Are Leases and Debt Substitutes? Evidence form Capital Structure Adjustment

Transcription

1 Are Leases and Debt Substitutes? Evidence form Capital Structure Adjustment Shofiqur Rahman New Mexico State University Department of Finance College of Business, Rm. 317 Las Cruces, NM , USA (575) Harikumar Sankaran New Mexico State University Department of Finance College of Business, Rm. 222 Las Cruces, NM , USA (575) January 2016 ABSTRACT Traditional finance theory suggests that leases and debt are substitutes. Ang and Peterson (1984) however, present a puzzle by showing that leases and debt are instead complements to each other. A large body of literature has developed since Ang and Peterson (1984), but the empirical evidence is inconclusive. We view the leasing puzzle from the perspective of the trade-off theory of capital structure. Specifically, we examine the relation between leases and debt using firm s deviation from the target leverage. We find that firms that are underlevered (overlevered) exhibit higher (lower) lease intensity. Our results support the hypothesis that leases and debt are substitutes and are robust to alternative measure of both lease intensity and target leverage. JEL classification: G32 Keywords: Capital Structure, Operating leases, Debt financing Corresponding author: shofi@nmsu.edu. The author retains responsibility for any remaining errors.

2 1. Introduction Leases and debt are important sources of financing for U.S. firms. 1 Typical capital structure theory suggests that leases and debt are substitutes (i.e., an increase in debt financing is associated with a decrease in lease financing). However, the empirical evidence on such relationship is mixed. For example, Ang and Peterson (1984) were the first to present leasing puzzle by showing that leases and debt are complements. 2 In contrast, a number of studies argue that leases and debt are substitutes (e.g., Marston and Harris, 1988; Krishnan and Moyer, 1994; Adedeji and Stapleton, 1996; Beattie, Goodacre, and Thomson, 2000; Yan, 2006). In this paper, we examine the substitution hypothesis using a novel experimental setting provided by firms adjustment toward capital structure. Investigating change in lease intensity while firms move towards the targets enables us to capture the impact of firms deliberate debt rebalancing. Capital structure literature provides empirical evidence that firms have target debt ratios and that, if they deviate from the targets, firms gradually adjust back to the target capital structures (e.g., Leary and Roberts, 2005; Frank and Goyal, 2007). 3 In the presence of financing frictions, such deviation from the target sometimes creates leverage deficits that may result in a firm issuing further debt. If leases and debt are substitutes, firms would cut on lease financing as they increase the debt level. Specifically, under (over) levered firms are expected to reduce (increase) lease intensity as they move up (down) towards the target debt ratios. One way to empirically test this substitution hypothesis is to test the association between lease intensity and the extent of deviation (hereafter leverage deficit) from the target capital structure. A positive relation between lease intensity and leverage deficits supports the substitutability hypothesis, while no relation or a negative relation goes in favor of leases and debt complements. 1 Based on Compustat, an average firm reports yearly debt amount of $456 million and lease amount of $114 million during the year between 1980 and Moreover, Leasing accounts for about one third of new equity investment (Beatty, Liao, and Weber, 2010) and leasing is comparable to long term debt in terms of importance (Eisfeldt and Rampini, 2009). 2 See also Lewis and Schallheim (1992) that supports that leases and debt are complements. 3 For example, Bradley, Jarrell, and Kim (1984), Stulz (1990), Rajan and Zingales (1995), Hovakimian, Opler, and Titman (2001), DeAngelo, DeAngelo, and Whited (2011), Graham and Harvey (2001), and etc. find that firms do have target capital structures.

3 We employ a two step estimation process. First, as in Fama and French (2002), we estimate the target debt ratio using yearly regressions. Second, we calculate leverage deficits by deducting current debt level from the estimated target debt ratios. A positive value for leverage deficit indicated under leverage and a negative value indicates overleverge. In order to test the prediction that lease intensity and leverage deficits are positively associated, we regress lease intensity on leverage deficits (i.e., target debt ratio minus current debt ratio) along with a battery of control variables similar to those used in Sharpe and Nguyen (1995) and Beatty, Liao, and Weber (2010). Using lease intensity measured as the ratio of the proportion of net, property, plant, and equipment financed through leases to total capital financed through both leases and own capital as the dependent variable, we find that increase in leases is associated with an increase in leverage deficits. Specifically, a 1 percent increase in leverage deficits (i.e., reduction of debt) leads to, ceteris paribus, an increase in lease intensity ratio by about percent. Our results are also economically significant. Using the mean lease amount of $ million for our sample firms, the movement of leverage deficits from 25 th percentile to 75 th percentile can be translated into an increase of $2.81 million of leased assets for an average firm. Overall, our findings are consistent with the hypothesis that leases and debt are at least substitutes, if not perfect substitutes. We include firm and industry fixed effects in all of our regression models to capture time-invariant and unobservable omitted variable effects. We also analyze the impact of financial constraints in the relationship between lease intensity and leverage deficits. Consistent with our prediction, we find that the leases and debt substitutability is more pronounced in financial constrained firms. Our measures are robust to alternative measures of lease intensity and alternative method of target debt estimation. Overall, our paper contributes to the existing literature in two different ways. First, we add to the lease puzzle explained by Ang and Peterson (1984) by providing empirical evidence in support of substitutability hypothesis. Second, we exploit firms capital structure adjustment framework to investigate the association between leases and debt. We argue that such natural environment enables us to capture the impact of firms deliberate debt rebalancing on lease financing. While Myers, Dill, and Bautista (1976) and

4 Ross, Westerfield, and Jaffe (1990) modeled leases in the light of optimum capital structure theory, to our knowledge, we are the first to provide empirical analysis using firms capital structure adjustment in explaining lease and debt substitutability. The remainder of our paper is structured as follows: we discuss related literature and develop our main hypothesis in Section 2; Section 3 describes our data, variable definition, and research design. In Section 4, we discuss our empirical findings. Section 5 contains our additional analysis and robustness checks; and Section 6 concludes the paper. 2. Literature and hypothesis development Firm values are irrelevant to the choice of capital structure in a perfectly competitive market (e. g., Modigliani and Miller, 1958). However, in the presence of capital market frictions, leverage deficits can play a major role in corporate financing policies such as the choice between lease and debt. Static-trade-off theory of capital structure implies that firms have target capital structures and, using cost and benefit tradeoff, firms optimize their capital structure. 4 It also suggests that leases and debt are substitutes. Ang and Peterson (1984) were the first to explain leasing puzzle by showing that leases and debt are complements. Since then a body of literature provides evidence in support to the substitution hypothesis (e.g., Marston and Harris, 1988; Krishnan and Moyer, 1994; Adedeji and Stapleton, 1996; Beattie, Goodacre, and Thomson, 2000; Yan, 2006). We do not yet have a conclusive answer to the puzzle. If leases and debt are substitutes, in the event of capital structure adjustment, under levered firms would cut on lease financing in response to increasing debt level in order to move up to the target and viceversa. One possible way to observe this adjustment of lease and debt is through the impact of leverage deficits. Specifically, when an under levered firm wants to increase the debt level, it reduces the gap 4 Existing studies suggesting that firms have leverage targets include Hovakimian, Opler, and Titman (2001), DeAngelo, DeAngelo, and Whited (2011), and Graham and Harvey (2001).

5 between target leverage and current leverage (i.e., leverage deficits) and cut on lease financing given substitution hypothesis is true. We, therefore, test the substitutability of leases and debt by hypothesizing that lease intensity is positively associated with leverage deficits. 3. Sample selection, variable measurement, research design 3.1 Sample selection We start with all firm-years in the Compustat database during a sample period between 1980 and We apply several sample selection criteria in constructing the final sample. First, we exclude financial firms (SIC ) and utilities (SIC ) because the financial statements of these firms are different and are subject to different regulations that may impact their capital structure and lease decisions. Second, we require all firm-years in our sample to have nonnegative assets, sales, and share prices. Third, we drop observations that have missing operating lease intensity ratio. Finally, we lose few observations while estimating target leverage and eliminate observations for missing control variables that are used in all of our regression models. As a result of imposing all these sample selection criteria, our final sample consists of 111,876 firm-year observations. Table 1 presents the sample selection process in details. Insert Table 1 here 3.2 Measuring lease intensity A comprehensive measure of lease intensity is the ratio of leases to total capital in which total capital includes both leased and own capitals. In this paper, we try to approximate this ratio with annual operating lease ratio (OLR), which is estimated as the ratio of the proportion of net, property, plant, and equipment financed through leases to total capital financed through both leases and own capital. Since operating leases are off-balance sheet items and operating leases are unavailable, we add current rental expenses to the discounted values of future lease obligations. Using COMPUSTAT variables, we measure operating lease intensity (OLR) as in the following equations:

6 Total lease = rental expense + PV(future rental commitments) (1) Lease intensity (OLR) = Total lease Total lease + PPENT (2) where, rental expense (XRENT) is the total current lease obligation. Following existing studies (e.g., Graham, Lemmon, and Schallheim, 1998; Lim, Mann, and Mihov, 2003; Robicheaux, Fu, and Ligon, 2008; Devos and Rahman, 2014), we use a discount rate of 10% to calculate the present value of future five years and thereafter lease obligations (MRC1, MRC2, MRC3, MRC4, MRC5, and MRCTA). Our measure is different from those of Sharpe and Nguyen (1995) and Eisfeldt and Rampini (2009) in that we augment their measure by adding future rental expenses. We also use two alternative measures of lease intensity as robustness checks. First, we employ lease intensity measure of Beatty et al., (2010) calculated as capitalized lease expenditure (lagged MRC1 10) divided by the sum of PPENT and capitalized lease expenditure. Second, we employ lease intensity measure of Sharpe and Nguyen (1995) calculated as net capital lease (PPENLS) divided by lagged PPENT. Since the variable PPENLS is mostly missing in COMPUSTAT, we lose a significant number of observations from our sample. Therefore, we don t rely on this measure while interpreting the regression estimates. 3.3 Measuring target debt ratios and leverage deficits As firms have target capital structure and, based on the costs and benefits trade-off, they can be classified as either under or over levered, estimating target leverage is the first step in our analysis. Following previous studies (e.g., Rajan and Zingales, 1995; Fama and French, 2002; Kayhan and Titman, 2007; Byoun, 2008; Warr, Elliott, Koëter-Kant, and Öztekin, 2012), we employ two-step estimation procedure. In the first step, we estimate target leverage by running yearly regressions (e.g., Eq. (3)). In the second step, we use Eq. (4) to calculate leverage deficits (i.e., target leverage minus current leverage), which is used in Eq. (5) to test the association between lease intensity and leverage deficits.

7 Lev i,t+1 = βx i,t + ε i,t (3) LevDef i,t = Lev i,t+1 Lev i,t (4) where, Lev i,t represents forward looking target leverage ratio for both book leverage (BLev) measured as total debt (DLTT+DLC) scaled by book value of assets (AT) and market leverage (MLev) measured as total debt (DLTT+DLC) scaled by debt plus market equity (DLTT+DLC+(PRCC_f*CSHO)). The independent variable X i,t refers to the firm specific characteristics that include operating income (OI) measured as the operating income (OIBDP) scaled by book value of assets, market-to-book ratio (MB) where market value is measured as the total equity (SEQ) minus deferred taxes (TXDITC) plus market value of equity (PRCC_f*CSHO), log of total asset, fixed assets measured as plant, property, and equipment (PPENT) scaled by total assets, research and development measured as the research & development expense (XRD) scaled by sales (SALE), dividend measured as common stock dividend (DV) scaled by total assets, company's likelihood of bankruptcy measured as: [3.3 OIADP + SALE RE (ACT-LCT)]/AT, marginal tax rate that equals to the statutory tax rate if the firm reports no net operating tax loss carry forward and positive pre tax returns, and industry median debt ratio calculated as the median debt ratio of the industry where industry is defined as the first two-digit SIC code. 3.4 Research design In this paper, we investigate whether leases and debt are substitutes. This substitution hypothesis can be confirmed if we find that over levered firms (i.e., firms with lower leverage deficits) have lower lease intensity and vice-versa. Specifically, we test the hypothesis that leverage deficit is positively associated with lease intensity. We estimate the following baseline equation:

8 OLR i,t = α 0 + β 1 LevDef i,t + β 2 NoDiv i,t + β 3 Earnings i,t + β 4 7 S&PRatings i,t + β 8 TaxRate i,t + β 9 SmallTaxLCF i,t + β 10 LargeTaxLCF i,t + β 11 Loss i,t + β 12 Size i,t + Firm dummy + Industry dummy + ε i,t (5) where, the dependent variable (OLR) is the ratio of the proportion of net, property, plant, and equipment financed through leases to total capital financed through both leases and own capital. The subscripts i and t index firm and year, respectively. The regression model includes firm and industry fixed effects with standard errors clustered by years (Petersen, 2009). Our main variable of interest is LevDef, which refers to leverage deficits measured as forward looking target leverage minus current leverage ratio (i.e., Eq. (4)). A positive (negative) value for LevDef is considered under (over) levered. In order for substitution hypothesis to hold, we argue that firms with higher leverage deficits will more likely to have higher operating lease ratios and, therefore, expect the coefficient on the LevDef variable to be positive. Our model controls for several firm level characteristics that may potentially affect lease ratio. Since leasing decision is dependent on whether a firm is financially constraint (e.g., Sharpe and Nguyen, 1995), we include NoDiv, Earnings, and variables related to S&P ratings as controls and expect the coefficients to be positive. NoDiv is a dummy variable equals to 1 if the firm paid no dividend (DVC) in that year and zero otherwise. Earnings is the operating income before depreciation (OIBDP) divided by sales (SALE). All S&P ratings are indicator variables equal to 1 if the firm has S&P Domestic Long Term Issuer Credit Rating (SPLTICRM) and zero otherwise. The rating groups are partitioned according to: AAA through AA, A+ through A, BBB+ through BBB, BB+ through D and Unrated. Taxes are another primary drivers of lease decision. Firms with lower marginal tax rate have lower depreciation tax shield and, hence, more likely to lease than buy (e.g., Smith and Wakeman, 1985). Tax rate is the tax expense (TXT) divided by pre-tax income (PI). Small tax-loss CF is a dummy variable equals 1 if firm had a positive tax loss carry-forward (TLCF) not exceeding current year OIBDP and zero otherwise. Large tax-loss CF is a dummy variable equals to 1 if firm had a positive tax loss carry-forward (TLCF) exceeding current year

9 OIBDP and zero otherwise. As in Beatty et al. (2010), we also control for loss and firm size. Because loss firms are possibly in lower marginal tax rate and are less likely to enjoy depreciation tax shield, these firms are more likely to lease. In addition, bigger firms are less likely to be financially constrained and, therefore, less likely to lease. Specifically, the coefficients on Loss and Size are expected to be positive and negative, respectively. 4. Results and discussions 4.1 Target leverage estimation results We present the descriptive statistics for all the variables used in the target leverage estimation in Table 2 for a sample of 111,876 firm-year observations. 5 The average (median) book leverage is (0.168) and the average (median) market leverage is (0.130). We use several firm characteristics that are commonly used in leverage estimation as control variables. Insert Table 2 here The average (median) firm in our sample has operating income of (0.109), market-to-book ratio of (1.406), a depreciation and amortization of (0.039), fixed assets of (0.207), and a ratio of R&D to sales of (0.002). The average (median) firm in our sample has AZ index of (1.837) and size of (4.638). We also control for industry debt ratios. The mean for industry median book (market) leverage is (0.151). We estimate the targets using Eq. (3) for both book leverage (BLev) and market leverage (MLev). These yearly regressions are based on non-missing observations for all control variables. Given that our sample consists of data from 1980 to 2013, we have 34 estimates for each independent variable. We calculate the mean slope coefficients by averaging the slopes for the 34 annual regressions. As in Fama and 5 All variables are winsorized at 1 percent and 99 percent tails except for DIV and RND that are winsorized at 0 and 99 percent tails..

10 French (2002), the time series standard error is calculated as time series standard deviations of the regression coefficients divided by (34). In Table 3, we present the mean coefficient estimates, mean adjusted R-square, and p-values for the variables. The adjusted R-squares for book (market) leverage estimation is (0.2910). Insert Table 3 here All the coefficients are statistically significant at 1% level except for the tax rate (Tax) in book leverage (BLev) regression. We report negative association between debt ratio and market-to-book, depreciation, research and development expenses, dividend, and Altman Z-scores. These results are consistent in both book leverage and market leverage estimations. All other variables have a positive effect on debt ratios. Consistent with the results in past studies (e.g. Frank and Goyal, 2004; Byoun, 2008), the coefficients for industry median are positive for both debt ratios. 4.2 Univariate results Now that we have forward looking leverage target for each firm-year observations. We then calculate leverage deficits (i.e., target leverage minus current leverage) using Eq. (4). Starting from a debt ratio well below the target debt ratio (i.e., positive value for LevDef) and as a firm moves up towards the target, the value for leverage deficits gradually decreases. This leverage deficit variable can even show a negative value if the firm continues to move up and becomes over levered. More specifically, under (over) levered firms have positive (negative) values for LevDef variable. As the first attempt to show that higher leverage deficits firms (e.g., under levered) have higher lease intensity, we construct decile ranks for firms based on their book leverage and market leverage ratio in two-digit SIC code. We calculate mean lease ratios across deciles. Figure 1 presents the results. Insert Figure 1 here

11 As the figure shows, lowest deciles (i.e., firms well above the target leverage) based on both book- and market leverage deficits firms have lower lease intensity. In contrast, the highest deciles firms have higher lease intensity. These findings shed some lights on the argument that firms increase the debt level in an effort to move up to the target leverage by cutting lease financing, suggesting that leases and debt are substitutes. Insert Table 4 here In Table 4, we present the descriptive statistics and univariate comparison for all variables used in the baseline regression (e.g., Eq. (5). Firms are divided into two groups based on the leverage deficits. As the table shows, the average firm in our sample has operating lease intensity of 0.328, tax rate of 0.209, and size of Univariate comparison shows that under levered firms (e.g., LevDef>0) have higher lease intensity than do over levered firms. We also report a difference of in lease intensity in market leverage deficits sub-sample. These differences are significant at 1% level of significance. All other firm characteristics, except for Earnings and tax rate, are statistically different in both samples. Collectively, the univariate results also suggest that leverage deficits are positively associated with lease intensity. 4.3 Multivariate results In this section, we test the association between lease intensity and leverage deficits in multivariate settings, controlling for a number of firms characteristics. The results are reported in Table 5. Colum (1) and (3) reports the baseline regressions for both book leverage- and market leverage deficits. Using fixed effects to mitigate the impact of unobserved and time-invariant firm factors and industry factors, we show both in Column (1) and (3) that the coefficients on LevDef are positive and statistically significant (0.123, p = 0.000) for book leverage deficits and (0.084, p = 0.000) for market leverage deficits. We also estimate our regressions using firm-year fixed effect in column (2) and (4). The coefficients are consistently positive and statistically significant. Collectively, the results in Table 5 supports the prediction that the leverage deficits are positively associated with lease intensity (i.e.. leases and debt are substitutes).

12 Insert Table 5 here We also estimate the baseline model using under- and over levered variable (e.g., UnderLev_dummy) instead of leverage deficits, which is a continuous variable. We define UnderLev_dummy as an indicator variable equals 1 if LevDef is positive or zero otherwise. The results are reported in Table 6. Using fixed effects to mitigate the impact of unobserved and time-invariant firm factors and industry factors, we show both in Colum (1) and (3) that the coefficients on UnderLev_dummy are positive and statistically significant (0.040, p = 0.000) for book leverage and (0.028, p = 0.000) for market leverage. The results echo those in Table 5 and further support that leases and debt are substitutes. Insert Table 6 here Our results also have economic significance. For instance, a 1 percent increase in leverage deficits (i.e., reduction of debt) leads to, ceteris paribus, an increase in lease intensity by about percent. We also analyze the economic significance of our findings by multiplying the coefficient estimates for the LevDef variable by the change in the leverage deficits when moving from the 25 th percentile to the 75 th percentile (an increase of LevDef by 0.2). 6 The economic effect of LevDef on lease intensity can be interpreted as an increase in lease intensity by 2.46 percent. Using the mean lease amount of $ million for our sample firms, the effect can be translated into an increase of $2.81 million of leased assets for an average firm. Overall, our findings are consistent with the hypothesis that leases and debt are at least substitutes, if not pure substitutes. 5. Additional analysis and robustness checks 5.1 Impact of financial constraints 6 Based on an unreported table but available upon request, the values for leverage deficits at the 25th and 75th percentile are and 0.12, respectively. In addition, moving from 25th to 75th percentile indicates that the firm is moving towards under levered area or further deep into under levered area.

13 A number of studies (e.g., Sharpe and Nguyen,1995; Eisfeldt and Rampini, 2009) argue that financially constrained firms have limited access to the debt and, in turn, are more likely to resort to lease financing. Besides, leases provide lessor with assets repossession capacity in the eve of bankruptcy. Motivated by these findings, we test the prediction that the association between lease intensity and leverage deficits are more pronounced among financially constrained firms. We, therefore, introduce an interaction term, LevDef Fin-Con, into our baseline regression. Given a variety of proxies available for financial constraints (e.g., Kaplan and Zingales, 1997; Li, 2011), we use only two proxies for financial constraint namely, zero dividend and KZ index. The results are reported in Table 7. Insert Table 7 here As in column (1) and (2), the coefficients on LevDef Fin-Con are positive and statistically significant (0.097, p = 0.000) for zero dividend and (0.096, p = 0.000) for KZ-index. We find similar results in column (3) and (4) for market leverage. Collectively, these results support the prediction that the association between lease intensity and leverage deficits are more pronounced in financially constrained firms. Specifically, the leases and debt substitutability is more pronounced in financial constrained firms. 5.2 Alternative method of target estimation- Blundell Bond GMM two-step estimation In order to mitigate the concern that dynamic panel may produce biased speed of capital structure adjustment (e. g., Baltagi, 2001), we estimate target capital structure using a two-step system generalized method of moments (GMM) of Blundell and Bond (1998). The calculated leverage deficits are then used to re-estimate Eq. (5) to test the association between lease intensity and leverage deficits. The results are presented in Table 8. Insert Table 8 here As the table shows, the coefficients on LevDef are positive and statistically significant (0.182, p = 0.000) for book leverage and (0.772, p = 0.000) for market leverage. The results are similar to what we find in

14 previous tests. Therefore, our results are robust and not driven by any particular method of leverage target estimation. 5.3 Alternative measures of lease intensity We also use two alternative measures of lease intensity ratios as robustness checks. First, following Beatty et all. (2010), we measure lease intensity as capitalized lease expenditure (lagged MRC1 10) divided by the sum of PPENT and capitalized lease expenditure. Second, we employ lease intensity measure of Sharpe and Nguyen (1995) calculated as net capital lease (PPENLS) divided by lagged PPENT. We reestimate our baseline regression and the results are presented in Table 8. Insert Table 9 here For capitalized lease ratio and as shown in column (1) and (3), the coefficients on LevDef are positive and statistically significant (0.133, p = 0.000) for book leverage and (0.075, p = 0.000) for market leverage. However, for CLR, the coefficients in column (2) is insignificant and in column (4) is significant only in 10% level of significance. One potential reason for this weak result could be the number of observation used in the regression models. As the variable net capital lease (PPENLS) is sparsely populated in Compustat, we lost almost 93,425 firm-year observations. 6. Conclusion A large body of finance literature suggests that firms do have target capital structure. In a perfect world, firms move back to the target capital structure instantaneously if they deviate from the target. However, in the presence of financing frictions, capital structure adjustment is a function of costs and benefits trade-off. The lags in capital structure adjustment and the need for alternative sources of financing gives us an opportunity to test if lease financing acts as a substitute for debt financing. Typical finance theory suggests that leases and debt are substitutes. A number of studies also document that leases and debt are complements instead. Yet, our understanding regarding the relation

15 between leases and debt are incomplete. If leases and debt are considered substitutes, an increase in leverage level will be associated with a decrease in lease ratio. We find strong support for the substitution hypothesis. We also document that leases and debt substitution is more pronounced in financially constrained firms. Overall our results are robust to alternative measures of lease intensity and of target estimation.

16 References Adedeji, A., & Stapleton, R. C. (1996). Leases, debt and taxable capacity. Applied Financial Economics, 6(1), Beattie, V., Goodacre, A., & Thomson, S. (2000). Operating leases and the assessment of lease debt substitutability. Journal of Banking & Finance, 24(3), Beatty, A., Liao, S., & Weber, J. (2010). Financial reporting quality, private information, monitoring, and the lease-versus-buy decision. The Accounting Review, 85(4), Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of econometrics, 87(1), Bradley, M., Jarrell, G. A., & Kim, E. (1984). On the existence of an optimal capital structure: Theory and evidence. The journal of Finance, 39(3), Byoun, S. (2008). How and when do firms adjust their capital structures toward targets?. The Journal of Finance, 63(6), DeAngelo, H., DeAngelo, L., & Whited, T. M. (2011). Capital structure dynamics and transitory debt. Journal of Financial Economics, 99(2), Devos, E., & Rahman, S. (2014). Location and lease intensity. Journal of Corporate Finance, 29, Eisfeldt, A. L., & Rampini, A. A. (2009). Leasing, ability to repossess, and debt capacity. Review of Financial Studies, 22(4), Fama, E. F., & French, K. R. (2002). Testing trade off and pecking order predictions about dividends and debt. Review of financial studies, 15(1), Frank, M. Z., & Goyal, V. K. (2004). The effect of market conditions on capital structure adjustment. Finance Research Letters, 1(1), Frank, M.Z., Goyal, V.K., (2007). Capital Structure Decisions: Which Factors are Reliably Important? Unpublished Working Paper. University of Minnesota and Hong Kong University of Science and Technology, Twin Cities, MN and Hong Kong. Graham, J. R., & Harvey, C. R. (2001). The theory and practice of corporate finance: Evidence from the field. Journal of financial economics, 60(2), Graham, J. R., Lemmon, M. L., & Schallheim, J. S. (1998). Debt, leases, taxes, and the endogeneity of corporate tax status. Journal of Finance, Hovakimian, A., Opler, T., & Titman, S. (2001). The debt-equity choice. Journal of Financial and Quantitative analysis, 36(01), Kaplan, S. N., & Zingales, L. (1997). Do investment-cash flow sensitivities provide useful measures of financing constraints?. The Quarterly Journal of Economics,

17 Kayhan, A., & Titman, S. (2007). Firms histories and their capital structures. Journal of Financial Economics, 83(1), Krishnan, V. S., & Moyer, R. C. (1994). Bankruptcy costs and the financial leasing decision. Financial Management, Leary, M. T., & Roberts, M. R. (2005). Do firms rebalance their capital structures?. The journal of finance, 60(6), Lewis, C. M., & Schallheim, J. S. (1992). Are debt and leases substitutes?. Journal of Financial and Quantitative Analysis, 27(04), Li, D. (2011). Financial constraints, R&D investment, and stock returns. Review of Financial Studies, hhr043. Lim, S. C., Mann, S. C., & Mihov, V. T. (2003). Market Evaluation of Off-Balance sheet financing: You can run but you can't hide. In EFMA 2004 Basel Meetings Paper. Marston, F., & Harris, R. S. (1988). Substitutability of leases and debt in corporate capital structures. Journal of Accounting, Auditing & Finance, 3(2), Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance and the theory of investment. The American economic review, Myers, S. C., Dill, D. A., & Bautista, A. J. (1976). Valuation of financial lease contracts. Journal of Finance, Rajan, R. G., & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. The journal of Finance, 50(5), Robicheaux, S. H., Fu, X., & Ligon, J. A. (2008). Lease financing and corporate governance. Financial Review, 43(3), Ross, S., Westerfield, R., and Jaffe, J. (1990). Corporate Finance, Second Ed. Boston, MA: Irwin. Sharpe, S. A., & Nguyen, H. H. (1995). Capital market imperfections and the incentive to lease. Journal of Financial Economics, 39(2), Smith Jr, C. W., & Wakeman, L. M. (1985). Determinants of corporate leasing policy. Journal of Finance, Stulz, R. (1990). Managerial discretion and optimal financing policies. Journal of financial Economics, 26(1), Warr, R. S., Elliott, W. B., Koëter-Kant, J., & Öztekin, Ö. (2012). Equity mispricing and leverage adjustment costs. Journal of Financial and Quantitative Analysis, 47(03), Yan, A. (2006). Leasing and debt financing: substitutes or complements?. Journal of Financial and Quantitative Analysis, 41(03),

18 Operating lease intensity (OLR) OLR (Blev deciles) OLR (Mlev deciles) Leverage Deficit Deciles Figure 1: Operating lease intensity over leverage deficit deciles. The figure shows means of firms lease intensities across leverage deficits (book and market) deciles. The sample includes 111,876 firm-years from Compustat between 1980 and Every year, the leverage deficits (target minus actual leverage ratio) of firms are assigned into 10 groups based on two digits SIC code. The lower (higher) deciles are over (under) levered relative to moderately leveraged. The solid (dashed) line represents means of operating lease ratios (OLR) for groups of firms across book (market) leverage decile ranks.

19 Table 1 Sample selection criteria This table provides the sample selection criteria. The sample includes US public firms from COMPUSTAT and covers 111,876 observations for the period between 1980 and Selection process # of observations Number of firm-year observations in COMPUSTAT between 1980 and ,967 Less: Firm-years with (SIC ) and (SIC ) 100, ,213 Negative assets, sales, and share price 64, ,454 Missing values for variables used in target leverage estimation 7, ,881 Missing values for operating lease intensity measure 15, ,194 Missing values for control variables 18,318 Total number of firm-years excluded from the sample 207,091 Final sample over ,876

20 Table 2 Sample summary statistics This table presents the summary statistics for the full sample used to estimate leverage targets. BLev refers to total debt (DLTT+DLC) scaled by book value of assets (AT). MLev refers to total debt (DLTT+DLC) scaled by debt plus market equity (DLTT+DLC+(PRCC_f*CSHO)). OI is the operating income (OIBDP) scaled by book value of assets. MB is the market-to-book ratio of the assets. Market value of asset is the total asset minus total equity (SEQ) minus deferred taxes (TXDITC) plus market value of equity (PRCC_f*CSHO). LnA is the log of total asset. DEP is depreciation and amortization expenses (DP) scaled by total assets. FA is plant, property, and equipment (PPENT) scaled by total assets. RND is the research and development expense (XRD) scaled by sales (SALE). D_RND is a dummy variable that equals one if the firm reports missing XRD and zero otherwise. DIV is common stock dividend (DV) scaled by total assets. Altman Z-score is company's likelihood of bankruptcy measured as: [3.3 OIADP + SALE RE (ACT-LCT)]/AT. Marginal tax rate equals to the statutory tax rate if the firm reports no net operating tax loss carry forward and positive pre tax returns. Industry median debt ratio is calculated as the median debt ratio of the industry where industry is defined as the first two digits SIC code of the firm. All variables that are scaled by total assets are winsorized at 1 st and 99 th percentile. Variables Obs. Mean Median Std. Dev. Q1 Q3 BLev 111, MLev 111, OI 111, MB 111, LnA 111, DEP 111, FA 111, RND 111, D_RND 111, DIV 111, AZ 111, Tax 111, Ind. Median BLev 111, Ind. Median MLev 111,

21 Table 3 Parameter estimates from cross-sectional regression on determinants of debt ratio This table presents the mean and the standard deviation of parameter estimates from the yearly regressions on a sample consisting 111,876 firm-year observations. The dependent variables are BLev measured as total debt (DLTT+DLC) scaled by book value of assets (AT) and MLev measured as total debt (DLTT+DLC) scaled by debt plus market equity (DLTT+DLC+(PRCC_f*CSHO)). OI is the operating income (OIBDP) scaled by book value of assets. MB is the market-to-book ratio of the assets. Market value of asset is the total asset minus total equity (SEQ) minus deferred taxes (TXDITC) plus market value of equity (PRCC_f*CSHO). LnA is the log of total asset. DEP is depreciation and amortization expenses (DP) scaled by total assets. FA is plant, property, and equipment (PPENT) scaled by total assets. RND is the research and development expense (XRD) scaled by sales (SALE). D_RND is a dummy variable that equals one if the firm reports missing XRD and zero otherwise. DIV is common stock dividend (DV) scaled by total assets. Altman Z-score is company's likelihood of bankruptcy measured as: [3.3 OIADP + SALE RE (ACT-LCT)]/AT. Marginal tax rate equals to the statutory tax rate if the firm reports no net operating tax loss carry forward and positive pre tax returns. Industry median debt ratio is calculated as the median debt ratio of the industry where industry is defined as the first two digits SIC code of the firm. The significance levels of 10%, 5%, and 1% are represented by *, **, and *** respectively. Independent Variable Book Leverage (BLev) Market Leverage (MLev) Coef. p-value Coef. p-value Constant 0.088*** *** OI 0.027*** *** MB *** *** LnA 0.014*** *** DEP *** *** FA 0.136*** *** RND *** *** D_RND 0.028*** *** DIV *** *** AZ *** *** Tax *** Ind. Median Debt Ratio 0.405*** *** Average-R Observations 111, ,876

22 Table 4 Univariate analysis- post-target estimation This table presents the univariate comparison of means based on leverage deficits for lease intensity and other control variables used in our based regression. LevDef refers to leverage deficits measured as target leverage minus leverage ratio. A positive (negative) value for LevDef is considered under (over) levered. OLR represents lease intensity measured as the sum of rental expense (XRENT) and present value of rental commitments for the next five years and after (MRC1, MRC2, MRC3, MRC4, MRC5, and MRCTA) divided by the sum of rental expense, present value of rental commitments for the next five years and after, and property, plant, and equipment (PPENT). A discount rate of 10% is used to discount all the future rental commitments. NoDiv is a dummy variable equals to 1 if the firm paid no dividend (DVC) in that year and 0 otherwise. Earnings is the operating income before depreciation (OIBDP) divided by sales (SALE). All S&P ratings are indicator variables equal to 1 if the firm has S&P Domestic Long Term Issuer Credit Rating (SPLTICRM) and 0 otherwise. The rating groups are partitioned according to: AAA through AA, A+ through A, BBB+ through BBB, BB+ through D and Unrated. Tax rate is the tax expense (TXT) divided by pre-tax income (PI). Small tax-loss CF is a dummy variable equals 1 if firm had a positive tax loss carry-forward (TLCF) not exceeding current year OIBDP and 0 otherwise. Large tax-loss CF is a dummy variable equals to 1 if firm had a positive tax loss carry-forward (TLCF) exceeding current year OIBDP and 0 otherwise. Loss is an indicator variable equals to 1 if the firm has negative income before extraordinary items (IBC) in a particular year and 0 otherwise. Size is the natural log of total assets (AT). The mean differences are based on p-value. The significance levels of 10%, 5%, and 1% are represented by *, **, and *** respectively. Book Leverage (BLev) Market Leverage (MLev) Whole LevDef > 0 LevDef < 0 Diff LevDef > 0 LevDef < 0 Diff sample (Under Levered) (Over Levered) (Under-Over) (Under Levered) (Over Levered) (Under-Over) OLR *** *** NoDiv *** *** Earnings AAA to AA *** *** A+ to A *** BBB+ to BBB *** *** BB+ to D *** *** Tax rate Small tax-loss CF *** *** Large tax-loss CF ** *** Loss *** *** Size *** ***

23 Table 5 Association between leverage deviations and lease intensity This table presents the association between leverage deficits and lease intensity. The dependent variable is OLR representing lease intensity measured as the sum of rental expense (XRENT) and present value of rental commitments for the next five years and after (MRC1, MRC2, MRC3, MRC4, MRC5, and MRCTA) divided by the sum of rental expense, present value of rental commitments for the next five years and after, and property, plant, and equipment (PPENT). A discount rate of 10% is used to discount all the future rental commitments. LevDef is measured as target leverage minus leverage ratio. The definitions for all other variables are in Table 4. For models 1 and 3, the p-values are calculated based on robust standard errors clustered by firm year. For models 2 and 4, the p-values are calculated based on robust standard errors clustered by firm. *, **, and *** denote significance levels of 10%, 5%, and 1% respectively. Book Leverage (BLev) Market Leverage (MLev) (1) (2) (3) (4) Coef. p-value Coef. p-value Coef. p-value Coef. p-value Constant 0.447*** *** *** *** LevDef 0.123*** *** *** *** NoDiv 0.046*** *** *** *** Earnings AAA to AA ** * * A+ to A *** *** *** BBB+ to BBB *** ** *** *** BB+ to D 0.015*** *** *** *** Tax rate Small tax-loss CF Large tax-loss CF 0.032*** *** *** *** Loss 0.020*** *** *** Size *** *** *** *** Firm FE Yes No Yes No Industry FE Yes Yes Yes Yes Year FE No Yes No Yes R-square Observations 111, , , ,876

24 Table 6 Association between leverage deviations and lease intensity: under and over levered This table presents the association between leverage deficits and lease intensity. The dependent variable is OLR representing lease intensity measured as the sum of rental expense (XRENT) and present value of rental commitments for the next five years and after (MRC1, MRC2, MRC3, MRC4, MRC5, and MRCTA) divided by the sum of rental expense, present value of rental commitments for the next five years and after, and property, plant, and equipment (PPENT). A discount rate of 10% is used to discount all the future rental commitments. UnderLev_dummy is an indicator variable equals 1 if LevDef measured as target leverage minus leverage ratio is positive or 0 otherwise. The definitions for all other variables are in Table 4. For models 1 and 3, the p-values are calculated based on robust standard errors clustered by firm year. For models 2 and 4, the p-values are calculated based on robust standard errors clustered by firm. *, **, and *** denote significance levels of 10%, 5%, and 1% respectively. Book Leverage (BLev) Market Leverage (MLev) (1) (2) (3) (4) Coef. p-value Coef. p-value Coef. p-value Coef. p-value Constant 0.423*** *** *** *** UnderLev_dummy 0.040*** *** *** *** NoDiv 0.045*** *** *** *** Earnings AAA to AA ** * ** A+ to A *** *** BBB+ to BBB *** ** *** * BB+ to D 0.013*** *** *** ** Tax rate Small tax-loss CF Large tax-loss CF 0.032*** *** *** *** Loss 0.019*** *** Size *** *** *** *** Firm FE Yes No Yes No Industry FE Yes Yes Yes Yes Year FE No Yes No Yes R-square Observations 111, , , ,876

25 Table 7 Impact of financial constraints on the association between leverage deviations and lease intensity This table presents the association between leverage deficits and lease intensity by the importance of financial constraints. The dependent variable is OLR representing lease intensity measured as the sum of rental expense (XRENT) and present value of rental commitments for the next five years and after (MRC1, MRC2, MRC3, MRC4, MRC5, and MRCTA) divided by the sum of rental expense, present value of rental commitments for the next five years and after, and property, plant, and equipment (PPENT). A discount rate of 10% is used to discount all the future rental commitments. LevDef is measured as target leverage minus leverage ratio. Fin-Con represents firm financial constraints measured using two proxies- zero dividend and KZ-Index. Zero dividend is an indicator variable equals 1 if the firm paid no dividend (DVC) in that year and 0 otherwise. KZ-Index is measured as 1.002(CF/TA) (DIV/TA) 1.315(CA/TA) LEV Q, where CF is the cash flow, CA is the cash balances, LEV is the total debt, Q is the ratio of the market-to-book value of the firm's assets, and TA is the total assets. KZ values are coded high and low based on ranks in two digits SIC code in each firm year. The definitions for all other variables are in Table 4. The p- values are calculated based on robust standard errors clustered by firm year. *, **, and *** denote significance levels of 10%, 5%, and 1% respectively. Book Leverage (BLev) Market Leverage (MLev) (1) (2) (3) (4) Zero dividend KZ- Index Zero dividend KZ- Index Coef. p-value Coef. p-value Coef. p-value Coef. p-value Constant 0.306*** *** *** *** LevDef Fin-Con 0.097*** *** *** *** LevDef 0.045*** *** *** *** Fin-Con 0.080*** *** *** *** Earnings AAA to AA *** *** *** *** A+ to A *** *** *** *** BBB+ to BBB *** *** *** *** BB+ to D *** *** *** *** Tax rate Small tax-loss CF *** ** *** ** Large tax-loss CF 0.044*** *** *** *** Loss 0.037*** *** *** *** Firm FE Yes Yes Yes Yes Industry FE Yes Yes Yes Yes R-square Observations 111, , , ,675