2004 V32 4: pp. 619–643

REAL ESTATE ECONOMICS

The Wealth Effects of Sale and Leasebacks: New Evidence Lynn M. Fisher∗ This paper investigates the phenomenon of sale and leasebacks as one way in which firms may use financial contracts to rearrange their organizational architecture. A theoretic model links the length of initial leaseback period to incentives to make noncontractible future investments in the lease relationship and predicts that firms choose shorter leases when landlords make relatively important investments. Using a sample of 71 sale and leaseback events from the 1990s, we document a significant mean abnormal return of 1.3% for shareholders of seller/lessee firms announcing relatively short leasebacks. The evidence suggests that firms may use sale and leasebacks to optimize their claims to real estate.

In this study, we investigate integration decisions by firms who engage in the sale and leaseback of commercial real estate. In a sale and leaseback, the firm sells an asset but simultaneously enters into a lease for its continued use. Sale and leasebacks have historically been considered financial contracts. This study differs from prior research, however, because we dispense with the assumption that leasebacks are all long-term financial leases.1 In particular, we hypothesize that contractual hazards first proposed by Coase (1937) influence the firm’s decision to use long-term leasebacks (continued integration of the real estate within the firm) versus short-term leasebacks (nonintegration). As applied to commercial real estate markets, this work suggests that in certain cases it is efficient for investors other than the user of an asset to own commercial real estate. Prior studies have found that announcements of sale and leasebacks are associated with positive wealth effects for seller/lessee firms and that these wealth gains are attributable to the reallocation of tax benefits from the ownership of a durable asset to a firm who values the benefits more highly than the seller (Slovin, Sushka and Polonchek 1990, Rutherford 1990, 1992, Alvayay, Rutherford and Smith 1995, Ezzell and Vora 2001). Alvayay, Rutherford and ∗ 1

Massachusetts Institute of Technology, Cambridge, MA 02139 or [email protected].

Financial leases are defined by Copeland and Weston (1983) as leases that are separable from maintenance activities, are noncancelable and which fully amortize the leased asset.

620 Fisher

Smith (1995), however, provide evidence that the value of taxes expropriated from the government through sale and leasebacks was reduced by the Tax Reform Act of 1986. Absent significant tax consequences, the extant financial theory suggests that announcements of sale and leasebacks should be similar to other announcements of debt. We develop a model of the sale and leaseback in which the length of the initial leaseback influences the incentives of both the seller/lessee and the buyer/lessor to make future noncontractible investments in activities related to the real estate. Lease length matters when the returns to investment are dependent on the continuation of the leasing relationship and the investors anticipate that some of the returns may be appropriated by their contracting partner at the end of the lease. We show both that the choice of a leaseback period is endogenous to the decision to enter into a sale and leaseback and that the optimal choice depends on the relative importance of one party versus the other in producing joint wealth. The model predicts that firms optimally choose shorter lease lengths when there are positive wealth gains to be captured relative to continued ownership of the asset. Using a sample of 71 sale and leaseback events involving commercial real estate between 1990 and 2000, we use standard event study methodology to document that there were no abnormal returns to the shareholders of seller/lessee firms for our full sample on the day of a sale and leaseback announcement (day 0). When we divide the sample into short (less than or equal to 15-year) or long (greater than 15-year) leasebacks, we document a mean abnormal return of 1.3% for shareholders of seller/lessee firms in the short subsample. The mean abnormal return associated with short leasebacks is significantly different from zero at the 5% level of significance and is also different from the mean abnormal return to lessee firms announcing long leasebacks at the 1% level. In multivariate analysis, we control for alternative explanations of wealth gains including the tax hypothesis. In all cases, a categorical set of indicators for the length of the initial leaseback retain their magnitude and significance in explaining abnormal returns to lessee firms. We conclude that short leasebacks are used when the buyer/landlord is expected to make relatively important contributions to the value of the relationship, and that the use of sale and leasebacks may efficiently reorganize a firm’s claim to its real estate. This paper proceeds as follows. Next we briefly review related literature about the nature of the firm. Then we develop a theoretical model that demonstrates the relationship between lease length and value of the sale and leasebacks; in an Appendix we derive some our theoretical results. Given our theoretical predictions, we introduce our sample and methodology and report our empirical results. The last section concludes.

The Wealth Effects of Sale and Leasebacks 621

Related Literature Beginning with Coase (1937), financial economists have long tried to understand the nature of the firm. Our research is related to recent work in transaction cost economics and property rights theory that explain the decision of firms to own certain assets. The theory of vertical integration suggests that firms should own assets when potentially incomplete contracts expose them to future opportunistic behavior by their contracting partners (Klein, Crawford and Alchian 1978, Williamson 1979). If real estate needs to be highly specialized to a particular manufacturing process, for example, the firm faces a hold-up problem when it anticipates that the landlord will expropriate some of the value of the specialized leased space from the firm in the future.2 There may be excess value for the landlord to appropriate once the firm has invested in modifications to the property whenever the space has greater value in its use to the firm than in an alternate use. When the returns to investment are highest in the current use of the real estate, the investment is said to be specific to the lease relationship. The theory of vertical integration suggests that a firm maximizes value by owning the assets in which it optimally makes specific investments. While the reasoning behind the theory of vertical integration is somewhat anecdotal, recent property right models beginning with Grossman and Hart (1986) provide more completely specified models that consider the organization of the firm. The property right models assume that contracts are incomplete and focus on how the structure of contracts, and especially the ownership of assets, affects the incentives of either party to undertake future, unforeseen or noncontractible investments that generate value in context of the relationship. The models of Hart (1995) and Whinston (2003), for example, focus on the relative marginal contribution of each party in generating surplus future value as a determinant of the optimal degree of integration between contracting parties. With respect to sale and leasebacks, we create a property rights model that investigates the incentives created by different lengths of the initial leaseback period.3 We assume that the tenant benefits from any investments made during the initial lease term, but the remaining surplus from the relationship is

2

The hold-up problem arises from the fact that the firm may not, in fact, invest in a project if some of the returns are expected to be captured by the landlord.

3

There is good precedent for expecting lease length to vary with hold-up problems. See Klein, Crawford and Alchian (1978), Mulherin (1986) and Joskow (1987), among others. We also assume that options to renew the contract are priced in the initial contract, but that a right to renew does not avoid the renegotiation of the exact terms of the contract when the first lease expires.

622 Fisher

distributed between the parties through renegotiation at the end of that period. Therefore, the choice of an initial lease length ex ante alters the incentives of either party to invest in the relationship ex post. Longer leases provide favorable conditions for tenant investment while shorter leases create some additional incentives for the landlord to invest.4 Depending on the marginal contribution of each party to the total value of the lease, de-integrating the firm with respect to the ownership of its real estate may be wealth enhancing. Financial economists have offered several additional explanations for why financial contracting influences firm value. In particular, Myers, Dill and Bautista (1976), Lewellen, Long and McConnell (1976) and Alvayay, Rutherford and Smith (1995) hypothesize that the wealth effects of sale and leasebacks that utilize financial leases differ from stock price reactions to debt announcements because of the tax consequences of sale and leasebacks. When buyers and sellers have different tax rates and abilities to utilize asset depreciation allowances, sale and leasebacks allow parties to generate wealth gains by expropriating wealth from the government. The main body of empirical evidence about sale and leasebacks supports the hypothesis that financial leases are perfect substitutes for secured debt except for their tax consequences under U.S. tax laws prevailing until 1986. Table 1 presents a summary of the existing empirical evidence about sale and leasebacks (market reactions to various debt issues are included for comparison). Changes in tax laws during the late 1980s and early 1990s may have reduced the tax advantages of sale and leasebacks. Alvayay, Rutherford and Smith (1995) examine a set of real estate sale and leasebacks from 1982–1989. They predict that tax changes in the Tax Reform Act of 1986 had a negative impact on the tax benefits of sale and leasebacks, and their event study found supporting evidence. Most recently, Ezzell and Vora (2001) provide evidence that lessee firms’ tax rates are negatively related to average cumulative abnormal returns for lessee firms announcing sale and leasebacks between 1984 and 1991. The study did not separate the sample or otherwise examine how changes in tax rules affected the observed wealth gains, however. It should be noted that subsequent to all of these sample periods in 1993, depreciation recovery periods for real estate

4 That financial leases can allow tenants to exercise considerable control over real estate analogous to ownership of the asset seems likely in the case of sale and leasebacks. Anecdotal evidence suggests that “the seller-lessee brings the property to the table and consequently possesses more significant bargaining power than an average tenant in a typical lease negotiation. Accordingly, a seller-lessee can use its leverage to negotiate leases that allow it to control maintenance and alterations, have substantial and varied assignment and subletting rights, enjoy lengthy initial and renewal terms and control the operation of the property” (Fodor and Kiely 2003, p. 4).

1984–1991

Ezzell and Vora (2001)

b

44 firms

45 firms

33 firms

28 SLBs of real property 1982–1986 17 SLBs of real property 1987–1989 44 SLBs of which 29 represent land and real estate

59 SLBs of real property 14 SLBs of aircraft 41 SLBs of real property 33 SLBs of various type assets

459 straight debt 189 mortgage 172 straight debt

Number of Events and Type

n/a $102 $82 $176 $324

$156

+.85% (1.98); +2.29% (2.28) +1.59% CAR (2.88) Day −1 Equity AAR + 0.95% (1.97); Monthly Equity CAR + 3.83% (2.60); Monthly Bond CAR + 1.41% (1.65). +0.8 CAR (2.1)

2.63% CAR (p-value < 0.001)

$104

n/a

−0.06% (−0.44); −0.20% (−1.67) −.23% (−1.40)

−0.3 CAR (−0.3)

Mean SLB Size ($M)

AAR or CAR for days (−1, 0)a

Two Day Average Abnormal Return unless otherwise noted. CARs are also reported for the 2 day interval (−1, 0). T-statistic in parenthesis. Full sample includes 360 industrial firms of which 78 issued debt.

1982–1989

Alvayay, Rutherford and Smith (1995)

a

1982–1989

Rutherford (1992)

41 firms

1975–1986

1980–1987

73 firms

1972–1982

Mikkelson and Partch (1986) Slovin, Sushka and Polonchek (1990)

Rutherford (1990)

216 industrial and utility 78 industrialb

1964–1981

Eckbo (1986)

Number of Firms and Type

Sample Period

Author

Table 1
Summary of prior results for studies of debt and sale-and-leaseback (SLB) announcements.

The Wealth Effects of Sale and Leasebacks 623

624 Fisher

were lengthened once again, which might be expected to further reduce the tax benefits available to be traded in sale and leasebacks.5 Smith and Warner (1979), Krishnan, Sivarama and Moyer (1994) and Barclay and Smith (1995b) note that financial leases may mitigate costs to the lessor in default relative to the costs of a secured lender. Since the lessor is technically the owner of the asset, the lessor has better priority than a lender who holds a mortgage lien if the lessee goes into default or bankruptcy. Krishnan, Sivarama and Moyer (1994) and Sharpe and Nguyen (1995) provide empirical evidence that firms with high costs of capital are more likely to lease. In addition, Barclay and Smith (1995a) find that firms suffering greater asymmetric information problems (and therefore a higher cost of capital) are more likely to utilize short-term debt financing. If credit-constrained firms are more likely to lease and to utilize shorter debt maturities, then controlling for whether selling/leasing firms are credit constrained may be an important robustness check of any results linking leaseback length to wealth gains. Our sample of real estate sale and leasebacks provides an opportunity to empirically examine the integration decision of the firm for several reasons. Most importantly, the nature of the transaction requires firms to explicitly choose a lease length, and as mentioned previously, we observe both short and long leaseback arrangements by public firms over real estate in the 1990s. Second, while market contracts including leases are well known to incur other contracting costs, namely agency costs, there are good reasons to believe that these costs are mitigated in the case of commercial leases over durable real estate assets, which allows us to focus mainly on the integration decision.6 Third, by definition, the seller/lessee already owns the real estate and continues to occupy the space after the sale and leaseback deal is in place. Therefore, the sale and leaseback event is not confounded by the disposition of the seller/lessee firm’s use of the asset. Finally, sale and leasebacks offer interesting evidence about the relationship between capital structure and the theory of the firm as recently investigated by Zingales (2000) and Smith (2001).

5

Our sample of sale and leasebacks includes only two observations that occur prior to 1993 and only one of these has sufficient information to calculate a proxy for its tax rate, so we are unable to investigate how tax benefits of sale and leaseback potentially changed after 1993. 6

Recent theory and empirical evidence suggest that there exist sources of economic gains from commercial real estate leasing which may dominate expected agency costs. According to Benjamin, de la Torre and Musumeci (1998, p. 223) the “primary factors in favor of leasing are the abilities of landlords and property managers to eliminate free-rider problems, to exploit economies of scale, and to specialize in the valuation, maintenance and disposal of commercial property.”

The Wealth Effects of Sale and Leasebacks 625

A Sale and Leaseback Model In this section, we develop a theoretic model to guide our investigation. The model is closely related to recent work by Hart (1995) and Whinston (2003) in which contracts are assumed to be incomplete. At time 0 in our model, the seller/lessee and the buyer/lessor decide to enter into a sale and leaseback agreement. They agree on all contractible events and settle on a market transfer price, which the buyer pays the seller at time 0. The parties simultaneously create a lease that is a single-tenant, triple net lease with a fixed rental payment at the beginning of the lease period. The length of the initial leaseback period is x, where we normalize the lease relative to the remaining economic life of the improved real estate, 0 < x ≤ 1.7 One issue with the lease is that some future investments by the seller/lessee and the buyer/lessor cannot be written into the contract ex ante.8 For example, the lessee may be unable to fully anticipate future needs with respect to business operations at this location. The lessee may need to make modifications to the real estate itself or invest in marketing for a new product that that will be sold from this location. A lessor, on the other hand, may consider investment in new technologies to improve the productivity of the building. Hart (1995) describes an incomplete contract as one that will “have gaps, missing provisions, or ambiguities, and so situations will occur in which some aspects of the uses of non-human assets are not specified” (p. 29). Even if some investments are anticipated, their exact magnitude and affect on business revenues may not be describable ex ante. As is common in the literature, we assume that both the tenant and the landlord can observe each other’s investment level and cost functions, but, even if anticipated, some investments are either too complicated to write into a contract or unverifiable to a third party and therefore the investments are noncontractible. Even more simply, some investments may simply not be anticipated at time 0. We denote the noncontractible level of investment by the tenant and landlord by i ∈ R+ and e ∈ R+ , respectively, and the cost of investment by ci (i) and ce (e). We assume that the cost functions are convex in their arguments.9 Noncontractible investments are assumed to include a whole range of investments in either the real estate itself or the activities of the firms at the particular location.

7

We assume that renegotiation is costly and therefore a series of “spot” contracts is sufficiently costly to be ignored.

8

Some future investments are contractible and are therefore able to be written into the lease. For example, tenant improvements are a common investment included in the initial lease. Whinston (2003) shows that neither sunk nor contractible investments affect the likelihood of integration. For clarity, we suppress contractible investments in the following model assuming that appropriate transfer pricing has already been established. 9 See, for example, the assumed functional form in Whinston (2003).

626 Fisher

While the noncontractible investments in question can be made any time during the initial leaseback period, we assume that the timing of the investments are exogenously determined, and to simplify the model, we assume that all investments are made immediately following the sale and leaseback agreement at time 0. We also assume for simplicity that renegotiation occurs at time x. Each party’s discount rate is assumed to be 0, and there is no risk of default. Let the revenue generated during the remaining life of the real estate asset from the investment i when the tenant and landlord continue in their leasing relationship be denoted by R(i) and the revenue from the buyer/lessor’s noncontractible investment be denoted by S(e), where R(0) = 0 and S(0) = 0. Denote total revenues from the relationship after noncontractible investments have been made by π = R(i) + S(e) and let these revenues be evenly distributed over the remaining life of the real estate. We assume that ∂R/∂i ≥ 0 and ∂S/∂e ≥ 0. Whether or not the parties realize the total revenue stream from their noncontractible investments may depend on whether or not the parties continue in their relationship after time x. Denote by r (i) ≥ 0 the tenant’s total revenue from investment if the investment is made in the current location but realized in the tenant’s next-best market lease, and by s(i, e) ≥ 0 the landlord’s revenue from noncontractible investments made now but realized in his next-best leasing arrangement. The extent to which R(i) differs from r (i) and S(e) from s(i, e) captures the degree to which investments are specific to the sale and leaseback relationship. Using the prior example of an investment in marketing for retail goods sold at the site, suppose that the advertisements included the address of the tenant’s current location when the investment was made. The payoff to this investment would be reduced if the tenant were to move to a new site. The revenues may not be diminished completely since awareness of the product was generated, but even if she relocates right across the street, the expected revenues will be reduced by the cost of the inaccurate information about the tenant’s location. If the tenant’s noncontractible investment is in real estate improvements, then there may be no value of this investment in an alternate lease for the tenant. Notice that the buyer/lessor’s investment, e, does not affect the payoff of the tenant’s outside option, r, since the tenant’s next-best alternative involves another landlord and building. On the other hand, the landlord’s alternate revenue stream may be influenced by the investments made by both himself and the prior tenant (to the extent that the tenant’s investments were in the real estate).10

10 Prior to the sale and leaseback agreement, had the seller/tenant merely hired a manager to operate the property, the owner/tenant could have fired the manager in the event of disagreement and hired a new manager with a payoff from the continued use of the asset of r (i, e). The manager’s payoff in such scenario would have been unaffected by the owner/tenant’s investments, s(e).

The Wealth Effects of Sale and Leasebacks 627

We assume that each party’s investment is at least as valuable in the context of the current relationship as compared to alternative trading relationships in the sense that ∂R/∂i ≥ (∂r /∂i) + (∂s/∂i) and ∂S/∂e ≥ ∂s/∂e, and that π ≥ r + s, so that it is efficient for the parties to renew the lease at time x. We are concerned with the value of the next-best alternatives r and s, not because we expect the tenant and landlord to fail to renew the lease, but because the value of each party’s alternative affects the surplus to be negotiated over at time x, and therefore affects the incentives of each party to undertake noncontractible investments. Although we assume that the parties create the initial sale and leaseback agreement under competitive conditions, once specific investments have been made, they find themselves in a bilateral monopoly whenever they must renegotiate the contract because each party can “hold up” the other through a threat of nonrenewal. We assume that both parties anticipate renegotiation and Nash bargaining over any surplus from their relationship (and that each has equal bargaining power). Define the total surplus value generated by noncontractible investments as V ≡ π − r − s. The choice of lease length alters the proportion of the surplus that each party expects to claim in renegotiation in the following way. Since we assume that leaseback involves a triple net lease with the rent paid to the landlord at time 0, the tenant realizes all of the revenue generated by noncontractible investments made by herself and the landlord during the first lease term. Recall that the main characteristic of this model is that the sale and leaseback contract is incomplete in some way. The exact magnitude of the surplus that results from decisions taken in these unspecified situations will not be known at time 0 and therefore cannot be explicitly written into contract. Therefore, each party recognizes that at time x they will negotiate over the remaining revenue from noncontractible investments made during the initial lease and split any surplus above and beyond the “threat point” of their outside options. We have assumed that revenues are evenly distributed through time; therefore the remaining surplus is (1 − x)V . The main implication of the model is that the choice of x influences the value of the relationship through the incentives it creates for each party to undertake noncontractible investments. The total revenue from noncontractible investments is therefore rewritten as π = R(i(x)) + S(e(x)). To see this point more clearly, consider that immediately following time 0 and given a lease of length x, the tenant optimally and noncooperatively chooses i according to   1 max xπ + (1 − x) r + V − ci (i). i 2 Here, the tenant’s objective function is comprised of the proportion of noncontractible total revenue that is realized during the initial lease period, plus for

628 Fisher

the remaining period, the value of her investment in the next-best alternative plus one-half the surplus value of the relationship, less her costs of investment. Substituting V ≡ π − r − s, we obtain   1 ∂R 1 ∂s ∂ci ∂r (1 + x) + (1 − x) − = . (1) 2 ∂i 2 ∂i ∂i ∂i Denote by i(x) the solution to (1). For the landlord, the problem is   1 max(1 − x) s + V − ce (e). e 2 The landlord’s objective function is comprised of the payoff that he negotiates at time x. The payoff includes the remaining value of his investment in the next-best alternative plus one-half of the relationship’s surplus value, less the cost of his investment. The landlord’s optimal (and noncooperative) investment level solves   1 ∂s ∂ce ∂S (1 − x) + = . (2) 2 ∂e ∂e ∂e Denote by e(x) the solution to (2). Notice that i(x) is increasing in the length of the lease. Correspondingly, the lessor’s optimal noncooperative investment level, e(x), is decreasing in lease length. In addition, given the assumptions of the model, if x = 1, then e(x) = 0, which means that a long-term lease over the remaining economic life of the real estate evokes no noncontractible investment by the lessor. The socially optimal levels of seller/lessee and buyer/lessor investment, i ∗ and e∗ , solve max R(i) − ci (i) + S(e) − ce (i). i,e

Our model is again consistent with the idea that a long-term lease is analogous to ownership of the asset by the seller/lessee. Referring to (1), if x = 1, the seller/tenant invests in a socially optimal manner and i(x) = i ∗ . If x < 1, however, the seller/lessee’s choice of noncontractible investment level is generally less than the socially optimal level. Importantly, we find that for any choice of x, the buyer/lessor’s choice of noncontractible investment level, e(x), is generally less than the socially optimal level, e∗ . This result occurs because the lessor’s

The Wealth Effects of Sale and Leasebacks 629

payoff is always reduced by having to wait until the end of the lease to negotiate and receive a return on investment, and even then the lessor only receives a proportion of the relationship-specific return. These claims are derived in the Appendix. Now consider the value of a sale and leaseback. When a firm owns and occupies its real estate prior to time 0 (in the sense of fee simple ownership), we assume that there are no contracts involving any other parties with respect to the real estate. Denote the value that accrues to the firm from noncontractible investment in the real estate when the firm decides not to sell at time 0 by W R = R(i ∗ ) − ci (i ∗ ).11 Define W0 as the joint wealth resulting from a sale and leaseback agreement with respect to the same real estate at time 0, W0 = R(i(x)) − ci (i(x)) + S(e(x)) − ce (e(x)). The change in total wealth attributable to the real estate that results from a sale and leaseback agreement is, therefore,   W0 − W R = [R(i(x)) − ci (i(x))] − R(i ∗ ) − ci (i ∗ ) + [S(e(x)) − ce (e(x))]. (3) Equation (3) yields the first empirical prediction of the sale and leaseback model. Remark: If a sale and leaseback has an initial lease of length x = 1, then there is no change in total wealth from the sale and leaseback agreement relative to continued ownership of the asset by the seller/lessee. This observation follows directly from the fact that when x = 1, i(x) = i ∗ and e(x) = 0. Therefore, when x = 1, (3) is equal to 0. Recall that i(x) is increasing in x, while e(x) is decreasing in x. A decrease in x is therefore expected to increase the landlord’s investment level while decreasing the tenant’s optimal and noncooperative choice of investment. Given that we expect the parties to optimize the sale and leaseback contract, the model yields a second empirical prediction:

11 Notice that here, as throughout this model, we have suppressed the value of any contractible or sunk investments.

630 Fisher

Result: The seller/lessee and buyer/lessor optimally choose x < 1 if and only if the net marginal gain of an increase in the buyer/lessor’s investment is greater than the net marginal loss from the seller/lessee’s decreased investment. This result can be derived by taking the total differential of (3). The change in total wealth is positive with respect to a decrease in x if and only if     ∂S  ∂e  ∂ce  ∂e  ∂R ∂i ∂ci ∂i − ≥ − .     ∂e ∂ x ∂e ∂ x ∂i ∂ x ∂i ∂ x If the change in total wealth is negative for any x < 1, then by the reasoning above, the seller/lessee and buyer/lessor are always better off by choosing a contract with x = 1. We conjecture that the exact nature of marginal returns and costs are likely to vary with the identity of the lessee and lessor and perhaps with the use of the real estate. Consistent with Hart (1995), the implications of the model suggest that optimal organizational form may depend on the relative importance of investments by one party versus another. If the buyer/lessor makes important noncontractible investments in the relationship, the choice of a relatively short lease as opposed to a long-term financial lease is wealth enhancing. Below we investigate the empirical implications of the model. Data and Methodology The Sample For the years 1990–2000, 158 announcements of real estate sale and leasebacks were identified from the Dow Jones Interactive service. Given this set of events, further searches were performed across other publications and the Lexis-Nexis index to identify the first public announcement of the sale and leaseback. An event qualified for this sample if the public announcement referred to a sale and leaseback involving particular (identified) assets owned by the selling firm prior to the agreement. Further screening required that firms report the length of the initial leaseback period and that the lessee firm was a public firm for whom CRSP data was available. Twelve sale and leaseback announcements are associated with other announcements for lessee firms on the same day (day 0). In particular, six announcements are concurrent with earnings reports, four are associated with announcements of multipart financing deals and two sale and leaseback announcements are concurrent with other business announcements.12 12 Both of the lessee firms in this category are financial institutions selling/leasing office buildings. No significance in these correlations was discerned in further analysis.

The Wealth Effects of Sale and Leasebacks 631

Given that we believe these confounding events to be idiosyncratic, we retain the events in our sample, but identify the observations for further tests to rule out some type of systematic information content. The sample is comprised of 71 events involving 69 different lessee firms. Many events involve multiple buildings or properties. These events fall in a time frame that is almost completely distinct from prior studies. Ezzell and Vora’s (2001) sample has the most recent observations and covers the period 1984–1991; only two events in our sample occur prior to 1994. Summary statistics are reported in Table 2. The average reported value of the deals is $39 million, which is significantly smaller than the size of transactions reported in prior studies (see Table 1). The mean value of sale and leaseback transaction relative to the market capitalization (value to equity or V/E) of the lessee firm is 36%, which is greater than the 22% found in Slovin, Sushka and Polonchek (1990).13 Of the publically traded buyer/lessors identified in the sample, 25 are Real Estate Investment Trusts (REITs). Five of the lessee firms are financial institutions. We have also identified the type of real estate space involved in these transactions from the newspaper announcements and SEC filings (see Table 2, Panel C). In the context of our theoretic model, we suspect that the type of real estate may proxy for the nature of the landlord’s versus the tenant’s future investment activity. For example, Smith (2001) states, “If office buildings are less firm-specific than manufacturing or research facilities, office buildings should be leased more frequently” (p. 10). Interpreting “less firm-specific” from the viewpoint of the lessee firm in our model, we might expect the value of the lessee’s investment in the business or asset to be relatively less valuable than the landlord’s for certain types of properties. Referring to Panel C of Table 2, office and hotel sale and leasebacks demonstrate the shortest initial lease periods of 12 years on average, while retail leases are quite long, averaging 20 years. The average initial lease terms for R&D and manufacturing, light industrial and warehouse and assisted living space fall in an intermediate range, while restaurants and entertainment properties are more closely aligned with retail space in terms of average lease lengths. Panel D also reports summary statistics for the sample according to whether the reported leaseback period is for 15 years or less, between 15 and 25 years or greater than 25 years. Our model suggests that altering the length of a leaseback results in a trade-off between landlord and tenant investment incentives, and that lease length alone may fall short of providing optimal incentives when both parties make important investments. Recent work by Wheaton (2000) and others on retail 13

The median transaction value to market capitalization for our sample is 21%, however.

632 Fisher

Table 2
Summary statistics. Panel A: Market capitalization of lessee firms ($M) Mean Median

$2,511 $92

Panel B: Distribution of transaction values Value ($M) $0–$20 $21–$30 $31–$40 $41–$50 $51–$100 $101–$150 >$150

Frequency 38 7 4 1 8 4 3

Panel C: Statistics by property type

Type Full Sample Office Retail R&D/Manufacturing Hotel Assisted Living Distribution/Warehouse Restaurant Entertainment Land

N 71 13 11 16 7 8 6 5 3 2

Mean Market Cap. ($M) 2,511 11,968 133 134 1,145 379 1,201 88 131 16

Value ($M) 39.08 56.33 37.92 10.88 142.32 19.23 27.07 8.36 55.90 3.45

L/A 61% 62%∗ 65% 62% 58% 44% 55% 63% 66% 52%

Term 15 12 20 15 12 14 15 17 17 18

# REIT Lessor 23 1 5 1 6 8 0 4 0 0

Mean Market Cap. ($M) 1187 259 3613

Value ($M) 50.51 25.36 43.60

L/A 0.76 0.65 0.58

Term 26 20 12

# REIT Lessor 2 5 18

Panel D: Statistics by lease length

Type Long Intermediate Short

N 6 19 46

Panel A reports the mean and median market capitalization for the lessee firms in the sample. Panel B desribes the distribution of sale and leaseback events according to their reported value. Panel C describes the type of commercial real estate sold and leased as reported in the Wall Street Journal, Lexis-Nexis or in SEC filings. Panel D reports sample statistics for short (less than or equal to 15 years), intermedate (greater than 15 and less than 25) and long leases (greater than or equal to 25). The value of the sale and leasebacks events are reported for 65 of the 71 events in the Wall Street Journal, Lexis-Nexis or SEC filings. Mean value to equity is the average of the ratio of transaction value to lessee firm market capitalization. Market capitalization is taken from CRSP for each firm for the month end prior to the event date. Also reported in Panel C and D are the mean transaction value, the mean Liabilities/Assets, average initial lease terms and the number of deals that report a REIT as the buyer/lessor. ∗ Liabilities/Assets reported for non-financial firms only.

The Wealth Effects of Sale and Leasebacks 633

leasing suggests that landlord and tenant investments are highly specialized to their relationship. An interesting interpretation of retail leases may be that optimal incentives for the retail tenants may be created through the use of long-term leases while incentives for the landlord are introduced through other means, namely percentage leases. Methodology Standard event study methodology is employed to discern whether the surprise announcement of a sale and leaseback is accompanied by a significant change in stock prices on average for lessee firms. We calculate abnormal and cumulative abnormal returns (CARs) over windows of varying length around the event. The abnormal return for each firm around the event date is calculated as the residual from a market model. Daily returns for 150 days (from day −250 to day −101 relative to the event day 0) are used to estimate market models. Abnormal returns are cumulated over various intervals and the mean abnormal return and CARs are reported below. Mean standardized abnormal returns and mean standardized cumulative abnormal returns are used to calculate univariate test statistics (Mikkelson and Partch 1986). We also utilize multiple regression to further sort our results and competing hypotheses. Tests for heteroskedasticity in the errors are rejected in all of the specifications reported below. Empirical Analysis Given our sample of sale and leaseback events, we want to test the predictions of our property rights model that long-term financial leases are associated with no change in shareholder wealth, while shorter leasebacks are accompanied by positive wealth events for shareholders. To do so, we must decide in a practical sense what short and long mean. The longest leaseback in the sample is for 30 years, and so we might regard 30 years as the lease length corresponding to x = 1 in the sale and leaseback model. As a practical matter, most real estate investment analyses use 10 years as a sufficiently “long” period of analysis when considering buy and hold strategies. Due to the fact that we are partially constrained by the number of observations in the tails of the lease length distribution and the fact that our reported lease lengths tend to cluster around 5-year multiples, we evaluate several possible definitions, using as a cut-off 10-, 15-, 20- and 25-year lease lengths. Univariate Results We begin by examining abnormal and cumulative abnormal returns around event day 0. For the sample as a whole, there are no average abnormal or cumulative abnormal returns that significantly differ from zero (Table 3). In

634 Fisher

Table 3
Average (cumulative) abnormal returns for lessee firms. Interval

Mean

z-stat.

p-value

(−3, 0) (−2, 0) (−1, 0) 0 (0, 1) (0, 2) (0, 3)

0.0111 0.0055 −0.0046 0.0005 0.0004 −0.0019 −0.0019

0.9631 0.8440 0.3537 0.6254 0.0835 0.1689 0.1689

(0.34) (0.40) (0.72) (0.53) (0.93) (0.87) (0.87)

Univariate statistics are reported for the average cumulative abnormal returns (CARs) to shareholders of lessee firms around announcements of sale and leasebacks of commercial real estate. Seventy-one sale and leaseback events involving 69 lessee firms were identified from the Dow Jones Interactive Service for the years 1990–2000. The abnormal returns for each event day are calculated as the prediction error of individual firm market models estimated over days (−250, −101) using return data from CRSP. Abnormal returns are cumulated over various intervals, and the mean of the sample CARs are reported below. Z-statistics and p-values for tests of whether the mean of the standardized CARs is equal to zero are also reported.

Table 4, we proceed with our univariate analysis by dividing the sample according to various definitions of Short and Long lease lengths (less than or equal to 10 or 15 years and less than 20 years). We find that lessee firms reporting Short leasebacks under all three definitions experience significant and positive abnormal returns on the day of the sale and leaseback announcement. In particular, when we classify a Short leaseback as less than or equal to 15 years, the lessee firms realize a 1.3% abnormal return. We find that the lessee firms announcing correspondingly Long (greater than 15 year) leasebacks experience negative abnormal returns on day 0. A t-test for whether the means of the abnormal returns for these two subsamples are equal is rejected at the 1% level of significance. The average abnormal returns for Long (greater than 15-year) leases are significantly negative (at the 10% level), which does not appear consistent with the prediction of our model. In the final panel of Table 4, we divide the Long subsample further and examine the abnormal returns to lessee firms reporting leasebacks between 15 and 25 years of length and 25 years or longer. The intermediate length leasebacks are associated with significantly negative average abnormal returns, while the longest leasebacks are not associated with abnormal returns. It should be noted that only six lessee firms announce leasebacks that are 25 years or longer, but that the separation of these abnormal returns from the intermediate leasebacks increases the magnitude and statistical

The Wealth Effects of Sale and Leasebacks 635

Table 4
Day 0 average abnormal returns for lessee firms according to leaseback length. Abnormal Returns Lease Term Short Long Short Long Short Intermediate Long

10 years 15years [0, 15] (15, 25) [25, 30]

Mean

z-stat.

N

0.0152 −0.0061 0.0130 −0.0226 0.0130 −0.0309 0.0037

2.06 −0.62 2.07 −1.76 2.07 −1.89 −0.23

22 49 46 25 46 19 6

Tests for Equality in Subsamples t-stat. −1.74 −2.51

Univariate statistics are reported for the abnormal returns to shareholders of lessee firms from announcements of sale and leasebacks of commercial real estate according to whether the initial leaseback is for a Short or Long term (given various definitions). Seventy-one sale and leaseback events involving 69 lessee firms were identified from the Dow Jones Interactive Service for the years 1990–2000. We also report t-statistics for tests of whether the mean abnormal returns of the two subsamples (Short vs. Long) are equal.

significance of the average abnormal returns associated with intermediate length leasebacks.14 Regression Results We undertake multivariate analysis, reported in Table 5, to discern whether shorter leasebacks are still correlated with wealth gains once we control for other potential determinants of abnormal returns. In particular, we noted previously that some announcements of sale and leasebacks were concurrent with other announcements about lessee firms. In a previous section, we also identified two hypotheses from the corporate finance literature that may explain stock market reactions to the information in sale and leaseback announcements. In particular, credit-constrained firms may benefit from a relatively lower cost of capital provided by sale and leasebacks as opposed to other forms of financing. The tax hypothesis suggests that sale and leasebacks may be valuable when the contract allows parties to trade the tax benefits associated with durable assets. 14 Parsing abnormal returns by property type is tempting, but given the small sample sizes, it is difficult to say much in terms of statistical significance about these estimates. Average abnormal returns calculated according to the type of property are positive for the property types that have average lease terms that are less than or equal to 15 years and negative for average lease terms that are greater than 15 years (see mean lease terms in Table 2). The relationship between average lease lengths and property types may be explainable in the context of our model if certain property types systematically benefit more or less from the specific investment activity of the tenant versus the landlord.

−0.0309 0.0439 0.0346

Intercept Short Long Other Debt Ann. Liabilities/Assets REIT Lessor Tax Rate Adj. R2 p-value of F statistic N

−2.55 2.82 1.40 1.48

t-stat.

0.0954 0.02 71

−0.0331 0.0433 0.0368 0.0430

Coeff.

2 −2.28 2.90 1.30 1.59 0.88

t-stat.

0.0924 0.03 71

−0.0475 0.0449 0.0346 0.0469 0.0219

Coeff.

3

−0.57

−0.0082 0.0861 0.04 71

−2.30 2.85 1.42 1.56

t-stat.

−0.0312 0.0443 0.0376 0.0464

Coeff.

4 −2.21 2.95 1.32 1.69 0.94 −0.66

t-stat.

0.0846 0.06 71

−0.0463 0.0462 0.0353 0.0512 0.0236 −0.0096

Coeff.

5

−2.38 2.59 1.28 1.06

t-stat.

0.0320 0.43 0.0967 0.09 45

−0.0461 0.0560 0.0436 0.0451

Coeff.

6

This table reports OLS regression results when using the abnormal return for event day 0 for lessee firms as the dependent variable. Independent variables include a dichotomous variable for lease lengths: short (less than or equal to 15 years), intermedate (greater than 15 and less than 25) and long (greater than or equal to 25). Intermediate is the omitted category. Another dichotomous variable, Other Debt Announcement, is equal to 1 if the announcement of the sale and leaseback was accompanied by an announcement of other debt arrangements on day 0. Liabilities/Assets are from the lessee’s year-end financial statement prior to the event. REIT Lessor is a dichotomous variable equal to 1 if the buyer/landlord is identified as a publically traded REIT. Finally, we obtain information from COMPUSTAT to compute a proxy for the lessee firm’s tax rate as operating income before depreciation divided by taxes paid in the year prior to the sale and leaseback. The adjusted R2 and p-value of the F-statistic are reported for each regression.

2.37 2.83 1.30

t-stat.

0.0796 0.02 71

Coeff.

OLS Regressions

1

Table 5
OLS regressions relating firm and leaseback characteristics to abnormal returns.

636 Fisher

The Wealth Effects of Sale and Leasebacks 637

Finally, we know that a subset of lessors are REITs, and we want to control for any source of systematic variation that may arise when seller/lessees do deals with this class of lessors. The dependent variable in our regressions is the day 0 abnormal return (AR) for lessee firms, and we estimate the following regression, AR = β1 (lease length) + β2 (confound ) + β3 (constrained ) + β4 (REIT ) + β5 (tax) + ε.

We utilize several categorical indicators for lease length. We experimented with using the actual term of the initial lease (and its log) in the regressions, and the results generally correspond with the results reported in Table 5. However, lease lengths generally cluster around multiples of 5 in our sample, and upon comparison we find that the dichotomous indicators Short, Intermediate and Long offer greater explanatory power in our analysis. The indicator Short is equal to 1 if the initial lease term is reported to be 15 years or less and equal to 0 otherwise. The indicator Intermediate is equal to 1 if the lease length is between 15 and 25 years in length and Long is equal to 1 if the lease length is at least 25 years. Throughout our regression analysis, intermediate is the omitted category. Specification 1 in Table 5 simply verifies our conclusions from Table 4 and suggests that lease length helps to explain about 8% of the variation in abnormal returns in the sample. We then examine potential noise around the event. We categorize concurrent announcements on day 0 according to whether lessee firms (1) announce new debt financing in addition to the sale and leaseback of real estate, (2) announce firm earnings and (3) announce other events concerning the lessee firm. None of these indicators alone or in combination with others prove to be significantly different from zero in regressions that include the indicators for lease length, but the announcement of new debt (as distinct from the sale and leaseback financing) generally improves the explanatory power of the model and the addition of this indicator is reported in specification 2. Removing these observations from the sample and rerunning specification 1 does not significantly change the results. To capture whether or not a firm might be considered financially constrained, we obtain the ratio of lessee firm liabilities to assets for the year end prior to the sale and leaseback event from the firms’ financial statements because data for all of the firms were not available on COMPUSTAT. The coefficient on the measure of liabilities to assets is positive, which suggests that firms with greater financial constraints may benefit more from sale and leasebacks.

638 Fisher

However, the coefficient is not significantly different from zero. Also, notice that the coefficient on Short is no longer sufficient by itself to overcome the negative intercept of the model and result in a positive prediction of abnormal returns. Further examination reveals, however, that for all but one firm in the Short subsample, when the Liabilities/Assets coefficient multiplied by the firm’s L /A ratio is added to the intercept, the coefficient on Short results in a positive predicted abnormal return for Lessee firms. In results not reported in Table 5, we enter the market capitalization of the lessee firm as an alternate proxy for credit constraints under the assumption that smaller firms may face more credit constraints in accessing financial markets. Running the regression when using market capitalization did not add any additional information, nor did it alter the signs or significance of other independent variables. In specification 4, we control for whether or not the lessor is identified as a REIT. Assuming that the identity of the landlord and tenant may add information about the importance of their respective investments, we seek to identify whether there is some systematic information to be had in the identity of the buyer/lessor. The indicator for REIT, however, is never significant in explaining abnormal returns from day 0. In all specifications where we included the REIT indicator, the coefficient on the indicator was negative. In results not reported in Table 5, we also consider other characteristics of the lessee firms, in particular whether or not the firm was a financial institution that may have additional regulatory constraints on its liabilities and assets. There are five such lessee firms in the analysis, and the indicator is never significant and did not alter the sign or significance of any of the other coefficients. We also investigate whether the type of property in conjunction with lease lengths might influence the observed abnormal returns, but with no significant results. Ultimately, we estimate the full model, absent taxes, in specification 5. For all reported L /A ratios, lessee firms reporting short initial leasebacks are predicted to experience positive abnormal returns. As a final test, we are able to identify a proxy for the lessee firm’s tax rate using information from COMPUSTAT. Following Ezzell and Vora (2001), we create a proxy for the tax rate by dividing operating income before depreciation by the taxes paid for the year prior to the sale and leaseback announcement. We expect that lessee firms with lower tax rates may be less able to utilize the deprecation tax benefits of real estate ownership. Therefore, events for seller/lessee firms with lower tax rates are expected to be associated with greater wealth gains, all else equal, if sale and leasebacks allow parties to trade tax benefits and expropriate revenue from the government. Tax information is only available for 45 of our sample firms, and in specification 6 of Table 5 we report no significant evidence of a link between taxes and wealth gains, although the sign of the coefficient is positive as predicted.

The Wealth Effects of Sale and Leasebacks 639

Table 6
Firm organization and credit constraints. 1

2

OLS Regressions

Coeff.

t-stat.

Coeff.

t-stat.

Intercept Short Short × High L /A Intermed × High L /A Long Other Debt Ann. High L /A Adj. R2 p-value of F Statistic N

−0.0401 0.0445

−2.66 2.88

0.0309 0.0458 0.0129

1.13 1.57 0.91

−0.0495 0.0577 0.0048 0.0313 0.0532 0.0407

−2.63 2.61 0.28 1.20 1.79 1.36

0.0931 0.03 71

0.0890 0.05 71

This table reports OLS regression results when using the abnormal return for event day 0 for lessee firms as the dependent variable. Independent variables include a dichotomous variable for lease lengths: Short (less than or equal to 15 years), Intermedate (greater than 15 and less than 25) and Long (greater than or equal to 25). Intermediate is the omitted category. Another dichotomous variable, Other Debt Announcement, is equal to 1 if the announcement of the sale and leaseback was accompanied by an announcement of other debt arrangements on day 0. High Liabilities/Assets is equal to 1 if the firm’s L /A ratio is greater than the sample median. High L /A is also interacted with the proxies for lease length. All of the lessee firms with long leasebacks have a L /A ratio greater than the sample median. The adjusted R2 and p-value of the F-statistic are reported for each regression.

We are unable to explain the consistently negative abnormal returns reported for leasebacks with intermediate length leases. In Table 6, we attempt to discern if credit constraints can provide any additional explanation by creating an indicator that is equal to 1 if the lessee firm’s Liability/Assets ratio is greater than the median of the full sample. In particular, in specification 2 of Table 6, we allow the coefficient of this indicator to vary according to lease length. Of note, all six of the Long leasebacks have L /A ratios that exceed the sample median; therefore, no interaction term is entered into the regression for this category. We are still unsuccessful in explaining the negative returns associated with intermediate lease lengths, and although not statistically significant, the coefficients for high L /A indicator are consistently positive. Discussion of Results In summary, announcements of sale and leasebacks where the initial lease length is less than or equal to 15 years are consistently associated with positive wealth effects for the shareholders of lessee firms. The results suggest that firms

640 Fisher

should own real estate in which they make highly specialized investments, but that more generic real estate may optimally be owned by investors who are not the end users of the asset. While the signs of other potential explanatory variables are consistent with expectations, none are statistically significant or alter our primary findings. In particular, our results are consistent with the findings of Alvayay, Rutherford and Smith (1995) that the tax benefits of sale and leasebacks may have been reduced by changes in the U.S. tax code relative to the environment which prevailed during the early to mid-1980s. The longest leasebacks in the sample exhibit behavior consistent with our prediction for true financial leases, but we are unable to explain the significantly negative returns to lessee firms who enter into leasebacks for between 15 and 25 years. The model predicts that these firms would have clearly been better off by choosing sufficiently long leases, and perhaps by choosing even shorter ones. We are unable to say definitively how the leases in these events were classified on the lessee firms’ financial statements. There may be some incentive for lessee firms to negotiate deals that can be construed as operating as opposed to capital leases. In particular, leasebacks construed as operating leases may remove debt and depreciation associated with the real estate from the firm’s balance sheet. Among other restrictions, operating leases must last for less than 75% of the remaining economic life of an asset. Therefore, if operating leases remove debt from liabilities and provide future flexibility in obtaining additional finance, then this incentive might influence the choice of leaseback length. The results of our empirical analysis suggest that firm managers need to balance the economic impact of altering the firm’s claims with the value of financial restructuring. Our study has additional limitations related to the lack of more detailed information about actual contracts, especially with respect to provisions that might alter the incentives of tenants and landlords. Our model of the sale and leaseback suggests that the choice of lease length involves a trade-off in incentives such that other contractual features may be valuable in optimizing the relationship. Finally, our sample size is relatively small, which limits the statistical power of tests when we attempt to stratify the sample. Conclusions The theory of the firm has long predicted that firms will own assets in which they make specialized investments. While prior studies investigated the financial advantages of sale and leasebacks, we investigate the phenomenon of sale and leasebacks as one way in which firms may contract to rearrange their organizational architecture. Our model predicts that firms optimally de-integrate the firm with respect to its real estate by using short leasebacks when the buyer/lessor’s

The Wealth Effects of Sale and Leasebacks 641

noncontractible future investments are relatively valuable. The model also predicts that the true financial leasebacks are analogous to continued ownership of the asset by the seller/lessee firm and that, absent any credit constraints or tax incentives, the wealth effects of announcements about the longest leaseback will be similar to the announcement of debt and close to zero. Empirically, sale and leasebacks with leases of less than or equal to 15 years are associated with 1.3% abnormal returns to lessee firm shareholders on the day of announcement and this wealth effect is larger than the gains reported in prior studies of sale and leasebacks during the late 1980s and early 1990s. Leasebacks for 25 years or longer do not affect the wealth of lessee firms in this sample, and further work is necessary to understand the apparently nonoptimal choice of intermediate lease lengths by some firms. Special thanks to Harold Mulherin, Austin Jaffe, Abdullah Yavas, Chris Muscarella, Ed Coulson, C.F. Sirmans, Bill Wheaton, David Geltner, participants of the 2002 AREUEA Annual Conference and Meeting, participants of seminars at Washington State University, Western Washington University, MIT, the University of Wisconsin and the University of Georgia and three anonymous referees for helpful comments and suggestions.

References Alvayay, J.R., R.C. Rutherford and W.S. Smith. 1995. Tax Rules and the Sale Leaseback of Corporate Real Estate. Real Estate Economics 23: 207–238. Barclay, M.J. and C.W. Smith, Jr. 1995a. The Maturity Structure of Corporate Debt. The Journal of Finance 50: 609–631. ———. 1995b. The Priority Structure of Corporate Liabilities. The Journal of Finance 50: 899–917. Benjamin, J.D., C. de la Torre and J. Musumeci. 1998. Rationales for Real Estate Leasing Versus Owning. The Journal of Real Estate Research 15: 223–237. Coase, R.H. 1937. The Nature of the Firm. Economica 4: 386–405. Copeland, T. and J.F. Weston. 1983. Financial Theory and Corporate Policy. Reading, MA: Addison-Wesley Publishing Co. Eckbo, B.E. 1986. Valuation Effects of Corporate Debt Offerings. Journal of Financial Economics 15: 119–151. Ezzell, J.R. and P.P. Vora. 2001. Leasing Versus Purchasing: Direct Evidence on a Corporation’s Motivations for Leasing and Consequences of Leasing. The Quarterly Review of Economics and Finance 41: 33–47. Fodor, S.S. and D.D. Kiely. 2003. Sale-Leasebacks Gain Popularity as a Financing Option. New York Law Journal 229: 4. Grossman, S.J. and O.D. Hart. 1986. The Costs and Benefits of Ownership: A Theory of Vertical and Lateral Integration. The Journal of Political Economy 94(4): 691–719. Hart, O. 1995. Firms, Contracts and Financial Structure. New York: Oxford University Press. Joskow, P.L. 1987. Contract Duration and Relationship-Specific Investments: Empirical Evidence from Coal Markets. The American Economic Review 77(1): 168–185.

642 Fisher

Klein, B., R.G. Crawford and A.A. Alchian. 1978. Vertical Integration, Appropriable Rents, and the Competitive Contracting Process. The Journal of Law and Economics 21: 297–327. Krishnan, V.S. and R.C. Moyer. 1994. Bankruptcy Costs and the Financial Leasing Decision. Financial Management 23: 31–42. Lewellen, W.G., M. Long and J.J. McConnell. 1976. Asset Leasing in Competitive Capital Markets. Journal of Finance 31: 787–798. Mikkelson, W.H. and M.M. Partch. 1986. Valuation Effects of Security Offerings and the Issuance Process. Journal of Financial Economics 15: 31–60. Mulherin, J.H. 1986. Complexity in Long-Term Contracts: An Analysis of Natural Gas Contractual Provisions. Journal of Law, Economics and Organization 2(1): 105–117. Myers, S.C., D.A. Dill and A.J. Bautista. 1976. Valuation of Financial Lease Contracts. Journal of Finance 31: 799–819. Rutherford, R.C. 1990. Empirical Evidence on Shareholder Value and the SaleLeaseback of Corporate Real Estate. Journal of American Real Estate and Urban Economics Association 18: 522–529. ———. The Impact of Sale-Leasebacks Transactions on Bondholder and Shareholder Wealth. Review of Financial Economics 2: 75–80. Sharpe, S. and H. Nguyen. 1995. Capital Market Imperfections and the Incentive to Lease. Journal of Financial Economics 39: 271–294. Slovin, M.B., M.E. Sushka and J.A. Polonchek. 1990. Corporate Sale-and-Leasebacks and Shareholder Wealth. The Journal of Finance 45: 289–299. Smith, C.W., Jr. 2001. Organizational Architecture and Corporate Finance. Journal of Financial Research 24(1): 1–13. Smith, C.W. and J.B. Warner. 1979. On Financial Contracting: An Analysis of Bond Covenants. Journal of Financial Economics 7: 117–161. Wheaton, W.C. 2000. Percentage Rent in Retail Leasing: The Alignment of Landlord and Tenant Interests. Real Estate Economics 28(2): 185–204. Whinston, M. 2003. On the Transaction Cost Determinants of Vertical Integration. Journal of Law, Economics and Organization 19(1): 1–23. Williamson, O. 1979. Transaction-Cost Economics: The Governance of Contractual Relations. Journal of Law and Economics 22: 233–261. Zingales, L. 2000. In Search of New Foundations. The Journal of Finance 55(4): 1623– 1653.

Appendix Claim: When x = 1, the seller/lessee’s noncooperative choice of investment level is equal to the socially optimal level. When x < 1, the seller/lessee’s choice of noncontractible investment level, i(x), is generally less than the socially optimal level, i ∗ . Derivation: The socially optimal levels of seller/lessee and buyer/lessor investment solve max R(i) − ci (i) + S(e) − ce (i). i,e

(a)

The Wealth Effects of Sale and Leasebacks 643

The first order condition for (a) with respect to i is ∂R ∂ci = . ∂i ∂i

(1 )

Denote by i ∗ the solution to (1 ). Now notice that when evaluated at x = 1, (1) reduces to (1 ) and i(x) = i ∗ . Assuming that ∂s/∂i ≥ 0, an inspection of (1) and (1 ) reveals that when x < 1, i(x) ≤ i ∗ . Claim: For any choice of x, the buyer/lessor’s choice of noncontractible investment level, e(x), is generally less than the socially optimal level, e∗ . Derivation: From the social function (a) above, we find that the first order condition with respect to e is, ∂S ∂ce = . ∂e ∂e

(2 )

Denote by e∗ the solution to (2 ). Then (2 ) is equivalent to (2) if and only if ∂S/∂e = ∂s/∂e. Therefore, when ∂S/∂e > ∂s/∂e, e∗ > e(x) for any choice of x.