Abstract
Prior empirical research has produced conflicting findings on the lease–debt relation. Although some studies provide evidence that leases substitute for debt, other studies have shown the lease–debt relation to be complementary. This study investigates the lease–debt relation in a sample of 233 restaurant and retail firms from 2006 to 2008. Using a comprehensive measure of leasing that includes both capital and operating leases and other control variables, the results show that leasing and debt are significantly and inversely related. On average, $1.00 of leasing displaced approximately $0.50 of debt. The results are robust to alternative specifications of the regression models and provide additional evidence that growth opportunities and firm size are important determinants of leasing.
Introduction
Leasing is a large and important source of financing for many restaurant companies. Restaurant companies use operating leases to acquire fixed assets such as land, buildings, and equipment, without making large upfront cash payments that are commonly associated with debt financing. Since both leases and debt are similar in their characteristics and involve a contractual claim of fixed cash flows, financial theories have suggested that leases and debt are substitutes. An increase in the use of leases should lead to a corresponding decline in the use of debt because leasing displaces debt capacity. However, the degree of substitutability between debt and leases remains an unresolved empirical issue. Ang and Peterson (1984) called this an unresolved puzzle. Some financial theories predict the degree of substitutability between debt and leases to be a ratio of one to one or less than one (but greater than zero; Myers, Dill, & Bautista, 1976), whereas other theories have predicted a trade-off greater than one, implying a complementary relation (Lewis & Schallheim, 1992). Thus far, the empirical evidence on the lease–debt substitutability has been mixed. Although there is empirical evidence that leases and debt are substitutes (Bayless & Diltz, 1986; Beattie, Goodacre, & Thomson, 2000; Marston & Harris, 1988), others find a complementary relation (Ang & Peterson, 1984; Bowman, 1980) or little or no evidence (Mehran, Taggart, & Yermack, 1999).
One explanation for the mixed findings has been the failure or difficulty in controlling for firm characteristics that influence both leasing and debt capacity. Smith and Wakeman (1985) argue that firm characteristics that affect debt capacity will also provide for more leasing. Hence, they argue for controlling for these factors in measuring the lease–debt substitutability. Measurement error is another limitation that has arisen from the difficulty in measuring operating lease liabilities because of limited data availability. Prior to 2000, databases such as S&P’s Compustat did not report operating lease payments beyond 5 years. As a result, some prior empirical studies have either ignored operating leases and focused only on capital leases (Adedeji & Stapleton, 1996; Ang & Peterson, 1984; Krishnan & Moyer, 1994) or used incomplete data in computing operating lease measures (Koh & Jang, 2009; Marston & Harris, 1988; Mehran et al., 1999; Sharpe & Nguyen, 1995). If the market views operating and capital leases to be similar to debt in their fixed claim obligations, then a comprehensive measure of leasing activity should include both capital and operating leases. Moreover, there is evidence that capital and operating leases behave similarly to debt in their effect on equity risk (Bowman, 1980; Ely, 1995; Imhoff, Lipe, & Wright, 1993). Beattie et al. (2000) noted that operating leases are an important source of finance and thus their omission in capital structure research studies poses a serious limitation on prior studies.
The purpose of this study is to investigate the degree of substitutability between leases and debt financing in a sample of 233 restaurant and retail firms from 2006 to 2008. Using firm lease disclosure data from annual 10-K reports, this study uses the discounted cash flow technique and constructive lease capitalization methodology of Imhoff, Lipe, and Wright (1991, 1997) to estimate the discounted present value of operating lease payments. This unrecorded amount of operating leases is then added to capital leases on the balance sheet to provide for a more comprehensive measure of leasing. Based on prior capital structure research, the lease–debt relation is then evaluated within a multiple regression framework including various proxy control variables that influence the leasing–debt decision. The methodology for capitalizing operating leases used in this study is not only similar in spirit to the measurement in the existing and proposed new lease accounting standard but also differs from prior research in using firm-specific assumptions. This study is also in response to a call for further research on this issue by Beattie et al. (2000) prior to the implementation of any new lease accounting standard.
The restaurant industry and retail sectors were chosen for this study to evaluate the lease–debt substitutability because of the increasing magnitude and pervasive use of operating leases as a primary source of financing in these sectors (Imhoff et al., 1997). Unlike other specialized industries such as manufacturing, restaurant and retail firms also share similar homogenous characteristics in the standardization of their location (stand-alone or located in malls, shopping centers, malls, etc.), which makes it more conducive and attractive for lessors to lease space for retail use (Goodacre, 2003). Furthermore, many restaurant companies are relatively small businesses without access to capital markets or bank financing and so they have to rely heavily on operating leases for their growth and expansion. Jamba Juice, Inc. (2010), for example, relies heavily on operating leases for its growth and expansion. The company leases all of the real estate for its 478 company-operated stores. In its fiscal 2009 annual report, the company disclosed that it does not intend to purchase the real estate for its restaurant sites in the future because the size and flexibility of its restaurant concept provides the company with a competitive advantage in its site selection process. Hence, its balance sheet reflects no long-term debt and only a quarter million dollars in capital leases for software licenses and equipment. The degree to which Jamba Juice, among many other restaurant companies, are substituting debt with leases remains largely an unanswered empirical question.
This study will contribute to the existing literature on the lease–debt substitutability research in the following ways. First, this study will use a more comprehensive measure of leasing to estimate the degree of substitutability between leases and debt financing. Previous research has generally been hampered in using a more comprehensive measure of leasing because of the availability of only partial lease data prior to 2000. Additional lease operating data is now reported on the Compustat database permitting a more comprehensive measure of leasing to be used. Moreover, this comprehensive measure of leasing is also in line with new lease accounting rules expected in late 2011 that will eliminate the classification of leases into operating and capital leases and require a single model that recognizes a right-to-use asset and an obligation to make a payment. Second, focusing on specific industry sectors has the advantage of controlling for industry factors and increasing the ability to detect an empirical relation within a group of firms that share similar leasing characteristics, enabling researchers to perform more intraindustry tests of differences. Research has shown that firms within a specific industry share similar capital structure characteristics with each other than with firms from other industries (Bradley, Jarrel, & Kim, 1984). Both the restaurant and retail sectors are large users of operating leases with some firms relying on them to a much greater extent than other firms within the same industry. Hence, more insights on capital structure decisions can be gained by focusing on a specific industry to test a hypothesis when firms within an industry face similar trade-offs. In addition, it reduces the likelihood that the results could be driven by unrelated industry factors (Imhoff et al., 1993). Finally, this timely and policy relevant ex ante research as advocated by Schipper (1994), when combined with ex post research, will make a new and significant contribution to the existing capital structure literature on leasing and debt financing. Financial statement users, investors, creditors, regulators, and academics will find the information useful and relevant in assessing and evaluating the lease–debt substitutability.
Literature Review
In perfectly competitive capital markets, in the absence of taxes, asymmetric information, and bankruptcy costs, the value of the firm is unaffected by its choice of debt, leasing, or equity financing (Modigliani &Miller, 1958). In the presence of these costly market imperfections, several financial theories have been proposed in the corporate finance literature to provide explanations for ways that capital structure decisions of companies (to use debt and leasing) affect firm value. These relevant capital structure theories include the static trade-off theory, the financial contracting cost (agency) theory, and the asymmetric information theory. The theoretical explanations of these theories lead to a set of predictions that identify factors influencing the relation between particular firm characteristics and debt and leases.
Static Trade-Off Theory
The static trade-off theory of capital structure suggests that firms will trade off the interest tax shield benefits of debt against the costs of financial distress such as bankruptcy. It implies that each firm has a target or optimal debt ratio and the capital structure of the firm is optimized when the marginal benefits of debt equals the marginal cost of debt. The theory suggests that leases and debt are substitutes because leasing involves a fixed claim obligation similar to debt and, thus, consumes debt capacity. Therefore, an increase in debt (leases) will lead to a corresponding decrease in leases (debt). This also implies that the marginal cost of leases (debt) will increase the likelihood of financial distress with any additional fixed claim obligations in the capital structure.
Myers et al. (1976) present the economic rationale for leasing and argue that leases displace debt and thus reduce debt capacity. Assuming leases and debt are substitutes, their models indicate that leasing can provide a tax incentive to both lessors and lessees when their tax rates differ. Their models show that the trade-off between leases and debt can be perfect (one to one) or less than one but greater than zero. In other words, a dollar of leases will displace a dollar of debt or less than a dollar of debt because of differences between the nature and terms of leases and debt contracts. Lewis and Schallheim (1992) also present tax incentives to use both debt and leasing within an optimal capital structure model. In contrast to Myers et al. (1976), they demonstrate that the substitution between leases and debt could be greater than one and go on to show the relation between leases and debt to be complementary. They argue that leasing is a mechanism for selling excess tax deductions that can motivate a firm to increase the proportion of debt in its capital structure relative to an identical firm that does not lease. When a lessee sells its nondebt tax shields to a lessor, it will reduce the potential redundancy of its tax shields and benefit the lessor through lower taxes. The lessor can pass on these gains to the lessee in the form of lower lease payments, which will then lead to a reduced marginal cost of debt for the lessee. Consequently, the lessee responds to this incentive by issuing additional debt that accounts for the positive relation between leases and debt. The benefits of leasing can be realized even when the marginal tax rates are the same between the lessor and lessee.
Empirical evidence on static trade-off theory
The empirical evidence on the lease–debt substitutability thus far has been mixed. Although there is some evidence of a complementary relation (Ang & Peterson, 1984; Bowman, 1980; Finucane, 1988), others find evidence that is largely consistent with the trade-off theory that leases and debt are substitutes (Adedeji & Stapleton, 1996; Bayless & Diltz, 1986; Beattie et al., 2000; Marston & Harris, 1988; Yan, 2006). In contrast, Mehran et al. (1999) document a complementary relation between debt and capitalized leases but no relation between debt and operating leases. Empirical research has yet to provide conclusive evidence on this relation. The exclusion of operating leases, failure to control for debt capacity or changing asset base, measurement error, and a lack of complete operating lease data could all potentially explain the mixed findings.
Empirical research in the hospitality industry has largely focused on the determinants of debt policy with few studies on leasing (Marler, 1993; Upneja & Dalbor, 1999). Upneja and Dalbor (1999) examined the relation between tax rates and leasing behavior for a sample of restaurant firms and find a positive relation between marginal tax rates, debt policy, and capital leases and a negative relation between operating leases and marginal tax rates. They also find a negative relation between financial distress and debt and leasing implying that restaurant firms in strong financial condition were less likely to use debt and leasing. In contrast, Koh and Jang (2009) find debt and financial distress to be positively related to operating leases among lodging firms, suggesting not only a complementary relation between debt and leasing but also the notion that lodging firms in strong financial condition were more likely to use operating leases. However, their measure of operating leases suffers from measurement error because the lease proxy is measured based on annual operating lease expense rather than the present value of operating lease payments or the “multiples” approach commonly used in practice by firms and analysts. These inconsistent findings on hospitality firms also suggest further research is needed to reexamine the lease–debt substitutability in light of new rules that will eliminate the distinction between capital and operating leases and, hence, a more comprehensive measure of leasing should be used to provide evidence on the factors that influence leasing.
Contracting Cost Theory
The existence of risky corporate debt in a firm’s capital structure can also create conflicts of interest between debt holders and stockholders over the firm’s investment opportunities. These conflicts of interest can impose agency or contracting costs, thereby increasing the cost of debt and reducing the value of the firm. Agency costs tend to increase with leverage and occur when managers engage in behaviors that benefit stockholders more than debt holders. These agency costs of debt include the problems of underinvestment, asset substitution and overinvestment. Thus, managers have incentives to reduce the cost of debt in debt agreements between debt holders and stockholders.
The value of a firm’s investment opportunity set depends on the value of its existing assets in place and future growth opportunities. Myers (1977) characterizes a firm’s investment opportunities as options and demonstrates that with high amounts of debt in a firm’s capital structure, taking on a positive net present value project can reduce stockholder wealth if the gains accrue primarily to debt holders. Debt holders recognize this behavior and will demand higher compensation for taking on additional risk. Consequently, stockholders have incentives to reject positive net present value projects to avoid a wealth transfer from themselves to debt holders. Myers (1977) called this an underinvestment problem, which can be mitigated by reducing debt, including restrictive covenants in debt, or by issuing short-term debt. This incentive to underinvest can also be reduced if the firm finances its new investments with fixed claims with high priority such as secured debt or capitalized leases (Stulz & Johnson, 1985). In contrast to the underinvestment problem, risky debt in the firm’s capital structure may also induce shareholders to substitute riskier assets for the firm’s existing assets. Unlike the underinvestment problem, this asset substitution problem transfers wealth from debt holders to stockholders but can be mitigated by including security provisions in the debt and issuing secured debt (Smith & Warner, 1979). Finally, the agency problem proposed by Jensen (1986) suggests that managers with too much free cash flow in hand tend to overinvest and financing with debt or leases can also mitigate this problem.
Empirical evidence on contracting cost theory
Financial contracting cost theory predicts that firms with higher growth opportunities will face severe underinvestment problem and be thus less likely to use debt financing, including secured debt and capitalized leases. Hence, a negative relation is expected between growth opportunities and debt and leasing. The empirical results from prior research provide support for the contracting cost hypothesis. Studies show that firms with higher market-to-book ratios have significantly low debt ratios (Graham, Lemmon, & Schallheim, 1998; Sharpe & Nguyen, 1995; Smith & Watts, 1992). Barclay and Smith (1995) found the market-to-book ratio to be negatively related to debt usage and positively related to capital leases, whereas Graham et al. (1998) show the use of all three financial instruments (debt, capital leases, and operating leases) to be negatively related to growth opportunities. Yan (2006) finds that the degree of substitutability between debt and leases is greater in firms having greater growth opportunities.
Empirical findings on the contracting cost theory in the hospitality industry show differences between lodging and restaurant firms. Although the findings of a negative relation between growth opportunities and debt usage for restaurant firms are consistent with prior research (e.g., Dalbor & Upneja, 2002), the findings for lodging firms are contrary to expectations. Dalbor and Upneja (2004) provide evidence of a significant and positive relation between long-term debt and the market-to-book ratio among lodging firms. They attribute this finding to the uniqueness of lodging firms in their fixed asset intensiveness and industry seasonality relative to other industries. Furthermore, they note that growth opportunities for lodging firms appear to be investments in renovations and additions to properties to make them more attractive to guests rather than investments in research and development. In a subsequent study on lodging firms, Tang and Jang (2007) affirmed Upneja and Dalbor’s (2004) finding of a positive relation between long-term debt and usage and growth opportunities. Although the influence of growth opportunities and debt usage has been explored in the hospitality literature, the association between leasing and growth opportunities is largely unexplored and therefore provides an opportunity to extend the research in the hospitality industry.
Asymmetric Information Theory
The basic premise of the asymmetric information theory or the pecking order theory is information asymmetry between managers and outsiders because managers have more information about the future performance of the firm than its outsiders (Myers & Majluf, 1984). Myers and Majluf argue that the capital structure of the firm will depend on its need for external financing. This is in contrast to the target debt level or ratio proposed in the static trade-off theory. Asymmetric information leads to a pecking order of financing in which internal financing is the most preferred method of financing followed by external financing where debt or leasing is preferred over equity. Because lease payments have a priority claim relative to debt in bankruptcy, the pecking order theory suggests that leasing can reduce the costs related to the adverse selection problem and therefore be the first choice in the pecking order of external financing alternatives. The pecking order theory helps explain why changes in debt structure may be driven by the need for external financing as opposed to achieving an optimal capital structure that is proposed in the static trade-off theory.
Empirical evidence on asymmetric information theory
Empirical evidence from Krishnan and Moyer (1994) suggests that capital lease financing becomes an increasingly attractive financing option in the pecking order of alternatives when there is an increase in financial distress. They provide evidence that firms with high debt leverage and facing the potential for financial distress were more likely to find leasing to be an attractive financing alternative because of its lower bankruptcy costs relative to secured debt. Sharpe and Nguyen (1995) noted that firms facing higher information asymmetry problems were more likely to lease in order to reduce the higher costs of external financing. On the other hand, Barclay and Smith (1995) find little or no support for the priority structure of debt across firms. Yan (2006) finds evidence that the lease–debt substitutability is more pronounced in firms that face a higher degree of information asymmetry.
Hospitality empirical research on the pecking order theory has focused mainly on explaining the debt behavior of lodging firms. Whereas Sheel (1994) found profitability to be an important variable in explaining the short-term leverage behavior of lodging firms, Gu (1995/1996) and Tang and Jang (2007) failed to find any evidence of the relation between profitability and debt usage among lodging firms. Dalbor and Upneja (2004) suggested growth opportunities as a better proxy variable for testing the pecking order theory given their finding of a positive relation between growth opportunities and long-term debt among lodging firms. Upneja and Dalbor (2001) also suggested that information asymmetry problems increase the likelihood that restaurant firms facing bankruptcy will obtain more short-term debt relative to long-term debt. Finally, Koh and Jang (2009) find that lodging firms with limited internal sources of financing were more likely to use leasing because of information asymmetry problems.
Research Design and Methodology
Leasing-Dependent Variable
Because proposed new accounting rules will require the capitalization of operating leases on the balance sheet and eliminate the distinction between operating and capital leases, a more comprehensive measure of leasing is used in this study that combines both operating and capital leases. Leasing is defined as the sum of book value of capital leases and present value of operating leases scaled by the market value of the firm. Market value of the firm is defined as the sum of market value of assets plus the present value of operating lease. The discounted cash flow method is used to estimate the present value of operating lease payments. The conversion of operating leases into capital leases uses lease footnote disclosures from annual 10-K reports along with assumptions about interest rates and remaining life of leases beyond 5 years. Appendix A provides an example of the lease footnote disclosures of Red Robin Gourmet Burgers, Inc. (2010), and Appendix B illustrates the mechanics of computing the present value of operating leases. The methodology is consistent with the constructive lease capitalization methodology developed by Imhoff et al. (1991, 1997). However, instead of a 10% incremental interest rate used in prior studies (Beattie, Edwards, & Goodacre, 1998; Imhoff et al., 1991; Marler, 1993), the incremental borrowing rate is allowed to vary by firm because different firms face different costs of borrowing. Similarly, unlike the uniform assumption of a 15-year remaining lease life used in prior studies (Imhoff et al., 1991; Marler, 1993), the total remaining operating lease life is also allowed to vary by firm in this study. Beattie et al. (2000) used a similar lease capitalization methodology, which was however adapted for U.K. firms because of different and inadequate operating lease disclosure requirements. Finally, capital leases are measured as the book value of capital leases divided by the market value of the firm.
Explanatory Variables
Financial theory predicts a negative relation between debt and leasing, indicating that leasing displaces debt capacity. The debt policy variable is measured as the book value of total debt (short term and long term) net of capital leases and divided by the market value of the firm. Debt usage is expected to be negatively related to leasing.
Financial contracting theory also predicts that firms with higher growth opportunities are more likely to face underinvestment problems. Hence firms with higher growth opportunities will use less leasing. Consistent with prior research, the market-to-book ratio is used to proxy for a firm’s investment opportunity set. The market-to-book ratio is defined as the market value of assets divided by the book value of assets. Market value of assets is defined as the book value of assets less the book value of equity plus the market value of equity. This ratio is expected to be negatively related to leasing.
Tax-based theories of optimal capital structure predict a negative relation between leasing and the corporate marginal tax rate (DeAngelo & Masulis, 1980). Graham et al. (1998) argued that previous studies have all used tax variables that suffered from endogeneity problems that could have induced spurious correlations with leasing activity. Using forward-looking estimate of before-financing corporate marginal tax rates, Graham et al. (1998) documented a negative relation between operating leases and tax rates consistent with the static trade-off hypothesis but they find no relation between capital leases and debt. They attribute this finding to the ambiguity in the distinction of capital lease as either true or nontrue leases. To test the relation between taxes and leasing in this study, the predicted coefficients of the marginal tax rates in Graham and Mills (2008) are used to compute the before-financing tax rates for each firm. This tax variable is expected to be negatively related to the comprehensive measure of leasing, but no prediction is made about its relation to capital leases.
The static trade-off theory predicts that firms with higher expected costs of financial distress will use fewer leases, which implies a negative relation between financial distress and leasing. Following prior research (Graham et al., 1998), financial distress is measured using Altman’s (1968) modified z score with a lower score indicating a higher potential for financial distress. Financial distress is expected to be negatively related to leasing.
The availability of collateral may also affect a firm’s financing policy because a firm can use its valuable assets to obtain more favorable financing at a lower cost and consequently increase its debt capacity. This suggests a positive relation between the collateral value of assets and leasing. Thus firms with assets that can be used as collateral have a higher propensity to lease. Collateral value of assets is measured as the net amount of property, plant, and equipment and scaled by book value of total assets.
Finally, firm size is expected to influence leasing policy. Large firms are more likely to finance with debt because these firms are more diversified, have more stable cash flows, and can easily exploit economies of scale in external financing. On the other hand, smaller firms will find external financing more costly and therefore these firms are more reliant on leasing. Firm size is measured as the log of the market value of the firm. Firm size is expected to be negatively related to leasing.
Research Design
Various multiple regression models were formulated to assess the degree of lease–debt substitutability and theoretical determinants of leasing. The main analysis employs pooled ordinary least squares (OLS) cross-sectional regressions in which the comprehensive lease ratio (capital leases plus operating leases) is regressed on the debt ratio and other control variables that proxy for theoretical determinants of leasing. This model is specified in the following form:
where CLR is the comprehensive lease ratio [(book value of capital leases + present value of operating leases)/market value of the firm], DR is the total debt ratio (total long-term debt + short-term debt - capital leases)/market value of the firm, GO is the growth opportunity variable measured as the market-to-book ratio (market value of assets divided by the book value of total assets), CVA is the collateral value of assets (net property, plant, and equipment/book value of total assets), TR is the before-financing book-simulated marginal tax rates graciously provided by John Graham and available at his website, ZS is the modified version of Altman’s (1968) z score as a measure of financial distress, Size is firm size (natural log of the market value of the firm), and ID is a dummy variable that is equal to one for a restaurant firm and zero otherwise.
To provide a basis for comparison with prior research, a censored Tobit regression model is specified for the full sample using the above equation in which only capital leases (excluding operating leases) are regressed on the explanatory variables. Because many firms are observed to have no capital leases on the balance sheet, this variable is left-censored at zero, and hence, a Tobit regression is an appropriate specification of the censored capitalized leases as the limited dependent variable in the regression model. In addition, an OLS regression is also specified for a restricted subsample of firms that report only capital leases.
The coefficient (β1) in Equation (1) measures the lease-to-debt displacement ratio rather than the debt-to-lease displacement ratio. The debt-to-lease displacement ratio indicates by how much debt capacity decreases with each additional dollar of leases. Adedeji and Stapleton (1996) and Beattie et al. (2000) noted that the debt-to-lease displacement ratio cannot be estimated in Equation (1) by taking an inverse of the debt coefficient because of the presence of the constant and other control variables in the regression. Therefore, they proposed that if evidence shows leases and debt to be substitutes, this debt-to-lease displacement ratio can be easily estimated by switching the debt ratio to a dependent variable and the lease variable as the explanatory variable in a regression analysis (see Adedeji & Stapleton, 1996; Ang & Peterson, 1984; Beattie et al., 2000).
Data and Sample Selection
The sample for the study was drawn from all active public restaurant and retail firms from S&P’s Compustat database during the 3-year period from 2006 to 2008. This was done to avoid using data for fiscal years prior to 2005 when a large number of restaurants and retail firms restated their historical financial statements for lease accounting errors. Firms that were subject to bankruptcies, subject to mergers/acquisitions, going private, and with lack of complete data were dropped from the sample. From an initial list of 366 firms (112 restaurant firms and 254 retail firms), 133 firms were deleted (48 restaurant firms and 85 retail firms) for the above reasons. The final sample comprised 233 firms (64 restaurant firms and 169 retail firms) and 699 firm-year observations from 2006 to 2008. Of these 699 firm-year observations, 357 firm-year observations are censored observations and the balance 342 firm-year observations are noncensored and form the subsample—that is, firms observed to have both capital and operating leases in the debt structure.
Results
Descriptive Statistics
The descript statistics for the overall sample of 233 firms are presented in Table 1. The firms had total average (median) assets of $3.6 billion ($695 million), generated sales of $7.1 billion ($1.3 billion), and earnings before interest, taxes, depreciation, amortization, and rent expense of $738 million ($180 million). The amount of capital leases (excluding firms with no reported capital leases) recognized on the balance sheet amounted to an average (median) of only $111 million ($13 million) representing 114 firms. The proportion of capital leases relative to the market value of the firm accounted for an average of less than 1% for the overall sample and only 2% for a subsample of firms that reported capital leases.
Descriptive Statistics
Note. This table presents a summary of the descriptive statistics for a number of variables. Some statistics are provided in dollar amounts and others presented as ratios. Where applicable, some statistics have been computed for a sum-sample of firms that reported capital leases. Definitions of variables are provided in Tables 2 to 7.
When operating leases are capitalized, the present value of operating lease liabilities is an average (median) of $835 million ($267 million) for the overall sample and $1.1 billion ($369 million) for the subsample of firms with capital leases. Given the considerable variation in the use of capital leases across firms, the magnitude of operating leases relative to capital leases is even more dramatic. Based on median values, the magnitude of off–balance sheet operating lease liabilities is 21 times larger than capital leases and accounted for 19% of market value. Similarly, operating lease liabilities are 28 times larger than capital leases and represent 18% of market value within the subsample of firms with reported capital leases. This magnitude of operating lease liabilities relative to capital leases is also much higher than the ratio of 9 and 13 reported in previous studies (Beattie et al., 1998; Marler, 1993). When operating lease liabilities are added to capital leases, the total lease liabilities (comprehensive lease ratio) is an average (median) amount of $889 million ($276 million) or 23% (20%) of market value. The results also show that total debt (including short-term debt), net of capital leases, amounted to a median value of $126 million or 8% of market value, which is substantially less than the ratio of operating lease liabilities to market value. Overall, there is considerable variation in the use of debt and leasing across firms with a substantial amount of off–balance sheet operating lease liabilities. Finally, collateral value of assets accounted for an average 40% of total assets with firms facing predicted average marginal tax rates of 31%, z scores of approximately 2.6, and estimated market-to-book ratios of 1.7.
Univariate Analysis
Table 2 shows the results of tests of differences (t tests) between restaurant and retail firms. Because of differences in sample sizes between the two groups, the equality of variances was first tested before tests of differences were performed. As an alternative, the nonparametric Wilcoxon–Mann–Whitney test of differences was also conducted. Given similarity in the results, only the parametric t test results are reported here. For capital leases, these tests were also performed within the subsample of firms that reported capital leases. The results show no significant differences in the comprehensive leasing ratio, debt ratio, and marginal taxes and only weak evidence of differences in the capital lease ratio. On the other hand, significant differences were observed between the two groups for growth opportunities (GO), collateral value of assets (CVA), modified z scores, and firm size. Restaurant firms are observed to have significantly higher growth opportunities and substantially higher collateral value of assets than retail firms. In contrast, retail firms are significantly larger, use more operating leases, and are less likely to face financial distress when compared with restaurant firms. The results are unchanged when operating leases are added to the denominator in the growth opportunity proxy, when the collateral value of assets measure included operating leases in the numerator and denominator, or when firm size was defined as the log of total assets instead of the market value of the firm.
Tests of Mean Differences
Note. This table presents t test differences in means between restaurant firms and retail firms. Comprehensive lease ratio is defined as the sum of the book values of capital leases plus the present value of operating leases and divided by the market value of the firm. Market value of the firm is defined as the book value of total assets less the book value of common equity plus the market value of equity plus the present value of operating lease liabilities. Capital lease ratio is defined as the book value of capital leases divided by the market value of the firm. Operating lease ratio is measured as the present value of operating leases divided by the market value of the firm. The total debt ratio is defined as the sum of total long-term debt plus short-term debt divided by the market value of the firm. Growth opportunities is the market-to-book ratio and defined as the market value of assets divided by the book value of total assets. Market value of assets is defined as the book value of total assets less the book value of common equity plus the market value of equity. Collateral value of assets is defined as net property, plant, and equipment divided by the book value of total assets. The before-financing book-simulated marginal tax rates are obtained from John Graham’s website. Missing tax rates are predicted based on the predicted book simulated coefficients estimated in Graham and Mills (2008). The measure of financial distress is the modified version of Altman’s (1968) z score. Firm size is measured as the natural log of the market value of the firm.
Correlation Analysis
The correlation analysis in Table 3 shows the relation between leasing and the explanatory variables to be in the direction as predicted for most of the variables used in the empirical analysis. The comprehensive lease ratio, which includes both capital leases and the present value of operating leases, is significantly and negatively related to the total debt ratio, a preliminary indication that debt and leasing are substitutes. Alternatively, the insignificantly positive relation between capital leases and debt suggests that they could be complements, a conflicting result that is contrary to theoretical expectations. The collateral value of assets proxy shows no correlation with comprehensive leasing but is significantly and negatively associated with the debt ratio implying a lack of support for higher debt capacity for these assets. Further examination of the data reveals an insignificantly negative correlation between leasing and collateral value of assets for retail firms but a significantly positive association for restaurant firms. The market-to-book ratio is also negatively and significantly related to leasing, consistent with the observation that firms with significant growth opportunities are less likely to use debt or leasing. Firm size is negatively associated with leasing, an indication that smaller firms were more likely to use operating leases. The significant negative relation between marginal tax rates and leasing is supportive of the tax hypothesis, whereas the insignificantly negative relation between marginal tax rates and debt indicates that firms with high debt levels have low marginal tax rates. Finally, z score as a measure of financial distress is negatively related to both leasing and debt, supporting the notion that that firms in strong financial condition are less likely to use leasing or debt. Financial distress is also significantly and positively correlated with firm size suggesting that larger (smaller) firms have higher (lower) z scores. A correlations analysis was also performed for the subsample of firms with reported capital leases. The results were similar in the direction predicted with one notable exception. In the subsample, the relation between capital leases and debt is negatively significant at the 10% level, suggesting that debt and leases are substitutes in contrast to the results for the overall sample.
Correlation Matrix
Note. This table presents the Pearson correlations between all the variables used in the empirical models excluding dummy variables. The comprehensive lease ratio (CLR) is defined as the sum of the book values of capital leases plus the present value of operating leases and divided by the market value of the firm. Market value of the firm is defined as the book value of total assets less the book value of common equity plus the market value of equity plus the present value of operating lease liabilities. Capital lease ratio (CL) is measured as the book value of capital leases divided by the market value of the firm. The total debt ratio (DR) is defined as the sum of total long-term debt plus short-term debt less capital leases and scaled by the market value of the firm. Growth opportunities (GO) is the market-to-book ratio and defined as the market value of assets divided by the book value of total assets. Market value of assets is defined as the book value of total assets less the book value of common equity plus the market value of equity. Collateral value of assets (CVA) is defined as net property, plant, and equipment divided by the book value of total assets. The before-financing book-simulated marginal tax rates (TR) are obtained from John Graham’s website. Missing tax rates are predicted based on the predicted book simulated coefficients estimated in Graham and Mills (2008). The measure of financial distress (ZS) is the modified version of Altman’s (1968) z score. Firm size (Size) is measured as the natural log of the market value of the firm.
p < .01. **p < .05.
Multiple Regression Analysis
The multivariate regression analysis was performed with several alternative specifications of the empirical model. To provide a basis for comparison with previous studies, a cross-sectional Tobit model regression is first specified with capital leases as the dependent variable for the overall sample of 233 firms (699 observations) along with an OLS regression for the subsample of 114 firms (342 observations) with capital leases. A pooled OLS cross-sectional regression is then specified with the comprehensive lease ratio as the dependent variable using the overall sample. All OLS regressions were estimated with heteroskedasticity and autocorrelation consistent standard errors. Multicollinearity was also assessed and found to be within conventional limits (variance inflation factors <2) in all regressions.
Censored Tobit and OLS regression
The Tobit regression results in Table 4 show a significantly positive relation between capital leases and the debt ratio. Contrary to expectations, this implies that leases and debt are complementary, consistent with previous Tobit regression results (Ang & Peterson, 1984; Mehran et al., 1999). Mehran et al. (1999) attributed their finding of a significantly positive relation to omitted variable bias. Beattie et al. (2000) also find a positive but insignificant relation using a similar Tobit model. In contrast, the OLS estimates using a subsample of firms with only capital leases (excluding operating leases) show a significantly negative relation supporting the notion that capital leases and debt are substitutes consistent other research findings that used a subsample of firms with capital leases (Adedeji & Stapleton, 1996; Krishnan & Moyer, 1994). It should also be noted that the OLS results are subject to bias in this case and should be interpreted with caution because the capital lease dependent variable is left-censored at zero and Models 3 and 4 (Table 4) exclude firms with no reported capital leases (357 observations). The explanations for these conflicting findings are discussed below.
Tobit and OLS Regression for Capital Leases
Note. This table presents the results of Tobit and pooled ordinary least squares (OLS) cross-sectional regressions where the dependent variable in each model is the capital lease ratio (CL) measured as the book value of capital leases divided by the market value of the firm. Market value of the firm is defined as the book value of total assets less the book value of common equity plus the market value of equity plus the present value of operating lease liabilities. Models 1 and 2 are based on the overall sample (including firms with no reported capital leases), whereas Models 3 and 4 are based on a subsample that excludes firms with no reported capital leases. The total debt ratio (DR) is defined as the sum of total long-term debt plus short-term debt less capital leases and scaled by the market value of the firm. Growth opportunities (GO) is the market-to-book ratio and defined as the market value of assets divided by the book value of total assets. Market value of assets is defined as the book value of total assets less the book value of common equity plus the market value of equity. Collateral value of assets (CVA) is defined as net property, plant, and equipment divided by the book value of total assets. The before-financing book-simulated marginal tax rates (TR) are obtained from John Graham’s website. Missing tax rates are predicted based on the predicted book simulated coefficients estimated in Graham and Mills (2008). The measure of financial distress (ZS) is the modified version of Altman’s (1968) z score. Firm size (Size) is measured as the natural log of the market value of the firm. Industry (ID) is measured using a dummy variable that is equal to one if the firms is a restaurant firm and zero otherwise. Year dummies (Y07 and Y08) for 2007 and 2008 are included in the model with 2006 as the base year. OLS regressions are estimated with heteroskedasticity-consistent robust standard errors with t statistics in parentheses.
p < .10. **p < .05. ***p < .01.
The results across all models in Table 4 also show that growth opportunities, as measured by the market-to-book ratio, are significantly negative, suggesting that firms with higher growth opportunities view capital leases as debt and thus are less likely to use them. These findings support the financial contracting cost theory and are consistent with prior empirical findings (Graham et al., 1998). Collateral value of assets is positively related to capital leases supporting the notion that firms with valuable assets are more likely to lease. Marginal tax rates are negatively associated with capital leases, suggesting that firms with high marginal tax rates are less likely to use capital leases. This finding is also in contrast to the significantly positive relation found by Upneja and Dalbor (1999) between marginal tax rates and capital leases among restaurant firms. The coefficients on financial distress as measured by the modified z score provide weak or no evidence that firms in strong financial condition (high z scores) are less likely to use capital leases. Although Upneja and Dalbor (1999) find a significantly negative relation between financial distress and capital leasing, they failed to control for firm size in their empirical model increasing the possibility that z scores could have been a proxy for firm size since larger firms tend to have higher z scores than smaller firms. The results for the effect of firm size on capital leasing are also in conflict. Whereas the Tobit model suggests that larger firms are more likely to use capital leases, the OLS model implies the opposite effect. Finally, dummy variables that proxy for industry sector and time show no significant differences between restaurant and retail firms.
Resolving conflicting results
To resolve the above conflicting findings on the relation between debt and leases, further analysis was undertaken to understand the sample-specific characteristics that could be driving the observed results. Therefore, to provide a common basis for comparison, total debt, capital leases, operating leases, and firm size proxies were first scaled by total assets. The descriptive means for leasing were then computed to assess the magnitude and importance of capital leases relative to total leasing. The computed means were based on the presence of capital leases and/or debt in the capital structure or lack thereof. Finally, the data were sorted by the debt ratio and then by the capital lease ratio. The results of this analysis are presented in Table 5. These results are similar when median ratios are used in place of means or when total debt or firm market values were used as alternative scaling variables.
Descriptive Mean Ratios of Selected Variables in Regression Analysis
Note. This table provides additional descriptive means for the total debt ratio, capital lease ratio, operating lease ratio, and firm size. Panel A provides the relative proportion of capital leases and operating leases to total leasing and total assets. Total leasing is measured as the sum of capital and operating leases. Panel B provides the summary statistics based on firms that have no reported and no reported capital leases, firms with debt but no capital leases, and the sub-sample of firms with reported capital leases. Panel C data is computed by first sorting the sample by the total debt ratio and then split into four quartiles. Quartile 1 of Panel C is further split into firms with no debt. Panel D data is first sorted by capital leases and then split into four quartiles. All ratios in Panels B to D are computed using total assets as the denominator. Firm size is measured as the log of total assets.
A number of key findings emerge from the data analysis. First, the results (Table 5, Panel A) show the relative insignificance of capital leases as a source of financing. Relative to total leasing (operating leases plus capital leases) for the overall sample, capital leases accounted for only 4.4% of total leasing whereas operating leases made up the balance 95.6%. Even more dramatic is the relative insignificance of capital leases in financing assets. For the overall sample, capital leases accounted for a mean ratio of only 1.6% of assets whereas operating leases represented a significant 43.6%. Second, the results in Panel B show more than half the sample firms (357 firm-year observations) show no reported capital leases in the capital structure, and of these, 111 observations indicated no reported total debt or capital leases. As it turns out, these firms were observed to be the smallest in size but the largest users of operating leases. Third, the results in Panels C and D provide additional evidence to show that the positive correlation between debt and leases is being induced by these sample firm characteristics. When firms are sorted by the total debt ratio, the results in Panel C show smaller firms (with no debt or capital leases) with a greater magnitude of operating leases. On the other hand, larger firms are observed to have high debt and capital lease ratios but use less operating leases. Similarly, if the data are sorted by the capital lease ratio, the results again show that large firms use the most debt and capital leases but the least amount of operating leases. As an additional check, if an OLS regression is performed separately for a restricted sample of firms in the bottom and top quartiles of Table 5 Panel D, the results (Quartiles 1 and 2) would still show the debt coefficient to be significantly positive (0.123 with t statistic of 2.31) for firms with no capital leases and significantly negative (−0.021 with t statistic of −2.25) for firms with high capital lease ratios (Quartile 4).
Framing these findings within the context of prior research findings, Marston and Harris (1988) also find that high-debt firms use more of both leasing and debt than low-debt firms at the cost of reducing their ability to finance with debt. The above findings are also similar to those of Adedeji and Stapleton (1996) in the United Kingdom. Within their sample, the researchers show that firms with no capital leases were largely small firms with relatively low debt ratios compared with firms with capital leases. They attributed their Tobit model finding of a positive relation between debt and leases as having been caused by the distribution of values of these variables (debt and firm size) between firms with and without capital leases. When Adedeji and Stapleton (1996) split their sample and focus their analysis on a subsample of firms with capital leases, their results provide stronger support for the theoretical prediction that finance leases are debt are inversely related. In summary, the conflicting positive relation between debt and capital leases in the Tobit model appears to be driven by a relatively large number of firms with no reported capital leases in the capital structure. Many of these small firms also reported no debt in their capital structure. Given that proposed new lease accounting rules will eliminate the distinction between capital leases and operating leases and therefore treating them as similar to debt financing, the next stage of the regression analysis proceeded with an OLS regression using the comprehensive lease ratio (including both operating and capital leases) as the dependent variable
Pooled OLS Regression Using Comprehensive Lease Ratio
The OLS regression results using the comprehensive ratio are presented in Table 6. Models 1 and 2 are based on the overall sample whereas Models 3 and 4 are based on the subsample of firms reporting capital leases. Consistent with theory, the relation between debt and total leasing is highly significant and negative. The results provide evidence that leasing substitutes for debt and supports the trade-off theory of capital structure. If capital leases are excluded from the equation, the relation between operating leases and debt would still be highly significant and negative, implying that operating leases and debt are indeed substitutes. This is because of the relatively large magnitude of operating leases relative to capital leases. These results are also consistent with those of Beattie et al. (2000) in the United Kingdom but stand in contrast to Mehran et al.’s (1999).
OLS Regression Using Comprehensive Lease Ratio
Note. This table presents the results of ordinary least squares (OLS) cross-sectional regressions where the dependent variable in each model is the comprehensive lease ratio (CLR), which is defined as the sum of the book values of capital leases plus the present value of operating leases and divided by the market value of the firm. Models 1 and 2 are based on the overall sample (including firms with no reported capital leases), whereas Models 3 and 4 are based on a subsample that excludes firms with no reported capital leases. Market value of the firm is defined as the book value of total assets less the book value of common equity plus the market value of equity plus the present value of operating lease liabilities. The total debt ratio (DR) is defined as the sum of total long-term debt plus short-term debt less capital leases and scaled by the market value of the firm. Growth opportunities (GO) is the market-to-book ratio and defined as the market value of assets divided by the book value of total assets. Market value of assets is defined as the book value of total assets less the book value of common equity plus the market value of equity. Collateral value of assets (CVA) is defined as net property, plant, and equipment divided by the book value of total assets. The before-financing book-simulated marginal tax rates (TR) are obtained from John Graham’s website. Missing tax rates are predicted based on the predicted book simulated coefficients estimated in Graham and Mills (2008). The measure of financial distress (ZS) is the modified version of Altman’s (1968) z score. Firm size (Size) is measured as the natural log of the market value of the firm. Industry (ID) is measured using a dummy variable that is equal to one if the firms is a restaurant firm and zero otherwise. Year dummies (Y07 and Y08) for 2007 and 2008 are included in the model with 2006 as the base year. All OLS regressions are estimated with Newey–West heteroskedasticity autocorrelation consistent (HAC) robust standard errors with t statistics in parentheses.
p < .10. **p < .05. ***p < .01.
The results also show growth opportunities to be significantly and negatively associated with leasing, consistent with the notion that firms with higher growth opportunities are less likely to use leasing. The coefficient on the collateral value of assets provide no evidence at the 5% level that firms with valuable assets can obtain more favorable financing at a lower cost and consequently increase their debt capacity. Given the relatively large magnitude of operating leases relative to total leasing, predicted before-financing marginal tax rates show a significantly negative association with leasing, consistent with prior research findings on the relation between operating leases and tax rates (Graham et al., 1998) and providing some evidence of the tax hypothesis that low–tax rate firms lease more than high–tax rate firms. Although the relation between financial distress and leasing is significantly negative in Models 1 and 2 (using all observations) at the 10% level, the results are however insignificant in Models 3 and 4 because of a smaller subset of firms. The significant negative relation between firm size and leasing shows a pervasive use of leasing among smaller firms and to a lesser extent among larger firms. The evidence is also weak in Models 1 and 2 to show that restaurant firms have a lower magnitude of total lease usage compared with retail firms. Finally, the inclusions of year dummy variables indicate that sample firms significantly increased the magnitude of leasing in 2008 relative to the base year 2006.
The empirical evidence thus far has suggested that leasing and debt are substitutes consistent with prior studies in the United States and the United Kingdom (Adedeji & Stapleton, 1996; Beattie et al., 2000; Marston & Harris, 1998). When leasing is used as the dependent variable in the regression models in Table 6, the estimated debt coefficient provides a measure of the lease-to-debt displacement ratio and not the debt-to-lease displacement ratio. Given evidence that leases and debt are substitutes, Adedeji and Stapleton (1996) and Beattie et al. (2000) proposed and estimated the debt-to-lease displacement ratio by running regressions with debt as the dependent variable and leasing as the independent variable. Similarly, the OLS regressions in Table 6 are reestimated with debt as the dependent variable and leasing as a predictor. This estimate was also made using the subsample of firms with observed capital leases. The total leasing ratio is negatively significant at the 1% level with an average estimated debt-to-lease displacement ratio of −41% in Models 1 and 2 and −50% in Models 3 and 4 of Table 6 for an average of −45% across all four models. Based on the analysis of uncensored observations, these results indicate that, on average, $1.00 of total leasing (capital leases plus operating leases) displaced approximately $0.50 of debt during the 2006 to 2008 period. To determine the magnitude of substitutability between capital leases and operating leases, each of the regressions were reestimated with separate lease ratio variables, one for capital leases and one for operating leases. The coefficients on operating leases were all negatively significant at the 1% level and identical to the comprehensive leasing estimates reported above (average of 51% for Models 3 and 4 in Table 6). Similarly, the capital lease coefficients were also negatively significant at the 1% level but lower than the operating lease estimates (average of 46% for Models 3 and 4 in Table 6). These estimates are in sharp contrast and differ from those of Beattie et al. (2000), who found extreme variability and a lack of any significant relationship between the capital lease ratios and debt among U.K. firms.
Sensitivity Analysis
To test whether the results in Table 6 were driven by the measures used in this study, several alternative specifications were developed. First, the variables were scaled by total assets (assets + PV of operating leases) instead of market value of the firm. The results were qualitatively similar, indicating that the results were not influenced by the choice of deflator. The results are also similar when total long-term debt is used instead of total debt. Second, as an alternative definition, the collateral value of assets was measured by adding operating leases to the numerator and denominator of the proxy variable. The results for collateral value of assets in all four models in Table 6 were highly significant at the 1% level whereas the industry dummy was negatively significant at the 5% level for the subsample, with all other results remaining unchanged. This increase in significance was driven by a positive bias in the correlation between collateral value of assets and the comprehensive lease ratio since both variables use the same deflator. A similar reasoning applies to the growth opportunity proxy when operating leases are added to the numerator and denominator of the market-to-book ratio. Third, when the cross-sectional regressions were performed by year for the overall sample and subsample, again the signs and significance of the coefficients were similar to those in Table 6 with the exception of the collateral value of assets and industry dummy variables, which were insignificant in all regressions. Fourth, instead of using a mix of book-simulated and predicted book-simulated marginal tax rates, only predicted marginal tax rates were used for all firms. This is because only 476 out of 699 book-simulated marginal tax rates were available and the balance 223 missing observations had to be predicted using the predicted coefficients in Graham and Mills (2008). Predicted marginal tax rates were found to be highly correlated and negatively significant with leasing at the 1% level in all four models. The results were again similar to those in Table 6, with the exception of the z score coefficient, which was insignificant in all four models. Finally, the regressions were performed separately for each industry sector using all sample observations. The results for both sectors were again similar to the main results in Table 6 for debt, growth opportunities, and firm size, with the exception of collateral value of assets, z scores, and marginal tax rates. For the restaurant industry, the collateral value of assets is positively significant at the 1% level in Models 1 and 2 but negatively insignificant for retail firms. Unlike retail firms, restaurant firms have more valuable tangible assets that can support higher amounts of leases that are often viewed as debt. This is also consistent with previous lodging research that found a significant positive relation between debt and the collateral value of assets (Dalbor & Upneja, 2004). The relation between financial distress and leasing is insignificant for retail firms, but it is significantly negative at the 5% level for restaurant firms in all four models indicating that restaurant firms in financial distress were more likely to use leases. The association between marginal tax rates and total leasing for the overall sample was also found to be insignificant for restaurant firms compared with retail firms.
Fixed effects regressions
The use of a pooled OLS cross-sectional regression assumes that the effect of each explanatory variable on the dependent variable has remained constant over time. Pooled OLS will produce biased and inconsistent estimators if unobserved factors are correlated with the explanatory variables, resulting in omitted variable bias. Therefore, to mitigate this potential problem, a fixed effects regression is specified as an alternative model specification. Fixed effects estimation removes any time constant unobserved firm characteristics by holding constant the average differences across firms and observable or unobservable predictors before model estimation. This is done by subtracting each firm’s average observation from the individual firm-year observation. The fixed effects model was also compared with the random effects model using the Hausman (1978) test, and the null hypothesis of no systematic differences was rejected at the 1% level indicating that the fixed effects model specification was appropriate. Serial correlation in the residuals was tested using a modified Durbin–Watson (Bhargava, Franzini, & Narendranathan, 1982) and the Baltagi–Wu (1999) locally invariant test. Although exact critical values of these statistics are unavailable in the literature, values less than 1.5 are generally accepted as reliable indicators of positive serial correlation. The test statistic yields a value of at least 1.52 for the modified Durbin–Watson and a value of 2.35 for the Baltagi–Wu test. Given these values, it is safe to assume some serial correlation in the panel data. Therefore, to control for serial correlation in the data, fixed effects panel regressions were specified with Newey–West heteroskedasticity robust and autocorrelation consistent standard errors clustered by firms (Wooldridge, 2009). Stock and Watson (2008) recommended using these cluster robust standard errors as the appropriate correction for serial correlation.
The results of this analysis are presented in Table 7. Again, with the exception of marginal tax rates, collateral value of assets, and z scores, the findings are largely similar to the pooled results in Table 6, providing further support for the robustness of the results. More important, the results provide strong support and conclusive evidence that leases and debt are substitutes in the restaurant and retail industry.
Fixed Effects Regression Using Comprehensive Lease Ratio
Note. This table presents the results of fixed effect regressions where the dependent variable in each model is the comprehensive lease ratio (CLR), which is defined as the sum of the book values of capital leases plus the present value of operating leases and divided by the market value of the firm. Models 1 and 2 include all 699 observations, while Models 3 and 4 includes only firms with capital leases. Market value of the firm is defined as the book value of total assets less the book value of common equity plus the market value of equity plus the present value of operating lease liabilities. The total debt ratio (DR) is defined as the sum of total long-term debt plus short-term debt less capital leases and scaled by the market value of the firm. Growth opportunities (GO) is the market-to-book ratio and defined as the market value of assets divided by the book value of total assets. Market value of assets is defined as the book value of total assets less the book value of common equity plus the market value of equity. Collateral value of assets (CVA) is defined as net property, plant, and equipment divided by the book value of total assets. The before-financing book-simulated marginal tax rates (TR) are obtained from John Graham’s website. Missing tax rates are predicted based on the predicted book-simulated coefficients estimated in Graham and Mills (2008). The measure of financial distress (ZS) is the modified version of Altman’s (1968) z score. Firm size (Size) is measured as the natural log of the market value of the firm. Industry (ID) is measured using a dummy variable that is equal to one if the firms is a restaurant firm and zero otherwise. Year dummies (Y07 and Y08) for 2007 and 2008 are included in the model with 2006 as the base year. All regressions are estimated with Newey–West heteroskedasticity autocorrelation consistent (HAC) standard errors with t statistics in parentheses.
p < .10. **p < .05. ***p < .01.
Summary and Conclusions
This study investigated the relation between leasing and debt to determine whether they are substitutes or complements. Using a more comprehensive measure of total leasing, the results of this study provide evidence that leasing and debt are substitutes in the restaurant and retail industry. A comprehensive measure of leasing also displayed a stronger level of substitutability than observed in previous studies. Leasing consumed debt capacity on a less than dollar-for-dollar basis. On average, $1 of leasing displaced approximately $0.45 of debt. The magnitude of total leasing is also significantly greater in the retail industry than in the restaurant industry. Restaurant firms are observed to use more capital leases whereas retail firms reported a greater magnitude of operating leases in their respective capital structures. Moreover, restaurants firms were significantly smaller in size, had higher collateral values of assets, lower z scores, and higher growth opportunities in comparison with retail firms.
The results of this study also provide contrasting evidence to show that the complementary relation observed in prior studies is largely due to sample-specific characteristics. The initial complementary relation between capital leases and debt was driven by more than half the sample firms without any reported capital leases including firms with no reported debt in the capital structure. It was also observed that many of these firms were small in size with low debt ratios and little or no capital leases. In addition, these small firms were also among the largest users of operating leases unlike larger firms that reported higher debt and capital leases ratios and lower amounts of operating leases. A subsample analysis of firms with reported capital leases further confirmed the inverse relation between debt and capital leases, a result that is also robust to alternative specifications of the empirical models. Additionally, growth opportunities and firm size were found to be important determinants of leasing, consistent with prior research findings and supporting the trade-off theory and financial contracting cost theory. There is also some evidence that the collateral value of assets is positively related to leasing, implying that restaurant firms can use their valuable assets to obtain more favorable lease financing at a lower cost. On the other hand, there is mixed evidence on the relation between leasing and marginal tax rates and financial distress. There are two possible explanations for these mixed findings between the pooled OLS and fixed effects regressions. First, a large number of predicted book-simulated tax rates were used to “fill” in missing marginal tax rates, thus yielding a less reliable proxy. Second, the z-score proxy ignores operating leases and thus may fail to fully capture financial distress. Overall, this study has made a significant contribution to the literature on leasing by providing relevant information to show the inverse relation between leases and debt and extending previous restaurant leasing research in the hospitality industry.
Implications for Future Research
In August 2010, the International Accounting Standards Board and the Financial Accounting Standards Board jointly issued an Exposure Draft of proposed new rules on lease accounting that would fundamentally change the accounting and reporting of leases. In particular, these rules will eliminate the distinction between capital leases and operating leases and require lessees to apply a single right-to-use model in which all leases will be recognized as assets and liabilities in the financial statements. The financial implications of these new rules will be significant for the restaurant and retail sectors given the widespread and pervasive use of off–balance sheet operating leases in these sectors. Evidence from industry reports and the author’s own research suggests that these rules are expected to negatively affect financial statement ratios and presentation especially leverage, capital, and performance ratios. In addition, anecdotal evidence and research by the author also suggests that debt covenant agreements, incentive compensation plans, lease versus buy decisions, cost of capital, and credit ratings could also be affected.
The impact of the proposed new rules on the capital structure and future financial policy decisions will provide a basis for conducting further research on leasing. Future research should investigate the impact of the new rules on the aforementioned areas just before and after the rules are implemented as well as assess its impact on firm values. Given the importance of operating leases and its magnitude relative to total leasing, future capital structure studies should include a comprehensive measure of leasing especially when using data from prior years to investigate leasing-related research questions and larger samples over a longer time horizon to determine whether marginal tax rates and z scores are important determinants in the restaurant industry. Finally, it should be noted that the estimates and assumptions made in this study are subject to certain limitations. Many features of operating leases have been excluded such as residual value guarantees, renewal terms, and contingent rentals. For example, the exclusion of contingent rentals and residual value guarantees would understate the present value of operating lease. Similarly, interest rate and lease term assumptions used in capitalizing leases will differ from one lease to another lease. Capitalizing operating leases over shorter lease terms, excluding renewal terms, and using higher discount rates would produce lower present values. To the extent that such information on lease terms is unavailable and unaccounted for, the present value estimates in this study are understated. This study is also limited in its investigation of the lease–debt substitution over a relatively short time horizon.