not bring the requisite skills to the job and, hence, prefer to play a less confrontational role rather than a critical monitoring role. An argument that has recently gained support and was first advocated by Franks et al. (2001) relates to the inability of the UK regulatory system to enforce the duties of directors, which causes non-executive directors to be passive, leading to higher entrenchment of executive directors (see Weir et al., 2002; Ozkan and Ozkan, 2004, for evidence on the role of non-executive directors in UK companies).4
In conducting PCA, we pick the first factor, hereafter called GOVERNANCE, which accounts for the highest percentage of variation. This factor is a linear combination of the variables CONCENTR, NON-EXEC and BOARDSIZE. The respective factor loadings are 0.336, –0.524 and –0.782, which show that ownership concentration contributes positively to the governance measure whereas board size and the ratio of non-executive directors contribute negatively. The positive loading of CONCENTR and the negative loading of BOARDSIZE are in line with Shleifer and Vishny (1997) and Yermack (1996), respectively. The negative sign of NON-EXEC is inconsistent with the proposition in Byrd and Hickman (1992) but in line with that in Franks et al. (2001).
3.3. Definitions of leverage
In examining the factors that are correlated with financial leverage, we use two definitions of leverage. First, we measure leverage as total debt divided by total assets (book leverage). Second, we also use an alternative leverage measure, which is the ratio of total debt to the sum of book debt and the market value of equity (market leverage). Although they are conceptually different and there is no consensus on the preferred definition, they are often used interchangeably in the empirical literature. Early empirical research on capital structure focuses on book leverage by arguing that corporate debt is mainly supported by the assets a firm has in place rather than its growth opportunities (Myers, 1977). This is further supported by a more recent survey by Graham and Harvey (2001) who provide evidence that managers pay more attention to book values when setting financial policies. It is also argued that book values are less volatile than market equity values and, hence, are a better guide to financial structure. Finally, Barclay et al. (2006) show that book leverage is preferred to market leverage in regressions of financial leverage as using the market value of equity in the denominator might spuriously correlate leverage with explanatory variables such as Tobin's q. However, there are studies that argue against book leverage in favour of market leverage. For example, Welch (2004) argues that the book value of equity, which can even be negative, is primarily a ‘plug number’ to balance the left-hand side and the right-hand side of the balance sheet. He further argues that the significant effects reported in the literature of profitability and fixed assets in determining leverage may arise from the accounting rules implying that the book value of equity increases with historical cash flows and decreases with asset depreciation.
In light of the lack of consensus as to whether to use book or market leverage in regressions, we run our regressions using both definitions. Similar to Barclay et al. (2006), we argue that the use of both definitions can be justified as they measure debt in relation to firm value. While the book leverage ratio expresses total debt relative to the value of assets in place (i.e. backward looking), the market value leverage ratio should be seen as expressing the ratio of total debt to the market value of the firm (i.e. forward looking).
3.4. Control variables
Following Rajan and Zingales (1995), we use the following firm characteristics as control variables in the empirical analysis: asset tangibility, growth opportunities, firm size and profitability.5 We also include industry and time dummies to control for industry-specific and time-specific effects.
4. Results
4.1. Univariate analysis
Table 3 presents univariate mean and standard deviation comparisons of various firm-specific characteristics by leverage quartiles. To do so, we divide firms into quartiles on the basis of their leverage ratio and test whether the characteristics of companies differ across low-leverage firms (first quartile) and high-leverage firms (fourth quartile). In line with the findings of the majority of the extant empirical literature, we find that firms with low leverage ratios are usually smaller, and have lower tangible assets and higher market-to-book ratios. The differences in means are statistically (at the 1 per cent level) and economically significant. The results using book leverage, for example, suggest that the mean market-to-book (tangibility) ratio of low-leverage firms is 80 per cent (69 per cent) greater (lower) than that for high-leverage firms. We also observe that low-leverage firms usually have smaller boards and a lower ratio of non-executive directors. However, the level of ownership concentration does not seem to differ significantly across the first and the fourth leverage quartiles.
| Variable | Q1 | Q2 | Q3 | Q4 | t-test |
|---|---|---|---|---|---|
| TANGIBILITY | 13.32 | 21.52 | 28.79 | 42.97 | −13.79*** |
| 13.25 | 20.89 | 27.63 | 44.79 | −15.05 *** | |
| [17.41] | [20.05] | [21.77] | [28.31] | ||
| MKTBOOK | 2.95 | 2.07 | 1.83 | 1.64 | 7.37*** |
| 3.161 | 2.34 | 1.62 | 1.36 | 10.18*** | |
| [2.63] | [1.47] | [1.39] | [0.80] | ||
| SIZE | 9.51 | 10.50 | 11.40 | 12.17 | −14.74*** |
| 9.43 | 10.60 | 11.44 | 12.19 | −15.64 *** | |
| [1.56] | [1.91] | [2.15] | [2.32] | ||
| PROFITABILITY | −0.06 | 0.01 | 0.05 | 0.08 | −7.39*** |
| −0.07 | 0.01 | 0.07 | 0.07 | −7.23 *** | |
| [0.25] | [0.22] | [0.19] | [0.15] | ||
| EXECOWNER | 15.04 | 12.85 | 10.14 | 8.37 | 4.59*** |
| 14.98 | 13.27 | 10.74 | 7.41 | 5.22*** | |
| [16.91] | [16.99] | [14.61] | [14.33] | ||
| CONCENTR | 38.36 | 36.68 | 34.98 | 35.61 | 1.59 |
| 38.36 | 35.18 | 36.03 | 36.04 | 1.38 | |
| [18.99] | [19.89] | [17.77] | [18.71] | ||
| NON-EXEC | 47.92 | 47.73 | 50.43 | 52.02 | −3.15*** |
| 47.79 | 48.16 | 50.00 | 52.15 | ||
| [15.25] | [13.14] | [13.22] | [13.14] | −3.44 *** | |
| BOARDSIZE | 6.10 | 6.33 | 7.06 | 7.58 | −7.70*** |
| 6.07 | 6.43 | 7.15 | 7.42 | ||
| [1.73] | [1.83] | [2.39] | [2.40] | −7.17 *** |
- Notes: This table provides univariate mean comparisons of several firm-specific characteristics by book leverage quartiles (normal font) and market leverage quartiles (italics). It also provides standard deviation comparisons by book leverage quartiles (in square brackets). The t-statistic is for a difference of means from the first to the fourth quartiles. Definitions for all the variables are provided in Table 1.***indicates that the difference in means is statistically significant at the 1 per cent level.
Table 3 also shows that the mean executive ownership in low-leverage firms is higher than that in high-leverage firms. The difference is significant at the 1 per cent level. The difference is economically significant as well. Executive ownership in low-leverage firms has a mean value of 15.04 per cent, which compares to 8.37 per cent for high-leverage firms. This suggests that the mean ownership in low-leverage firms is 80 per cent higher than the mean executive ownership in high-leverage firms. The significantly different executive ownership levels are consistent with the managerial entrenchment hypothesis that executive directors become entrenched after a specific level of executive ownership and, hence, ceteris paribus, choose to have a level of leverage that is lower than optimal.
However, one needs to be cautious in interpreting the latter finding as anecdotal evidence for a negative relationship between executive ownership and leverage. First, univariate analysis does not effectively control for a potential non-linearity in that relationship. Second, as mentioned earlier, it is likely that the relationship between executive ownership and leverage also depends on the corporate governance environment in which firms operate. To address these issues, we provide a detailed preliminary investigation on the relationship between executive ownership, corporate governance and leverage. Specifically, we split the sample into two groups by labelling the upper 45 per cent in terms of the governance measure, GOVERNANCE, as ‘high-governance firms’ and the lower 45 per cent as ‘low-governance firms’.6 Then, in Table 4 we examine how changes in executive ownership influence leverage for the two subsamples under two alternative definitions of leverage.
| Book leverage | Market leverage | |||
|---|---|---|---|---|
| High-governance firms | Low-governance firms | High-governance firms | Low-governance firms | |
| 1. EXECOWNER < 10% | 14.75 | 21.17 | 9.11 | 12.39 |
| 2. 10% < EXECOWNER < 20% | 10.02 | 14.22 | 5.38 | 8.15 |
| 3. 20% < EXECOWNER < 30% | 14.95 | 19.75 | 8.15 | 10.43 |
| 4. 30% < EXECOWNER < 40% | 11.73 | 19.43 | 6.34 | 9.63 |
| 5. 40% < EXECOWNER < 50% | 18.78 | 10.08 | 9.98 | 4.90 |
| 6. EXECOWNER > 50% | 12.28 | 8.37 | 6.49 | 3.19 |
- Notes: This table examines how leverage varies with changes in executive ownership and corporate governance. We split the sample into two groups by labelling the upper 45 per cent in terms of GOVERNANCE as ‘high-governance firms’ and the lower 45 per cent as ‘low-governance firms’. Analytical definitions for the variables EXECOWNER and GOVERNANCE are provided in Table 1. The number of high-governance firms in groups 1, 2, 3, 4, 5 and 6 is 202, 91, 58, 26, 22 and 31, respectively. The numbers of low-governance firms in groups 1, 2, 3, 4, 5 and 6 are 346, 30, 19, 18, 7 and 10, respectively.
The results from this investigation point to a non-linear relationship between executive ownership and leverage. Although strong inferences cannot be drawn from this exercise, there is some evidence suggesting that, if any, the non-linear relationship between executive ownership and leverage is more pronounced in low-governance firms. In particular, when executive ownership is between 10 and 20 per cent, the average book leverage ratio (market leverage ratio) is 14.22 per cent (8.15 per cent) for the subsample of low-governance firms. As we move to the next subgroup, the average value of book leverage (market leverage) increases to 19.75 per cent (10.43 per cent) and it seems that the relationship between executive ownership and leverage becomes negative at higher levels of ownership. For example, the average book leverage ratio (market leverage ratio) drops to 10.08 per cent (4.90 per cent) when executive ownership lies between 40 and 50 per cent and to 8.37 per cent (3.19 per cent) when executive ownership is greater than 50 per cent.
4.2. Regression analysis
Table 5 presents the regression results where book leverage is regressed on a set of firm characteristics, including corporate governance, executive ownership and their interactions. We start by estimating a baseline model that includes the firm characteristics suggested by Rajan and Zingales (1995). This model is estimated using a pooled ordinary least-squares (OLS) estimator (model 1), a fixed effects estimator (model 2) and an OLS lagged estimator where leverage is measured at time t whereas the independent variables are measured at t – 1 (model 3). In line with the findings of prior empirical research on leverage, the results indicate that the estimated coefficients on asset tangibility and firm size are positive and statistically significant (see e.g. Rajan and Zingales, 1995). In addition, as expected, the estimated coefficient on the market-to-book ratio is negative and statistically significant at the 1 per cent level in models 1 and 3, which supports the view of Myers (1977) that high-growth firms prefer lower leverage to avoid potential agency problems related to underinvestment. However, we do not observe a significant relationship between the market-to-book ratio and leverage under the fixed effects estimation. Finally, there is some evidence that profitability exerts a negative influence on leverage, which is in line with the pecking order explanation of capital structure.
| Independent variables | Estimation method | |||||
|---|---|---|---|---|---|---|
| OLSModel 1 | Fixed effectsModel 2 | OLS-laggedModel 3 | OLSModel 4 | Fixed effectsModel 5 | OLS-laggedModel 6 | |
| Constant | −0.092 (−2.98)*** | — | −0.076 (2.29)** | −0.158 (−4.28)*** | — | −0.157 (3.82)*** |
| TANGIBILITY | 0.198 (9.40)*** | 0.294 (6.74*)** | 0.178 (7.92)*** | 0.198 (9.54)*** | 0.295 (6.74)*** | 0.178 (8.12)*** |
| MKTBOOK | −0.003 (−2.84)*** | 0.001 (1.23) | −0.005 (−3.56)*** | −0.002 (−2.05)** | 0.001 (1.24) | −0.004 (−2.79)*** |
| SIZE | 0.022 (10.4)*** | 0.038 (5.89)*** | 0.022 (9.35)*** | 0.028 (10.3)*** | 0.039 (5.89)*** | 0.028 (9.37)*** |
| PROFITABILITY | −0.046 (−3.26)*** | −0.093 (−7.72)*** | −0.021 (−1.43) | 0.021 (0.55) | −0.093 (−7.68)*** | −0.031 (−2.03)** |
| EXECOWNER | — | — | — | 0.134 (2.14)** | 0.041 (0.63) | 0.171 (2.59)*** |
| EXECOWNER_SQ | — | — | — | −0.248 (−2.57)*** | −0.057 (−0.56) | −0.296 (−2.86)*** |
| GOVERNANCE | — | — | — | 0.016 (3.09)*** | 0.008 (1.98)* | 0.016 (2.84)*** |
| GOVERNANCE *EXECOWNER | — | — | — | −0.153 (−2.50)** | −0.101 (−2.15)** | −0.144 (−2.01)** |
| GOVERNANCE *EXECOWNER_SQ | — | — | — | 0.276 (2.81)*** | 0.166 (1.84)* | 0.248 (2.14)** |
| Industry dummies | Yes | — | Yes | Yes | — | Yes |
| Time dummies | Yes | Yes | Yes | Yes | Yes | Yes |
| R 2 | 27.19 | 11.45 | 26.78 | 29.76 | 11.67 | 27.71 |
| Number of firms | 956 | 883 | 875 | 956 | 883 | 875 |
| Number of observations | 4293 | 4220 | 3297 | 4293 | 4220 | 3297 |
- Notes: This table provides the results from several regressions predicting leverage. We use a pooled ordinary least-squares (OLS) approach to estimate models 1 and 4, a fixed effects approach to estimate models 2 and 5, and an OLS lagged approach to estimate models 3 and 6. The dependent variable in all models is the book value of total debt to the book value of total assets. The independent variables, except for GOVERNANCE, are defined in Table 1. GOVERNANCE is an index variable that evaluates the effectiveness of the corporate governance environment in which firms operate and is derived after using principal component analysis (see Section 3.2 for details). t-statistic values are reported in parentheses.***, ** and *indicate coefficient is significant at the 1, 5 and 10 per cent levels, respectively.
In models 4–6, we extend our baseline model by including executive ownership, the square of executive ownership, and the corporate governance measure in the model. The level and the squared terms of executive ownership are included together to test the hypothesis that there is a non-linear relationship between leverage and executive ownership. The inclusion of the governance measure in the model helps test the direct effect of the quality of corporate governance on leverage. Finally, the interaction terms are used to test the hypothesis that corporate governance also influences the capital structure decision indirectly, through the effects on the incentives of managers to issue debt.
In line with our expectations, the results reveal that the relationship between executive ownership and leverage is non-monotonic. In particular, when the level of executive ownership is low, an increase in executive ownership has the effect of aligning the interests of managers and shareholders, leading to a relatively higher leverage ratio. However, at higher levels of executive ownership the entrenchment effect sets in to result in lower leverage. Our findings suggest a turning point at about 29 per cent. That is, the leverage ratio increases with executive ownership up to 29 per cent of executive ownership and then decreases for ownership levels above that. We also observe a positive and significant association between our measure of corporate governance measure, GOVERNANCE, and leverage. This finding supports the view that well-governed firms, other things being equal, face a lower cost of external finance and, hence, can afford higher leverage in their capital structures.
Furthermore, the results reveal that the estimated coefficients on interaction terms are statistically significant. The negative coefficient on the interaction term between GOVERNANCE and EXECOWNER indicates that, ceteris paribus, the alignment effect of executive ownership is less pronounced in well-governed firms. This lends support to the proposition that the role of leverage as a disciplining and a monitoring device is reduced in well-governed firms as a result of lower expected agency costs. In other words, executive ownership plays a relatively less important role as an incentive mechanism in firms with strong monitoring mechanisms. We also observe a significantly positive coefficient on the interaction term between corporate governance and the square of executive ownership, providing support for the view that the entrenchment effect of executive ownership on leverage becomes weaker as the effectiveness of corporate governance increases. This is possibly because managers in well-governed firms are less able to expropriate wealth by pursuing a low-leverage policy. We test the joint significance of both interaction terms included in specifications 4–6 by utilizing a Wald test. The results reject the null hypothesis that both interaction terms equal zero, supporting the specification of these models.7 We obtain similar results using the market definition of leverage.
The economic magnitude of these effects is also significant. For example, using the estimated coefficients in model 6 in Table 5, for a firm with the median executive ownership of 4.04 per cent, an increase in the governance measure, GOVERNANCE, from –0.59 (the first quartile) to 0.71 (the third quartile) is associated with an increase of about 1.4 percentage points in book leverage.8 With a mean sample book leverage of 16.79 per cent, this suggests about an 8.34 per cent increase in leverage on average (1.4/16.79 = 8.34). Using the first quartile (0.36 per cent) and the second quartile (17.80 per cent) executive ownership, instead of median ownership, yields respectively an 11.98 per cent increase and a 1.4 per cent reduction in book leverage ratio on average.
In Table 6, we provide the results of further tests for the existence of an interaction between executive ownership and corporate governance in determining leverage. To do so, we split the sample into ‘low-governance’ firms and ‘high-governance’ firms and examine whether the non-linear relationship observed between executive ownership and leverage earlier holds for both subgroups. We perform this exercise both for book leverage (Panel A) and market leverage (Panel B). The models are estimated using the OLS lagged approach. The results confirm the existence of a non-linear impact of executive ownership for the sample of low-governance firms. The estimated coefficients of EXECOWNER and EXECOWNER_SQ are statistically significant. However, in support of our earlier findings and argument, we do not observe a significant impact exerted by executive ownership on leverage for high-governance firms. These findings provide further support for the existence of an interaction between corporate governance and executive ownership in influencing leverage.9
| Panel A | Panel B | |||
|---|---|---|---|---|
| Low-governance firms(Model 7) | High-governance firms(Model 8) | Low-governance firms(Model 9) | High-governance firms(Model 10) | |
| Constant | −0.159 (−2.98)*** | −0.040 (−0.68) | −0.094 (−2.49)** | −0.063 (−1.50) |
| TANGIBILITY | 0.169 (5.18)*** | 0.224 (7.22)*** | 0.122 (5.28)*** | 0.159 (6.74)*** |
| MKTBOOK | −0.005 (−2.33)** | −0.005 (−3.28)*** | −0.009 (−5.40)*** | −0.006 (−5.52)*** |
| SIZE | 0.027 (7.67)*** | 0.015 (3.14)*** | 0.017 (6.90)*** | 0.012 (3.25)*** |
| PROFITABILITY | −0.028 (−1.12) | −0.025 (−1.34) | −0.049 (−2.71)*** | −0.027 (−2.04)** |
| EXECOWNER | 0.331 (3.28)*** | 0.050 (0.59) | 0.198 (3.32)*** | 0.010 (0.16) |
| EXECOWNER_SQ | −0.506 (−3.12)*** | −0.115 (−0.87) | −0.322 (−3.43)*** | −0.059 (−0.65) |
| Time dummies | Yes | Yes | Yes | Yes |
| Industry dummies | Yes | Yes | Yes | Yes |
| R 2 | 0.303 | 0.253 | 0.329 | 0.266 |
| Number of firms | 403 | 388 | 403 | 388 |
| Number of observations | 1586 | 1382 | 1586 | 1382 |
- Notes: This table provides the results from ordinary least-squares (OLS) lagged regressions predicting leverage. We split the sample into two groups by labelling the upper 45 per cent in terms of GOVERNANCE as ‘high-governance firms’ and the lower 45 per cent as ‘low-governance firms’. In Panel A, the dependent variable is the book value of total debt to the book value of total assets. In Panel B, the dependent variable is the book value of total debt to the sum of the book value of total assets and the market value of total equity. The independent variables are defined in Table 1. All regressions include industry dummies. t-statistic values are reported in parentheses.*** and **indicate coefficient is significant at the 1 and 5 per cent levels, respectively.
4.3. Capital structure dynamics and robustness
So far, our empirical specification has been static. In this section, we present the results from the dynamic panel data estimations. In the context of our analysis, a dynamic panel data framework is useful for two reasons. First, it helps control for the endogeneity problem that may arise due to unobserved heterogeneity where unobservable firm characteristics may be highly correlated with regressors. To this end, the dynamic model complements the OLS lagged approach, which only controls for endogeneity due to reverse causality. Second, it enables us to examine the dynamic nature of the capital structure decisions of firms. A dynamic specification recognizes that firms may have target leverage ratios that cannot be achieved instantaneously. Instead, due to adjustment and other costs firms adjust partially to the desired leverage level (see Ozkan, 2001, for a useful discussion on these issues). The dynamic empirical specification we adopt is given by
(1)where Y represents our proxy for corporate leverage and Xk is a vector of the explanatory variables. The terms ni and nt represent time-invariant firm-specific and firm-invariant time-specific effects respectively, while b refers to the adjustment parameter, which takes values between 0 (indicating no adjustment at all towards the target) and 1 (indicating immediate adjustment to the target).10
For the estimation of equation (1) we use the GMM estimator proposed by Arellano and Bond (1991) given that OLS and fixed effects are likely to yield inconsistent estimates when used to estimate dynamic models (see Bond, 2002, for a detailed discussion). The GMM estimator involves the use of instruments dated [t – 2] or earlier for the lagged dependent variable and the endogenous regressors as well as a first difference transformation. These two characteristics control for the possibility that the results are driven by reverse causality or unobserved heterogeneity problems. However, the consistency of the GMM estimator depends on the validity of instruments used, which in turn depends on the absence of higher order serial correlation in the idiosyncratic component of the error term. Therefore, we report the Sargan test of overidentifying restrictions, under the null that instruments are valid, and two further tests for the existence of first and second order serial correlation in the first differenced residual (denoted as m1 and m2, respectively).
Table 7 presents the results from the dynamic leverage model. In all specifications we use instruments dated [t – 2]. The Sargan test supports the validity of instruments, whereas the m1 and m2 tests confirm the existence of serial correlation of order one but not of order two. Consistent with the dynamic capital structure hypothesis, the results indicate that firms partially adjust towards an optimal leverage ratio, with the coefficient of adjustment being close to 0.6. The dynamic panel data regressions also show that size and growth opportunities remain as two of the most important determinants of leverage. Their coefficients are positive and statistically significant in both dynamic models.
| Independent variables | Model 11 | Model 12 | ||
|---|---|---|---|---|
| Constant | −0.002 (−0.37) | 0.003 (0.59) | ||
| LEVERAGEt −1 | 0.423 (5.86)*** | 0.351 (4.90)*** | ||
| TANGIBILITY | 0.256 (1.17) | 0.170 (0.75) | ||
| MKTBOOK | −0.010 (−1.91)* | −0.009 (−2.14)** | ||
| SIZE | 0.109 (2.87)*** | 0.065 (1.74)* | ||
| PROFITABILITY | 0.115 (1.75)* | 0.046 (0.88) | ||
| EXECOWNER | — | 0.495 (2.04)** | ||
| EXECOWNER_SQ | — | −0.692 (−2.16)** | ||
| GOVERNANCE | — | 0.043 (2.36)** | ||
| GOVERNANCE*EXECOWNER | — | −0.316 (−1.96)** | ||
| GOVERNANCE*EXECOWNER_SQ | — | 0.419 (1.13) | ||
| Observations | 2125 | 2125 | ||
| Wald (joint) | 46.40*** | 0.00 | 51.98*** | 0.00 |
| Sargan | 21.40 | 0.13 | 36.72 | 0.19 |
| m1 test | −5.197*** | 0.00 | −5.110*** | 0.00 |
| m2 test | 0.092 | 0.93 | −0.105 | 0.92 |
- Notes: This table reports the results from the GMM (in first differences) estimator. The dependent variable is the book value of total debt to the book value of total assets. The independent variables, except for GOVERNANCE, are defined in Table 1. GOVERNANCE is an index variable that evaluates the effectiveness of the corporate governance environment in which firms operate and is derived after using principal component analysis (see Section 3.2 for details). For the estimation, levels dated [t − 2] were used as instruments. Time dummies were used in all specifications. For the estimation we used asymptotic standard errors robust to heteroscedacity. We report a Wald test which evaluates the join significance of all regressors in each model. We also report the Sargan test, which is a test of overidentifying restrictions, asymptotically distributed as a χ2 under the null of valid instruments. m1 and m2 are tests for the absence of first order and second order correlation in the residuals. These test statistics are asymptotically distributed as N (0, 1) under the null of no serial correlation.***, ** and * indicate coefficient is significant at the 1, 5 and 10 per cent levels, respectively.
The effects of managerial incentives and internal corporate governance on the financing policy of the firm remain significant. Specifically, there is supporting evidence for both the alignment and the entrenchment effects of executive ownership. Moreover, the corporate governance measure, GOVERNANCE, enters the equation with a positive sign (see model 12). In addition, the GMM estimations indicate that corporate governance also affects the leverage decision indirectly. Specifically, it seems that the alignment effect of executive ownership is less pronounced in firms that operate under a strong corporate governance regime. However, in contrast to the results obtained using the static specification, the dynamic analysis does not suggest that the entrenchment effect of executive ownership varies with corporate governance (the coefficient of the interaction term between the square of executive ownership and corporate governance is statistically insignificant in model 12).
In summary, the results from the dynamic panel data regressions support our earlier findings that corporate governance and managerial incentives are both important in determining the capital structure decision of firms. It also seems that the impact of managerial incentives on leverage, in particular the alignment effect of executive ownership, varies with the effectiveness of the corporate governance environment in which firms operate. Finally, consistent with recent studies on the subject, our GMM results confirm the dynamic nature of the capital structure decision of firms.11
5. Conclusion
In this paper, using a sample of 956 UK listed firms during the period 1999–2004, we provide an empirical analysis of the relationship between leverage, corporate governance and managerial incentives. We use PCA to construct a simple corporate governance measure, which represents a score based on the existence of perceived good governance attributes, such as ownership concentration, non-executive directors and board size. The econometric specification used in this study allows the test of the hypothesis that internal governance influences leverage both directly, through reducing the expected agency costs of debt, and indirectly, through influencing managerial incentives towards leverage. It also enables us to investigate the hypothesis that firms adjust only partially towards an optimal leverage ratio.
Our empirical findings strongly suggest that the internal corporate governance structure matters in determining leverage. In particular, it seems that firms with strong corporate governance are able to raise more external debt due to a reduction in the expected agency costs of the manager–shareholder conflict. Moreover, managerial incentives play a significant role in determining leverage. We provide strong evidence that the relationship between leverage and executive ownership, used as a proxy for managerial incentives, is non-monotonic. Most importantly, our results support the view that internal corporate governance and managerial incentives interact in determining leverage. Specifically, it seems that although managerial incentives play a significant role – though changing at different levels of executive ownership – the exact nature of these effects depends on the corporate governance mechanisms that firms have in place. In firms with weak corporate governance, both the alignment and the entrenchment effects of executive ownership are strongly observed. However, in firms with strong governance, executive ownership does not significantly influence the capital structure decision of firms. Finally, these findings also imply that the role of leverage in reducing the costs of the manager–shareholder agency conflicts may be significant only for firms with weak governance mechanisms in place.
The results of our analysis matter for future research. A fruitful research direction, for example, is the investigation of the impact of managerial incentives and corporate governance not only on leverage but also on the effort of the firm to maintain a target liquidity level (Almeida et al., 2004; Faulkender and Wang, 2006; Almeida and Campello, 2007). Some specific questions that require attention by researchers include: (i) To what extent do managers treat external financing resources like leverage as substitutes for internal financing resources such as cash? (ii) Do managers adopt particular capital structure policies that help firms to establish and retain debt capacity? A few recent studies provide some preliminary evidence on the relevance of these questions. Specifically, Yun (2008) finds that external corporate governance (e.g. the threat of a takeover attempt) significantly influences the choice between cash and lines of credit. Lemmon and Zender (2008) show that focusing solely on leverage and ignoring debt capacity can lead to misleading conclusions regarding the reliability of the traditional capital structure theories to explain actual corporate behaviour. Finally, Gan (2007) shows that establishing debt capacity is mainly important for firms that are unable to collateralize.
Another avenue for future research is the analysis of the extent to which managerial incentives can help explain persistence and convergence, two prominent but otherwise puzzling features of leverage ratios. Behavioural and corporate governance explanations may prove useful in this context given recent evidence by Lemmon et al. (2008) that the previously identified determinants of capital structure choice (e.g. size, profitability, growth opportunities) are inadequate to explain why leverage ratios are sticky but, still, exhibit some degree of convergence. Finally, it is important to note that future empirical research should also pay more attention to empirical specifications that allow the examination of debt and equity adjustments separately. As mentioned earlier, examining the adjustment process by using leverage ratios can be misleading because such a framework does not allow to show whether observed leverage changes reflect debt or equity adjustments.
Footnotes
– Yit–1), where Yit is the actual leverage ratio at time t, while 
– Yit–1 can be interpreted as the desired change in leverage.
