Off-balance-sheet activities, earnings persistence and stock prices: Evidence from operating leases*

Weili Ge University of Washington Business School University of Washington Mackenzie Hall, Box 353200 Seattle, WA 98195 (206) 221-4835 geweili@u.washington.edu November 22, 2006

Abstract
This paper examines the implications of the off-balance-sheet treatment of operating leases for future earnings and stock returns. The property rights granted by an operating lease contract generate both future benefits (off-balance-sheet capital investment) and future obligations (offbalance-sheet financing liabilities) for the lessee. The change in the off-balance-sheet capital investment can be viewed as a form of growth in net operating assets and also a form of offbalance-sheet accruals. By examining the footnote disclosure on operating leases, this paper shows that, after controlling for current earnings, greater off-balance-sheet operating lease activities lead to lower future earnings. This finding is consistent with diminishing marginal returns to investment in operating lease activities. Additional tests show that investors seem to incorrectly estimate the implications of off-balance-sheet lease activities for future earnings. A long-short investment strategy that exploits this misestimation generates significant future abnormal stock returns. These results suggest that the accrual anomaly documented in prior research extends to off-balance-sheet lease accruals.

* This paper was awarded First Prize in the 2005 Chicago Quantitative Alliance Annual Academic Competition. I would like to thank the members of my dissertation committee: Patricia Dechow (Chair), Eugene Imhoff, Richard Sloan, Lu Zheng (Finance), and Ji Zhu (Statistics). I thank the helpful comments and suggestions from Carol Anilowski, Judson Caskey, Peter Demerjian, Ilia Dichev, Rajib Doogar, Mei Feng, Michelle Hanlon, Raffi Indjejikian, Andrew Jackson, Charles Lee, Reuven Lehavy, Lian Fen Lee, Feng Li, Sarah McVay, Karl Muller, Shiva Rajgopal, Scott Richardson, Cathy Shakespeare, D. Shores, Terry Shevlin, Charles Wasley, Franco Wong, the University of Michigan doctoral students and the seminar participants at AAA Midyear Conference (FARS), Barclays Global Investors, University of Chicago, University of California at Berkeley, UCLA, University of Illinois, Massachusetts Institute of Technology, University of Michigan, Northwestern University, Ohio State University, University of Pennsylvania, University of Texas at Austin, University of Texas at Dallas, University of Utah, Washington University in St. Louis, University of Washington, Yale University, and Chicago Quantitative Alliance 12th Annual Conference. I gratefully acknowledge the financial support from the Paton Fellowship Fund, the Deloitte & Touche Foundation, the Robert Neary Scholarship, and the Harry H. Jones Endowment Fund for Research on Earnings Quality at the University of Michigan.

I. INTRODUCTION This paper investigates whether the information disclosed in the operating lease footnotes can be used to predict future earnings and future stock returns. The property rights granted by an operating lease contract represent both future benefits and future obligations for the lessee. The transaction can therefore be viewed as creating both off-balance-sheet operating assets and offbalance-sheet financing liabilities. Prior research shows that growth in net operating assets (accruals) and the raising of external financing to fund such growth are both associated with lower future earnings and stock returns.1 However, the existing literature has focused on onbalance-sheet activities. This paper corroborates and extends previous research by examining the implications of off-balance-sheet operating lease activities for future firm performance. Operating leases are similar to mortgages and other financing plans in which an asset is obtained with financing that requires pre-specified future payments that include principal and interest. In substance, most operating leases represent assets and liabilities of the lessee company (Imhoff, Lipe and Wright 1991).2 Under Generally Accepted Accounting Principles (GAAP), when a firm classifies a lease as “operating,” it is not recognized on the balance sheet. Imhoff and Thomas (1988) provide evidence suggesting that lessees engage in costly restructurings of capital leases to avoid recognition of these leases on the balance sheet.3 The SEC, in a June 15, 2005 staff report to Congress and the President on off-balance-sheet activities,

See, for example, Abarbanell and Bushee (1997, 1998); Fairfield, Whisenant and Yohn (2003a); Richardson, Sloan, Soliman and Tuna (2005, 2006); and Bradshaw, Richardson and Sloan (2005). 2 In a Special Report, Leases: Implementation of a New Approach, published by the FASB and other G4+1 organizations, it is suggested that each separate right arising out of a lease contract represents an asset and each separate obligation represents a liability that lessees need to recognize and account for individually. 3 As discussed in the 2004 AICPA conference on SEC developments, “lease accounting is a great example of accounting for the form of a transaction over its substance.”

1

1

recommends that the accounting guidance for leases be reconsidered, since many lease arrangements are structured to avoid crossing the “bright lines” in the accounting standards.4 Operating leases are a prevalent type of off-balance-sheet financing and one of the largest sources of corporate financing. The 2005 SEC staff report estimates that undiscounted total noncancelable future cash flow obligations associated with operating leases for U.S. companies are approximately $1.25 trillion. A recent survey by the Equipment Leasing Association (ELA) states that eight out of ten companies in the U.S. lease some or all of their equipment. Moreover, according to Compustat data, operating lease liabilities in 2004 accounted for 39.7 percent of total fixed claims on average, while capital lease obligations accounted for just 1.5 percent, and long-term debt accounted for 58.8 percent of total fixed claims.5 The extensive use of operating leases indicates that many companies lease assets such as office space or stores through operating leases rather than capital leases. In other words, the unrecorded assets and liabilities from operating leases are of the same order of magnitude as on-balance-sheet assets and liabilities. Given the pervasive use and materiality of operating leases, it is important to examine the implications of the off-balance-sheet operating lease activities for future earnings and valuation. In this paper, operating lease activities are measured as the change in the present value of future non-cancelable operating lease obligations (see Imhoff et al. 1991). This measure captures the level of new financing through operating leases. I use this measure to proxy for growth in capital investment in operating leases or off-balance-sheet lease accruals. The results indicate that increases in operating lease activities lead to lower future earnings (return on

4

See the SEC website: http://www.sec.gov/news/studies/soxoffbalancerpt.pdf. In July 2006, the FASB formally added reconsidering lease-accounting to its agenda. 5 Total fixed claims are defined as the sum of the book value of long-term debt, the book value of capital leases, and the present value of future operating lease obligations.

2

assets), after controlling for current earnings. The finding is consistent with diminishing marginal returns to the investment in operating leases. Additional tests investigate whether investors fully anticipate the negative relation between off-balance-sheet operating lease activities and future earnings. The analysis based on the Mishkin (1983) framework indicates that the stock market behaves as if the operating lease activities have positive implications for future earnings. A long-short investment strategy based on off-balance-sheet lease activities generates significant one-year-ahead abnormal hedge returns. The abnormal returns based on operating lease activities are robust to the inclusion of Fama-French (1993) risk factors and the momentum factor (Jegadeesh and Titman 1993; Carhart 1997). Further examination reveals that firms with greater changes in off-balance-sheet lease activities are likely to be experiencing greater changes in on-balance-sheet accruals and external financing. Firms tend to grow with both on-balance-sheet and off-balance-sheet operating assets. However, the information in off-balance-sheet lease activities has incremental explanatory power in the prediction of future earnings and stock returns. These results are also robust to the inclusion of the variables that might be potentially related to the leasing decision (e.g., marginal tax rate). This paper makes three contributions to the existing literature. The first relates to the literature on recognition and disclosure in financial reporting. This paper highlights the importance of incorporating disclosed but not recognized items into predictions of future earnings. Prior research has shown the link between various types of disclosed information and future earnings. For example, Rajgopal, Shevlin and Venkatachalam (2003) document that one leading indicator, order backlog, predicts future earnings. The findings of my paper suggest that the mandatory disclosure of future operating lease obligations also helps predict future earnings.

3

In addition, this paper finds that stock prices do not correctly impound the operating lease information contained in footnote disclosures, corroborating the findings of Imhoff, Lipe and Wright (1993). The second contribution of the paper is to provide more comprehensive definitions of capital investment, accruals and external financing. An increase in the present value of future operating lease obligations is equivalent, in substance, to growth in net operating assets.6 A large body of research documents that firms that increase capital investments or experience asset growth realize lower future earnings and stock returns (e.g., Abarbanell and Bushee 1997, 1998; Titman, Wei and Xie 2004; Cooper, Gulen and Schill 2005). Recent research on accruals (e.g., Richardson et al. 2005, 2006; Dechow and Ge 2005) measures accruals as the change in net operating assets. This paper points out that the new capital investment in off-balance-sheet operating leases should be included as part of total change in capital investment as well as total accruals. Prior research also documents that external financing activities are associated with lower future firm performance (see Ritter 2003). This relation holds for different types of corporate financing activities, including equity offerings and public debt offerings (e.g., Ritter, 1991; Bradshaw et al. 2004; Cassar 2005). However, existing studies focus on the external financing activities recognized in the financial statements, and ignore a major category of corporate financing: off-balance-sheet financing. This paper shows that the negative relation between external financing activities and future firm performance, specifically earnings and stock returns, also holds for off-balance-sheet financing through leases. This paper extends prior research on external financing by suggesting that incorporating both on-balance-sheet and off-balance-sheet
For example, JetBlue plans to expand its fleet by three planes. JetBlue can obtain the right to use three new planes through either purchasing or off-balance-sheet leasing. These three new planes are real operating assets regardless of whether they are included on the balance sheet.
6

4

financing enables one to simultaneously examine the relation between a firm’s complete set of corporate financing activities and future firm performance. The third contribution of the paper is to help discriminate between the competing explanations for the Sloan (1996) result that the accrual component of earnings is less persistent than the cash flow component. One stream of research suggests that the lower persistence of the accrual component of earnings is due to subjective estimation and low reliability of accruals (e.g., Xie 2001; Dechow and Dichev 2002; Richardson et al. 2005, 2006). Another stream of literature attributes the lower persistence of accruals to firm growth and the associated diminishing marginal returns to increased investment (e.g., Fairfield, Whisenant and Yohn 2003a). An examination of the relation between off-balance-sheet operating lease activities and future earnings provides new evidence regarding the underlying cause. The “accruals” created by operating lease activities are less prone to accrual earnings management, because they are not recognized on balance sheet and less likely to be manipulated by managers to boost contemporaneous earnings. They are also objectively computed from the mandatory lease footnote disclosure of future lease obligations. Moreover, off-balance-sheet accruals are less subject to the “denominator effect” documented by Fairfield et al. (2003b), because off-balancesheet assets are not included in the denominator. 7 Therefore, the finding that operating lease accruals, which involve less subjective estimation, lead to lower future earnings is more consistent with the diminishing marginal returns explanation, suggesting that accrual accounting distortions are an incomplete explanation for the lower persistence of accruals. However, I also find that on-balance-sheet accruals are less persistent than off-balance-sheet lease accruals in

Fairfield et al. (2003b) suggest that the lower persistence of the accrual component of earnings is not due to earnings management, but rather to the growth in invested capital, which is in the denominators of accruals and the profitability measure.

7

5

predicting one-year-ahead earnings, suggesting that accounting distortions drive some of the lower persistence of on-balance-sheet accruals. The remainder of the paper is organized as follows. Section II discusses prior research and my predictions. Section III describes the sample formation, and section IV presents the empirical results. Section V describes additional analyses and robusteness tests. Section VI concludes. II. BACKGROUND ON LEASES AND PREDICTIONS Leasing is a common way to obtain the use of productive assets. The Equipment Leasing Association (ELA) states that $208 billion (31 percent) out of the $668 billion of productive assets acquired by businesses in the U.S. were acquired through leasing in 2003.8 Leases can be classified as either capital leases or operating leases from the lessee’s perspective. Capital leases are similar to purchases by the lessee and require balance sheet recognition of an asset and an obligation. In contrast, operating leases are off-balance-sheet activities for the lessee and are reflected in the income statement as rent expense.9 Graham, Lemmon and Schallheim (1998) find that operating leases account for a much larger part of firms’ capital structures than capital leases. The high use of operating leases is partly due to the benefit of balance sheet management arising from operating leases. To illustrate, when SFAS No. 13 on leases was implemented, the terms of most leases were structured to avoid balance sheet recognition (Imhoff and Thomas 1988). In addition, Imhoff et al. (1993) provide evidence that compensation committees do not adjust reported ROA (return on assets) to reflect operating leases when they establish executive

8 9

See the Equipment Leasing Association’s website: http://www.elaonline.com/industrydata/overview.cfm. Synthetic leases are operating leases. A synthetic lease is treated as an operating lease for financial reporting purposes and can still enjoy the tax benefit of the asset ownership. Sale/leaseback transactions are also considered operating lease activities as long as the associated leases are not reported on the balance sheet.

6

cash compensation. Brealey and Myers (2003) also suggest that off-balance-sheet leasing can be used to circumvent restrictive covenants. The “bright line” nature of lease accounting results in different accounting treatment for economically similar arrangements.10 For example, a lease that has payments equal to 90 percent of an asset’s fair value will be classified as a capital lease and recognized on the balance sheet, while a lease that has payments equal to 89 percent will be classified as an operating lease. Both academic researchers and practitioners have reached the consensus that, in substance, many offbalance-sheet operating leases represent both “assets” and “liabilities” (see Lipe 2001). An increase in operating leases can be viewed as an increase in operating assets that are financed by the lessor. Note that an increase in operating leases also suggests an increase in the obligation to make future payments. At the inception of each lease, the unrecorded lease asset and unrecorded lease liability both equal the present value of the future lease payments (Imhoff et al. 1991). Changes in the “assets” resulting from operating leases can be viewed as changes in “off-balance-sheet capital investment” and therefore also a form of “off-balance-sheet accruals.” Accruals are defined as the change in net operating assets (see Richardson et al. 2005, 2006). I expect off-balance-sheet operating lease activities to be negatively associated with future earnings (return on assets), ceteris paribus. Many studies have shown that increased capital investments are followed by lower future firm performance. For example, Abarbanell and Bushee (1997) find that increases in industry-adjusted capital expenditure lead to decreases

10

To be a capital lease, a lease has to meet one or more of the following criteria: the lease term is longer than 75% of the estimated economic life of the equipment; the lease gives title of the asset to the lessee at the end of the lease; the lease contains a bargain option to buy the equipment at the end of the lease; or the present value of the lease payments (the sum of the payments at any given time during the course of the lease) is larger than 90% of the fair market value of the asset.

7

in future earnings. Fairfield et al. (2003a) and Richardson et al. (2005) find that growth in longterm net operating assets is negatively associated with one-year-ahead return on assets.11 Prior research conjectures that lower operating performance following increased capital investment results from diminishing marginal returns to increased investment. Fairfield et al. (2003a) argue that diminishing marginal returns occur when firm managers exploit the most profitable investment opportunities before less profitable investment opportunities.12 Their argument is based on Stigler’s (1963) assertion that competition will always equalize return on investments in all industries.13 Note that diminishing marginal returns do not necessarily suggest managers taking negative NPV projects. More generally, diminishing marginal returns occur when firms increase production, supply increases, and profits start to fall. A more severe form of diminishing marginal returns is over-investment that is associated with negative NPV projects.14 Titman et al. (2004) suggest that those managers who are empire builders might invest for their personal benefit rather than for shareholders’ benefit. A few other related studies provide evidence supporting managers over-investing in assets (e.g., Richardson 2006; Li 2004; Richardson and Sloan 2003). If diminishing marginal returns apply to capital investments in operating leases, then an increase in operating leases will be followed by a decrease in firm operating performance (return

In addition, Titman et al. (2004) document that firms with increased capital investment experience lower future benchmark-adjusted returns. Cooper et al. (2005) investigate the growth effect for total assets and document a strong negative relation between growth in firms’ total assets and future stock returns. Some event studies find that stock prices respond favorably to announcements of capital expenditure plans (e.g. McConnell and Muscarella 1985). However, firms might only announce the capital expenditure plans that will be viewed favorably. Trueman (1986) suggests that management might use a high level of capital investment to signal favorable information. 12 Fairfield et al. (2003a) also suggest that the lower persistence of the accrual component of earnings might result from conservative accounting. However, Richardson et al. (2006) show that conservative accounting does not explain the empirical regularity. 13 As pointed out by Richardson et al. (2006), Stigler’s arguments are for industry-level return on investments. His argument applies to firm-level returns assuming that either the firm dominates the industry or firm-level performance is highly correlated in the same industry. 14 Richardson (2006) defines over-investment as “investment expenditure beyond that required to maintain assets in place and to finance expected new investments in positive NPV projects.”

11

8

on assets) on average. For firms with similar levels of current operating performance, the firms that experience larger increases in operating leases have lower future operating performance than similar firms with less investment in operating leases. Richardson et al. (2005, 2006) argue that accounting distortions also cause lower future earnings following long-term on-balance-sheet accruals. However, the “accruals” created by operating lease activities are less prone to earnings management because they are not recognized on balance sheet and are unlikely to be manipulated by managers to boost earnings. Therefore, I expect increases in operating leases to be associated with lower future earnings, due to diminishing marginal returns to investment. P1: Increases in off-balance-sheet operating leases are associated with lower future earnings, after controlling for contemporaneous earnings. The next hypothesis concerns the extent to which stock prices reflect the implications of operating leases for future earnings. A large body of research suggests that investors do not correctly price the implications of asset growth for future earnings. Titman et al. (2004) and Abarbanell and Bushee (1998) document a negative relation between capital expenditure and future abnormal stock returns. Prior research on accruals also finds that investors do not correctly price short-term operating accruals (e.g., accounts receivable) and long-term operating accruals (e.g., changes in PPE). See Sloan (1996), Fairfield et al. (2003a), and Richardson et al. (2005). Several recent papers attempt to attribute the mispricing of accruals to sample selection biases (Kraft, Leone and Wasley 2005) and risk factors (Khan 2005; Zach 2004). By identifying the role of off-balance-sheet operating lease accruals in the forecasting of future earnings, this study provides a new setting in which to corroborate and extend prior evidence. I expect off-balance-sheet lease activities to be negatively associated with future abnormal stock returns. As discussed earlier, the existing evidence in the literature is largely

9

consistent with the mispricing of on-balance-sheet capital investment, accruals and any type of external financing. If the above results are generalizable to off-balance-sheet activities, investors would fail to incorporate the negative implications of off-balance-sheet leasing for future earnings in a timely manner. This would result in a negative relation between operating leasing and future abnormal stock returns. It is worth noting that information on operating leases is disclosed in footnotes to financial statements. As pointed out by Hirshleifer and Teoh (2003), investors have limited attention and cognitive processing power. Less salient information that requires more cognitive processing is less likely to be used by investors and more likely to be priced incorrectly. If investors fail to correctly price the accounting information recognized on the financial statements, I hypothesize that they are less likely to correctly price the information in operating leases, which are not recognized on the financial statements. Therefore, my prediction regarding future stock returns is as follows: P2: Increases in off-balance-sheet operating leases are associated with lower future

abnormal stock returns.

III. DATA AND SAMPLE SELECTION Measuring Off-balance-sheet Operating Lease Activities Data on operating leases are obtained from Compustat. SEC registrants are required by SFAS 13 to report the minimum payments of non-cancelable operating leases for the following five years and a total amount of payments for the “thereafter years,” the years after the fifth year.15 For example, in its 10-K filing for the 2002 fiscal year, Starbucks reports minimum future rental payments under non-cancelable operating lease obligations (in thousands) of $248,016 for year 2003, $243,519 for year 2004, $232,641 for year 2005, $219,384 for year
15

SFAS 13 is effective for fiscal years ending on or after December 1978 (see Imhoff and Thomas 1988).

10

2006, $203,395 for year 2007, and $863,874 for the thereafter years. See Appendix A for an example of footnote disclosure. Prior to year 2000, Compustat tabulated only the future rental payments for the next five years (Compustat Item #96, #164, #165, #166, and #167), but not the payment for the thereafter years (thereafter number).16 The issues related to the thereafter number are discussed in more detail in the next section. Ten percent is used to measure the cost of debt; this rate has been used by Standard & Poor’s to capitalize operating leases.17 The present value of the scheduled minimum future operating lease cash flows for the next five years is calculated as follows:18
OPLEASE t = RENTt +1 RENTt + 2 RENTt +3 RENTt + 4 RENTt +5 + + + + 1 .1 1.12 1.13 1.14 1.15

Off-balance-sheet financing through operating leases is measured as the change in the present value of future operating lease obligations (∆OPLEASE). ∆OPLEASEt = OPLEASEt - OPLEASEt-1 ∆OPLEASE is deflated by average total assets (Compustat Item #6) to measure the amount of operating lease financing relative to the existing asset base. I also use ∆OPLEASE to proxy for new capital investment in operating leases or off-balance-sheet lease accruals. At the inception of a lease contract, the unrecorded asset and unrecorded liability resulting from the lease both equal the present value of the future lease payments (see Appendix B).19 As illustrated in the figure of Appendix B, after the inception of a lease, the unrecorded asset is less
16

Compustat has collected the future lease payments beyond the next five years only since 2000 (Compustat Item #389). 17 SFAS 13 requires a lessee to use the incremental borrowing rate to determine present value of lease payments. However, the information about the incremental borrowing rate is not available from Compustat. The results are similar when I use 8%, 12% or short-term average borrowing interest rate (Compustat Item 105) as the discount rate to calculate the present value of operating leases. 18 The sample period of this paper is from 1988 to 2003. Since the thereafter portion of future lease payments is available only for years 2000-2003, to be consistent, I calculate future operating lease liabilities based on the next five years’ lease payments for all time periods. 19 This is assuming there is no material down payment in the lease contract. The unrecorded liability will be less than the unrecorded asset if there is a material down payment.

11

than the unrecorded liability until the lease expires. The unrecorded lease asset equals the cost (PV of future lease obligations) less accumulated depreciation; the unrecorded lease liability equals the cost (PV of future lease obligations) less the accumulated paid interests. The difference between the unrecorded lease asset and unrecorded lease liability depends on the percentage of the total lease life expired and the interest rate. Note that companies usually have a portfolio of leases; information on the details of each lease in the portfolio is not disclosed. Appendix C provides an example to illustrate how off-balance-sheet operating lease accruals differ from off-balance-sheet operating lease financing. A company is assumed to enter a new five-year lease contract each year from Year 0 to Year 5, and then no longer takes new leases from Year 6 to Year 10. Year 0 to Year 4 is the growth stage of the company; Year 5 to Year 6 is the steady state (e.g., the company takes a new lease when another lease expires), and Year 7 to Year 10 is the declining stage. It appears that the operating lease accruals are smaller than the operating lease financing when the firm is growing; equal to the operating lease financing when the firm is at a steady stage, and larger than the operating lease financing when the firm is declining. This suggests that using ∆OPLEASE to proxy for off-balance-sheet lease accruals is likely to overstate the lease accruals for growing firms, and understate the lease accruals for declining firms. Since the magnitude of off-balance-sheet lease accruals tends to be equal to or smaller than the magnitude of off-balance-sheet financing, using ∆OPLEASE to proxy for operating lease accruals might understate the magnitude of the coefficient.20 The other proxy for off-balance-sheet lease activities is the change in the following year’s minimum rental payment disclosed in the footnotes. This proxy differs from ∆OPLEASE in three ways. First, an increase in the following year’s rent payment would normally lead to

20

Moreover, the Spearman correlation between capital lease assets and capital lease liabilities is 0.840, suggesting that lease liability is a reasonable proxy for lease assets.

12

higher next year’s rent expense and thus have a direct impact on next year’s earnings. Second, there is no need to estimate the interest rate for this proxy. Third, using ∆RENT is useful in that the lack of data on rent payments beyond the following five years is no longer a concern. However, ∆RENT does not capture the overall off-balance-sheet financing through operating leases. The change in the following year’s rent payment is calculated as: ∆RENTt = RENTt+1 - RENTt ∆RENT is also deflated by average total assets. Data on the following year’s rent (RENTt+1) are obtained from Compustat Item #96. Note that the information for RENTt+1 is disclosed in year t’s footnotes, and the information for RENTt is disclosed in year t-1’s footnotes. When calculating ∆OPLEASE and ∆RENT, I replace the missing values with zero and delete those observations with zero changes in operating leases.21 Rent Payments for the Thereafter Years As discussed in the previous section, operating lease payments more than five years into the future have been available on Compustat only since 2000. To be consistent, I calculate future operating lease obligations based on the next five years’ lease payments for all years. In this section, I investigate how the lack of thereafter lease payments would affect the empirical analysis. Omitting the thereafter number would understate the off-balance-sheet lease liabilities, and thus affect the change in the present value of the future rental payments (∆OPLEASE). To facilitate the understanding of the potential impact of thereafter rental payments, I incorporate the thereafter portion of lease payments (Compustat Item #389) into calculating future operating lease obligations for firm-year observations from 2000 to 2003 using the
21

Missing data on operating leases could either result from data unavailability or lease obligations of zero. As a robustness check, I delete the observations with missing data on all of the following five years’ rents and replace the remaining missing fields with zero. The number of observations remains similar, and the empirical results are extremely similar.

13

approach developed by Imhoff et al. (1991)22. I examine the correlation of ∆OPLEASE before and after the inclusion of the rental payments for thereafter years. The Pearson (Spearman) correlation is 0.903 (0.938) between the two measures of ∆OPLEASE (not tabulated). Both of these correlations are significant at less than a one percent level. The high correlation mitigates the concern that the lack of the thereafter number would significantly alter the tenor of the results. It suggests that the exclusion of the thereafter lease payments is likely to reduce the power of the tests. As a robustness test for the main results in the paper, I also include the thereafter portion of future lease payments in calculating ∆OPLEASE for firm-year observations after 2000. The results remain extremely similar.

Calculating Total Accruals, Net External Financing, and Stock Returns Other financial data are obtained from the Compustat annual database. Stock return data are obtained from the CRSP daily and monthly stock returns files.23 The resulting sample covers all firm-years with available data on Compustat and CRSP for the period 1988-2004.24 The analysis is restricted to observations after the release of SFAS 95 in order to calculate the measure of accruals from the statement of cash flows (Hribar and Collins 2002). In addition, there are fewer missing operating lease observations in more recent years. I remove firm-year observations lacking Compustat data necessary to calculate the primary financial statement variables used in the tests. Data from the statement of cash flows is used to calculate total accruals. Total Accruals (TACC) is calculated as the difference between earnings (Compustat item #123) and free cash
22 23

See Appendix A for an example for the calculation of OPLEASE using the Imhoff et al. (1991) approach. The sample is not restricted to NYSE/AMEX firms; therefore, it does not have the exchange listing bias suggested by Kraft Leone and Wasley (2005). 24 The actual years examined are 1988-2003 for one-year-ahead performance tests and 1988-2002 for two-yearahead performance tests.

14

flows (FCF). Free cash flows are calculated as CFO + CFI. 25 CFO is cash flows from operations (CFO, Compustat item #308). CFI is cash flows from investing activities (Compustat item #311), as reported on the statement of cash flows. Free cash flows reflect the impact of cash spent on PPE, acquisitions and other investments that have been capitalized as assets on the balance sheet. It also reflects cash received from the sale of divested assets and other investments.26 Therefore, free cash flow matches the flow in earnings better than CFO because earnings include capital charges such as depreciation and amortization charges that are ignored in CFO. Total Accruals, as outlined above, is similar to the measure used in recent papers on total accruals (e.g., Dechow and Ge 2005; Richardson et al. 2006; Dechow et al. 2005). Following Bradshaw et al. (2004), I measure the net amount of cash flow received from external financing activities (CFF) as: CFF = ∆EQUITY + ∆DEBT ∆EQUITY is defined as net cash received from the sale (and/or purchase) of common and preferred stock less cash dividends paid (Compustat item #108 less Compustat item #115 less Compustat item #127). ∆DEBT represents net cash received from the issuance (and/or reduction) of debt (Compustat item #111 less Compustat #114 plus Compustat item #301). Note that throughout the paper all variables are scaled by average assets. I refer to these variables by the numerator’s name for simplicity. Stock returns are measured using compounded buy-hold size-adjusted returns, inclusive of dividends and other distributions. Returns are calculated for a twelve-month period beginning four months after the end of the fiscal year. The size-adjusted return is calculated by deducting
25

I subtract the cash portion of discontinued operations and extraordinary items (Compustat item #124) from free cash flows to calculate total accruals per Hribar and Collins (2002). 26 The equipment acquisition in a capital lease does not show up as a capital expenditure in CFI under GAAP. SFAS No. 95 requires firms to disclose non-cash simultaneous financing and investing activities either in a narrative or in a schedule, which is included as a separate section of the statement of cash flows.

15

the value-weighted average return for all firms in the same size-matched decile, where size is measured as the market value at the beginning of the return accumulation period. For delisted firms during the future return window, the remaining return is calculated by first applying CRSP’s delisting return and then reinvesting any remaining proceeds in the appropriate sizematched portfolio. The one percent tails of all financial statement variables are trimmed in order to remove extreme outliers. The final sample with non-missing financial statement data consists of 59,235 firm-year observations.27 IV. Descriptive Statistics Table 1 provides summary statistics of the financial variables used in the analysis. Panel A reports descriptive statistics. All variables are scaled by average assets. ∆OPLEASE has a mean of 0.007, indicating that, on average, companies increase future operating lease obligations. The average annual growth in future operating lease obligations is about 0.7 percent of total assets. The mean of TACC is 0.024, suggesting the average firm’s on-balance-sheet accruals are around 2.4 percent of total assets. The mean values for CFF, ∆EQUITY, and ∆DEBT are 0.049, 0.033, and 0.016, respectively, consistent with an overall propensity for raising capital. The amount of net operating lease financing is less than that of financing recognized on the balance sheet (0.007 versus 0.033 and 0.016). One possible reason is that companies usually take a portfolio of leases, and increase or reduce their lease transactions more smoothly than they do their equity or debt issuances, which are more lump sum in nature. The mean of operating lease liabilities as a percentage of total fixed claims is 0.438; the mean of
27

EMPIRICAL RESULTS

The results remain similar if each regression is estimated using only observations with data available for that regression. For the stock return tests, I do not require the observations to have data for non-missing one-year-ahead earnings.

16

long-term debt as a percentage of total fixed claims is 0.562 (not tabulated). Capitalized leased PPE (∆PPE_CAPLEASE) and the change of capitalized lease obligations (∆CAPLEASE) both have means and medians close to zero, suggesting that capital leases are not as important as operating leases as a financing mechanism.28 This is consistent with the findings in Graham et al. (1998). Panel B provides both Spearman and Pearson correlations. First, ∆OPLEASE is positively correlated with TACC (Spearman=0.234) as well as with ∆PPE (Spearman=0.305), suggesting that firms tend to grow their on-balance-sheet and off-balance-sheet operating activities at the same time. This is consistent with the finding of Feng, Gramlich and Gupta (2005) that the change in the number of special purpose entities is positively related to onbalance-sheet total accruals. Second, there is a positive correlation between ∆OPLEASE and CFF (Spearman=0.140), indicating that off-balance-sheet and on-balance-sheet financing activities are complements rather than substitutes. Third, there is a stronger positive correlation between ∆OPLEASE and ∆DEBT (Spearman=0.119) than the correlation between ∆OPLEASE and ∆EQUITY (Spearman=0.056). This correlation indicates that a firm raising more onbalance-sheet debt is likely to also raise more off-balance-sheet debt. Debt and leases do not appear to be substitutes, which is also found by Ang and Peterson (1984). Moreover, consistent with prior research, there is a negative correlation between ∆EQUITY and ∆DEBT (Spearman=0.045), indicating refinancing activities.

∆PPE_CAPLEASE and ∆CAPLEASE have low standard deviations; thus, the statistical power is likely to be low for empirical analyses using these two variables. ∆PPE_CAPLEASE is also a noisy variable. Many companies do not separately disclose PPE under capital leases. Compustat sometimes includes leasehold improvement related to operating leases as PPE under leases (Compustat Item #159). Moreover, Compustat stopped collecting data on PPE under leases after 1997.

28

17

[Insert Table 1 about here]

Industry Distribution across ∆OPLEASE Deciles Table 2 reports the industry distribution of the sample across ∆OPLEASE deciles. The industry classification scheme is based on Frankel, Johnson and Nelson (2002). Panel A reports the percentage of firms in each industry group for each ∆OPLEASE decile. The extreme ∆OPLEASE deciles have a higher presence in Durable Manufacturers, Computers, Retail, and Services. For example, Computers consists of 25 percent for the lowest decile and 19 percent for the highest decile of ∆OPLEASE; Retails consists of 12.4 percent of the lowest decile and 26.6 percent for the highest decile. Panel B presents the percentage of firms in each ∆OPLEASE decile within each industry group. Looking across ∆OPLEASE deciles, the extreme ∆OPLEASE deciles have a relatively larger presence in Computers, Retail, and Services. Low ∆OPLEASE deciles also have a larger presence in Pharmaceuticals, while higher ∆OPLEASE deciles have a larger presence in Transportation. Note that leasing firms are not only concentrated in a few industries. There are also certain industry variations in ∆OPLEASE in the sample. In order to investigate whether the results are industry driven, I reperform the empirical analyses using industry indicator variables. Additionally, I adjust all main financial variables by the industry medians. The results continue to hold with both specifications. [Insert Table 2 about here] Operating Lease Activities and Future Earnings Table 3 provides a regression analysis of future earnings performance on change in operating lease obligations. Earnings are measured as earnings before extraordinary items

18

(Compustat item #123) deflated by average total assets. Contemporaneous earnings are included in the regression to control for the autocorrelation of earnings. I conduct all of the regression analyses following the Fama and MacBeth (1973) procedure of estimating annual cross-sectional regressions and reporting the time series averages of the resulting regression coefficients. I estimate the following regression: EARNINGSt+1 = β0 + β1 EARNINGSt + β2 ∆OPLEASEt Prediction P1 predicts β2 to be negative. The result in Column (1) of Panel A indicates that there is a statistically significant negative relation between ∆OPLEASE and future earnings. The coefficient estimate on ∆OPLEASE is -0.159, indicating that, after controlling for contemporaneous earnings, higher operating lease activities are associated with lower one-yearahead earnings, consistent with prediction one.29 Figure 1 shows the time-series properties of earnings for firm-years in the extreme deciles when ranked by operating leases. Year 0 stands for the year in which firms are ranked into extreme lease deciles; the plots demonstrate mean earnings in the three years before and after Year 0. Consistent with the results in Table 3, earnings appear to mean-revert quickly for both portfolios. This figure illustrates the lower persistence of earnings for both top decile lease firms and bottom decile lease firms. [Insert Figure 1 about here] The earlier results in Table 1 suggest that firms with greater changes in off-balance-sheet activities tend to have higher on-balance-sheet accruals. Previous research (e.g., Richardson et al. 2005) has shown that on-balance-sheet accruals reduce earnings persistence. Therefore, I next investigate whether the negative relation between operating leasing and earnings persistence
29

The results are robust to a few sensitivity tests: 1) adjusting the standard error of the coefficient estimates for autocorrelation in the annual coefficient estimates (t=3.13); 2) controlling for lagged change in earnings; 3) including operating leases as part of total assets in the deflator;

19

is incremental to the relation between on-balance-sheet accruals and future earnings. The results are reported in Panel A of Table 3. In Column (2) of Panel A, I include on-balance-sheet PPE accruals (growth in PPE) in addition to ∆OPLEASE. Recall that ∆OPLEASE and ∆PPE are positively correlated (Spearman=0.305). The coefficient on ∆PPE is negative and statistically significant (-0.067), consistent with the findings in Richardson et al. (2005). In Column (3) and (4) of Panel A, total accruals (TACC) is included as an additional control variable. ∆OPLEASE remains negatively related to future earnings after controlling for onbalance-sheet total accruals, consistent with Prediction P1. The magnitude of the coefficient on ∆OPLEASE declines from -0.159 to -0.075, suggesting that part of the negative relation between ∆OPLEASE and future earnings can be explained by on-balance-sheet accruals. Note that the coefficient on total accruals (TACC) is more negative than ∆OPLEASE (-0.120 versus -0.075 in Column 4); the difference is significant at the 10 percent level (not tabulated), indicating that low reliability of on-balance-sheet accruals also contributes to lower earnings persistence. Previous analysis in Table 1 indicates that firms doing more on-balance-sheet external financing have more off-balance-sheet external financing. Therefore, I next investigate whether the negative relation between operating leasing and future earnings still exists after controlling for on-balance-sheet financing variables. The results are reported in Panel B of Table 3. Specifically, regression (2) includes CFF and regression (3) includes ∆EQUITY and ∆DEBT. ∆OPLEASE remains negatively related to future earnings after controlling for on-balance-sheet external financing. The magnitude of the coefficient on ∆OPLEASE (-0.098) is larger than ∆DEBT (-0.051), indicative of a stronger negative relation between off-balance-sheet lease financing and future earnings than the relation between on-balance-sheet debt and future

20

earnings. It appears that firms that use leases tend to perform worse than firms that use debt financing. The coefficient on ∆OPLEASE is significantly more negative than the coefficient on ∆DEBT (p-value < 0.01; not tabulated). There are two possible reasons. The first is that leasing reduces the lessor’s bankruptcy costs compared to financing with debt. Leasing contracts have a higher priority than debt in bankruptcy (see Eisfeldt and Rampini 2005).30 A firm with poor earnings is more likely to get financing through operating leases than debt. Moreover, purchasing an asset enables the buyer to have accelerated depreciation, which is a tax advantage relative to an operating lease that tends to have evenly spread payments. However, if the taxable income is low, the lessee would not be able to use the tax advantage; it is better to let the lessor own the asset and make better use of depreciation tax shields (Brealey and Myers 2003; Scholes, Wolfson, Erickson, Maydew and Shevlin 2004). Then the lessor can pass on some of the tax benefits to the lessee in the form of low lease payments. Therefore, operating leases are likely to be associated with low expected future taxable income.31 [Insert Table 3 about here] Table 4 replicates the analyses in Table 3 replacing the dependent variable with two-yearahead earnings. It is possible that new capital investments in operating leases have not affected earnings in one year, while the rent expenses have. Table 4 investigates the sensitivity of the results to using two-year-ahead earnings. The results suggest that ∆OPLEASE remains

30

In the event of bankruptcy, if the bankruptcy court decides that the asset is “essential” to the lessee’s business, the lessor is entitled to receive lease payments in accordance with the original lease agreement because these payments are classified as administrative expenses, which are satisfied first in the bankruptcy code. Otherwise, the lessor can immediately recover the possession of the equipment and file a claim against the lessee for economic losses incurred. However, other outstanding creditor claims have lower priority and are not guaranteed to be met. This leasing advantage only applies to those leases when the lessor retains ownership of the asset. These leases are called “true” leases from a legal and tax point of view. Operating leases are usually true leases (see Krishnan and Moyer 1994 and Sharpe and Nguyen 1995). 31 Note that the results hold if the on-balance-sheet accruals and external financing variables are both included in the same regression, as shown in a later table, Table 9.

21

negatively associated with two-year-ahead earnings, even after controlling for on-balance-sheet accruals or on-balance-sheet net external financing. This finding gives further support for prediction P1, consistent with diminishing marginal returns to increased investment in operating leases. [Insert Table 4 about here] Additional tests that use ∆RENT to proxy for operating leasing activities also show that ∆RENT is negatively associated with one-year-ahead earnings (not tabulated). As expected, the coefficient estimate on ∆RENT is significantly negative (-0.551; t-statistic=-4.65). After controlling for on-balance-sheet accruals and external financing, ∆RENT remains negatively related to next period earnings.32 Stock Return Tests In this section, I investigate whether investors fully anticipate the implications of operating lease activities for future earnings. ∆OPLEASE should not be able to predict future abnormal stock returns if investors correctly price the implications of operating leases for future earnings. Prior research has shown that investors fail to fully anticipate the lower earnings persistence resulting from accruals and external financing, consistent with the naïve investor hypothesis. If investors also incorrectly estimate the implications of operating leasing for future earnings, there will be a negative relation between ∆OPLEASE and future abnormal returns. Table 5 reports the results of the regression analyses of future size-adjusted returns on ∆OPLEASE. As reported in Panel A, the coefficient on ∆OPLEASE is significantly negative.

32

∆RENT can be considered as the short-term component of ∆OPLEASE. As expected, ∆RENT is insignificantly associated with two-year-ahead earnings (untabulated) because ∆RENT is the change of the following year’s rent. Moreover, ∆RENT remains negatively related to one-year-ahead earnings after controlling for the change in the rents from the second year to the fifth year.

22

This is consistent with Prediction 2, indicating the mispricing of off-balance-sheet lease accruals. The coefficient magnitude of -0.843 suggests that an increase in off-balance-sheet operating lease accruals equal to one percent of average assets results in a -0.843 percent size-adjusted stock return in the subsequent year.33 In Columns 2-5, the coefficient magnitude on ∆OPLEASE becomes smaller as more on-balance-sheet accruals are included as controlling variables. However, ∆OPLEASE remains significant. Panel B reports the results after controlling for onbalance-sheet external financing. ∆OPLEASE appears to be negatively related to future stock returns after controlling for those external financing activities reflected in financial statements. The results in Column (5) in Panel A and Column (4) in Panel B suggest that the results continue to hold after control for book-to-market.34 [Insert Table 5 about here] I also conduct the Mishkin (1983) test to test the potential market mispricing of the information in operating leasing.35 The main system of equations of the analysis is as follows:
EARNINGS t +1 = γ 0 + γ 1 EARNINGS t + γ 2 ∆OPLEASE t + ν t +1 ABNORMALRETURN t +1 = β ( EARNINGS t +1 − γ 0 − γ 1 EARNINGS t − γ 2 ∆OPLEASE t ) + et +1
* *

If investors underestimate the negative implications of ∆OPLEASE for future earnings, then γ2 < γ2*. The results are reported in Table 6. The analysis based on the Mishkin framework
This result still holds if the stock return variable is trimmed or winsorized. This mitigates the concern regarding the outlier problem addressed in Kraft et al. (2005). However, the distribution of stock returns is positively skewed. Core (2005) suggests that the approach used in Kraft et al. (2005) is inappropriate because trimming skewed stock return results in biased estimates. See also Teoh and Zhang (2005) and Kothari, Sabino and Zach (2005). 34 Supplementary analyses using ∆RENT to proxy for operating lease activities show that the coefficient on ∆RENT is -2.34 with t-statistic equal to -3.96 (not tabulated). This result suggests that an increase in ∆RENT equal to one percent of average assets results in a -2.34 percent abnormal stock return over the subsequent year. After controlling for ∆PPE, TACC, as well as on-balance-sheet financing variables, ∆RENT continues to be negatively related to future stock returns. 35 To test the market efficiency as to the variables in the model, the Mishkin approach does not require a complete specification of the relation between variables in the forecasting equation and earnings at t+1 (Sloan 1996). This approach does maintain the assumption that stock prices are efficient as to those variables that are correlated with the predictor variables and are omitted from the forecasting equation (Fairfield et al. 2003a). However, this assumption does not affect the inference regarding overall market efficiency.
33

23

explicitly tests how investors value operating leases, complementing previous stock return regression analysis. In Panel A of Table 6, γ2 is significantly negative (-0.201) while γ2* is significantly positive (0.395). The results reject the null that the investors correctly price the operating leasing information for one-year-ahead earnings, consistent with the conclusions from Table 5. The investors appear to overvalue the growth in operating leases relative to its ability to predict one-year-ahead earnings. Panel B includes PPE accruals in the regressions. The valuation coefficient on ∆PPE is significantly higher than the forecasting coefficient, suggesting that investors overprice growth in PPE. This finding is consistent with Fairfield et al. (2003a). It appears that the mispricing of ∆OPLEASE is incremental to the mispricing of ∆PPE. I measure the degree of mispricing using MISPRICING = |γ*- γ|. MISPRICING equals to 0.442 (=|0.369-(-0.073)|) for ∆PPE and 0.420 for ∆OPLEASE. It does not appear that ∆OPLEASE is more mispriced than ∆PPE. Panel C adds total accruals, and Panel D includes net external financing in the analyses. Consistent with previous research and the findings in Table 5, the market appears to incorrectly impound the information contained in total accruals and external financing on future earnings. Moreover, the mispricing of operating leases still holds after controlling for the mispricing of on-balance-sheet accruals and external financing. Off-balance-sheet lease accruals do not appear to be more mispriced than on-balance-sheet accruals or external financing (for example, MISPRICING = 0.365 for ∆OPLEASE and MISPRICING = 0.304 for TACC). [Insert Table 6 about here] Two observations emerge from Table 6. First, investors do not appear to ignore operating lease information disclosed in footnotes. Rather, investors value these operating lease activities as if they have positive implications for future earnings. This finding is in contrast of

24

those of a few previous studies (e.g., Landsman 1986; Barth 1994), which found that disclosed footnote information is at least partially reflected in stock prices. Second, growth in off-balancesheet operating leases is not more mispriced than growth in on-balance-sheet operating assets. It is well known that analysts (e.g., Standard & Poor’s analysts) adjust balance sheets for operating leases. For example, Graham and Dodd (1988) recommend that coverage ratios incorporate the unrecorded liabilities. Therefore, these two inferences are not entirely surprising. Table 7 reports mean future stock returns for portfolios of firms formed on ∆OPLEASE. ∆OPLEASE is ranked in each calendar year and assigned to ten portfolios based on the ranks. I calculate the mean annual stock returns for each decile. The hedge returns are calculated as the difference between the extreme deciles. Table 7 also reports t-statistics for the significance of the hedge returns for each variable, based on the time-series of annual hedge returns following the Fama and Macbeth (1973) procedure. I start by examining the future raw returns following ∆OPLEASE. The lowest decile of ∆OPLEASE has a mean raw return of 28.6 percent, while the highest decile of ∆OPLEASE has a mean of 15.1 percent. The mean of the annual hedge returns is 13.5 percent. The finding is consistent with the regression analysis results, in support of Prediction P2. The second column reports the size-adjusted returns when the size portfolios are based on market value of equity deciles of NYSE, AMEX and NASDAQ firms; the third column reports the size-adjusted returns when size portfolios are based market value deciles within the sample.36 The mean of annual hedge returns is 13.0 percent and 9.0 percent respectively. Column (4) reports control firmadjusted returns, which are the differences in returns between a sample observation and a control
36

The size-adjusted returns in Column (2) tend to be positive. This is because the sample of the paper already excludes those firms with zero changes in operating leases. Smaller firms are more likely to use leases (Eisfeldt and Rampini 2005). Therefore, size-adjusted returns of the sample firms are likely to be positive, because the size portfolios are based on all public firms. However, using this size-adjusted return does not influence cross-sectional analysis. Column (3) shows size-adjusted returns within the sample.

25

firm matched on size and book-to-market; the control firm observation has a market value between 0.70 and 1.30 times the treatment firm’s market value and has the closest book-tomarket ratio within the matched size subset. The procedure is specified in Barber and Lyon (1997). The mean of annual hedge returns based on control firm-adjusted returns is 8.4 percent. Table 7 also reports hedge returns using alpha from Fama and French’s (1993) three-factor model and four-factor model (Jegadeesh and Titman, 1993; Carhart, 1997).37 It appears that the hedge returns to ∆OPLEASE are robust to the inclusion of these potential risk factors.38 I next investigate whether the hedge returns to ∆OPLEASE still exist conditional on total accruals (TACC). The results are presented in Panel B of Table 7. I rank stocks independently on ∆OPLEASE and TACC and report the future size-adjusted stock returns for the portfolio combinations. By reading across the columns in Panel B, the ∆OPLEASE trading strategy generates positive hedge returns, holding total accruals constant. Two hedge returns across two of the three total accrual groups are statistically significant. The evidence suggests that ∆OPLEASE trading strategy earnings abnormal returns incremental to the accruals strategy, consistent with the regression results in Table 5. [Insert Table 7 about here] V. ADDITIONAL ANALYSIS AND ROBUSTNESS CHECKS Potential Determinants of Lease Decision This section investigates whether the negative relation between operating lease activities and future earnings can be explained by the endogeneity of the leasing decision. The finance
37

Monthly returns for each of the 10 portfolios based on are regressed, over the 160 months in the sample period, on mimicking returns to the three Fama and French (1993) factors – the market factor (MKT), size (SMB), book-tomarket (HML), and the returns to the momentum factor (UMD) (Jegadeesh and Titman 1993). I multiply monthly hedge returns by 12 to obtain annual hedge returns. 38 However, part of the hedge returns might be due to barriers to arbitrage (Mashruwala, Rajgopal and Shevlin 2006; Lev and Nissim 2005; Bushee and Raedy 2005).

26

literature has focused on studying the determinants of leasing decisions following Miller and Upton (1976). The findings of prior research are consistent with the conjecture that low tax rate firms, firms in financial distress, and growth firms tend to lease more (Sharpe and Nguyen 1995; Graham et al. 1998; Eisfeldt and Rampini 2005). In this section, I first study the univariate relation between operating lease activities and firm characteristics related to marginal tax rate, financial distress and growth. Then I use regression analyses to assess the association between operating lease activity and future earnings, after controlling for the above firm characteristics. Note that the focus of previous research is on the level of operating lease liabilities as part of capital structure, while this paper focuses on the change in operating lease liabilities. In a leasing contract, the ability to transfer ownership rights can create value for both parties: leasing allows low tax rate firms to sell tax shields to high tax rate lessors, who value the tax benefits more highly (see Graham et al. 1998). According to IRS, only the use of true leases allows tax benefit transfers from low marginal-tax-rate firms to high marginal-tax-rate firms.39 Operating leases are predominantly true leases. Thus, the level of operating leases is expected to be negatively associated with tax rate, and Graham et al. (1998) provide consistent evidence. Table 8 Panel A reports the means of marginal tax rate by ∆OPLEASE deciles. Marginal tax rate is estimated using the simulation approach developed by Shevlin (1990) and Graham (1996b).40 The finding is partially consistent with previous research. On average, low ∆OPLEASE firms have the lowest marginal tax rate (10.3%) among all deciles; marginal tax rate increases to 20.6% at Rank 6-7 and then declines to 17.7% for the highest ∆OPLEASE firms. Low lease firms have poor past performance, which is usually associated with low tax rates.
39

Another example of off-balance-sheet activities that allow tax benefit transfers is off-balance-sheet R&D partnerships (Shevlin 1987; Beatty, Berger and Magliolo 1995). 40 The marginal tax rate data is obtained from John Graham’s website: www.duke.edu/~jgraham. I would like to thank John Graham for providing the data.

27

One of the differences between leasing and secured lending is treatment in bankruptcy. For a true lease, in Chapter 11, a lessee can assume the lease and continue to make payments or reject the lease and return assets. Leasing contracts have high priority in bankruptcy relative to debt. Leasing and ex ante measures of financial distress are expected to be positively correlated. Financial distress (SHUMWAY) is measured using the model developed by Shumway (2001):41 eα 1 + eα α = −13.303 − 1.982 * NI + 3.593 * TL − 0.467 * SIZE − 1.809 * RET + 5.791 * SIGMA SHUMWAY = Panel A of Table 8 shows the means of Shumway score based on ∆OPLEASE deciles. Low ∆OPLEASE firms appear to have a higher probability of bankruptcy (0.56%) than high ∆OPLEASE firms (0.098%). This is inconsistent with the idea that high financial distress firms tend to lease more. However, low ∆OPLEASE firms have poor past performance (e.g., earnings and past sales growth). These firms are likely to be in financial distress and are downsize their business and let lease contracts expire without replacing them. Young, fast-growing, innovation-intensive firms face severe information asymmetry problems. Firms facing high costs of external funds can economize on the cost of funding by leasing (Sharpe and Nguyen 1995). The analysis in Table 8 Panel A suggests that high ∆OPLEASE firms have a lower book-to-market ratio, suggesting that investors have high expectations of future growth for these firms (e.g., Lakonishok, Shleifer and Vishny 1994). This is consistent with the idea that growth firms tend to use more leases than non-growth firms.
41

The variables included in the model are: net income scaled by total assets (NI), total liabilities scaled by total assets (TL), relative size measured as the logarithm of each firm's size relative to the total size of the NYSE and AMEX market (SIZE), past market-adjusted return (RET), and the idiosyncratic standard deviation of each firm's stock returns (SIGMA). SIGMA is calculated by first regressing each stock’s monthly returns in t-1 on the valueweighted NYSE/AMEX index return during the same time period. SIGMA is the standard deviation of the residual of the regression. I use the parameter estimates provided in Shumway (2001) to estimate the probability of bankruptcy.

28

The correlations in Panel B of Table 8 reconcile the findings in this paper with those in prior research regarding the relation between operating lease activities, marginal tax rate and financial distress. This paper studies the change in operating lease liabilities, while the focus of prior research is the level of operating lease liabilities as part of capital structure. Table 8 Panel B shows that ∆OPLEASE and OPLEASE are positively correlated; however, the magnitude of the Pearson Correlation is only 0.123, indicating that the endogeneity issue associated with the level variable (OPLEASE) does not necessarily apply to the change variable (∆OPLEASE). Interestingly, both ∆OPLEASE and TACC are positively correlated with marginal tax rate and negatively correlated with Shumway score. In contrast, OPLEASE, the level variable, is negatively correlated with marginal tax rate and positively correlated with Shumway score, which is consistent with the findings in prior literature. Note that the lease-versus-buy decision is already conditional on the company’s decision to grow assets and is more likely to be captured by studying the level variable (OPLEASE). Differently, ∆OPLEASE represents the company’s decision to grow assets and increase capital investments in the form of operating leases. As evidenced in Table 8, marginal tax rate, Shumway score, and book-to-market ratio still vary based on the ranks of the operating lease variable. To investigate whether the negative relation between operating lease activities and future firm performance is driven by these variables, I include them in the regression analysis. The results are presented in Table 9. In Panel A of Table 9, I first present limited regressions controlling for each of the potential determinants variables. In the final column, I present a complete regression with all the control variables. Across all columns, ∆OPLEASE is negatively associated with one-year-ahead earnings. This finding gives further support to Prediction P1, suggesting that the negative relation between ∆OPLEASE and future income goes beyond those variables that are potentially

29

correlated with the leasing decision. Note that in Column (4) of Panel A, both TACC and CFF are included as control variables. Both coefficient estimate magnitudes and t-statistics on these two variables are lower than those in previous tables. This is because TACC and CFF are highly correlated. Referring to Table 1 Panel B, the Spearman correlation between TACC and CFF is 0.459, consistent with the intuition that firms raise external financing and grow their assets at the same time. Panel B of Table 9 investigates whether ∆OPLEASE can still predict stock returns after including the additional control variables: marginal tax rate, Shumway score, and book-tomarket. Similar to the findings in Panel A, ∆OPLEASE continues to be negatively associated with one-year-ahead size-adjusted stock returns. [Insert Table 9 about here] Future Profit Margin and Asset Turnover The main results suggest that earnings decline following high operating lease activities. This section investigates how the two multiplicative components of earnings based on DuPont analysis vary with the rank of ∆OPLEASE. DuPont analysis decomposes earnings in the following way:42

EARNINGS EARNINGS SALES = X = PROFIT MARGIN X ASSET TURNOVER . ASSETS SALES ASSETS To better understand what drives the lower future earnings for high leasing firms, I study oneyear-ahead changes in profit margin and asset turnover based on ∆OPLEASE. The results are reported in Table 10. Consistent with the regression analysis, the first two columns suggest that high ∆OPLEASE firms have lower change in one-year-ahead earnings. In the next four columns, high ∆OPLEASE firms appear to have a lower change in one-year-ahead gross margin [(Sales42

See Soliman (2004) for a detailed literature review on Dupont analysis.

30

Cost of Goods Sold)/Sales] and a lower change in one-year-ahead profit margin (∆ Profit Margint+1) than low ∆OPLEASE firms, even though the relations are not always monotonic. The last two columns show that high ∆OPLEASE firms experience a lower change in one-year-ahead asset turnover (∆Asset Turnovert+1) and the relation is nearly monotonic. Taken together, the evidence presented indicates that those high ∆OPLEASE firms experience both declines in future profit margins and operating efficiency, leading to declines in future earnings.
[Insert Table 10 about here] VI. CONCLUSION

This paper investigates the relation between off-balance-sheet activities, earnings persistence and stock prices by focusing on operating leases. Previous research in the area of capital investment, accruals and external financing has focused on the investment and external financing activities that are recognized in the financial statements. However, the property rights granted by an operating lease contract represent future benefits and future obligations. Thus the change in the off-balance-sheet asset can be viewed as off-balance-sheet capital investment and, therefore, off-balance-sheet accruals. Likewise, the change in the off-balance-sheet liability can be viewed as a source of off-balance-sheet financing. I show that, similar to on-balance-sheet accruals and sources of external financing, offbalance-sheet operating lease activities are negatively associated with future earnings and stock returns. Additional tests reveal that information about off-balance-sheet lease activities disclosed in footnotes has incremental explanatory power in the prediction of future earnings and stock returns beyond on-balance-sheet accruals and external financing. Investors appear to value operating lease activities as if they are positively associated with future operating performance.

31

I also investigate the multiplicative components of earnings and find that firms with high operating leases activities have both declines in future profit margins and future asset turnover compared to low operating lease firms. Moreover, the negative relation between operating lease activities and future firm performance continues to hold after controlling for the potential determinants of leasing decisions. This paper sheds insight into the current debate on the explanations for the accrual anomaly found by Sloan (1996). Taken together, my findings give support to the explanation of diminishing marginal returns to investment. Earnings management and accrual estimation error appear to be an incomplete explanation. Furthermore, the stock market mispricing of offbalance-sheet operating lease information documented in this paper is likely part of a larger phenomenon of mispricing related to off-balance-sheet information disclosure. For example, Landsman and Ohlson (1990) document that the stock market is inefficient in incorporating information regarding net pension liability. In a similar vein, Picconi (2004) provides evidence suggesting that firms might take advantage of investors’ incomplete processing of pension information footnote disclosure and manage earnings. Future research might investigate how mispricing of off-balance-sheet information varies based on the type and nature (e.g., complexity, readability; see Li 2006) of the disclosure.

32

References
Abarbanell, Jeffery and Brian Bushee, 1998, Abnormal returns to a fundamental analysis strategy. The Accounting Review 73 (1): 19-45. Abarbanell, Jeffery and Brian Bushee, 1997, Fundamental analysis, future earnings, and stock prices. Journal of Accounting Research 35 (1): 1-24. Ang, James, and Pamela P. Peterson, 1984, The leasing puzzle. Journal of Finance 39: 1055-1065. Barber, M. Brad and John D. Lyon. 1997. Detecting long-run abnormal stock returns: the empirical power and specification of test statistics. Journal of Financial Economics 43:341-72. Barth, Mary, 1994, Fair value accounting: evidence from investment securities and the market valuation of banks. The Accounting Review 69: 1-25. Beatty, Anne, Philip G. Berger and Joseph Magliolo, 1995, Motives for forming research & development financing organizations. Journal of Accounting and Economics 19: 411-442. Bernard, Victor and Katherine Schipper, 1994, Recognition and disclosure in financial reporting. Working paper. University of Michigan and University of Chicago. Bradshaw, Mark T., Scott A. Richardson, and Richard G. Sloan, 2005, The relation between corporate financing activities, analysts’ forecasts and stock returns. Working paper, Harvard Business School. Brealey, A. Richard and Stewart C. Myers, 2003, Principles of corporate finance. Irwin, McGraw-Hill, 7th. Bushee, Brian and Jana Smith Raedy, 2005, Factors affecting the implementability of stock market trading strategies. Working paper, University of Pennsylvania. Carhart, Mark M., 1997, On persistence in mutual fund performance, Journal of Finance 52, 57-82. Cassar, Gavin, 2005, External financing and firm operating performance. Working paper. University of Pennsylvania. Cooper, Michael, Huseyin Gulen, and Michael J. Schill, 2005. What best explains the corss-section of stock returns? Exploring the asset growth effect. Working paper. Purdue University. Core, John E., 2005, Discussion of an analysis of the theories and explanations offered for the mis-pricing of accruals and accruals components, Working paper, University of Pennsylvania. Dechow, Patricia M. and Catherine M. Schrand. 2004. Earnings quality. The Research Foundation of CFA Institute. Charlottesville, Virginia. Dechow, Patricia M., Scott A. Richardson, and Richard Sloan. 2005. The persistence and pricing of the cash component of earnings. Working paper, University of Michigan. Dechow, Patricia M. and Weili Ge, 2005. The persistence of earnings and cash flows and the role of special items: implications for the accrual anomaly. Review of Accounting Studies. Forthcoming. Dechow, Patricia M. and Ilia Dichev, 2002. The quality of accruals and earnings: The role of accrual estimation errors. The Accounting Review 77 Supplement: 35-59. Desai, Hemang, Shivaram Rajgopal, and Mohan Venkatachalam, 2004, Value-glamour and accruals mispricing: one anomaly or two? Accounting Review 79 (2): 355-385.

33

Diamond, Michael, Earl Stice, and James Stice, 2000, Financial Accounting: Reporting and analysis. SouthWestern College Publishing. Eisfeldt, Andrea and Adriano A. Rampini, 2005, Leasing, ability to repossess, and debt capacity. Working paper, Northwestern University. Fairfield, Patricia M., J. Scott Whisenant, and Teri Lombardi Yohn, 2003a, Accrued earnings and growth: Implications for future profitability and market mispricing, The Accounting Review 78 (1): 353-371. Fairfield, Patricia M., J. Scott Whisenant, and Teri Lombardi Yohn, 2003b, The differential persistence of accruals and cash flows for future operating income versus future profitability, Review of Accounting Studies 8 (2-3): 221243. Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in the return on bonds and stocks, Journal of Financial Economics 33, 3-53. Fama, Eugene F. and Mcbeth, J. D., 1973. Risk, return and equilibrium – empirical tests. The Journal of Political Economy 81, 607-636. Feng, Mei, Jeffrey D. Gramlich, and Sanjay Gupta, 2005, Off-balance-sheet entities: what motivates firms to sponsor them and how sponsorship impacts accruals, leverage and return on equity. Working paper, University of Pittsburgh. Frankel, Richard M., Marilyn F. Johnson, and Karen K. Nelson. 2002, The relation between auditors’ fees for nonaudit services and earnings management. The Accounting Review 77 (Supplement): 71-105. Graham, Benjamin and David L. Dodd, 1988, Security Analysis. New York, NY: McGraw Hill, 5th. Graham, John R., 1996, Proxies for the marginal tax rate. Journal of Financial Economics 42: 187-221. Graham, John R., Michael L. Lemmon, and James S. Schallheim, 1998, Debt, leases, taxes, and the endogeneity of corporate tax status. The Journal of Finance 53 (1): 131-162. Hirshleifer, David, Kewei Hou, Siew Hong Teoh, Yinglei Zhang. 2004, Do investors overvalue firms with bloated balance sheets? Journal of Accounting & Economics 38: 297 Hirshleifer, David and Siew Hong Teoh, 2003, Limited attention, information disclosure, and financial reporting. Journal of Accounting and Economics 36: 337-386. Hribar, Paul and Daniel Collins, 2002, Errors in estimating accruals: implications for empirical research. Journal of Accounting Research 40, 105-134. Imhoff, Eugene A., Robert C. Lipe, and David W. Wright, 1997, Operating leases: Income Effects of constructive capitalization. Accounting Horizons 11 (2): 12-32. Imhoff, Eugene A., Robert C. Lipe, and David W. Wright, 1993, The effects of recognition versus disclosure on shareholder risk and executive compensation. Journal of Accounting, Auditing, & Finance 8 (4): 335-368. Imhoff, Eugene A., Robert C. Lipe, and David W. Wright, 1991, Operating leases: Impact of constructive capitalization. Accounting Horizons 5 (1): 51-63. Imhoff, Eugene A. and Jacob K. Thomas, 1988, Economic consequences of accounting standards: The lease disclosure rule change. Journal of Accounting and Economics 10: 277-310. Jegadeesh, N., and S. Titman, 1993. Returns to buying winners and selling losers: Implications for stock market efficiency. Journal of Finance, 48, 65-92.

34

Khan, Mozaffar, 2005, Are accruals really mispriced? Evidence from tests of an intertemporal capital asset pricing model. Working paper. MIT. Kraft, Arthur, Andrew Leone, and Charles Wasley, 2005, An analysis of the theories and explanations offered for the mis-pricing of accruals and accrual components. Working paper, University of Rochester. Kothari, S. P., Jowell S. Sabino and Tzachi Zach, 2005, Implications of survival and data trimming for tests of market efficiency. Journal of Accounting and Economics 39: 129-161. Krishnan, V. Sivarama and R. Charles Moyer, 1994, Bankruptcy costs and the financial leasing decision. Financial Management 23, 31-42. Lakonishok, Josef, Andrei Shleifer and Robert W. Vishny, 1994, Contrarian investment, extrapolation, and risk. Journal of Finance 49 (5): 1541-1578. Landsman, W., 1986, An empirical investigation of pension fund property rights. The Accounting Review 61: 662691. Lev, Baruch and Doron Nissim, 2005, The persistence of the accruals anomaly." Contemporary Accounting Research (forthcoming). Li, Donglin, 2004, The implications of capital investment for future profitability and stock returns – an overinvestment perspective. Dissertation, University of California, Berkeley. Lipe, Robert C., 2001, Lease accounting research and the G4+1 proposal. Accounting Horizons 15 (3): 299-310. Mashruwala, A. Christina, Shivaram Rajgopal, and Terry Shevlin, 2006, Why is the accrual anomaly not arbitraged away? The role of idiosyncratic risk and transaction costs. Journal of Accounting and Economics 42: 3-33. McConnell, J. John and Chris J. Muscarella, 1985, Corporate capital expenditure decisions and the market value of the firm, Journal of Financial Economics 14: 399-422. Miller, Merton H., and Charles W. Upton, 1976, Leasing, buying, and the cost of capital services, Journal of Finance 31: 761-786. Mishkin, Frederic S., 1983, A rational expectations approach to macroeconometrics: Testing policy effectiveness and efficient markets models. Chicago, IL: University of Chicago Press for the National Bureau of Economic Research. Nailor, Hans and Andrew Lennard, 2000, Special Report: Leases: Implementation of a new approach. Financial Accounting Standards Board. Rajgopal, Shivaram, Terry Shevlin, and Mohan Venkatachalam, 2003, Does the stock market fully appreciate the implications of leading indicators for future earnings? Evidence from order backlog, Review of Accounting Studies 8: 461-492. Richardson, Scott A., 2006, Over-investment of free cash flow, Forthcoming, Review of Accounting Studies. Richardson, Scott A., Richard G. Sloan, Mark T. Soliman, and A. Irem Tuna, 2006, The implications of accounting distortions and growth for accruals and profitability, Forthcoming, The Accounting Review. Richardson, Scott A., Richard G. Sloan, Mark T. Soliman, and A. Irem Tuna, 2005, Accrual reliability, earnings persistence and stock prices, Journal of Accounting and Economics, 39: 437-485.

35

Richardson, Scott A. and Richard G. Sloan, 2003, External financing, capital investment and future stock returns. Working paper, University of Pennsylvania. Ritter, Jay R., 1991, The long run performance of initial public offerings, Journal of Finance, 46: 3-27. Scholes, S. Myron, Mark A. Wolfson, Merle M. Erickson, Edward L. Maydew, and Terry J. Shevlin, 2004. Taxes and business strategy: a planning approach (3rd edition). Prentice Hall. Securities and Exchange Commission (SEC), 2005, Report and recommendations pursuant to Section 401(c) of the Sarbanes-Oxley Act of 2002 on arrangements with off-balance sheet implications, special purpose entities, and transparency of filings by issuers. Washington, D.C. Shevlin, Terry, 1990, Estimating corporate marginal tax rates. Journal of the American Taxation Association, Spring: 51-67 Shevlin, Terry, 1987, Taxes and off-balance-sheet financing: Research and development limited partnerships. The Accounting Review 62 (3): 480-509. Sloan, Richard G., 1996, Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review 71 (3): 289-315. Sharpe, Steven A. and Hien H. Nguyen, 1995, Capital market imperfections and the incentive to lease. Journal of Financial Economics 39: 271-294. Shumway, Tyler, 2001, Forecasting bankruptcy more accurately: A simple hazard model. The Journal of Business 74 (1): 101-124. Teoh, Siew Hong, and Yinglie Zhang, 2005, Loss firms, asset pricing controls, and accounting anomalies, Working paper, Ohio State University. Titman, Sheridan, K. C. John Wei, and Feixue Xie, 2004, Capital investments and stock returns. Journal of Financial and Quantitative Analysis 39: 677-700. Trueman, Brett, The relationship between the level of capital expenditure and firm value, Journal of Financial and Quantitative Analysis 21 (2): 115-129. Xie, Hong, 2001, The mispricing of abnormal accruals, The Accounting Review 76 (3): 357-373. Zach, Tzachi, 2004, Evaluating the “accrual-fixation” hypothesis as an explanation for the accrual anomaly. Working paper, Washington University in St. Louis.

36

Figure 1 Time series properties of earnings based on operating leases
0.04 0.03 Mean Earnings 0.02 0.01 0 -0.01 -0.02 -0.03 -0.04 Event Year -3 -2 -1 0 1 2 3 Low Lease Portfolio High Lease Portfolio

Year 0 is the year in which firms are ranked and assigned in equal numbers to ten portfolios based on ∆OPLEASE. Low Lease Portfolio is the bottom decile lease firms and High Lease Portfolio is the top decile lease firms. EARNINGS is earnings before extraordinary items (Compustat item 123). ∆OPLEASE is the change of the present value of the next five years’ minimum rent commitment under operating leases (Compustat item 96, 164, 165, 166, and 167). The present value is calculated using 10% discount rate.

37

Appendix A: An example of calculating the unrecorded liability resulting from off-balance-sheet lease activities *********************************************************************
Starbucks Corp. (2002 10-K filing) FOOTNOTE 10: LEASES The Company leases retail stores, roasting and distribution facilities and office space under operating leases expiring through 2025. Most lease agreements contain renewal options and rent escalation clauses. Certain leases provide for contingent rentals based upon gross sales. Minimum future rental payments under non-cancelable lease obligations as of September 29, 2002, are as follows (in thousands): Fiscal year ending 2003 2004 2005 2006 2007 Thereafter Total minimum lease payments 248,016 243,519 232,641 219,384 203,395 863,874 2,010,829

********************************************************************** Calculating present value of future operating lease obligations (OPLEASE): Fiscal 2002 Year 1 Year 2 Year 3 Year 4 Year 5 Thereafter
Present value of future operating lease obligations (OPLEASE)

Present Value 225,469 201,255 174,787 149,842 126,292 406,67443 1,284,320

248,016 243,519 232,641 219,384 203,395 863,874

43

Number of years thereafter = 863,874/203,395=4.24; round to five years; then the thereafter annual payment is 863,874/5=172,774. The present value of a five-year annuity of $172,774 at 10% = 406,674.

38

Appendix B: An example of capitalizing a five-year operating lease over time Assume that at the end of Year 0, the company enters a five-year lease contract. The interest rate is 10%. The future minimum payments are as follows: Lease payments Present value at Year 0 Year 0 909 Year 1 1,000 826 Year 2 1,000 751 Year 3 1,000 683 Year 4 1,000 Year 5 1,000 621 Total present value of future lease $3,791 payments (OPLEASE) at Year 0
Balance Sheet End book value of asset Year 0 Year 1 Year 2 Year 3 Year 4 Year 5 3,791 3,033 2,275 1,516 758 0 End book value of liability 3,791 3,170 2,487 1,736 909 0 End book value of equity -137 -213 -219 -151 0 Income Statement Interest expense Depr. +Int. exp. 1137 1075 1007 932 849 Impact on operating income44 -137 -75 -7 68 151 Cash Flow Statement Cash flow from operating activities (CFO) -1,000 -1,000 -1,000 -1,000 -1,000

Depreciation

758 758 758 758 758

379 317 249 174 91

Impact of Capitalizing Leases on Balance Sheet
4,000 3,500 Present Value of the unrecorded debt on operating lease asset

Asset / Liability

3,000 2,500 2,000 1,500 1,000 500 0 0 1

Net Book Value of the unrecorded operating lease asset

2

3

4

5

Life of Lease

44

Impact on operating income is calculated as (Lease Payment – Depreciation Expense – Interest Expense) (Imhoff et al. 1997).

39

Appendix C: An example of firm growth and operating leases Assume that at the end of each year, from Year 0 to Year 5, the company enters a new five-year lease contract. The company no longer takes new leases from Year 6 to Year 10. The interest rate is 10%.
Balance Sheet Firm status Year Year 0 Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10 No. of leases at year beginning 0 1 2 3 4 5 5 4 3 2 1 End book value of asset 3,791 6824 9098 10615 11373 11373 7582 4549 2275 758 0 End book value of liability 3,791 6961 9448 11183 12092 12092 8301 5131 2645 909 0 End book value of equity -137 -349 -568 -719 -719 -719 -582 -370 -151 0 Depr. Income Statement Interest expense 379 696 945 1118 1209 1209 830 513 265 91 Depr. +Int. exp. 1137 2212 3219 4151 5000 5000 3863 2788 1781 849 Impact on operating Income -137 -212 -219 -151 0 0 137 212 219 151 Cash Flow Statement Lease payment 1000 2000 3000 4000 5000 5000 4000 3000 2000 1000

Growth Stage Steady State Declining Stage

758 1516 2275 3033 3791 3791 3033 2275 1516 758

Firm status

Year Year 0 Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10

∆Off-balance-sheet asset 3791 3033 2275 1516 758 0 -3791 -3033 -2275 -1516 -758

Growth Stage

Steady State Declining Stage

∆Off-balance-sheet liability (∆OPLEASE) 3791 3170 2486 1736 909 0 -3791 -3170 -2486 -1736 -909

(∆Off-balance-sheet asset - ∆Offbalance-sheet liability) / | ∆Offbalance-sheet liability| 0% -4% -9% -13% -17% 0% 0% 4% 9% 13% 17%

Off-balance-sheet accruals

Off-balance-sheet financing

40

Table 1 Summary statistics for 59,235 firm-year observations from 1988 to 2003 Panel A: Descriptive statistics
Category

Variable
EARNINGS

Mean
-0.024 0.007 0.003 0.023 0.000 0.024 0.049 0.033 0.016 0.000

Median
0.027 0.001 0.001 0.007 0.000 0.024 0.002 0.000 0.000 0.000

Std Dev
0.188 0.034 0.010 0.080 0.008 0.183 0.173 0.144 0.101 0.007

25%
-0.044 -0.004 -0.001 -0.010 0.000 -0.052 -0.035 -0.012 -0.025 0.000

75%
0.071 0.012 0.005 0.043 0.000 0.112 0.074 0.011 0.039 0.000

Min
-1.898 -0.262 -0.097 -0.396 -0.203 -1.336 -0.390 -0.236 -0.389 -0.064

Max
0.347 0.277 0.085 0.589 0.093 0.826 1.428 1.427 0.727 0.077

Offbalancesheet Onbalancesheet Accruals
Onbalancesheet External Financing

∆OPLEASE ∆RENT ∆PPE ∆PPE_CAPLEASE TACC CFF ∆EQUITY ∆DEBT ∆CAPLEASE

Panel B: Spearman / Pearson correlation
EARNINGS ∆OPLEASE ∆RENT ∆PPE
∆PPE_CAPLEASE

TACC

CFF

∆EQUITY

∆DEBT

∆CAPLEASE

EARNINGS ∆OPLEASE ∆RENT ∆PPE
∆PPE_CAPLEASE

0.083 0.145 0.108 0.292 0.028 0.469 -0.202 -0.260 -0.036 0.003 0.776 0.305 0.061 0.234 0.140 0.056 0.119 0.063

0.062 0.816

0.167 0.222 0.224

0.009 0.038 0.034 0.050

0.520 0.191 0.191 0.436 0.026

-0.350 0.120 0.136 0.255 0.013 0.374

-0.380 0.073 0.086 0.079 0.005 0.149 0.812

-0.057 0.100 0.109 0.324 0.015 0.427 0.552 -0.039

0.007 0.076 0.076 0.160 0.066 0.070 0.060 0.018 0.077

0.300 0.055 0.225 0.175 0.113 0.121 0.057 0.096 0.530 0.315 0.063 0.316 0.142

TACC CFF ∆EQUITY ∆DEBT ∆CAPLEASE

0.050 0.032 0.000 0.036 0.081 0.459 0.085 0.448 0.065

0.532 0.698 0.069 -0.045 0.003

0.094

41

Table 1 Continued The sample covers 59,235 firm-year observations for the period 1988-2003. EARNINGS is earnings before extraordinary items (Compustat item 123). ∆OPLEASE is the change of the present value of the next five years’ minimum rent commitment under operating leases (Compustat item 96, 164, 165, 166, and 167). The present value is calculated using 10% discount rate. ∆RENT is the change in the first year’s operating lease payment (Compustat item 96). ∆PPE is the change in PPE other than capitalized leased PPE (Compustat item 8 – Compustat item 159). ∆PPE_CAPLEASE is the change of capitalized leased PPE (Compustat item 159). TACC is total accruals, calculated as EARNINGS – CFO-CFI. CFO is cash flow from operations (Compustat item 308). CFI is cash flow from investing (Compustat item 311). CFF is net external financing reflected on balance sheet, calculated as the sum of ∆EQUITY and ∆DEBT. ∆EQUITY is net equity financing measured as the proceeds from the sale of common and preferred stock (Compustat item 108) less cash payments for the purchase of common and preferred stock (Compustat item 115) less cash payments for dividends (Compustat item 127). ∆DEBT is net debt financing measured as the cash proceeds from the issuance of long-term debt (Compustat item 111) less cash payments for long-term debt reductions (Compustat item 114) less the net changes in current debt (Compustat item 301). ∆CAPLEASE is the change of capitalized lease obligations (Compustat item 84). All variables are scaled by average total assets (Compustat item 6). All correlations greater than 0.01 in absolute magnitude are significant at less than 0.01 levels.

42

Table 2 Industry composition for decile portfolios sorted by operating leases Panel A: Percentage of the firms in each industry group for each ∆OPLEASE decile (Column)
Industry groups Agriculture Mining & Construction Food & Tobacco Textile and Apparel Lumber, Furniture, & Printing Chemicals Refining & Extractive Durable Manufacturers Computers Transportation Utilities Retail Services Banks & Insurance Pharmaceuticals Total Lowest 0.2 1.5 1.6 1.4 2.6 1.6 1.1 21.2 25.0 5.5 1.0 12.4 14.7 4.9 5.6 100% 2 0.3 1.9 2.0 2.3 3.2 2.5 3.0 29.5 18.8 4.2 1.1 11.2 10.0 4.2 6.0 100% 3 0.4 2.4 2.7 2.3 4.1 3.6 5.2 29.0 14.9 4.3 1.2 9.6 8.8 6.4 5.3 100% 4 0.4 3.6 3.3 1.9 4.8 3.9 6.7 26.9 9.9 5.3 1.7 7.5 6.9 13.2 3.9 100% 5 0.5 2.9 2.6 1.7 4.6 3.3 6.1 24.5 8.3 6.5 1.8 6.6 7.7 18.7 4.2 100% 6 0.3 2.6 3.0 2.0 5.1 3.3 4.8 27.5 10.7 6.1 1.6 9.6 8.0 10.5 4.9 100% 7 0.4 2.5 2.6 2.1 4.6 3.2 4.7 25.3 13.1 5.9 1.4 12.1 9.5 7.5 5.3 100% 8 0.3 1.7 1.9 2.2 3.6 2.2 2.8 25.3 15.9 5.9 1.3 13.9 12.1 5.9 4.9 100% 9 0.4 1.4 1.6 1.7 3.3 1.6 1.9 20.4 17.5 6.2 1.0 19.0 14.0 5.0 5.0 100% Highest 0.1 0.9 1.2 1.3 1.5 1.0 1.1 14.2 19.0 7.4 0.8 26.6 16.9 4.0 4.3 100%

Panel B: Percentage of the firms in each ∆OPLEASE decile for each industry group (Row)
Industry groups Agriculture Mining & Construction Food & Tobacco Textile and Apparel Lumber, Furniture, & Printing Chemicals Refining & Extractive Durable Manufacturers Computers Transportation Utilities Retail Services Banks & Insurance Pharmaceuticals Lowest 6.6 7.0 6.9 7.2 6.8 6.0 3.0 8.7 16.3 9.5 7.5 9.7 13.5 6.1 11.3 2 8.1 8.7 9.0 11.9 8.5 9.3 8.2 12.1 12.3 7.3 8.8 8.7 9.2 5.3 12.1 3 11.1 11.3 12.0 12.2 11.0 13.9 13.8 11.9 9.7 7.6 9.2 7.4 8.1 8.0 10.7 4 13.1 16.9 14.7 10.2 13.0 14.9 17.9 11.0 6.5 9.2 13.4 5.9 6.4 16.4 8.0 5 14.7 13.5 11.7 9.1 12.2 12.8 16.4 10.1 5.4 11.3 13.8 5.1 7.1 23.3 8.5 6 8.6 12.0 13.4 10.6 13.7 12.6 12.9 11.3 7.0 10.7 12.1 7.5 7.4 13.1 10.0 7 13.1 11.9 11.3 11.3 12.2 12.1 12.5 10.4 8.6 10.2 10.9 9.4 8.8 9.4 10.7 8 10.1 7.8 8.6 11.9 9.7 8.6 7.5 10.4 10.4 10.4 10.4 10.8 11.2 7.3 9.9 9 11.6 6.7 7.1 8.9 8.9 6.3 5.1 8.4 11.4 10.8 7.8 14.8 12.9 6.2 10.1 Highest 3.0 4.1 5.3 6.8 4.0 3.7 2.8 5.8 12.4 12.9 5.9 20.7 15.5 4.9 8.7 Total 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100%

43

Table 2 Continued

The sample covers 59,235 firm-year observations for the period 1988-2003. Firms-year observations are ranked annually and assigned in equal numbers to decile portfolios. Industry classifications are compiled using the following SIC codes: Agriculture: 0100-0999; Mining: 1000–1299, 1400–1999; Food & Tobacco: 2000–2199; Textiles and Apparel: 2200–2399; Lumber, Furniture, & Printing: 2400-2796; Chemicals: 2800–2824, 2840–2899; Refining & Extractive: 1300–1399, 2900–2999; Durable Manufacturers: 3000–3569, 3580-3669, 3680-3999; Computers: 3570–3579, 3670–3679, 7370–7379; Transportation: 4000–4899; Utilities: 4900– 4999; Retail: 5000–5999; Services: 7000–7369, 7380–9999; Banks & Insurance: 6000–6999; Pharmaceuticals: 2830–2836.

44

Table 3 Time-series means and t-statistics for coefficients from annual cross-sectional regressions of future earnings for 59,235 firm-year observations from 1988 to 2003 Panel A: Off-balance-sheet accruals (∆OPLEASE) and future earnings
One-year-ahead Earnings (1) -0.014 (-3.98) 0.749 (44.12) -0.159 (-4.46) (2) -0.012 (-3.73) 0.758 (45.00) -0.131 (-3.84) -0.067 (-3.97) (3) -0.011 (-3.25) 0.808 (48.82) -0.075 (-2.19) -0.120 (-10.57) -0.124 (-12.01) 45.62% 45.94% 46.47% 46.69% (4) -0.011 (-3.43) 0.812 (48.36) -0.087 (-2.51) -0.100 (-5.28)

Intercept EARNINGS ∆OPLEASE ∆PPE TACC TACC_NET OF PPE Adjusted R-Square

Panel B: Off-balance-sheet financing (∆OPLEASE) and future earnings
One-year-ahead Earnings

(1)
Intercept EARNINGS ∆OPLEASE CFF ∆EQUITY ∆DEBT Adjusted R-Square 45.62% -0.014 (-3.98) 0.749 (44.12) -0.159 (-4.46)

(2)
-0.011 (-3.26) 0.719 (46.44) -0.098 (-2.92) -0.092 (-8.04)

(3)
-0.011 (-3.20) 0.712 (46.12) -0.098 (-2.93) -0.121 (-9.16) -0.051 (-6.12)

46.17%

46.23%

The sample covers 59,235 firm-year observations for the period 1988-2003. EARNINGS is earnings before extraordinary items (Compustat item 123). ∆OPLEASE is the change of the present value of the next five years’ minimum rent commitment under operating leases (Compustat item 96, 164, 165, 166, and 167). The present value is calculated using 10% discount rate. ∆RENT is the change in the first year’s operating lease payment (Compustat item 96). ∆PPE is the change in PPE (Compustat item 8). TACC is total accruals, calculated as EARNINGS – CFO-CFI. CFO is cash flow from operations (Compustat item 308 – Compustat item 124). CFI is cash flow from investing (Compustat item 311). TACC_NET OF PPE equals (TACC – ∆PPE). CFF is net external financing reflected on balance sheet, calculated as the sum of ∆EQUITY and ∆DEBT. ∆EQUITY is net equity financing measured as the proceeds from the sale of common and preferred stock (Compustat item 108) less cash payments for the purchase of common and preferred stock (Compustat item 115) less cash payments for dividends (Compustat item 127). ∆DEBT is net debt financing measured as the cash proceeds from the issuance of long-term debt (Compustat item 111) less cash payments for long-term debt reductions (Compustat item 114) less the net changes in current debt (Compustat item 301). All variables are scaled by average total assets (Compustat item 6).

45

Table 4 Time-series means and t-statistics for coefficients from annual cross-sectional regressions of future earnings for 50,167 firm-year observations from 1988 to 2002 Panel A: Off-balance-sheet accruals (∆OPLEASE) and future earnings
Two-year-ahead Earnings (1) -0.015 (-3.11) 0.610 (30.04) -0.138 (-5.57) (2) -0.014 (-2.72) 0.617 (29.98) -0.105 (-4.02) -0.072 (-3.40) (3) -0.012 (-2.58) 0.672 (32.59) -0.049 (-1.82) -0.126 (-7.99) -0.131 (-8.45) 30.42% 30.70% 31.49% 31.66% (4) -0.012 (-2.56) 0.674 (32.40) -0.059 (-2.13) -0.105 (-4.48)

Intercept EARNINGS ∆OPLEASE ∆PPE TACC TACC_NET OF PPE Adjusted R-Square

Panel B: Off-balance-sheet financing (∆OPLEASE) and future earnings
Two-year-ahead Earnings

(1)
Intercept EARNINGS ∆OPLEASE CFF ∆EQUITY ∆DEBT Adjusted R-Square 30.42% -0.015 (-3.11) 0.610 (30.04) -0.138 (-5.57)

(2)
-0.011 (-2.48) 0.568 (29.56) -0.056 (-1.94) -0.122 (-10.06)

(3)
-0.011 (-2.43) 0.558 (29.39) -0.056 (-1.94) -0.171 (-11.66) -0.050 (-5.29)

31.43%

31.70%

The sample covers 50,167 firm-year observations for the period 1988-2002. EARNINGS is earnings before extraordinary items (Compustat item 123). ∆OPLEASE is the change of the present value of the next five years’ minimum rent commitment under operating leases (Compustat item 96, 164, 165, 166, and 167). The present value is calculated using 10% discount rate. ∆RENT is the change in the first year’s operating lease payment (Compustat item 96). ∆PPE is the change in PPE (Compustat item 8). TACC is total accruals, calculated as EARNINGS – CFO-CFI. CFO is cash flow from operations (Compustat item 308 – Compustat item 124). CFI is cash flow from investing (Compustat item 311). TACC_NET OF PPE equals (TACC – ∆PPE). CFF is net external financing reflected on balance sheet, calculated as the sum of ∆EQUITY and ∆DEBT. ∆EQUITY is net equity financing measured as the proceeds from the sale of common and preferred stock (Compustat item 108) less cash payments for the purchase of common and preferred stock (Compustat item 115) less cash payments for dividends (Compustat item 127). ∆DEBT is net debt financing measured as the cash proceeds from the issuance of long-term debt (Compustat item 111) less cash payments for long-term debt reductions (Compustat item 114) less the net changes in current debt (Compustat item 301). All variables are scaled by average total assets (Compustat item 6).

46

Table 5 Time-series means and t-statistics for coefficients from annual cross-sectional regressions of future size-adjusted return for 56,755 firm-year observations from 1988 to 2003 Panel A: Off-balance-sheet accruals (∆OPLEASE) and future stock returns
One-year-ahead Size-adjusted Return
(1) Intercept
0.066 (1.45) -0.843 (-3.21)

(2)
0.078 (1.64) -0.606 (-3.05) -0.554 (-3.04)

(3)
0.070 (1.55) -0.550 (-2.69)

(4)
0.076 (1.66) -0.483 (-2.75) -0.563 (-3.17)

(5)
0.035 (0.67) -0.481 (-2.55) -0.511 (-3.02)

∆OPLEASE
∆PPE TACC TACC_NET OF PPE Book to Market Adjusted R-Square

-0.314 (-2.85) -0.275 (-2.51) -0.281 (-2.47) 0.065 (3.78) 1.26%

0.14%

0.39%

0.56%

0.68%

Panel B: Off-balance-sheet financing (∆OPLEASE) and future stock returns
One-year-ahead Size-adjusted Return
(1) Intercept
0.066 (1.45) -0.843 (-3.21)

(2)
0.081 (1.89) -0.673 (-2.41) -0.313 (-4.89)

(3)
0.083 (1.90) -0.660 (-2.40)

(4)
0.043 (0.95) -0.646 (-2.28)

∆OPLEASE
CFF ∆EQUITY ∆DEBT Book to Market Adjusted R-Square

-0.190 (-1.57) -0.436 (-7.36)

-0.151 (-1.28) -0.414 (-6.68) 0.059 (4.38) 1.67%

0.14%

1.12%

1.25%

The sample covers 56,755 firm-year observations for the period 1988-2003. EARNINGS is earnings before extraordinary items (Compustat item 123). ∆OPLEASE is the change of the present value of the next five years’ minimum rent commitment under operating leases (Compustat item 96, 164, 165, 166, and 167). The present value is calculated using 10% discount rate. ∆RENT is the change in the first year’s operating lease payment (Compustat item 96).

47

Table 5 Continued
∆PPE is the change in PPE (Compustat item 8). TACC is total accruals, calculated as EARNINGS – CFO-CFI. CFO is cash flow from operations (Compustat item 308). CFI is cash flow from investing (Compustat item 311). TACC_NET OF PPE equals (TACC – ∆PPE). CFF is net external financing reflected on balance sheet, calculated as the sum of ∆EQUITY and ∆DEBT. ∆EQUITY is net equity financing measured as the proceeds from the sale of common and preferred stock (Compustat item 108) less cash payments for the purchase of common and preferred stock (Compustat item 115) less cash payments for dividends (Compustat item 127). ∆DEBT is net debt financing measured as the cash proceeds from the issuance of long-term debt (Compustat item 111) less cash payments for long-term debt reductions (Compustat item 114) less the net changes in current debt (Compustat item 301). Annual returns are calculated from the start of the fourth month subsequent to the fiscal year-end. The sizeadjusted return is calculated by deducting the value-weighted average return for all firms in the same sizematched decile, where size is measured as the market value at the beginning of the return accumulation period. For delisted firms during the future return window, the remaining return is calculated by first applying CRSP’s delisting return and then reinvesting any remaining proceeds in the appropriate size-matched portfolio.

48

Table 6 Nonlinear Generalized Least Squares Estimation of the Stock Price Reaction to Information in Operating Leases about Future Earnings Panel A
EARNINGS t +1 = γ 0 + γ 1 EARNINGS t + γ 2 ∆OPLEASE t + ν t +1 ABNORMALRETURN t +1 = β ( EARNINGS t +1 − γ 0 − γ 1 EARNINGS t − γ 2 ∆OPLEASE t ) + et +1 Forecasting Coefficients Valuation Coefficients Parameter Coefficient Estimate Parameter Coefficient Estimate (t-statistic) (t-statistic) 0.762 0.911 γ1 γ1 * (215.73) (52.12) -0.201 0.395 γ2 γ2 * (-10.52) (4.22) 1.30 β (46.68)
* *

Test of market efficiency: Null Hypothesis EARNINGS: γ1 = γ1* ∆OPLEASE: γ2 =γ2* * All γ = γ Likelihood Ratio Statistic 72.34 39.52 120.26 Marginal Significance Level