You are on page 1of 41

Financial Contracts and Bargaining

Matan Tsur November, 2012

Abstract An entrepreneur considers whether to invest in a project, the prot of which depends on the outcome of bargaining with prospective buyers. I study how the nancial contract, in the form of a security, between the outside investor and the entrepreneur inuences the entrepreneurs strategic bargaining position with the buyers. In the case of a single prospective buyer, I consider the securities that the entrepreneur designs to maximize her payo. I characterize the entrepreneurs maximal payo and construct a security that takes a simple form and implements it uniquely. In the case of many prospective buyers, I consider a debt security and characterize the equilibria payos. Debt signicantly strengthens the entrepreneurs bargaining position when there are few buyers, but the bargaining advantage diminishes with the number of buyers. The more buyers there are, the lower the share of the surplus extracted by the entrepreneur. When bargaining oers are frequent, the advantage may vanish, in which case debt nancing is equivalent to selfnancing. Finally, I identify conditions under which outside nancing can increase the entrepreneurs incentive to invest and overcome the hold-up problem, leading to socially ecient investment.

I thank Alessandro Lizzeri and Tomasz Sadzik for their guidance and comments. Helpful comments given by Guillaume Frechette, Douglas Gale, Ariel Rubinstein, Emanuel Vespa, Tom Cunningham, Michael Richter, Alistair Wilson, Daniel Martin, Isabel Trevino, Yunmi Kong, Nicholas Kozeniauskas, Severine Toussaert are gratefully acknowledged. New York University, Department of Economics, 19 W. 4th Street, New York, NY.

Introduction

An entrepreneurs return from an investment often depends on the outcome of bargaining with suppliers, buyers or potential partners. In this paper, I study the interaction between nancing and bargaining. I consider an entrepreneur that seeks nancing from an outside investor, then produces a good and bargains with prospective buyers. My aim is to understand how the nancial contract with the outside investor inuences bargaining. In particular, I focus on the implications for contract design and for the entrepreneurs incentive to invest. The nancial contract consists of a security which determines how the revenue is split between the outside investor and the entrepreneur. Since the revenue is determined by the prices negotiated with the prospective buyers, the security also inuences the bargaining outcomes. There are no frictions between the entrepreneur and the outside investor: There is complete information, a single investor and there are no agency costs. The paper makes several contributions. First, I consider the securities that the entrepreneur designs to maximize her payo when there is a single prospective buyer. I identify the upper bound on the entrepreneurs payo and construct a security that uniquely implements this payo. Outside nancing enables the entrepreneur to extract a larger share of the surplus as compared to self-nancing. Second, I show that when the entrepreneur sells goods in dierent markets, a bargaining externality between buyers dilutes the eect of outside nancing. I characterize the equilibrium payos under a debt security. Debt allows the entrepreneur to extract a larger share of the surplus compared to self nancing, but the advantage is attenuated. The more buyers there are, the lower the share extracted. Third, when the entrepreneur also chooses the investment level, I identify conditions under which outside nancing can overcome the hold-up problem, thus leading to socially ecient investment. In the taxonomy of Harris and Raviv (1991), the present model falls under the category of models in which the main force determining the nancial contract is the eect on the input/output markets. The pioneering papers are Brander and Lewis (1986) and Fershtman and Judd (1987), and within the large literature that followed, several papers have stressed the link between nancing and bargaining (e.g., Bronars and Deere (1991); Perotti and Spier (1993); Sarig (1998)). In contrast, in this paper, bargaining is non-cooperative, and the bargaining outcome is determined through strategic interaction. The contribution of my paper is in the analysis of the strategic channel through which nancing inuences bargaining. To explain my results in further detail, consider, rst, a case with a single potential buyer. The outside investor invests in return for a portion of the future revenue, which is determined by the security. The entrepreneur designs a security to maximize her payo under the constraint that the outside investor agrees to nance the project with the security. I rst characterize the upper bound on the entrepreneurs payo. I then show that the entrepreneur can guarantee the maximal payo by designing a security which implements a bargaining game with a unique equilibrium. The degree to which the entrepreneur could be made liable plays an important role in determining this security. Under full liability, a debt security implements the entrepreneurs maximal payo uniquely. But when liability is limited, the bargaining game with debt has additional outcomes with lower prices. The reason is that under limited liability, the entrepreneur 2

is indierent between low prices that do not cover the debt, and as a result would have no incentive to bargain over a price increase. Thus, under limited liability, the bargaining game with debt has many outcomes in which the investor incurs losses, and in some cases she will not agree to nance the project through debt. When liability is limited I show that a simple security, consisting of an options scheme, can guarantee the entrepreneurs maximal payo. Since a security determines the entrepreneurs share for each price negotiated, a security will increase the entrepreneurs payo by increasing her share. The option scheme is constructed to incrementally increase the entrepreneurs share at dierent prices. It can be shown that when the incremental increases are suciently large, the option scheme can eliminate the low price equilibria, thereby uniquely implementing the entrepreneurs maximal payo. To illustrate the basic idea, note that in an equilibrium of the bargaining game, the buyer oers a price that makes the entrepreneur indierent between accepting and rejecting. A security that locally increases the entrepreneurs share at the continuation price, increases the price that the buyer must oer to make the entrepreneur indierent. With debt, for instance, the entrepreneurs payo is constant when the price is low. Providing the entrepreneur with some equity increases the entrepreneurs payo at every price, which eliminates some of the low price equilibria. But globally increasing the entrepreneurs payo, as does equity, may not be sucient. The security must provide incentives that are increasing in the price bargained. The security constructed creates these local incentive by providing the entrepreneur with options to increase her equity share that she only chooses to exercise when prices are suciently high. When the entrepreneur produces multiple goods that are sold in dierent markets, the entrepreneur may have to bargain with more than one buyer. In this case an additional factor inuences bargaining, which is the focus of the second part of the paper. There is no competition between the buyers and there are no direct externalities: Each buyers value does not depend on whether other buyers acquire their goods. The entrepreneur separately and simultaneously bargains with each buyer as follows: The entrepreneur makes oers simultaneously and privately to each buyer. The buyers then simultaneously accept or reject. Any buyer who accepts gets his good and pays the agreed amount. The game then continues with the remaining subset of buyers. These buyers simultaneously and privately make oers to the seller who chooses which, if any, to accept. The game continues indenitely. When the project is self nanced, an agreement with one buyer does not aect the other and the game has a unique equilibrium that is identical to Rubinsteins outcome (Rubinstein (1982)). However, a nancial obligation to the outside investor that is tied to the total revenue, links between the bargaining games. The reason is that an agreement with one buyer reduces the outstanding nancial obligation in subsequent periods, thereby changing the entrepreneurs bargaining posture. To focus the analysis on the bargaining externality, I x the security to be debt (with unlimited liability): The outside investor has a senior claim to the revenue, and the entrepreneur gets the residual after the debt is paid out. To illustrate the bargaining externality, consider a production company that makes a movie and sells the distribution rights. The previous case corresponds to a situation with a single global distributor, whereas now there are two separate distributors: a U.S. distributor and a 3

European distributor. In a bilateral bargaining game with debt (with unlimited liability), the entrepreneurs bargaining posture increases with the debt level, and so does the price payed by the buyer. Therefore, when the U.S. distributor expects the European distributor to pay a high price, the debt level in the subsequent period is low, and waiting is more favorable. But waiting is more costly if the European distributor pays a lower price. Hence, the highest price that each buyer is willing to pay decreases with the price in the other transaction. It is not clear whether this externality works in favor of the producer or the distributors. Put dierently, holding the total value and cost constant, is the producers payo higher when she bargains with a single global distributor as opposed to two separate distributors? How does the number of prospective buyers aect the advantage of debt-nancing compared to self-nancing? I characterize the equilibrium payos under a standard restriction of the o-equilibrium beliefs, with an arbitrary number of buyers. Even though there are multiple equilibria, the entrepreneurs revenue is unique, and the model makes sharp predictions. In equilibrium, the bargaining externality between buyers counterbalances the entrepreneurs bargaining advantage gained through outside nancing. The entrepreneur still extracts a larger share of the surplus when the project is nanced with debt as compared to self-nancing, but the advantage is attenuated. The more buyers there are, the smaller is the share extracted by the entrepreneur. Holding the total value and cost constant, the production company always prefers to bargain with fewer distributors. When bargaining oers are frequent, in some cases, the bargaining externality osets the eect of debt, and the entrepreneur is indierent between debt-nancing and self-nancing. Whether the advantage of debt is preserved in the limit, depends on the number of buyers and the ratio of the total value to total cost of investment. The results have implications that explain why a rm may prefer to concentrate transactions with few buyers, and why bundling projects is protable. The nal results show how the nancial contract inuences the entrepreneurs incentive to invest when she also chooses the investment level. In this case, a hold-up problem leads to underinvestment (compared to socially ecient investment) when the entrepreneur self-nances the project. But nancing through an outside investor may strengthen the entrepreneurs strategic bargaining position with prospective buyers, increasing the marginal benet from investment and the incentive to invest. Whether outside nancing, alone, can overcome the hold-up problem and restore the incentive to invest depends on the number of buyers and on the ratio of total value to total cost of investment. When the nancial obligation is debt and there are few buyers, the entrepreneur nds it optimal to choose the socially ecient investment level. But when there are many buyers and the ratio of total value to total cost of investment is high, investment is not socially ecient. Moreover, when bargaining oers are frequent, the investment level is identical to the one chosen when the entrepreneur self-nances the project, and the fundamental hold-up problem is unchanged. The rest of the paper is organized as follows: Section 2 summarizes the relevant literature. Section 3 presents an example that illustrates the main components in the analysis. Sections 4 and 5 study the single-buyer and multiple-buyers cases. Concluding comments are oered in section 6 and the Appendix contains proofs and technical derivations.

Literature

My model and results are related to several strands of literature. The point that a rm may choose a certain nancial structure or sign certain contracts to gain a competitive advantage when interacting with a third party is a central pillar of models pioneered by Fershtman and Judd (1987); Brander and Lewis (1986); Fershtman, Judd, and Kalai (1991). Several papers have emphasized the link between nancing and bargaining; for example, in Bronars and Deere (1991); Perotti and Spier (1993), the nancial obligation inuences the size of the pie that is bargained over. In numerous other papers, bargaining outcomes inuence rms decisions (Shleifer and Vishny (1989); Rajan (1992, 2012); Stole and Zwiebel (1996); Sarig (1998)Shleifer and Vishny (1989); Rajan (1992, 2012); Stole and Zwiebel (1996); Sarig (1998)). In those papers, bargaining power is determined either exogenously or endogenously by outside options. My work diers in that bargaining is non-cooperative, and the eect of nancing is endogenously determined by the strategic eect on bargaining. The strategic channel leads to new insights regarding the design of securities and how specic features of the project- e.g., number of buyers and the ratio of total value to total cost- inuence nancing decisions. Two recent papers on delegated bargaining by Bester and Sakovics (2001); Cai and Cont (2004) analyze a seller that delegates the bargaining assignment to an agent. Sakovics and Bester focus on the question of renegotiation, while Cai and Cont focus on aspects of moral hazard and adverse selection. The present paper focuses on a nancing relationship, thus, the questions, setup and analysis are quite dierent. In contrast to many papers on security design, such as Innes (1990); Demarzo and Due (1999); Repullo and Suarez (2004), in this paper, there are no frictions between the entrepreneur and outside investor. The only force governing the security chosen is the eect on bargaining. In DeMarzo, Kremer, and Skrzypacz (2005), securities aect the revenue from an auction. Another related literature is on multilateral bargaining. Rubinstein and Wolinsky (1985); Gale (1986) focus on a competitive environment- for example, when the production company bargains with two U.S. distributors over a single distribution right. In my paper, there is no competition; the European and U.S. distributors bargain over dierent rights. In Baron and Ferejohn (1989), a single pie is to be split between many players. The present work diers in that there are no externalities: When one buyer gets the good, it does not aect the others value. Horn and Wolinsky (1988b) model a single rm that bargains with union(s). Their bargaining game is similar to the one I consider, but the setting is quite dierent. In the present work, the nancial obligation of the seller is the link between the bargaining games. In several bargaining papers, an obligation of one player to an outside party enables some commitment that inuences bargaining. Examples are Haller and Holden (1997); Perry and Samuelson (1994), which model a bilateral bargaining game in which one player represents a constituency. According to Williamson (2005), the question of vertical integration is the paradigmatic problem in the economics of governance. A common explanation is that rms vertically integrate to overcome transaction costs that arise when contracts are incomplete and commitment problems hinder investment (see Bresnahan and Levin (2012), for a recent survey). In a cross-country comparative study, Acemoglu, Johnson, and Mitton (2009) empirically test how the interaction 5

between contractual costs and nancial innovation determines vertical integration. The work presented here identies conditions under which outside nancing can overcome this commitment problem and restore incentives to invest. It provides predictions that tie together how outside nancing and industry-specic features, such as the number of buyers and the ratio of total value to total cost of investment, may aect the benet of vertical integration. It also identies conditions in which downstream rms may benet from horizontal disintegration- e.g., when the distributor disintegrated into smaller units. Zingales (2000)emphasizes the somewhat ignored marriage between Corporate Finance and The Theory of the Firm, noting that nancing decisions are deeply intertwined with the rms structure and organization. The type of nancing that a rm obtains may directly impact dierent aspects of the inner structure and organization of the rm, while, at the same time, the structure of the rm aects the type of nancing that a rm obtains. This stands in contrast to the traditional Corporate Finance approach, which takes the rms organization and structure as given and analyzes how the rm chooses to nance projects. Rajan (2012) emphasizes the importance of this inter-linkage and analyzes a compelling link regarding how outside nancing enables dierent transformations of the inner structure and organization of the rm. My work brings to light a dierent link: Outside nancing impacts the structure of rms by inuencing the benet of vertical integration and horizontal factorization. Lack of commitment that distorts incentive to invest (a hold-up problem) has been the main component in the engine underlying numerous papers. The topics range from organizational design, property rights and ownership structure (Klein, Crawford, and Alchian (1978); Williamson (1979, 1998); Grossman and Hart (1986); Hart and Moore (1990); Aghion and Bolton (1992); Aghion and Tirole (1997); Rajan and Zingales (2001)) to dierent aspects of nancing decisions (Rajan (1992); Shleifer and Vishny (1989); Bronars and Deere (1991)). Several papers have analyzed how contractual schemes (Noldeke and Schmidt (1995); Che and Hausch (1999); Edlin and Reichelstein (1996); Aghion, Dewatripont, and Rey (1994)), long-term relationships (Baker, Gibbons, and Murphy (2002)) or dynamic investment decisions (Che and Sakovics (2004)) may mitigate or solve the hold-up problem. To my knowledge, none have studied the specic channel analyzed in this paper.

Example

The following example uses a simple two-period bargaining model to illustrate how precommitment to an outside investor inuences bargaining and the bargaining externality between buyers. Consider a case in which the required investment is I and there is a single buyer whose value is v . Bargaining lasts for two periods as follows: The buyer makes the rst oer; the entrepreneur makes an oer in the next period; if that oer is declined ,the good loses its value. The discount factor is . If the project is self-nanced, the entrepreneur incurs the investment cost I and bargains with the buyer. Since the game ends after two periods, the price can easily be determined by backward induction. In the second period, the entrepreneur makes a take it or leave it oer, 6

and, therefore, the buyer pays price v if the game reaches the nal stage. The entrepreneurs continuation value in the rst stage is v , which is the lowest price that she is willing to accept. Therefore the buyer pays v . The entrepreneurs prot is v I , and the buyer gets v v. If v < I , the entrepreneur does not undertake the project. To illustrate how precommitment to an outside investor may aect the bargaining outcome, consider two nancial contracts: Equity nancing : The outside investor invests and gets a share of the revenue. Precommitting to split a xed fraction with the outside investor does not inuence the price. The production companys continuation value is (1 )v ; the distributor oers a price v ; I , and the production

the production company gets (1 ) v and is indierent between accepting and rejecting. The outside investor agrees only if her share covers the cost v company can get, at most, (1 self-nancing. ) v v I . Thus, an equity share is not preferred to

Standard debt with limited liability : The outside investor invests I and has a senior claim to the revenue, implying that if the price covers the debt x x I . The continuation value is I (v (v accepts is x I ). The price is x = v + (1

I , the entrepreneur gets

I ), and the lowest price that the entrepreneur )I , and the entrepreneurs prot )(v I ). The point is that

increases to (v

I ), while the buyers prot decreases to (1

a delay in negotiations enables the entrepreneur to roll over the debt, which decreases her cost of delay and increases her bargaining position. To understand the eect of multiple buyers, consider the case in which the entrepreneur sells two goods to two buyers. Each buyer values only her good, and the value for each is
v 2.

Bargaining lasts for three periods, and the entrepreneur initially oers a price to each buyer. The entrepreneur nances the project with debt and the outside investor has senior claims to the revenue. If buyer 1 expects buyer 2 to pay x, then the debt in the subsequent period is I 0 = I buyer 1s continuation value (1 )( v 2 I 0 ) (1 )( v 2 x, and I + x) increases with the expected

price in the other transaction. The entrepreneurs nancial obligation creates a link between the transactions, in that the price each buyer is willing to pay decreases with the expected price that the other pays. This externality is the additional force that inuences prices. Considering an innite-horizon bargaining game, I rst focus on the single-buyer case and solve for the optimal nancial contract and investment level. The second part focuses on the multiple-buyers case. I x the nancial contract to be debt with liability and show that the bargaining model has a generically unique equilibrium with interesting properties.

4
4.1

Single Buyer
The Model

There are three players: an entrepreneur E , an outside investor, and a buyer B . The entrepreneur can produce a good for a buyer, and the cost of production is I . The good is worth v to the buyer. 7

There are two stages: In the rst stage, the entrepreneur oers the outside investor a nancial contract. If the outside investor declines, the game ends. If the outside investor accepts, investment is made, the good is produced and the game moves to the second stage. In the second stage, the entrepreneur and buyer bargain over a price. Financial Contract: The outside investor invests in the project in return for rights to the future revenue. The revenue is the price that the buyer pays and the price is veriable and contractible. For a revenue of x, a security s(x) is a function that species the portion of the revenue that the outside investor gets. The buyers value is not contractible. A nancial contract between the entrepreneur and the outside investor consists only of a security. I consider securities that are continuous and weakly increasing. If there is limited liability, then for each revenue level x: s ( x) x. The entrepreneur oers the outside investor a security s. If the outside investor decides to invest, she can costlessly monitor the entrepreneur: Any amount invested in the project cannot be diverted by the entrepreneur. In the second stage, the entrepreneur bargains with the buyer over the pie v . I denote the bargaining game over a pie v with security s as B (v, s). Bargaining game: All actions are observable, and the buyer knows the value and the security. Bargaining is non-cooperative, the bargaining game follows Rubinstein (1982) with a random initial proposer and specic payos. Agents discount time, with a discount factor of between bargaining rounds. Payos: Agents are risk-neutral. If the nancial contract is s, and the good is sold at date t for a price of x, the buyer gets uB (x, t) = the entrepreneur receives uE (x, t) = and the outside investor obtains
t t t

< 1

(v

x);

(x

s(x));

s ( x)

I . If an agreement is not reached, the buyer and

entrepreneur get nothing and the outside investor loses the initial investment. Strategies: The entrepreneurs strategy consists of a security oered and a bargaining strategy. The buyers strategy is a bargaining strategy for each security oered s. The outside investors strategy species whether to accept or reject any oer. All actions are observable. The solution concept is SP E . Discussion of the assumptions: 1. The security is an increasing function of the revenue. The restriction is standard; if the security were to decrease the entrepreneur would have incentives to articially inate the revenue and decrease the payment to the outside investor (Innes (1990)). One possible way in which the entrepreneur could do this here, is through an articial transaction with a dierent buyer in which no good is exchanged: the entrepreneur transfers money to that buyer; the buyer purchases some service from the entrepreneur; the revenue goes straight to the project. The scheme is protable when the security does not increase because 8

the entrepreneur neither gains nor loses money on the articial transaction (she gets the money from back), while the payment made to the outside investor decreases. 2. Bargaining is non-cooperative because the goal is to endogenously derive the strategic eect of precommitment to the outside investor on the entrepreneurs bargaining position. 3. Bargaining is dynamic while securities are restricted to be static. Debt contracts usually specify interest rates, maturity dates and term payments, which are not accounted for in the present model, but may inuence bargaining outcomes. (a) A maturity date leads to dierent bargaining dynamics, and the outcome is pinned down by backward induction. If bargaining rounds are frequent, a maturity date may be of a dierent order of magnitude as compared to the frequency of bargaining rounds- as a geological time scale is to a calendar time scale- in which case, an innite-horizon bargaining model may be a better approximation. (b) Term payments and interest rates may shift the cost of delay and inuence the bargaining position. In equilibrium, agreement is immediate and the outside investor may not demand such features. (c) The outcome of a bargaining model that accounts for these features is sensitive to the underlying assumptions, and small changes in the assumptions can lead to quite dierent outcomes. 4. The buyer knows the nancial contract. The nancial obligations of rms are usually not observable to buyers. But since the entrepreneur benets from disclosing the information, she would do so. 5. After deriving the main results, I further discuss what happens if the outside investor can design the security or can purchase the project and hire the entrepreneur; I also discuss the implications of allowing the entrepreneur and buyer to sign a contract.

4.2

SPE

The entrepreneur designs a security to maximize her payo, while making sure that the outside investor agrees. A security inuences both the price negotiated with the buyer and how that price is split between the entrepreneur and the outside investor. The entrepreneur takes both into account when designing the security. To better understand the inuence of a security on bargaining, consider, rst, a case in which the project is self-nanced. The entrepreneurs payo is just the price, and the bargaining game has a unique SP E , as shown in Rubinstein (1982). The SP E strategies are stationary: The entrepreneur always oers the price xE = The buyer always oers the price xB =
v 1+ v 1+

and accepts an oer x0 only if x0 x0 (v

xE . xB ) .

and accepts an oer x0 only if v

The prices xE and xB are the unique pair that makes each player indierent between accepting the others oer and waiting for her own oer.

Moving to a case in which the project is nanced through an outside investor, consider two dierent securities: an equity share s1 (x) = ax and a debt claim with limited liability s0 (x) = min[x, I ]. Each security implements a bargaining game with dierent SP E outcomes: If the security is equity, the entrepreneurs payo is uE (x, t) = game with equity has a unique SPE; the expected price is expected payo is (1 a) v 2.
t

(1

a)x. It is straight-

forward to check that the above strategies are also an SP E in this case. The bargaining
v 2,

and the entrepreneurs

If the security is debt with limited liability, the entrepreneur gets nothing if the price x is less then the debt I , but receives the entire residual x

I if the price covers the debt.


1 1+

When liability is limited, the bargaining game under a debt security has multiple SP E . The stationary strategies in which each party oers the prices xE = I + xB = I + nothing.2 The utility frontiers under both securities are depicted below:
Payoff B Payoff B

(v

I ) and

1+

(v

I ) are an SP E . In this case, the entrepreneurs expected payo is

1 2 (v

I ). But there also exists an SP E in which the buyer gets the entire value and pays

Utility frontier: equity

Utility frontier: debt with limited liability

I
v-I

v-x

(1-a)x

ax

(v-I)/2 v-y
(1-a)y

ay

Payoff E

(v-I)/2

Payoff E

As oers become frequent ( ! 1), the stationary SP E price oers converge to a single price.

Comparing the two securities, the price under equity is always v 2 , and the entrepreneur clearly prefers equity to debt when the low price SPE is played. But, if the high price SPE is played, the entrepreneur prefers debt to equity. The reason is that the high price under a debt security is greater than the SPE price under equity. The entrepreneur may prefer equity if her share is suciently high, but the lowest share that the outside investor is willing to accept in return for her investment is a v 2 = I . This case is depicted below:
is so because each party is indierent between accepting the others oer and waiting one period. the buyer oers a price of zero, and accepts only that price, undercutting the buyers demand by oering a slightly higher price is not a protable deviation. The reason is that the entrepreneurs payo is at for prices that do not cover the debt.
2 If 1 This

10

Payoff B

Utility frontier of debt and equity

v/2 ((v-I)/2

a(v/2)

((v-I)/2

Payoff E

The rst lemma establishes an upper bound on the entrepreneurs SP E payo: The entrepreneur can do no better than get half the social surplus. Thus, as long as the high price SP E is played, debt beats any other security. Lemma. In an SP E of the entire game, the entrepreneurs payo is at most 1 2 (v A formal proof is given in the Appendix. To illustrate the idea, not that a security determines the utility frontier. The restriction that the security increases with the price, implies that the slope of the utility frontier can not be too at (it must be steeper than -1), as depicted in the right panel:
Payoff B Payoff B

I ).

Utility frontier general security


s(x)

Utility frontier general security

v-x

x-s(x)

s(x) s(x) s(x) s(x)

Payoff E

Payoff E

The intuition underlying the upper bound is that if investment is made, then the bargaining game has at least one SP E in which the outside investors payo covers the debt- i.e., the security and the SP E price x are such that s(x) I .3 Otherwise, the outside investor would never agree to invest. Thus, there must be an SP E in which the price is within the shaded domain in the gure below. The proof shows that as long as the security increases, the highest SP E payo that the entrepreneur can get in this domain is: liability.
3 To

1 2 (v

I ). The upper bound relies

only on the restriction that the security increase and, therefore, also holds for the case of full
simplify the intuition, assume, ! 1, and the price oers converge.

11

Payoff B

Utility frontier

s(x)

Payoff E

It follows that when the project is nanced with debt and the high price SP E is played, the entrepreneur achieves the maximal payo. But under limited liability, the bargaining game with debt has many outcomes in which the investor incurs losses, and in those cases the outside investor will not agree to nance the project through debt. The next proposition shows that the entrepreneur can overcome this problem by designing a security, consisting of a simple option scheme, that uniquely implements her maximal payo. Thus, in every SP E the entrepreneur achieves her maximal payo. Proposition. In every SP E of the entire game, the entrepreneurs payo is 1 2 (v investment is made. A formal proof is given in the Appendix. Given some value d < v , the proof constructs a security that implements a bargaining game that has a unique SP E in which the entrepreneur gets 1 2 (v d d) and the outside investor gets I . The security takes the following simple form (see gure below): I ) and the

s ( x)

min[1 x, 2 x + d2 , ..., n x + dn , d] n < n


1

< ... < 1 < 1 and 0 < d2 < d3 ... < dn < d

s
s(x)

xB

xE

price

The security is constructed to incrementally and locally increase the entrepreneurs share at dierent prices. The key point is that in order to eliminate low price equilibria, the incremental increases need to be suciently large. The intuition is that in an equilibrium of the bargaining game, the buyer oers a price that makes the entrepreneur indierent between accepting and rejecting. A security that locally increases the entrepreneurs share at the continuation price, 12

increases the price that the buyer must oer to make the entrepreneur indierent. This, in turn, further increases the entrepreneurs continuation value. The proof show that through this process, when the incremental increases are suciently large, the option scheme can eliminate the low price equilibria. The security creates these local incentives by providing the entrepreneur with options to increase her equity share that she only exercises when the prices are suciently high. Below is a graphical illustration of the construction. Start with a debt claim s0 (x) = min[x, I ]. There exists an SP E in which the entrepreneur and buyer oer x E I + payo is
1 2 (v 1 1+

(v

I ). The problem is that there also exists an SP E in which the entrepreneur

I ) and x B I +

1+

(v

I ), and the entrepreneurs expected

gets nothing. The idea is to modify s0 in a way that maintains the good SP E but gets rid of the bad SP E . Step 1: Modify s0 to eliminate the degenerate outcome. Consider the security s1 (x) := min(x, I ) where < 1, depicted below. Security s1 allocates an equity share < 1 to the outside investor and provides the entrepreneur with an option to buy back all the equity for a price of I . Since the entrepreneur has some equity, her payo strictly increases with any price which eliminates a SP E in which the buyer
pays nothing. Additionally, if is suciently large, then s1 (x E ) = s1 (xB ) = I , and the good

SP E from the previous step also exists in the bargaining game with this security. However, the bargaining game under this security may have additional SP E . Specically, the SP E in the bargaining game without debt, or Rubinsteins SP E , may still exist in this game if the
R pair of price oers from Rubinsteins SP E , xR E and xB , both lie on the rst segment. This

happens when xR E

v 1+

The reason that Rubinsteins SP E also exists in this case is that when s1 (xR E ) = xE and s 1 ( xR B ) = xB , each player is indierent between accepting the others oer and waiting one period for her own oer: v (1 xE ) xB = = (v (1 xB ) ) xE

is less than the break point denoted by x1

I ,

as depicted below.

s
s0(x) I I s1(x) s1(x)

xB

xE

price

xR B

xR E

x1 xB

xE

price

In case the investment falls short of Rubinsteins price, I < we have that the price xR E does not fall on can be eliminated. But in the case that I

v 1+ , for a suciently high < 1 the rst segment, xR E > x1 , and Rubinsteins SP E v R R 1+ , both prices xE and xB always fall on the rst

segment (because for any : xR E x1 ), and Rubinsteins SP E always exists in the bargaining game with this security. The gure below depicts both cases: 13

s
I s1(x)

s
I s1(x)

R xR B I xE

xB

xE

price

R R x B xE I

xB

xE

price

Step 2: Modify s1 to eliminate Rubinsteins SP E and maintain the good outcome. Consider security s2 (x) = min[1 x, 2 x + d2 , I ] depicted below (1 > 2 and d2 > 0). Let
R x1 be the rst break point, as long as the security breaks before the price xR E ( xE < x 1 )

Rubinsteins SP E is eliminated.

s
I*

x1

xR

price

The crux of the construction is that if the dierence in slopes is suciently large- specically, if rst segment and the sellers oer xE lies on the second segment; alternatively, xB x1 xE . The pair of prices is a stationary SP E when each party is indierent between accepting the other partys oer to waiting for her own oer. As the dierence between the slopes of the rst and second segments increases, so does the compensation that the entrepreneur demands if he is oered a price on the rst segment, x < x 1 , and expects a price tomorrow on the second segment, x0 > x 1 . And if the dierence is suciently large, no two prices can compensate both players demands and make each indierent. In order to implement the good outcome, the slope of the second segment must be suciently high so that s2 (x B ) = I , which places a lower bound on 2 , as depicted below:
1 1 1 2

- then there does not exist a stationary SPE in which the buyers oer xB lies on the

14

s
I*

xR

x* j

price x* E

Recapping the construction steps: To eliminate the degenerate SP E in which the buyer gets everything, we set 1 < 1. To eliminate Rubinsteins SP E , the rst break point is set before the Rubinstein price: x 1 < xR E. To eliminate a stationary SP E that has price oers on both segments, we set a suciently large change of slope:
1 1 1 2

<

To ensure that s implements the good outcome, 2 cannot be too small. A sucient condition for the bargaining game B (v, s) to have a unique SPE is that i) x s ( x)

strictly increases, and ii) there exists a unique price pair that satises the indierence condition: v xB xE = = (v ( xE xB ) s(xE )) (1) (2)

s ( xB )

(see Osborne and Rubinstein (1990)). If there is a pair of prices on the second segment that satises the indierence conditions (1) and (2), we repeat the same process: Break again just before we reach the SP E price. The security is constructed to eliminate all pairs of prices that satisfy the above condition, aside
from x E , xB , and, therefore, implements a game with a unique SP E .

Loosely speaking, in order to eliminate SP E in which low prices are oered, we increase the curvature of s by adding segments. However, s must implement the highest possible payo, which requires s to be suciently steep. Thus, uniqueness imposes some curvature on the security s, while optimality restricts the degree of curvature. The proof constructs a security that balances both. Unique implementation implies that the incremental increases need to be suciently large. Thus, the second order eect, approximated by the second derivative of the security, is the key component that guarantees a unique outcome.

15

4.3

SPE Securities

An SP E does not pin down a unique security; for example, a debt security can be part of an SP E as long as the good SP E is played in the bargaining game. The only restriction that
4 SP E imposes is that the security attens out and s(x B ) = s ( xE ) = I .

In contrast to debt, the security constructed above has the robust feature that it implements the entrepreneurs maximal payo uniquely. In addition, this security can be implemented with standard options in the following way: The security initially allocates 1 shares to the outside investor and provides the en1

trepreneur with a sequence of options: Option i enables the entrepreneur to purchase i shares from the outside investor for a price of di . The entrepreneur can exercise a single option.

The number of options decreases with the ratio frequency of oers: If If


v I v I

v I

and increases with the discount factor or

> 1 + , a single option is sucient . 1 + , the number of options n satises,


I v v 1+

<1

n+1

(formulas are derived in ! 1 the breakpoints

the proof). The rst break point is xed x 1 =

, and as

converge to each other and s smooths out for all x > x 1 .

There are other securities that also achieve the entrepreneurs maximal payo uniquely, but the one constructed here takes a simple form. In the case of unlimited liability, the same SP E payos arise. The upper bound is the same: The entrepreneur can do no better than get
1 2 (v

I ) even if she could be made liable. The

entrepreneur can guarantee this payo with debt- s(x) = I . The dierence is that when liability is unlimited, a debt security implements the entrepreneurs maximal payo uniquely, whereas under limited liability, the bargaining game has multiple SP E .

4.4

Continuous investment level

The results extend in a straightforward manner to the case in which the entrepreneur chooses a continuous investment level I , and the buyers value v (I ) increases with investment. To accommodate for this case, assume that the entrepreneur oers a pair (I, s) that consists of an investment level and a security to the outside investor. The outside investor either agrees or declines. For any investment level, the entrepreneur oers a security to maximize her payo. The previous result implies that the entrepreneur can do no better or no worse than 1 2 (v (I ) I ). Thus, she chooses an investment level to maximize half of the social surplus, and the investment is ecient.
4 The

upper bound is not reached if s (st) increases; see remark in the proof.

16

4.5

Points for discussion

I discuss below what happens if some of the assumptions are relaxed: 1. The model does not allow for any communication between the entrepreneur and buyer before investment is made. The main result establishes that, on its own, a nancial contract between the entrepreneur and outside investor leads to socially ecient investment. Therefore, relaxing this assumptions and allowing the entrepreneur and buyer to contract does not improve social eciency. However, relaxing this assumption may lead to dierent conclusions regarding the role of nancial contracts. For instance, a hold-up problem does not arise if the entrepreneur and buyer can contract on the value. But when the the value is contractible, Che and Hausch (1999) show that if i) the entrepreneur makes the initial investment, ii) inecient outcomes are always renegotiated, and iii) investment is cooperative (as is the case here) and parties are risk-neutral, then a contract between the entrepreneur and buyer cannot lead investment levels higher than those that arise without a contract. Therefore, even if we allow the entrepreneur and outside investor to contract, a nancial contract with an outside investor is necessary to achieve eciency when parties cannot commit not to renegotiate inecient outcomes. 2. In any SP E , the expected price is
v +I 2 .

The entrepreneur could oer a security that

increases the price, but she does not benet from this because it requires giving up a larger portion to the outside investor. It follows from the above construction that there exists a security that implements a bargaining game with a unique SP E in which the buyer pays her entire value v and all the revenue goes to the outside investor. Therefore, if the outside investor were to design the security, she could extract the entire surplus, leaving the entrepreneur and buyer with nothing. Likewise, if the outside investor could purchase the project and hire the entrepreneur, then the entrepreneur could extract the entire surplus.

Multiple buyers, multiple goods

The previous section nds that when there is a single buyer, the entrepreneur can do no better than nance the project using debt with unlimited liability. This section considers a project in which the entrepreneur bargains with many buyers. The dierence is that, now, an agreement with one buyer may change the nancial obligation in subsequent periods, which adds an additional factor that inuences the bargaining game. To simplify the analysis, I consider debt-nancing (with unlimited liability). The project. The entrepreneur/seller bargains with multiple buyers. If the seller decides to undertake the project, she produces n > 1 goods for n buyers. There is no competition, each buyer demands a dierent good: Buyer i0 s value of good i is vi and is zero for all other goods. values are xed and do not depend on the investment level. The seller decides whether or not to undertake the project; the project is potentially protable: nv > I . 17 Values are symmetric: vi = vj v . The cost of the project is I . The number of goods and the

Bargaining procedure. Bargaining follows a simultaneous procedure.5 The seller makes oers simultaneously and privately to all buyers. The buyers then privately and simultaneously accept or reject. Any buyer who accepts gets his good and pays the agreed amount. The game then continues with the remaining subset of buyers. These buyers simultaneously and privately make oers to the seller who chooses which, if any, to accept. The game then continues indenitely. The discount factor between bargaining rounds is .

5.1

Debt Financing

The entrepreneur nances the project through an outside investor and the nancial contract is debt. The outside investor has a senior claim to the revenue: Any revenue goes rst to cover the debt. For example, if the debt is $100 and a single buyer purchases her good and pays $70, then the seller gets nothing that period, and the debt in the subsequent period is $30. The price each buyers pays is determined via bargaining. The bargaining game with debt: The bargaining game is a dynamic game with two state variables: the number of buyers nt and the debt level dt . If an agreement is reached with a buyer, the buyer pays, receives the good and leaves the game. If an agreement is reached at Pk period t with k buyers over prices x1 , ..., xk , then the revenue in that period is X i=1 xi . If the revenue does not cover the debt- i.e., X dt - the seller gets nothing in that period and the debt in the subsequent period is dt+1 = dt payed out, dt+1 = 0 and the seller gets the residual X period, the seller bargains with the remaining nt+1 nt X . If the revenue covers the debt, the debt is dt in that period. In the subsequent k buyers and a debt level of dt+1 .

Transactions are publicly observable, and the number of buyers and the debt level in each period are common knowledge. Let B (n, d, v ) be the simultaneous bargaining game with n buyers, an initial debt level of d and buyers values v . The solution concept is sequential equilibrium. To highlight the main issues, consider a case with two buyers and
0

buyer 2 to pay x < d, the debt in the subsequent period is d = d x, and her continuation value 1 d0 ) 1 d + x) increases with x (see previous section). Thus, each buyer takes into 2 (v 2 (v account the outcome of the other transaction. To see how this externality inuences the sellers revenue compared to the case in which there is a single buyer, consider a case in which the total value of the project is V = 200 and the debt is d = 100. When there is a single buyer, the seller gets 1 2 (V d) = 50 and the price is 1 2 (V + d) = 150. The seller cannot extract a revenue of 150 d0 ) = 37.5, and the highest price x=
1 d+ 2 (v 2 66 3 ; since v

1. If buyer 1 expects

when she bargains with two buyers, each of whom has a value of 100. If buyer 1 expects the other to pay 75, for example, her continuation value is 1 2 (v she would be willing to pay is: 68 3 4 (this price solves v she is willing to pay is 62.5. At the same time, if buyer 2 expects buyer 1 to pay 62.5, the most 62.5)). The highest x= 1 2 (v d + x) , price the seller can oer and have both buyers accept is x = which is less than what she gets when there is a single buyer. The next proposition places a standard restriction on the o-equilibrium and shows that even
5 The

this price makes each buyer indierent between accepting and rejecting. The seller gets 33 1 3,

procedure can be thought of as n Rubinstein bargaining games played simultaneously and independently.

18

though there are multiple pure strategy sequential equilibria, the sellers payo is unique. In this particular example, it is 33 1 3. Before stating the proposition, some more assumptions are needed. As shown in the previous section, under limited liability, there exist degenerate equilibria in which the seller is indierent between accepting and rejecting low prices. To rule out such outcomes, I assume unlimited liability as follows. In any sub-game with m buyers and a debt level d0 , the seller is solvent as long as the debt level is less then the total surplus- i.e. d0 < mv . Since nv > d, the game always starts with a solvent seller. If, in some period, the seller is insolvent, then the seller pays the debt and the game continues without debt. I also place a restriction on the o-equilibrium beliefs. Following a deviation to an o equilibrium oer by the seller, since oers are private, each buyer has conjectures about oers being made to other players. Thus, o equilibrium beliefs can have a large eect on the continuation game and this may create multiplicity. A buyer has passive beliefs if, when she receives an o-equilibrium price, she believes that other buyers received their equilibrium price. The need for such renements is common when agreements among pairs of players aect payos for other players (for instance in vertical contracting when there are several downstream rms, e.g., McAfee and Schwartz (1994); Horn and Wolinsky (1988a)). The main result shows that all pure-strategy sequential equilibria with passive beliefs are generically unique in the sense that they all give rise to the same aggregate price or total revenue. Proposition. In all sequential equilibria with pure strategies and passive beliefs of the game B (n, d, v ), agreement is immediate and the aggregate price is the same. The sellers prot is 8 < 1 (nv
2

vs (n, d, v ) =

(1 + )d) d)

if (n if (n

v 1) 1+

The proof is relegated to the Appendix. I sketch the main steps below. Step 1: The seller oers the same price to each buyer and each buyer accepts. The price is 8 <
v 1+ : d+v 1+ n

1+ 2(1+ n) (nv

v 1) 1+ < d.

xs (n, d) :=

if (n if (n

v 1) 1+ v 1) 1+

d < d.

Given passive beliefs, it directly follows that all buyers accept the sellers oer, and each buyer is indierent between accepting and rejecting, given that the other buyers accept. The main point is that this condition pins down a unique price function. To see this, consider a simple case with two buyers and condition then one of the following occurs: 1. Both prices cover the debt; x1 , x2 1. If prices x1 , x2 satisfy each buyers indierence

d. Each buyers continuation value is


v 2.

v 2,

and the

indierence condition implies that x1 = x2 =

This happens when

v 2

d.

19

2. Both prices fall short of the debt; x1 , x2 < d. The buyers indierence conditions: v v imply that x1 = x2 =
v +d 3 .

x1 = x2 =

1 (v 2 1 (v 2

d + x2 ) d + x1 )
v +d 3

This happens when

< d ()

v 2

< d.
v 2

3. Only one covers the debt; x1 < d x2 . This case cannot happen because x1 = x2 =
1 4v

and

1 2 d, v 2

which implies that

v 2

<d

v 2.

Therefore, either d x1 = x2 = v + 3 . condition: If (n


v 1) 1+

d, which implies that x1 = x2 =

v 2,

or

v 2

< d, which implies that

A similar argument is given for the general case. The function xs (n, d) satises the indierence

d, then xs =

v 1+

. The continuation game following n 1)xs = (n


v 1) 1+

1 accepted prices
v 1+

of xs has no debt (because (n

d ), and the continuation price is is the

determined via bilateral bargaining without debt. Notice that the price xs = indierent between accepting and rejecting. If (n
v 1) 1+ < d, then xs = d+v 1+ n .

sellers oer in a bilateral bargaining game without debt, and, therefore, each buyer is

In this case, the revenue from n

1 buyers does not

cover the debt, because (n 1) v 1+ < d () (n 1) d+v < d. 1+ n (n 1)xs

A buyer that rejects bargains with a seller that has a debt obligation of d0 = d and the price xs makes that buyer indierent as it solves v xs = (v d + (n 1)xs ).

1+

The RHS is the buyers continuation value in case n previous section).

1 buyers accept (follows from

Therefore, if prices x1 , ..., xn satisfy the above indierence condition, then xi = xs (n, d). Step 2: Each buyer makes an oer that is accepted, and the seller is indierent between accepting all oers and rejecting all oers. The proof that there is no delay is by induction. The crucial point is that the seller must be indierent between accepting all oers and rejecting all oers. Since the continuation price when the seller rejects all oers is pinned down by P the previous step, the indierence condition pins down the aggregate price uniquely: xi = xs (n, d) + (1 ) d. To see why this indierence condition holds, consider again a two-player case with 1.

20

If the seller strictly prefers accepting both to rejecting both, then the seller must be any buyer would oer a lower price).

indierent between accepting both and accepting one and rejecting the other (otherwise,

Both buyers oer the same price x. Otherwise, the buyer oering the higher price could reduce her price. From the sellers indierence condition, it follows that If x d, then 2x d=x d= 1 2 (v
v d+ 1 2v ! x = 2.

If x < d, then 2x

d + x) ! x =

v +d 3 .

v +d But if the seller rejects both prices, the price in the continuation game is max[ v 2 , 3 ].

Therefore, the seller must be indierent between accepting both and rejecting both.

The idea in the general case is similar. The seller must be indierent between accepting all oers and accepting a subset of the oers (otherwise, a buyer would decrease the price). Let k be the lowest number of oers for which the seller is indierent between accepting all oers and accepting k oers. If k > 0, we have the following: From the previous step, the continuation payos are strictly decreasing with the debt. k oers (otherwise, we have a contradiction to the equilibrium condition). There must be k + 1 maximal prices that are identical. Otherwise, the buyer oering the maximal price can reduce the oer, and it would still be a best response for the seller to accept. Denote the k + 1 highest prices by x. The seller must weakly prefer accepting k to k + 1 prices (otherwise, the equilibrium condition is violated), which places an upper bound on the maximal price: If x is too high, the seller would strictly prefer accepting all k + 1 price. The upper bound is used to show that the seller must weakly prefer rejecting all oers and getting the price xs (n, d) from all buyers in the next period to accepting k oers. It follows from steps 1 and 2 that in every pure strategy sequential equilibrium with passive beliefs, there is no delay; the seller oers the price xs (n, d) to each buyer and the aggregate price oered by buyers is X xi = nxs (n, d) + (1 ) d) .

The seller. therefore. is indierent between accepting all oers and accepting the highest

Therefore, the sellers prot is:

vs (n, d, v ) = which can be expressed as:6


2

X 1 (nxs (n, d) + xi ) 2 (1 + )d) d) if (n if (n

d) .

vs (n, d, v ) =
6 The

8 < 1 (nv :

v 1) 1+ v 1) 1+

d < d.

1+ 2(1+ n) (nv

sellers value is vEs (n, d) =

1 (nxs (n, d) 2

+ nxs (n, d) + (1

)d )

21

There are multiple pure-strategy sequential equilibria with passive beliefs, I construct a symmetric equilibrium in the Appendix. The important feature is that the sellers prot is unique and does not depend on which equilibrium is played. I next discuss some properties of this equilibrium, focusing rst on the eect of debt on bargaining, and how it changes with the number of buyers. Then, I focus on the entrepreneurs investment decision.

5.2

Equilibrium Properties

To understand the eect of debt on bargaining, I compare the sellers prot when the project is nanced through debt to the sellers prot when the project is self-nanced. If the seller nances the project on her own, costs are sunk and do not inuence bargaining. The unique outcome of the the simultaneous bargaining game is that each buyer pays a price identical to the one in a bilateral bargaining game without debt. The reason is that the bargaining games are independent of each other. The sellers expected prot is n v 2 I , which is less than the expected prot under debt. Similar to the case in which the seller bargains with a single buyer, precommitment to the outside investor strengthens the sellers bargaining position, but the extent of this eect depends on the number of buyers. To understand the inuence of the number of buyers, I hold the total value V nv and total cost I xed, but vary the number of buyers (think of slicing an orange into smaller pieces). The sellers expected prot under self-nancing
V 2

I depends only on the total value

and does not change. But the sellers prot under debt weakly decreases with the number of buyers. The reason is that the equilibrium prices are pinned down by the buyers indierence condition: Each buyer is indierent between accepting and rejecting, conditional on all other buyers accepting. As the number of buyers increases from n to n + 1, the total value is split into smaller parts and each buyers value decreases by a factor of
n n+1 .

But to satisfy the above


n n+1

indierence condition, the seller must decrease prices by a factor greater than

and, as a

result, the sellers revenue decreases with the number of buyers. The bargaining externality between buyers counterbalances the bargaining advantage gained through outside nancing. The seller prefers debt to self-nancing but the advantage of debt weakly decreases with the number of buyers. When oers become frequent- i.e., ! 1- the sellers payo is if I) if
(n 1) n (n 1) n V I V I

When either the number of buyers is low or the ratio of the total value to total cost of investment is low,
n 1V n I

8 <1V I V vs (n, I, ) = 2 : 1 (V n 1+n


1 2V

2 < 2.

< 2, and the sellers payo under debt nancing is

1 1+n (V n 1V n I

I ), which is greater 2 and the sellers

than her prot under self nancing payo is


1 2V

I . In this case, the seller prefers debt nancing to self-

nancing. But when there are many buyers and the ratio is high,

I under both debt nancing and self nancing. In this case, the bargaining

externality between buyers completely osets the bargaining advantage gained through outside 22

nancing and the seller is indierent between self-nancing and debt nancing. The payos are depicted bellow: The total value (represented by the circle) and the total cost are held xed, while the number of buyers varies. The entrepreneurs prot in each case is the shaded area. The left (high cost) panel corresponds to the case that the ratio any number of buyers vs (n, I ) =
1 1+n (V V I

I ); i.e. the seller always gets a portion of the


V I

2. In this case, for

social surplus that is proportional to the number of buyers. The right (low cost) panel corresponds to the case that the ratio as long as the number of buyers is small, such that vs (n, I ) =
1 1+n (V 1 2V n 1V n I

> 2. In this case,

< 2, the sellers prot is 2 the sellers payo

I ) ,which is greater than

1 2V

I , and the seller prefers debt to self


n 1V n I

nancing. But once the number of buyers is suciently large, is vs (n, I ) =

I , and the seller is indierent between self-nancing and nancing with

debt. The transition happens when the entrepreneurs portion of the social surplus is less than the shaded area under self-nancing.

Self Finance

Debt: 1 buyer

5.3

Investment level

The sellers only decision is whether or not to invest, as the number of buyers and the values are xed. The seller prots when the project is nanced with debt and, therefore, investment is always made. To better understand the eect of debt nancing on the incentive to invest, I rst consider a case in which the sellers investment decision determines the value v , and then a case in which the seller can also determine the number of buyers n. An example of the rst case is of a production company that sells distribution rights to a movie. The investment determines the quality of the movie, but the number of distributors is exogenous. An example of the second case is of a vertical relationship between a single input supplier and many producing rms. In this case. the input supplier is able to choose both the number of rms to supply to and the quality level. Denote the cost of producing n units of quality v by c(v, n). In the rst case, the number of buyers is xed and the seller invests in quality. Since the seller always nances the project with 23

debt, the sellers problem is to choose v to maximize max


v

vs (n, d, v ) d = c(n, v ) v 0. c(v ); let V := nv and C := c(n, v ) be

s.t.

The social optimum investment v maximizes7 nv

the social value and cost associated with the optimal investment level. When there are many buyers and the ratio of the total value to total cost of the optimal investment level is high, quality level v that is less than the social ecient level. But, otherwise, the seller chooses the ecient investment level v .
n 1 V n C

> 1 + , investment is not ecient, and the seller chooses a


n 1 V n C

1+

and

The reason is that as long as the sellers payo is proportional to the social surplus, the seller chooses an investment level to maximize the social surplus. But when payo from the optimal investment is not proportional to the social surplus, and investment is not ecient. In this case, as bargaining oers become frequent, the seller chooses the value to maximize V c(v, n). Therefore, the investment level is identical to when the project is self2 nanced and nancing with debt cannot improve on what can be achieved under self-nancing. When the seller can also choose quantity, the sellers problem is to choose n, v : max
n,v n 1 V n C

> 1 + , her

vs (n, d, v ) d = c(n, v ) n N.

s.t. v 0,

This case adds an additional source of distortion. Similar to a monopolist that accounts for the eect of quantity on prices, when the seller produces an additional unit, the price of all the other units changes. In this case, the investment level depends on the cost function. In general, incentives to invest are distorted when marginal costs increase (assume that n is continuous). But if the marginal cost in quantity is constant, then the investment level depends on the fundamentals, and in some cases, investment may be ecient. For example, if c(n, v ) = n + c(v ), where c(.) is convex, assume that the social optimum is n = N . Then, if
N 1 V N C

1 + , investment is ecient; otherwise it is not.

Conclusion

This paper focuses on how the form of the security used to nance a project aects the entrepreneurs bargaining posture with prospective buyers. The analysis of this dimension provides a novel perspective on security design and on the advantage of outside nancing. In the case of a single prospective buyer and limited liability, neither debt nor equity can guarantee the
7 Assume @ c(v,n) @v @ 2 c(v,n) @2v

that a solution exists:

> 0 and

> 0.

24

entrepreneurs maximal return, but a simple option scheme can. The option scheme provides the entrepreneur with increasing local incentives, designed to eliminate all low price equilibria, thereby uniquely implementing the entrepreneurs maximal payo. When the project delivers multiple goods that are sold in separate markets, the entrepreneur multilaterally bargains with dierent buyers. I limit the analysis to the case of debt-nancing (with unlimited liability). Debt still allows the entrepreneur to extract a larger portion of the surplus compared to selfnancing, and the advantage of debt is preserved in the multilateral setting, but is attenuated. The more buyers there are, the smaller is the share extracted by the entrepreneur. When bargaining oers are frequent and there are many buyers, the advantage may vanish and debt is equivalent to self-nancing. The results have implications that explain why a rm may prefer to concentrate transactions with few buyers, and why bundling projects together is protable. Additionally, when the entrepreneur chooses the investment level, outside nancing inuences the incentive to invest, and is some cases, may overcome the hold-up problem. In the basic hold-up problem the return from investment is determined via bargaining and costs are sunk. As a result, the entrepreneurs incentive to invest is reduced. Both parties could benet from an agreement in which the buyer pays a higher price and the entrepreneur increases the investment level, but such an agreement cannot be enforced and is not feasible. When the number of buyers is low, the entrepreneurs payo under debt-nancing is proportional to the social surplus, therefore she chooses the socially ecient investment level. Thus, through the eect on bargaining, outside nancing implicitly enables an agreement that cannot explicitly be enforced. But when the ratio of total value to total cost of investment (at the socially ecient investment level) is high and there are many buyers, the entrepreneurs payo is no longer proportional to the social surplus and the implicit agreement disintegrates. In this case, the hold-up problem cannot be alleviated by outside nancing. The results also have implications for the inuence that outside nancing may have on the inner structure and organization of rms. Vertical integration is one way in which rms can overcome commitment problems that reduce the value of investment. Thus, in some cases, the ability to obtain outside nancing reduces the incentive for vertical integration. In a cross country comparative study, Acemoglu, Johnson, and Mitton (2009) empirically test how the interaction between contractual costs and nancial innovation determines vertical integration. One of their goals is to empirically distinguish between two opposite eects that nancial innovation has on vertical integration. On the one hand, imperfect credit markets limits the number of small rms which leads to larger rms that tend to be vertically integrated (Rajan and Zingales (1998)). But on the other hand, lack of nancing may also prevent mergers between rms which leads to less vertical integration (McMillan and Woodru (1999)). The eect of nancing on bargaining analyzed here depends on industry-specic features, such as the number of buyers and the ratio of total value to total cost of investment. These features may distinguish this channel from others. On the other side, outside nancing may also increase the incentive for buyers to horizontally disintegrate. For example, a single global distributor can reduce the production companys bargaining advantage and extract a larger share of the surplus by disintegrating into a European distributor and a US distributor. As long as the production companys incentive to invest is not 25

distorted, distributors can extract a larger share of the surplus by localizing distribution and disintegrating into smaller units. The ratio of the total value to total cost may limit the extent of factorization, because when the ratio is high, the production companys incentive to invest may be distorted when there are too many distributors. But if the ratio is low, disintegration does not aect the production companys incentive to invest and, theoretically, distributors extract the entire surplus away from the production company by disintegrating into suciently small units. The are several issues, both theoretical and empirical, that are left for future work. In the single buyer-case, allowing for stochastic values may lead to a tighter characterization of the optimal security. Also, incorporating the eect of various dynamic features of debt securities, such as interest rate, term payments and maturity dates, may lead to interesting empirical questions.

7
7.1

Appendix
Optimal security single buyer
I ).

Lemma. In any SP E of the entire game, the entrepreneurs payo is at most 1 2 (v Proof. Let uE (x) := x

s(x) be the entrepreneurs utility from an agreement at time t = 0. It

is without the loss of generality to assume that uE (x) is weakly increasing, because any price oered where it is not, is strictly dominated by a lower price. Let m be the inmum of the buyers payo in a sub-game starting with his oer. The buyer will always oer a price xb v xb v
1 2

m, and will reject any price xE : v

xE < m. Therefore m and [ uE ( v m) + uE (v m)] =

the highest prices that the entrepreneur and buyer can oer in a SP E are xE v m. Since uE (x) is weakly increasing in price, the most the entrepreneur can get is UE [2v (1 + )m (s(v m) + s(v m))] Since E can get no more then uE (v
1 2

m) when he makes an oer, in any SP E the buyer m) and therefore: s(v m)] () m) s( v m)]

always oers a price x such that uE (x) uE (v v m s(v (1 since s is increasing, s(v (1 Which implies that (1 + )m Substituting this into 2 implies: UE 1 [v 2 s(v 26 m) )v [v (1
2

)m + s(v

m) s(v )v (1 v s(v

m) and we have that:


2

)m + s(v

m)

s( v

m)]

m).

m)]

There must be at least one SP E in which the outside investor recoups the investment. Since s is increasing it follows, 1 [s(v 2 m) + s(v m)] I

because otherwise, the outside investor loses money in every SP E . This implies that that s(v m) I and together with the previous inequality UE 1 (v 2 I)

Remark. If s is strictly increasing the the inequality is strict. The only investment level that achieves the upper bound is the optimal investment. Proposition. In every SP E of the entire game the entrepreneurs payo is 1 2 (v The previous lemma establishes Lower bound Lemma. For any d < v there exists 0 < n < ... < 1 < 1 and 0 < d2 < .. < dn < d such that if s(x) := min[1 x, 2 x + d2 , ...., n x + dn , d] the game B (v, s) has a unique SP E in which the entrepreneur gets 1 2 (v d) and the outside investor gets d.
1 2 (v

I)

I ) as an upper bound on upper bound on the en-

trepreneurs SP E payo. The proof shows that it is also a lower bound.

Proof. The proof will be by construction. Dene (i , di ) inductively: Algorithm: 0. set 1 < 1 and d1 = 0 1. given n , dn , x n :=
v (1+ )

dn (1+ ) (1 n ) 1 n
2

2. given n > 0, n+1 := max[1 3. given n+1 , x n :

, 0] and if n = 0 then n+1 = 0.

dn+1 := (n n+1 ) xn + dn =

(1

)(1
2

n )

x n +dn =

(1

)(1
2

n )

v+

dn

(1

)(1
2

n )

Since we set d1 = 0 the values n , dn , x n are determined by 1 denote them by n (1 ), dn (1 ), xn (1 ). For each 1 denote the security s(x, 1 ) := min [i (1 )x + di (1 ), d]. Claim 1: for a suciently high 1 there exists a SP E in B (v, s(., 1 ) in which the entrepreneur gets 1 2 (v d) and the outside investor gets d. A pair of prices (xE , xb ) is a stationary pair if v xb xE s ( xb ) = = 27 (v ( xE xb ) s(xE ))

As long as x

s(x) is strictly increasing, a pair of stationary strategies in which each player

oers the same price (xE , xb ) and rejects any oer less then the continuation value is a SP E when (xE , xb ) are a stationary pair The pair of prices, x E := d +
1 1+

(v

d) and x b = I +

1+

(v

d) , are a stationary pair if

s ( x j ) = s(xE ) = d. Stationary strategies that oer these price give rise to the desired payos.

Therefore, I will show that for a suciently high 1 there exists x < x j such that s(x, 1 ) = d. In words: the security s(x, 1 ) reaches the point d early enough, that is before x j . (i) As long as n > 0 the kink points are independent of 1 and for n x n = v v (1 + )
n

2:

+ 1)

To see this, take 1 < 1, using a recursion on the construction of dn , n : dn = (1 n )v (1


2

) +

dn 1 n

applying the recursion and because we set d1 = 0 we get: dn = (1 n )v (1


2

1+

+ ... +

n 2

= (1

n 1

)v (1+ )(1

n 1

by construction: v dn + (1 + ) (1 + ) 1 n

x n = and therefore

x n = v

v (1 + )

+ 1)

(ii) Therefore, since x n > x j < v , for suciently small there exists 1 and n such that x j (this implies that n > 0). This follows from the above formula: x n
1

the construction, 8n < 1 : lim1 !1 n (1 ) = 1. Let n := min[n : x n > x b ], notice that is that n is independent of 1 , as long as 1 > 0 .

n!1

lim x n = v and from

x n , let 0 be the lower bound such that if 1 > 0 : n > 0. The crucial point b < x

From the construction, 8n < 1 : lim1 !1 n (1 ) = 1 , lim1 !1 dn (1 ) = 0, and therefore


00 00 since x b > d there exists an such that if 1 > : n (1 )xb + dn (1 ) > d. Choose 1 > max[0 , 00 ] and we have that s(x b , 1 ) = min[n xb + dn , d] = d. Since s(x b , 1 ) = s(xE , 1 ) = d, the prices (xb , xE ) are a stationary pair, which concludes the

rst claim. Claim 2 : for 1 > max[0 , 00 ] the game B (v, s(x, 1 )) has a unique SP E Fix 1 > max[0 , 00 ] and a security s := s(x, 1 ). 28

It follows immediately from Rubinsteins proof (Osborne and Rubinstein (1990)) that B (v, s) has a unique SP E if i) x pair. Since (i) holds by construction, I show that B (v, s) has a unique stationary pair. A xE , xb are a stationary pair if: xE xb s ( xb ) = = (1 ( xE ) v + xb s(xE )) s(x) is strictly increasing, and ii) there exists a unique stationary

Together these imply: (1


2

) xb

s(xb ) + s ((1

)v + xb ) = (1

)v

Dene the function g to be the LHS: g (x) := (1 By construction g (x b) = g (x) = (1 (1 Both x and (1 (1
2

)x

s(x) + s ((1

) v + x)

)v , we want to show that this is the unique solution to )v < g (v ).


1+ 0

)v . The boundary conditions are: g (0) < (1


0

) + x lie on the rst segment as long as x < v v


0

(because then )v are on


2

)+ x<x 1 ). Denote by x the point at which s goes at: n x + dn = d.


1+

1. for all x s.t. x

: g is increases: The reason is that x and x + (1


0

the same segment, therefore s (x) = s ( x + (1 2. for all x such that This follows because x n + (1 x n+1
1+

)v ) = 1 < 1 and g = (1
0

1 )(1
1

) > 0.

and n is such that n > 0 . Which implies that x and x + (1 x + (1 )v x n + (1 )v > x n+1 . The inequality is true because: x n = ( v (1 + ) )(1 )+ dn+1 (1 + ) (1 n+1 ) (

< x x0 g decreases. For x x0 we have that x n

)v lie on dierent segments.

xx n

)v and as long as n > 1 we have that

dn ) = (1 (1 n )

)v

(1

If n = 1 we have that x + (1 Therefore s0 (x) = n and s0 ((1


2

)v > x 1 . )v + x) n+1 . Which means that g 0 (x) 1 )v ) = 0


2

n +

n+1 0.

3. For all x such that x0 < x g increases, because s0 (x) = s0 ( x + (1 The solution is unique because g ( 1+ v )v . Proof of proposition: Lemma 1 establishes an upper bound:
1 2 (v

) < (1

)v , and for all x > x b : g (x) > (1

I ). Lemma 2 establishes a lower bound:

1 2 (v

I ).

The reason is that the entrepreneur can make an oer that guarantees a continuation payo 29

that is arbitrarily close to 1 2 (v gets 1 2 (v I

I ). The oer consists of the security s, that is contracted to I (lemma 2). Since the outside investor

implement a continuation sub-game B (v, s) that has a unique SP E in which the entrepreneur ) and the outside investor gets I + always accepts the oer when > 0 , the lower bound follows. If = 0, the outside investor breaks even and the entrepreneur can do no better that 1 I ) (lemma 1). Therefore, such 2 (v an oer is a SP E of the entire game that achieves the bound.

7.2

Many Buyers

The rst lemma establishes that if there is a single buyer left, and the debt is 0 d < v then there is a unique SP E 8 Lemma. The game B (1, d, v ) with v > d has a unique SP E , the proposing party gets and the receiving party gets
1+ 1 1+

( v d)

(v

d)

1 Proof. The pair of prices x E = d + 1+ (v d) and xj = d + 1+ (v d) are the unique stationary

pair, and since preferences satisfy required assumptions, there is a unique SP E (Osborne and Rubinstein). The party that makes the initial proposal gets
1+ 1 1+

(v

d) and the other party

(v

d) . Proof of main proposition

7.2.1

Proposition. In all sequential equilibria with pure strategies and passive beliefs of the game B (n, d, v ), agreement is immediate and the aggregate price is the same. The proof is established by 4 Lemmas. Let passive beliefs: Lemma. The seller makes n oers that are all accepted. Proof. By induction on n. If n = 1 given by rst lemma. Induction step: assume the seller makes oers and m < n buyers accept. No agreement with any buyer is not a SP E outcome, because the seller has a deviation. Let t be the rst period in which any oer is accepted. By the induction hypothesis, the game ends in at most t + 2 periods. The remaining n m make positive prots. The seller either makes an oer at period t or period t + 1, by induction all remaining buyers accept the sellers oer, and no-one accepts a price x = v , (the buyer can always reject the oer, be the only buyer left, and makes positive prots, assuming full liability). The seller can oer a slightly higher price to all buyers that reject. Since beliefs are passive, those buyers expect to get a continuation price that is the same and therefore accept
8 With

be a pure strategy sequential equilibrium with

a single player the SP E and sequential equilibria are the same

30

Lemma. The seller makes n symmetric oers given by: xs (n, d) =

8 <

Proof. From previous lemma, the seller makes n oers (x1 , ..., xn ) that are all accepted. Let v (x i ) be the value for a that rejects, given that x 8 < :
v 1+ 1+ i

v 1+ : d+v 1+ n

if (n if (n

v 1) 1+ v 1) 1+

d <d

other oers are accepted: if P


j 6=i

v (x i ) =

(v

d+

j 6=i

xj )

if

xj

j 6=i

xj < d

First it must be that for all i: v

xi = v i ( x i ) . xi vi (x i ). If the inequality were strict

Each buyer weakly prefers accepting to rejecting v buyer should accept. Second, the oers are symmetric: 8i, j : xi = xj .

for some buyer, a seller could increase the price for that buyer, and under passive beliefs that

To see this let xm be the lowest price and xM is the highest price, then it must be that: P P j 6=M xj j 6=m xj . P If j 6=M xj d then vm (x m ) = vM (x M ) which implies that v xM = v xm and therefore xM = xm . P If j 6=m xj < d then v xm =
1+

(v

d+

subtracting the second form the rst implies that: xM

that xm = xM . P P v If xm = 1+ and v xM = 1+ (v j 6=M xj < d j 6=m xj then we have that v P v d + j 6=M xj ). The rst implies that xm = 1+ . We can rewrite the second as: v xM = P P P 1 d j 6=M xj ) () v = 1+ (v d+ j 6=m xj )+ 1+ xm + 1+ xM and since j 6=m xj 1+ (v d+ we have that v that xM
v 1+ 1+ 1 v + 1+ xm + 1+ xM or that v

j 6=m

xj ) and v xm =

xM =
1+

1+

(v

d+

( xM

xm ) which implies

j 6=M

xj )

xm + xM but since xm =

v 1+

we have

which implies that xM xm .

Third, the seller oers a unique price xs Let x be the price oered: either n a dierent price. If (n v
0

1 transactions cover the debt or not, each case pins down


v 1+

1)x

d the continuation value is vi (x) =


v 1+

(bilateral bargaining without debt), and

x = v i ( x) ! x = (n

. (v d + (n 1)x): v x = vi (x) () x =
d+v 1+ n . d+v 1+ n . v 1) 1+

If (n d =d

1)x < d, then in case a buyer rejects he bargains with a seller with a debt level of 1)x, and vi (x) =
v 1+ 1+ v 1) 1+

Uniqueness follows from: (n the seller oers x =

and if (n

v 1) 1+

d () (n

d+v 1) 1+ n

d. Therefore, if (n

< d he oers x =

Step 2: buyers propose The next lemma will be useful, in analyzing dierent deviation and the sub-games which they may lead to. 31

Transition lemma: Consider a case in which we start out at some game B (n, d)9 , and the the seller accepts k oers of price x then the following holds true: If x If x
1 1+

then if we start at a low debt game we stay in a low debt game, i.e. if B (n, d) is k, d kx) is also low debt. kx) is also high debt. then if we start at a high debt game we stay in a high debt game, i.e. if B (n, d) is k, d

low debt debt game then B (n


1 1+

a high debt game, then B (n

Proof. The lemma follows immediately from the denition: B (n k, d and vice versa in the high debt case kx) is low debt () (n k
v 1) 1+

kx () (n

v 1) 1+

v d + k ( 1+

x)

The transition lemma also implies that if accepting k oers of price x transitions the game from low debt to high debt then x < debt, it must be that x > same type of game. Lemma. All sub-games that begin with buyers proposals end immediately Proof. By induction on the number of buyers: if n = 1 there is a unique SP E . Induction step: assume that there are n buyers. And on the SP E path the seller accepts k < n k P oers, let Xk = xi be the total revenue. The continuation game has n k buyers, a debt of d0 = d
i=1 v 1+ v 1+

. And if it transitions the game from high debt to low


v 1+

. Also, if x =

then accepting any k oers leaves us in the

Xk and the seller proposes. It must be that the seller is solvent (otherwise he rejects k )v > d0 .

all): (n

From previous step, we know that the game ends in the next round. First, if the continuation game B (n price is
v 1+

k, d0 ) is low debt game, then we know that continuation


v k ) 1+

, and the sellers continuation value is: (n

d0 .
v 1+

Consider the deviation by a buyer whose oer is rejected, in which he oers the price x = If the seller accepts, the continuation game B (n d
0 v 1+

+ .

1, d0

x) has k

1 buyers, with debt

. By transition lemma, this game is also low debt, thus the seller still gets the same
v v+

price x =

from the remaining k ((n k

1 buyers.Thus, if d0 > x, the sellers continuation value 1) v 1+ d0 + v 1+ (1 + ) ), then the buyer is also better

from accepting is:

which is strictly better. If is suciently small, < o: v d0 x. Second, if the continuation game B (n price is
9 To

v 1+

x > (v

v 1+

). A similar argument can be made in the cases in which d0 = 0, or if k, d0 ) is a high debt game, the buyers continuation
(n k )v d 0 1+(n k) .

d0 +v 1+ (n k)

and the sellers continuation value is

relax notation, since v is xed, I denote a bargaining game only be the state variables.

32

Consider a deviation in which a buyer whose oer is rejected, oers a slightly higher price then his continuation price: x = continuation game with n
d0 +v 1+ (n k)

+ . x is high debt game for some (if

The seller is strictly better o for any > 0. To see this, there are two cases to consider: the k 1 buyers and debt of d0 its a high debt game for some , its a high debt game for all 0 < ), or for every > 0 the continuation game is a low debt Case 1: If B (n
0

1, d 0

x) is a high debt game for some > 0, it is a high debt game for k 1, d 0 xk ) =
(n k 1)v d0 +x 1+(n k 1)

all < then the sellers continuation value is: vE (n notice that
(n k 1)v d0 +x 1+(n k 1)

and

>

(n k )v d 0 1+(n k)

any . Also, for a suciently low epsilon the buyer is also strictly better o. Case 2: If B (n v k 1) 1+ (n of
d0 +v 1+ (n k)

() x >

d0 +v 1+ (n k)

and therefore the seller is strictly better o for

k
0

1, d0

d +v 1+ (n k)

moves us from a high debt game to a low debt game. By the transition lemma, it
d0 +v 1+ (n k)

x) is a low debt game for all > 0 then we have that 8 > 0 : 0 +v v + ! 1+ (n k 1) 1+ d(n k) , and therefore accepting a price () k )) ((1 + b)d0 (1 + b)d0 (n (n k k 1)v ) 1)v > >0 0 ()

must be that

>

v 1+

(1 + )(1 + (n

Also, the sellers continuation value in case he accepts is: (n greater then what he gets if he rejects if: (n Now, since x =
d0 +v 1+ (n k)

v 1) 1+

d0 + x which is

1)

v 1+

d0 + x >

(n k )v 1 + (n

d0 k)

+ then we have that the above hold true when:

(1 + )d which holds from the previous inequality. The next lemma pins down the SP E price

(n

1)v > 0

Lemma. The seller is indierent between accepting all oers and rejecting all oers (nxs (n, d) d) .

xi d =

Proof. The proof establishes a sequence of claims: Notice that from the previous stage, the outcome of the continuation game is uniquely pinned down by xs . Let vs (m, d0 ) be the sellers value entering a round in which he makes an oer, 33

their are m buyers left and the outstanding debt is d0 : vs (m, d0 ) := mxs (m, d0 )
0

d0

vs (m, d ) =

8 < mv :m

d0 1+ m v 1+

if (n d if (n

v 1) 1+

v 1) 1+ <d

and vs is strictly decreasing in the debt level. Claim 1: if a seller receives x1 , .., xn SP E oers, the seller must be indierent between accepting all oers or accepting a (st) subset of oers. The seller cant strictly prefer accepting a subset of oers. If the seller strictly preferred accepting all prices, then a buyer could lower the price and the seller would still accept. Let k be the lowest number of oers that the seller is indierent between accepting all oers and accepting k oers. Since vE (m, d0 ) is st decreasing in d0 , it must be that the seller is indierent between accepting all oers and accepting the k highest oers. WLOG, x1 ... xn . .. xk xk+1

Claim 2: If k > 0 then the k + 1 highest oers must all be the same: x1 = x2 = .. = xk+1 . Since k is lowest number of oers that the seller is indierent between accepting and rejecting, if j < k the seller strictly prefers accepting k highest oers to j highest oers. Therefore, if x1 > xk+1 then buyer 1 can decrease the price and still be accepted. Claim 3: k = 0 Let x be the k + 1 highest oers. The crux of the proof is that is that the seller can not strictly prefer accepting k + 1 oers to k oers, because then the seller would also (st) prefer accepting k + 1 to all oers. This places an upper bound on how high the price x can be: if x is too high the seller would prefer accepting k + 1 oers to k oers, and this upper bound leads to a contradiction. Formally, it must be that: (3)

vE (n

k, d

kx)

vE (n

1, d

(k + 1)x)

and lemma 1 provides the contradiction: Lemma1 : if vs (n k, d kx) vs ( n k 1, d (k + 1)x) for some k , then it must be:

vs (n, d)

vs ( n

k, d

xk )

Lemma1 : if vE (n

k, d

kx)

vE ( n

1, d

(k + 1)x) for some k , then it must be:

vE (n, d) Proof of lemma 1:

vE ( n

k, d

xk )

34

Proof. To prove lemma 1, assume that x and k are such that vE (n 1, d (k + 1)x), to show that vE (n, d) k, d k, d xk ) is high debt xk ) =
(n k)v (d xk) 1+ (n k)

k, d

kx)

vE (n

vE (n

k, d

kx), I break it down to three cases.

Case 1: B (n The value vE (n Case 1a: B (n The value vE (n

(k + 1), d (k + 1), d

x(k + 1)) is also high debt games. x(k + 1)) =


(n k+1)v (d x(k+1)) 1+ (n (k+1))

And (1) implies that x and as long as x


d+v 1+ n

d+v 1+ n .

v If B (n, d) is a low debt game, then v (n, d) = n 1+

d and v (n
d+v 1+ n

k, d

xk ) =

(n k)v (d xk) 1+ (n k)

the rst is greater then the second. (mathematica) and v (n k, d xk ) =


(d xk)+v 1+ (n k) d+v 1+ n

If B (n, d) is a high debt game, we have that v (n, d) = and again, as long as x Case 1b: B (n

the rst is greater then the second.

(k + 1), d

x(k + 1)) is a low debt game.


v 1+

By the transition lemma x (otherwise, since x


v 1+

otherwise moving from k to k + 1 buyers would leave us in

a high debt game. Applying the transition lemma again, it must that , B (n, d) is a high debt accepting k oers moves us to a low debt game).
v 1) 1+ <d

Since B (n, d) is high debt, it must be that (n From 1 it must be that vE (n (n () k, d kx) (n

vE ( n

(k + 1), d v 1+

(k + 1)x): (d (k + 1)x)

k )v (d xk ) 1 + (n k )

(k + 1))

x (k + 1), d

v + (n k )[(1 + )d + (2 + k n)v )] (1 + )((1 + (1 + k )(n k ))) x(k + 1)) is low debt game: (n x (1 + )d (n k 2)v (1 + )(1 + k ) ()
(1+ )d (n 1)v (1+ )(1+k)[1+ (1+k)(n k)]

Second since B (n implies:

v 2) 1+

x(k + 1) which

Notice that () d >


v 1+

(1+ )d (n k 2)v (1+ )(1+k)

>

v + (n k)[(1+ )d+(2+k n)v )] (1+ )((1+ (1+k)(n k)))

>0

(n

1)

Which is always true since B (n, d) is high debt game. Case 2: B (n k, d xk ) is a low debt game, and d xk > 0. This implies that vE (n k, d xk ) = (n
v k ) 1+

(d

kx).
v 1+

If B (n, d) is a low debt game, it must be that x = vE (n k, d xk ) = n k, d v 1+ d + (x

. To see this, we can rewrite (x v )k 1+

v )k = vE (n, d) 1+
v 1+

Since vE (n, d) vE (n that B (n (k + 1), d

xk ) we have that x

. The transition lemma implies


v 1+

x(k + 1)) is also high debt game (because x 35

). But since

vE (n

k, d (n

kx) k)

vE (n v 1+

(k + 1), d (d kx) x

x(k + 1)) it must be that (n v 1+


v 1+

(k + 1))

v 1+

(d

(k + 1)x) ()

Which implies that x =

v 1+

. Now, since x =

and both B (n, d) and B (n

k, d

xk ) are

low debt games we have that: vE (n, d) = vE (B (n and we are done with this case.
v If B (n, d) is a high debt game, then from transition lemma: x > 1+ , and since B (n k, d xk ) is low debt, the transition lemma implies that B (n (k + 1), d x(k + 1)) is also low debt.

k, d

xk ) () x =

v 1+

Since, vE (n (n

k, d k)

kx) v 1+

vE (n (d

(k + 1), d (n v 1+

x(k + 1)) it must be that (k + 1)) v 1+ (d (k + 1)x) ()

kx) x

Which is a contradiction. Case 3: B (n k, d xk ) is a low debt game kx


v 1+

If accepting k oers moves us to a game without debt. Then for E to not prefer accepting k + 1 oers it must be that x . The transition lemma implies that B (n, d) is a low debt game,
v vE (d, n) = (n 1+

which means that if E rejects all oers he gets: get by accepting k oers is 7.2.2 Existence
v n 1+

d) but the most he can

d which is strictly less.

Dene the following stationary strategies: For a sub-game with m buyers and debt d0 , the seller oers: xs (m, d0 ) = 8 <
v 1+ : d+v 1+ n

if (m if (m d0 .

v 1) 1+ v 1) 1+

d < d0

Denote the sellers value as vs (m, d0 ) = mxs (m, d0 )

When the seller receives oers he accepts a subset of oers to maximize his continuation value vE (m, d0 ), if indierent he accepts the highest number of oers. The buyers all oer the same price, in a sub-game with m and debt d: mxj (m, d0 ) d0 = = 36 vE (m, d0 ) () vE (m, d0 ) + d0 m

xj (m, d0 )

When a buyer receives an oer x, then if (m that v x (v


1+ 1+

v 1) 1+ v 1) 1+

d he accepts any price oer such < d he accepts any price oer

such that v rium oers.

x>

(v

) or x

v 1+

and if (m

d + (m

d+v 1) 1+ m )

Beliefs are passive, the buyers o-equilibria beliefs are that other buyers get the equilibProposition. The above strategy are a sequential equilibrium Proof. By construction, buyers beliefs are consistent with equilibrium play. Start with a seller: 1. It is never protable for the seller to oer a lower price. By induction on B (n, d) 2. The seller can deviate, and oer a higher price to some buyer, or a subset of buyers. Assume he oer k prices that are accepted, the subsuming sub-game has n d =n
0

k buyers, with a debt of


v 1+

k . There are two options, either B (n, d) is a low debt game or high debt game. , the
v 1+

If B (n, d) is a low debt game then by the transition lemma if k buyers accept a price of subsuming game is also a low debt game, and the seller gets a price that is less then this....

....show

d+v If B (n, d) is a high debt game then if k buyers accept a price of x = 1+ n the subsuming subgame, B (n k, , d kx) can either be low or high debt. If it is a high debt game, the sellers

value is vE (n B (n k, d

k, d

xk ), and since x =
d+v 1+ n

d+v 1+ n v 1+

we have that vE (n

k, d

xk ) = vE (n, d). If
v 1+

kx) is a low debt game, then the sellers continuation price is less then > k buyers

and

since B (n, d) is a high debt game: of x to the n

, and therefore he is better o oering a price

3.. The seller can deviate and accept a subset of oers. Assume the seller accepts 0 < k < n oers.Again there are two options:
v 3a. B (n, d) is a low debt game: the buyers oer a price of x b = ( 1+ ) + (1 n )d

Assume rst that accepting k oers of x move the game to a low debt game i.e. B (n k, d kx b) is low debt, the sellers total revenue from the n sell those units today for a revenue of (n k)
v 1+

k units tomorrow is (n + (n k ) (1 x b k)
)d n

v k ) 1+ but he can

, which is more. k, d xk ) if he have

Second, if accepting k oers of x moves the game to a high debt game, i.e. B (n is high debt, in which case his prots are vE (n accepts all oers he gets that
(n k)v (d xk) 1+ (n k) v (n 1+

k, d

d) and it holds true that for

(n k ) v ( d x k ) = ( 1+ (n k) b ) and (1 )d v x we b = ( 1+ ) + n

v n 1+

d () bn + bn2 ]v (1 + )d[ n(n (1 + b)n(1 + b(n k )) k ) + k (1 )] 0

n[ k

bkn

Since B (n, v ) is a low debt game, we have that (1 + )d (n LHS enumerator: n[k bkn bn+bn2 ]v (1+ )d[ n(n k )+k (1

1)v and therefore we have that

)] 37

n[k bkn bn+bn2 ]v (n 1)v [ n(n k )+k (1

)] = (1

)kv > 0

and we are done.


d+v 3b. if B (n, d) is a high debt game, the buyers oer a price x b = ( 1+ n ) + (1 d )n

then accepting a subset of k oers, can either move the game to a low debt one or a high debt game. If accepting a subset of prices moves the game to a low debt game, the notice that the continuation price is
v 1+

which is less then x b and therefore, the sellers total revenue is greater if he

accepts all prices today then rejecting a subset. If accepting a subset of prices moves the game to a high debt game, the we have that the sellers continuation value is (
(n k)v (d kx b) ) 1+ (n k) nv ( 1+ d n ).

and if the seller

accepts all oers he gets:

And the latter is greater when: k )v (d kx d+v b) () x b 1 + (n k ) 1+ n


d+v 1+ n

nv d 1+ n
d+v But since x b = ( 1+ n ) + (1

(n

d )n we have that x b

()

d n

<

d+v 1+ n

() nv > d

4. The buyers SP E oer is x b = xs (n, d)+(1 because the continuation price is always higher: x b < xs (n, d) () To see this: If B (n, d) is low debt, d (n
d+v n 1+ n

d )n . A buyer can deviate and oer a dierent

price and get rejected, in which case the continuation price is s (n, d), but this is not protable

d < xs (n, d) n and xs (n, d) =


v 1+ d+v 1+ n

v 1) 1+ , and xs (n, d) = v 1) 1+

and therefore the and we have that

above holds. If B (n, d) is high debt, d > (n > d () nv > d. And again this holds.

5. A buyer can deviate and reject the oer, but by construction a buyer is indierent between accepting and rejecting given that everyone else rejects.

References
Acemoglu, D., S. Johnson, and T. Mitton (2009): Determinants of Vertical Integration: Financial Development and Contracting Costs, The Journal of Finance, 64(3), 12511290. Aghion, P., and P. Bolton (1992): An Incomplete Contracts Approach to Financial Contracting, The Review of Economic Studies, 59(3), 473494. Aghion, P., M. Dewatripont, and P. Rey (1994): Renegotiation Design with Unveriable Information, Econometrica, 62(2), 257282. Aghion, P., and J. Tirole (1997): Formal and Real Authority in Organizations, Journal of Political Economy, 105(1), 129. Baker, G., R. Gibbons, and K. J. Murphy (2002): Relational Contracts and the Theory of the Firm, The Quarterly Journal of Economics, 117(1), 3984. 38

Baron, D. P., and J. A. Ferejohn (1989): Bargaining in Legislatures, The American Political Science Review, 83(4), 11811206. Bester, H., and J. Sakovics (2001): Delegated bargaining and renegotiation, Journal of Economic Behavior and Organization, 45(4), 459 473. Brander, J. A., and T. R. Lewis (1986): Oligopoly and Financial Structure: The Limited Liability Eect, The American Economic Review, 76(5), 956970. Bresnahan, T. F., and J. D. Levin (2012): Vertical Integration and Market Structure, Discussion Paper 17889, National Bureau of Economic Research. Bronars, S. G., and D. R. Deere (1991): The Threat of Unionization, the Use of Debt, and the Preservation of Shareholder Wealth, The Quarterly Journal of Economics, 106(1), 23154. Cai, H., and W. Cont (2004): Agency Problems and Commitment in Delegated Bargaining, Journal of Economics and Management Strategy, 13(4), 703 729. Che, Y., and D. B. Hausch (1999): Cooperative Investments and the Value of Contracting, American Economic Review, 89(1), 125147. Che, Y., and J. Sakovics (2004): A Dynamic Theory of Holdup, Econometrica, 72(4), 1063 1103. Demarzo, P., and D. Duffie (1999): A Liquidity-based Model of Security Design, Econometrica, 67(1), 65 99. DeMarzo, P. M., I. Kremer, and A. Skrzypacz (2005): Bidding with Securities: Auctions and Security Design, American Economic Review, 95(4), 936959. Edlin, A. S., and S. J. Reichelstein (1996): Holdups, Standard Breach Remedies, and Optimal Investment, American Economic Review, 86(3), 478501. Fershtman, C., and K. L. Judd (1987): Equilibrium Incentives in Oligopoly, American Economic Review, 77(5), 92740. Fershtman, C., K. L. Judd, and E. Kalai (1991): Observable Contracts: Strategic Delegation and Cooperation, International Economic Review, 32(3), 551559. Gale, D. (1986): Bargaining and Competition Part I: Characterization, Econometrica, 54(4), 785806. Grossman, S. J., and O. D. Hart (1986): The Costs and Benets of Ownership: A Theory of Vertical and Lateral Integration, Journal of Political Economy, 94(4), 691719. Haller, H., and S. Holden (1997): Ratication Requirement and Bargaining Power, International Economic Review, 38(4), 825851. Harris, M., and A. Raviv (1991): The Theory of Capital Structure, The Journal of Finance, 46(1), 297355. 39

Hart, O., and J. Moore (1990): Property Rights and the Nature of the Firm, Journal of Political Economy, 98(6), 11191158. Horn, H., and A. Wolinsky (1988a): Bilateral Monopolies and Incentives for Merger, The RAND Journal of Economics, 19(3), 408419. Horn, H., and A. Wolinsky (1988b): Worker Substitutability and Patterns of Unionisation, The Economic Journal, 98(391), 484497. Innes, R. D. (1990): Limited liability and incentive contracting with ex-ante action choices, Journal of Economic Theory, 52(1), 45 67. Klein, B., R. G. Crawford, and A. A. Alchian (1978): Vertical Integration, Appropriable Rents, and the Competitive Contracting Process, Journal of Law and Economics, 21(2), 297326. McAfee, R. P., and M. Schwartz (1994): Opportunism in Multilateral Vertical Contracting: Nondiscrimination, Exclusivity, and Uniformity, The American Economic Review, 84(1), 210230. McMillan, J., and C. Woodruff (1999): Dispute prevention without courts in Vietnam, Journal of Law, Economics, and Organization, 15(3), 637658. Noldeke, G., and K. M. Schmidt (1995): Option Contracts and Renegotiation: A Solution to the Hold-up Problem, The RAND Journal of Economics, 26(2), 163179. Osborne, M. J., and A. Rubinstein (1990): Bargaining and Markets. Academic Press. Perotti, E. C., and K. E. Spier (1993): Capital Structure as a Bargaining Tool: The Role of Leverage in Contract Renegotiation, American Economic Review, 83(5), 113141. Perry, M., and L. Samuelson (1994): Open- versus Closed-Door Negotiations, The RAND Journal of Economics, 25(2), 348359. Rajan, R. G. (1992): Insiders and Outsiders: The Choice between Informed and ArmsLength Debt, The Journal of Finance, 47(4), 13671400. (2012): Presidential Address: The Corporation in Finance, Journal of Finance, 67(4), 11731217. Rajan, R. G., and L. Zingales (1998): Financial Dependence and Growth, American Economic Review, 88(3), 55986. Rajan, R. G., and L. Zingales (2001): The Firm as a Dedicated Hierarchy: A Theory of the Origins and Growth of Firms, The Quarterly Journal of Economics, 116(3), 805851. Repullo, R., and J. Suarez (2004): Venture Capital Finance: A Security Design Approach, Review of Finance, 8(1), 75108. Rubinstein, A. (1982): Perfect Equilibrium in a Bargaining Model, Econometrica, 50(1), 97109. 40

Rubinstein, A., and A. Wolinsky (1985): Equilibrium in a Market with Sequential Bargaining, Econometrica, 53(5), 113350. Sarig, O. H. (1998): The eect of leverage on bargaining with a corporation, Financial Review, 33(1), 116. Shleifer, A., and R. W. Vishny (1989): Management entrenchment: The case of managerspecic investments, Journal of Financial Economics, 25(1), 123 139. Stole, L. A., and J. Zwiebel (1996): Organizational Design and Technology Choice under Intrarm Bargaining, The American Economic Review, 86(1), 195 222. Williamson, O. E. (1979): Transaction-Cost Economics: The Governance of Contractual Relations, Journal of Law and Economics, 22(2), 233261. (1998): The economic institutions of capitalism. Free Press. (2005): The Economics of Governance, The American Economic Review, 95(2), 118. Zingales, L. (2000): In Search of New Foundations, The Journal of Finance, 55(4), 1623 1653.

41

You might also like