Debt vs Foreign Direct Investment: The Impact of International Capital Flows on Investment in Environmentally Sound Technologies

Transcription

1 Debt vs Foreign Direct Investment: The Impact of International Capital Flows on Investment in Environmentally Sound Technologies J. O. Anyangah Abstract This paper employs the methods of mechanism design under informational asymmetry to examine the relationship between nancing arrangements and optimal investment in environmentally sound technology (EST). In a model incorporating both adverse selection and moral hazard, it is shown that the nancial structure has implication for agency costs and the optimal provisions of incentives. Relative to debt nance, Foreign Direct Investment (FDI) is characterized by less intense adverse selection, but a more pronounced moral hazard problem. Thus, when the scope of adverse selection consequences are su ciently signi cant vis-à-vis the moral hazard e ects, FDI outperforms debt nance in terms of the level of technological investment that is optimally induced. 1 Introduction International transfer of environmentally sound technologies (ESTs) has received much attention recently as a means to ameliorate leading environmental problems. Many researchers recognize, for example, that the transfer of energy ef- cient technologies from developed countries to developing countries can not only lead to local development, but also help the developed countries meet their commitment to reduce greenhouse gases in a cost e ective manner ( Fischer et al., 1998). However, for the process of technology transfer and its uptake to be successful, it is crucially important that adequate nancing be in place. Consider debt and foreign direct investment as alternative media through which Department of Economics, University of Lethbridge. address: joshua.anyangah@uleth.ca 1

2 the requisite nancing can be obtained. While debt typically precludes the foreign investor from any role in project implementation, foreign direct investment (FDI) may allow the investor to gain in uence over the domestic rm s broad strategic objectives and organizational direction potentially reducing the severity of any informational asymmetry. A simple and spontaneous argument may then lead to the conclusion that the agency problem could be largely mitigated under FDI than debt. But is this intuition correct? More generally, how can nancial structure a ect optimal incentives for investment in technology? The literature that has attempted to answer this question is scant. 1 This paper employs the methods of mechanism design under informational asymmetry to examine the relationship between nancing arrangements and the optimal investment in environmentally sound technology. In the model, a resource-constrained domestic agent must secure core capital from a risk neutral foreign investor (principal) in order to undertake a project. The project involves large-scale modernization ( greening ) of the production process at a rm owned by the agent. Two forms of nancing - Foreign Direct Investment (FDI) and debt nance - are considered. The central conclusion is that nancial structure has implication for the nature of investment decisions. More precisely, relative to debt nance, FDI diminishes the severity of the adverse selection problem, but exacerbates the moral hazard problem. Hence, whether or not agency costs are higher under FDI depends upon the signi cance of the adverse selection consequences vis-à-vis the moral hazard e ects. To discern the logic behind this result, one must understand the distinguishing characteristics of the two funding mechanisms and the concomitant informational structures. The basic di erence between debt and FDI revolves around the degree of control that can be exercised by the foreign investor over the project. Under debt nance, the two parties are presumed to interact strictly at arms length and any role played by the investor does not extend beyond providing liquidity. Here, the success or failure of the project depends upon the domestic rm s absorptive capacity, which is private information to the agent, and unobservable level of technological investment that is undertaken by the agent. By contrast, FDI enables the investor to intervene in the operations of the rm by providing managerial and organizational expertise. Thus, when examining FDI, it is argued that the success or failure of the project depends not only upon the 1 Although a number of studies have examined the implication of nancial policy, most have focused on the relationship between capital struture and the value of the rm in a domestic environment. These include work by Lewis and Sappington (1995) and Myers and Majluf (1984), Jensen and Meckling (1976) and Ross (1977). 2

3 domestic rm s absorptive capacity and the unobservable level of technological investment, but also on the amount of expertise supplied by the investor. The foregoing implies the following. Under debt nance, the foreign investor confronts a one-sided moral hazard problem and an adverse selection problem. To limit the agent s information rent, the level of investment is distorted away from its full information value. When FDI is the nancing channel, however, the optimal contract must not only take care of the moral hazard and adverse selection problems on the part of the domestic agent. It must additionally deal with a moral hazard incentive on the part of the investor because the value of the investor s expertise cannot be ascertained and therefore contracted upon. Consequently, the cost of inducing any given level of investment must include an incentive loss on the part of the investor since enhancing incentives for the agent implies a diminution of the investor s incentives. Another result is that the associated information rent is less pronounced. This arises because reducing the agent s incentive in a rational attempt to capture some of the information rents brings forth an o setting incentive gain on the part of the investor. In sum, the cost of extracting informational rent is lower under FDI. Hence, whether FDI leads to more investment as compared to debt nance depends upon the magnitude of the bilateral moral hazard e ect relative to the adverse selection one. Only in the special case where the adverse selection e ect is su ciently large vis-a-vis the moral hazard e ect will the agent invest more under FDI than under debt nance. This paper is in the same spirit as previous studies that have examined the problem of international technology transfer. Recent work in this realm include Choi (2001) and Tao and Wang (1998). Choi uses an incomplete contract model of the licensing relationship to explain the prevalence of royalty contracts. Tao and Wang focus on contractual joint ventures between multinationals and local rms in an environment characterized by weak enforcement of binding contracts. Although both studies employ the principal-agent framework, they assume away the crucially important role played by the absorptive capacity of the technology recipient in ensuring successful technology transfer. In these studies, imperfect information result from hidden action (moral hazard) only. By contrast, this paper is cast in an environment in which the foundation of imperfection information can be found in both moral hazard and hidden information (adverse selection). Among the earliest papers on international technology transfer are Gallini and Wright (1990). They focus on licensing contracts for newly patented inno- 3

4 vations when the licensor has private information about the economic value of the patent. One di erence with this paper is that nancial structure does not a ect agency problems between the contracting parties. The second di erence relates to the structure of information asymmetry. While our model assumes that the agent (technology recipient) is the one with a relevant private information, theirs is one of signalling by a privately informed principal in the spirit of Maskin and Tirole (1992) and Beaudry (1994). Also closely related to this paper are studies that have compared debt and foreign direct investment as forms of international investment under di erent institutional arrangements. Among the earliest papers in this realm is Razin et. al. (1996) who examine sources of market failure in the context of international capital ows. Their focus is on providing guidelines for e cient tax structure in the presence of market imperfection. Schnitzer (2002) considers the impact of sovereign risk on the structure of international capital ows. Neumann (2003) is perhaps closest to this paper. She compares FDI with debt nance when monitoring is costly and nds that FDI nancing is preferred to debt nance because costly monitoring increases investment, output and consumption for the domestic rm. There is no private information in Neuman s model and the moral hazard problem is one-sided. The rest of this chapter is structured as follows: The next section presents the basic elements of the model. Section 3 develops a benchmark solution. Section 4 analyzes the equilibrium when asymmetric information prevails. Under debt- nancing, the investor s interaction with the domestic agent does not go beyond providing liquidity. As a result, the investor is only constrained by the standard incentive compatibility and participation constraints. Under FDI, the investor participates in the project by providing managerial and organizational expertise. It is assumed that the investor s intervention enhances the project s success probability, but the value of this contribution can not be contracted upon. Thus, this version of the model imposes the additional constraint that the investor s participation be optimal given the contract. Section 5 concludes. 2 Basic Model There are two players in the model: a domestic agent and a foreign investor. 2 The agent owns a rm that currently employs a technology that involves an in- 2 The agent can be thought of as the government of a developing country that currently owns and operates a state entreprise such as an electric power utility. 4

5 e ciently high rate of fossil fuel use and emissions of greenhouse gases that are known to a ect the overall environmental quality in the region. By undertaking investment in a clean technology, the rm s rate of energy use could be significantly reduced. 3 The technological switch can be thought of as a large-scale modernization of the rm s production process that requires major purchases of EST. We assume that the agent s initial endowment, which we normalize to zero, is not su cient to meet the investment requirement. It is also assumed that the domestic capital markets can neither provide all the required capital nor the clean technology. These assumptions are necessary to ensure the that the agent will need core capital from the foreign capital market. The rm is otherwise assumed to be a large enterprise which can bene t from the international capital markets. The foreign investor is meant to be a foreign entity with deep pockets. The agent seeks funds from the foreign investor in order to undertake the technological switch. If the investor accepts to underwrite the cost of converting to the clean technology, then he has the option of channelling his funds through two media: Foreign Direct Investment (FDI) or a market loan (Debt). 4 It is assumed that FDI enables the foreign investor to control the rm s strategic direction. 5 With FDI, the investor e ectively obtains authority over the project by acquiring an equity stake in the domestic rm. Under a debt contract, however, the rm secures the external resources but retains control and management rights over the project. Once nancing has been secured, the execution of the project involves two successive steps. 1. In the rst stage, the agent embarks on plant modernization by buying and installing an EST from a foreign supplier. The amount of expenditure on the ETS is given by I 2 < During the second step, which corresponds to the main source of risk, the agent uses the new technology to produce and deliver a composite good to the market. The use of the EST generates a number of nancial bene ts: Reduced cost of raw materials and resources, improved compliance and reduced regulatory interference, reduced future liabilities and improved quality products 3 Throughout we use owner, rm, agent and manager interchangeably. 4 Throughout we use the masculine term he to refer to the investor and the femine term she to refer to the domestic rm/manager. 5 Control is de ned as owning 10% or greater of the voting shares of an incorporated rm, having 10% or more of the voting power for an unincorporated rm or development of a green eld investment. Thus, FDI can also be interpreted as a situation where the government privatizes a state-owned enterprise. 5

6 and reduced defects. The net nancial bene ts from the project are uncertain. In particular, if the technological switch is successful, the investment yields a net cash ow of H. If the technological switch is not successful, however, the corresponding net cash ow L. We assume that = H L > 0. A fundamental di erence between the two funding mechanisms relates to the degree of control exercised by the foreign investor over the project. If funding takes the form of debt, then it is assumed that the two parties interact at arms length. In this case, any role played by the investor does not extend beyond funding the project. By contrast, FDI enables the investor to be involved in the operations of the domestic rm. This may stem from the fact that FDI, by its nature, a ords the investor a greater degree of management control which is conducive to eliminating perquisite consumption and other wasteful activities. Also, FDI enables the investor to obtain better information because she is geographically closer to the rm. By having a representative from headquarters on the board of directors of the local establishment, the foreign direct investor is better able to in uence the rm s overall investment decisions and closely monitor the project than a lender. FDI may bring additional bene ts such as managerial expertise and technological skills. Formally, we account for these di erences by assuming that the probability that the project is successful depends upon the level of investment in the clean technology I, the rm s ability to absorb the clean technology (absorptive capacity), and the managerial and organizational expertise provided by the investor E. We denote the probability function by p(i; E; ), where = f0; 1g is the intervention parameter. Under debt contract, = 0 since the investor s involvement does not extend beyond providing liquidity. Under FDI, however, the foreign investor provides cash, but in addition participates in the project so = 1. The function p(i; E; ) satis es the following standard properties: p I (:) > 0; p II (:) < 0, p E (:) > 0; p EE () < 0, p IE (:) = p EI (:) = 0, p (:) > 0, p I (:) > 0, p E (:) > 0. (1) The rst four conditions say that investment and expertise increase the likelihood of success at a decreasing rate. The next assumption simply states that the probability function is separable in investment and expertise. The last three conditions say that rm s absorptive capacity enhances both the likelihood of 6

7 success and the productivity of both I and E. The assumption on p EI (:) has been imposed for analytical simplicity, but most importantly, to underscore the fact that the investor s input supplements the manager s investment I and is not absolutely essential for project success. To give more structure to the probability function and to obtain sharp results, we make the following assumption: (0; 1). 6 Assumption 1 p(i; E; ) = (q(i) + h(e)) = I + E where ; 2 Information: A crucial assumption is that the two parties have asymmetric information. The agent knows exactly the value of from the outset; the investor s knowledge of is limited. The investor only knows that parameter belongs to some compact set R +. Without loss of generality we take = [ ; ]. In addition, I and E, cannot be contracted upon. However, it is publicly known whether the project is a successful or not. Since the success probability is a function of, E and I, the model is characterized by both bilateral moral hazard and adverse selection. The investor has some a prior probability on which is associated with a continuous density function f() df ()=d, where F () is the distribution function of. As is standard in the incentive literature, we assume the following with respect to F (): 7 Assumptionh2 The distribution i of types F () satis es the monotone rate property: 0. F () f() The analysis covers a single period from point 0 when nancing, investment and production decisions are made to point 1, where returns are realized. Before investment activities can be undertaken, the rm and the investor must negotiate a contract specifying how the investor will be compensated for funding the project. Such a contract must be based on variables that are veri able by the investor or a third party. Since investment, expertise and type are unobservable (except under full information), the only possible basis for the contract is the realized cash ows i, i = H; L. One more point is worth noting here. The nature of the contractual arrangement between the investor and the rm will depend on the bargaining capabilities of the two parties. In this paper, we will assume that the investor has all the bargaining power when designing the contract. This may be a reasonable 6 A similar technology has been speci ed by Lewis and Sappington (2000). All qualitative results do not depend on this speci cation. 7 See, for example, Fudenberg and Tirole, 1991, p

8 assumption in situations where there is intense competition for foreign nancial resources among between host countries. 8 Thus, during contract negotiation, it can be appropriate to assume that the investor makes a take-it or leave-it o er to the rm. To complete the model set up, we assume that the international rate of interest is r so the opportunity cost of funds for both the investor and the rm is (1 + r) per unit. 3 E ciency Before proceeding to characterize the investor s optimum under information asymmetry, we consider, as a benchmark, a setting where no moral hazard or adverse selection problem arises because the investor shares the rm s knowledge of its absorptive capacity and is able to observe the level of investment; that is, both and I are observed by the investor. We also assume that E is observed by the rm and can therefore be contracted upon. Consider a debt contract. When the rm and the investor get together to negotiate a contract in this setting, they have two decisions to make: First, they must specify the loan amount b that the foreign investor will provide. Second, they must agree on the repayment schedule specifying the transfer that the rm must make for each possible realization of cash ows. We assume that the optimal contract calls for the rm to repay R when the project fails and and R + in the event that the project is a success. Here is a loading factor that can be heuristically interpreted as the incremental reward/punishment for success. Thus, if is negative in value, the rm is e ectively punished for success. On the other hand, if is positive, the agent is rewarded in the event of success. The rm s expected utility is U = p(i; )[ + ] + L R + (1 + r) [b I] while the investor s expected payo is V = p(i; )[R+]+[1 p(i; )]R (1+r)b. The investor s problem [L] reads: max p(i; ) + R (1 + r)b (2) (I;R;) subject to p(i; )[ ] + L R + (1 + r) [b I] 0, (3) 8 There is evidence that host-countries do compete for foreign direct investment. See, for example, Koray and Taylor (2000), and Hau er and Wooton (1999). 8

9 where the constraint is the usual ex ante participation constraint, which says that the rm s manager payo must be positive in order to guarantee his or her participation. Solving this problems is straightforward. The participation constraint of the rm is binding. Now consider FDI. Here, the investor acquires signi cant ownership and control rights over the project through an outright purchase of shares in the rm but retains the local manager to execute the investment project on its behalf. Thus, when the rm and investor get together to construct a contract, they must specify a purchase price and how the investor will be compensated for supplying the funds. Such a contract must be based on variables that are veri able by the investor or a third party. We assume that the investor implements the following linear payment scheme: s = ( ) i + v( ) 8 2 h i (), (), i = H; L: (4) In this formulation, v can be interpreted as the purchase price or lump sum transfer of funds. We assume that v accrues directly to the agent and does not therefore become part of the nal cash ows to be shared between the two parties. can be thought of as the manager s stake in the project. In other words, the investor acquires proportion (1 ) of the rm s shareholding and in return pays v. 9 The investor solves the following problem [D]: subject to max [p(i(); E(); ) + (I;E;;v) L + v()](1 ()) v() (5) [p(i(); E(); ) + L ]() + (1 + r)[v() I()] 0; (6) where again the constraint gives the participation constraint of the agent. The following result is immediate: Proposition 1 Suppose that investment in technology can be funded either through FDI or debt. Then under full information, the investment chosen will be the same regardless of the funding mechanism. Proof. The proof follows readily from problems [L] and [D] and the fact that constraints (3) and (6) are binding. Making use of the binding participation constraints, the rst-best levels of investment under debt and FDI can be given 9 v can also be interpreted as a base salary or an up-front payment. 9

10 by the solution to maxp(i; ) + L + (1 + r)i U and maxp(i; E; ) + I I L (1 + r)i U E, respectively. These imply that p I (I L ; ) (1 + r) = 0 under debt and p I (I D ; E; ) (1 + r) = 0 under FDI, where I L and I D are the rst-best level of investment. Given that p IE (:) = p EI (:) = 0 and from the fact that p(:) is concave in I, it follows from the rst-order conditions that I L = I D. Intuitively, the e cient investment to undertake is the amount that maximizes the total expected surplus. In this rst-best setting with no private information and moral hazard, the agent earns no rents (U = 0) and the optimal investment is equates the marginal return to the opportunity cost of funds. Since this cost is exogenously determined by the international capital markets independently of the nancial structure, the investor faces the same marginal cost of investment regardless of the medium used to obtain the required nancial resources. 4 Combining moral hazard and adverse selection When both moral hazard and adverse selection informational problems exist in the same context, the contract design problem can be analyzed within a mechanism design framework. By the revelation principle, there is no loss in generality in focusing on a direct mechanism in which the investor provides the rm with incentives that induce truthful behavior (e.g., Fudenberg and Tirole, 1991 and La ont and Tirole, 1993). In a direct mechanism, the investor o ers a standard screening contract C = fs(^) : ^ 2 g, prescribing a level of transfer s(^) conditional upon the manager s announcement ^. We assume that the investor can credibly commit not to renegotiate the contract. The investor selects s(^) to maximize her expected payo. In so doing, he takes into account the response of the privately informed rm. As in similar models, the optimal actions of the privately informed rm gives rise to two kinds of constraints that the investor must take into account when designing the mechanism. The rst kind ensures that the rm reports its type ^ truthfully and undertakes the optimal level of investment. These constraints are called the incentive compatibility constraints. The second kind of constraints are the individual rationality constraints. They require that the rm, whatever his type, gets its reservation payo, the payo that the rm would get by not participating in the project. The sequence of events is as follows: In the rst stage, nature draws a type 10

11 for the rm from a set of feasible types 2. Only the rm learns the true value of. In the second stage, the investor and the rm agree on a menu of contracts. The rm reports its type and then chooses a level of investment in the green technology given the sharing rule. Final project output is observed and the transfers implied by the menu of contracts are implemented. 4.1 Debt Finance Consider rst the case where the required liquidity is delivered through a debt contract. In this setting, = 0 and the success probability collapses to p(i; ). Recall from the previous section that the investor provides a loan in the amount of b dollars to the rm and that the optimal contract calls for the rm to repay R + and R() when the realized cash ows are H and L, respectively. Thus, the revelation mechanism in this situation is s(^) = fr(^); (^); b(^)g^!2[!,!]. A rm type that reports that her type is ^ when the true parameter is has a utility U(^; ): h U(^; ) = p(i(^); )[ + (^)] + L R(^) + (1 + r) b(^) i I(^). (7) Hence, incentive compatibility requires that h 2 arg max p(i(^); )[ + (^)] + L R(^) + (1 + r) b(^) ^ Let U() = U(; ) denote type s rent when reporting truthfully: i I(^) (8) U() = max ^ p(i(); )[ + ()] + L R() + (1 + r) [b() I()]. (9) A rst step in studying the mechanism design problem of the investor is to characterize the set of functions corresponding to an incentive compatible mechanism. Lemma 1 below characterizes this set. only if Lemma 1 If 2, a pair fu(), (:)g is incentive compatible if and du() d = p (I(); )[ + ()], (10) U() is convex on (11) and () is non-decreasing. (12) 11

12 In sum, the Lemma states that if the local incentive-compatible constraints are satis ed and (:) is nonincreasing, then these necessary conditions are suf- cient. Hence the in nite incentive constraints that the investor must take into account during the process of contract design collapse to a di erential equation and monotonicity constraint. A direct revealing mechanism is then characterized by (10) - (12). There are two standard approaches to studying the mechanism design problem of the investor. One alternative is to focus on the transfer schedule ((); R(); b()). The other option is to focus on the information rent U(). In the following, we will adopt the latter technique. Using the de nition of transfers as given in U(), we can rewrite the investor s nal wealth in terms of U() as V(U(:); (:); ) = p(i; ) + L + (1 + r)(i() U(). Since the investor does not know, his expected payo can be written as ev (U(:); (:); ) = Z p(i; ) + L + (1 + r)i U() df (). (13) Thus, the screening problem under a debt contract is as follows: subject to (10), (11), (12), max ev (U(); (); ) (14) fu();r();()g and U() ; (15) I 2 arg max p(i; )[ + ()] + L R() + (1 + r)(b() I()). (16) I Expression (14) states that the investor desires to maximize the expected total surplus less the rm s expected rent. Notice that since the rm s rent is costly, the investor has an incentive to optimally induce distortions away from e ciency in order to reduce the ability of the rm to command the rent. The participation constraint (15) for the rm ensures that it receives nonnegative expected payo, regardless of its type and report. Equation (16) requires n that the manager select the level of investment optimally given contract (R(^); (^); b(^)) : ^ o 2 [; ]. 12

13 The participation constraint must be binding somewhere. If it was not, then the investor could raise the rm s repayment commitment uniformly by a small amount, and thus recoup larger revenues, while still inducing participation of all types. It turns out that because U() is nondecreasing from (10), we only need to require the participation constraint to be satis ed at the lower end point =. Thus, the manager s participation constraint can be replaced by U() = 0 U() free (; given). (17) The investor s modi ed problem, therefore denoted by (L ), can be written as an optimal control problem: subject to (10), (11), (12), (16) and (17). max ev (U(); (); ) (18) fu();;()g Equilibrium contract and optimal investment We adopt a standard strategy to solve problem (L ). First, we ignore the monotonicity conditions (11) and (12), and solve the so called relaxed problem (L ). We then show that the solution derived in this way also satis es the monotonicity conditions and thus solves problem (L ). This will be true, for example, if the distribution of satis es assumption 1. The following proposition summarizes the solution to the investor s problem (L ). The formal proof can be found in the appendix. Proposition 2 The solution to (L ), denoted {(^),U L (); R(), I D ()}, is given by () = [1 F ()] f() + [1 F ()]) ; (19) Z 0 Z 0 U() = U() + p(i(); )d + ()d; (20) R() = p(i(); )[ + ] + L U() + (1 + r) [b I()] ; (21) and p I = (1 + r) + p I p p II p I (1 + r) p I [1 F ()]. (22) f() The rst expression gives the incremental reward for success. From the fact that F () 1, it is evident that this term is negative for all host types below 13

14 the upper end point <. This suggests that the most e cient type will be o ered a standard debt contract (with a non-contingent repayment schedule) while those below the upper end point will be subjected to a payment structure that includes a penalty in the event of success. To discern this rather counterintuitive outcome, one must understand the combination of agency problems in this situation. Under pure moral hazard, but without adverse selection, by di erentiating the total surplus, p(i(); ) + L (1 + r)i, the rst-order-condition with respect to is given by where di=d = compatibility condition on I: [p I (I; ) (1 + r)] di d = 0 (23) p I (I; )=[p II (I; )( + )] is obtained from the incentive p I (I; )[ + ] (1 + r) = 0. (24) Since the last two equations must in equilibrium hold simultaneously and given the fact that di=d > 0, it follows straightforwardly that the optimal contract in the absence of private information (but only moral hazard) will set to zero and commit the rm to a xed repayment schedule. When the rm has private information, however, a high type host, for whom success is most likely, will have an incentive to intentionally misreport her type as lower in order to earn information rent. And the incentive to misreport will be particularly more pronounced those states of the world in which the project is successful. To counter this incentive, the investor optimally lowers the attractiveness of contracts designed for low rms when L is the realized cash ow. This is achieved by requiring the rm to repay a higher amount in the event of success than without success. The e ect of this is to di erentially disadvantage high type rms who attempt to mimic the low types. 10 The second expression gives the amount of rent that is optimally ceded to the host in order to induce truthtelling. As in many screening models of this kind, this rent is positive for all types except the least e cient. The third expression gives the base repayment amount. It says that the rm will repay the investor an amount that is equal to the expected total surplus less the sum of her rent and cost of investment. In essence, the investor appropriates all 10 This result parallels Lewis and Sappington s (1997) argument that in a dynamic agency setting, early success should be optimally purnished to limit the agent s incentive to understate his ability 14

15 the proceeds from the project and then not only compensates the rm for her expenditure on investment, but also a ords her rent. The last expression gives the condition for optimal investment. Compared with equation (24), which gives the corresponding investment equation in the absence of private information, we can conclude that the second term on the RHS of (22) represents the e ect of the rm s private information. This term is strictly positive for all host types below the upper end point <. Since intense investment increases the informational rent of the host, the investor nds it optimal to lower investment in order to reduce this costly informational rent. An important consideration is the role played by the opportunity cost of investment, r, in determining the nature of distortion in investment. It is evident that an increase in r has two e ects: On the one hand, the distortion term on the right-hand-side of equation increases in signi cance. In other words, the informational rent that must be paid for any implementable level of investment and the investor s incentive to distort investment is increased following a rise in r. On the other hand, the second term on the left-hand-side of equation (22), becomes more negative and therefore increases in signi cance; that is, the marginal cost of investment increases. It follows that p I (:) must rise when r rises in order to maintain the equality in equation (22). Given the concavity of p(i; ), I must decrease as would be the case under pure moral. In sum, the e ect of a higher level of r on the size of the optimal investment incentives is unambiguously negative. The adverse selection e ects reinforce the moral hazard e ects to increase the distortion in investment. Thus, our model predicts that an increase in the opportunity cost of funds will exacerbate the agency problem. 4.2 Foreign Direct Investment (FDI) The previous section assumed away any direct contributions that the investor might make toward enhancing the success of the technological transfer. In that section the foreign investor was portrayed as a purely passive actor with no role beyond providing liquidity. This assumption was a plausible one given that standard debt contracts are often transacted at arms length. We now consider a situation in which the foreign investor funds the project and in return acquires an interest in the rm through an equity purchase. To make the problem interesting, we assume that the investor purchases su cient amount of stock or shares of the domestic rm as to gain in uence over the rm s broad strategic objectives and organizational direction. Thus, the investor s involvement does not end with the transfer of funds. Additionally, he exercises 15

16 control and makes decisions over the implementation of project. Since = 1, the success probability now becomes p(i; E; ). Endogenizing the investor s participation in the manner described above modi es the model by introducing a moral hazard incentive on the part of the investor. Thus, unlike the previous section, the model here features bilateral moral hazard and adverse selection. There is bilateral moral hazard because the value of the investor s expertise E cannot be ascertained and therefore contracted upon. 11 In addition to eliciting information from the domestic agent and inducing productive technological investment, the optimal contract must not only compensate the investor for providing funds. It must also reward the investor for the organizational and managerial skills that he brings to the project. The foregoing implies that any contract has to take into account both the investor s own incentive provision as well as the manager s incentive provision. 12 One can anticipate a tension in providing incentives to both parties, for providing the domestic agent with a ective incentive to undertake investment necessarily entails an incentive loss on the part of the investor. The assumption that the investor needs to be given adequate incentives to deliver appropriate level of expertise may on the surface look uncommon, but it is nevertheless a reasonable one. Since expertise is personally costly for the investor to supply, an investor with interest in a portfolio of projects may have an incentive to spend too little time on a particular project if his expected return from that project is too low. Let f(^); v(^)g^!2[!,!] be the revelation mechanism, where plays the dual role of motivating both the domestic agent and the investor whilst v elicits information. The investor s payo is obtained as: V ((); v(); ) = [p(i(); E(); ) + L + v()](1 ()) v(). (25) Ex ante, the investor does not know the manager s productivity. She sets 11 We believe that this is the rst study that has endogenized the investor s intervention in this form. Choi (2001) is, perhaps, a notable exception in this regard. As we show below, however, Choi does not incorporate adverse selection in the same context. 12 The framework we follow below is analogous to that employed in the literature on team production. In this setting one player designs the compensation scheme as a Stackelberg leader, and thereafter participates in the production as a team member (McAfee and McMillan, 1991). 16

17 f(); v()g to maximize her expected payo : E[V (U(:); (); )] = Z (1 ()) p(i(); E(); ) + L + v() v() df (). (26) The domestic agent s payo is comprised of the value of her equity stake less the cost of undertaking investment and is given by U(; ) = [p(i(); E(); ) + L ]() + (1 + r)[v I()] (27) Let U() = U(; ). Then using equation (27), we can rewrite the investor s objective function in terms of U() as ev (U(:); (); ) = Z p(i(); E(); )[ + L Hence, the screening problem under FDI, [D ] is as follows: U(:) (1 + r)i df (). (28) max () ev (U(:); (); ) (29) subject to U() 0 8 2, (30) du() d = p (I(); E(); )() 8 2, (31) I 2 arg max p(i(); E(); )() + (1 + r)[v I] (32) and E 2 arg max(1 ())p(i(); E(); ) E v, (33) where equation (30) is the participation constraint and equations (31) - (33) are the incentive compatibility conditions. 13 The main di erence between the investor s problem [D ] and problem [L ] in the preceding section is that the former has the additional constraint (33), which states that the investor s input into the project E must be chosen optimally given any contract. 13 This is a standard way of setting up problems characterized by doubled sided moral hazard. For more detailed overview of the double moral hazard literature, see, for example, Demski and Sappington (1991), Bhattacharya and Lafontaine (1995) and Kim and Wang (1998). 17

18 The optimal levels of investment and expertise given f(^); v(^)g, are de ned by equations (32) and (33). The two parties will undertake activity such that given f(^); v(^)g, the marginal bene t from the activity just equals the marginal bene t from undertaking the activity. In view of the inherent friction between motivating the domestic agent on the one hand, and inducing appropriate level of participation by the investor on the other, an interesting question is how the choice of I and E will change in response to changes in the sharing parameter ; that is, what are the signs of di/ d and de/ d? From the total di erentiation of the rst order conditions associated with (32) and (33) we obtain and di d = p I (I(); E(); ) > 0, (34) p II (I(); E(); )() de d = p E (I(); E(); )df () < 0. (35) (1 ())p EE (I(); E(); ) Equation (34) says that an increase in the manager s stake in the project will induce an increase in the level of investment. This makes intuitive sense: An increase in will increase the manager s share of the expected project s proceeds, thereby increasing the rate of return on investment. Equation (35) says that the amount of attention devoted to the project by the investor decreases with in equilibrium. Again the intuition for this is obvious. A higher level of reduces the investor s share of the project s cash ow, thereby decreasing the investor s return from labouring diligently for rm. Proposition 3 below characterizes the optimal level of investment in this setting. The construction of the proof follows along the lines of proposition 2. The appendix contains the derivations of the optimal investment rule. Proposition 3 Suppose that assumptions 1 and 2 hold. necessary for optimal level of investment under FDI will be given by Then the condition p I = (1 + r) + 1 () + 2 () [1 F ()] f() (36) where 1 () and 2 () are de ned as 1 () = [p E 1] p II [p E ] 2 (1 + r) (37) p EE [p I ] 2 18

19 and " 2 () = p I p p II p I p E p II p EE pe p I 2 (1 + r)# (1 + r) p I (38) respectively. Expression (36) says that the optimal level of investment is determined by equating the marginal bene ts (the left-hand-side) with the marginal costs (the sum of the terms on the right-hand-side). To better understand the implication of this condition, it is necessary to recall that the rst-best level of investment required the maximization of the total surplus which gave the marginal condition p I (1 + r) = 0 for all. Under bilateral moral hazard but without adverse selection, by maximizing the expected total surplus subject to the incentive contracts of I and E, the optimal investment rule is given by p I (1 + r) = [p E 1] p II [p E ] 2 (1 + r). (39) p EE [p I ] 2 Thus, the rst term 1 () gives the distortion in investment that derives from bilateral moral hazard. The second term 2 ()[1 F ]=f represents the distortion that arises from the rm s private information. Since both 1 () and 2 ()[1 F ]=f (see the appendix) are all positive, equation (36) leads to the conclusion that two-sided moral hazard and adverse selection reinforce each other as to increase the marginal cost of inducing investment. In this second-best outcome, the optimal level of investment is obtained by replacing the agent whose marginal cost is (1 + r) with a virtual type whose marginal cost is (1 + r) + 1 () + 2 ()[1 F ]=f. It is important to note that 1 () is always positive regardless of the type. Because of the tension between providing incentives to the agent on the one hand and motivating the investor to provide adequate expertise, on the other, the moral hazard problem is never resolved in this model. The magnitude of 2 ()[1 F ]=f is, however, dependent on type. In particular, 2 ()[1 F ]=f is equal to zero for all types at the upper boundary but is strictly positive for all types over the interval [, ). This result follows easily from the fact that F ( ) = 1 and is consistent with the main lesson of standard principal-agent models: to limit the informational rent paid to the agent, the principal optimally induces distortion in activity over some interval. In the context of our model, the level of investment is set below its full information value. This result relies on the agent s utility function satisfying the Spence-Mirlees 19

20 single-crossing property du=d d du=dv d^ = p I (I; ) di d + p E(I; ) de d + p It can be shown that 2 () 1 d (1 + r) d^ 0. (40) du=d di du=dv d. (41) Since di=d > sign du=d = sign 2 (). (42) du=dv Hence, if p I (I; )[di=d] + p E (I; )[de=d] + p 0, then is nonincreasing. In other words, as the agent s reported type increases, she is awarded a reduced stake in the project. This result can be understood by rst considering the special case in which there is only moral hazard and no private information. In this case, the optimal sharing parameter for any given levels of I and E is given by =(1 )=p E =p I (1 + r) = p E + rp E =p I, which implies )] = p E > 0 and is increasing in type. Intuitively, when no adverse selection considerations are present, is used strictly to limit the moral hazard incentive. If this sharing schedule is o ered in a setting characterized by private information, however, the agent with higher absorptive capacity (for whom success is most likely) will have an incentive to pass herself o a lower type since doing so would enable her to earn informational rents. To dampen the agent s incentive to understate her type, the investor lowers the attractiveness of contracts designed for the low type agents. This is done by raising (increasing the agent s equity stake). Higher equity stake for the agent is costly, from her perspective, because it lowers the investor s incentive to provide expertise. Hence, an agent who misreport her type must now undertake more costly investment to make up for the diluted incentives on the part of the investor. 4.3 Comparison of investment incentives under alternative funding mechanisms We now compare and contrast the level of investment that is optimally induced under the two funding mechanisms. In undertaking this exercise, we wish to uncover any qualitative di erences in the nature of optimal incentive provision when the project is nanced through a debt contract than when it is nanced through the issue of equity to the investor. 20

21 The following proposition summarizes the key conclusion. The proof is relegated to the appendix. Proposition 4 Suppose that assumption 1 and 2 holds. Then at the optimum FDI will yield a higher level of investment only when the adverse selection consequence of FDI exceeds the moral hazard e ects.. The logic of this result can be understood by rst supposing that funding is obtained through the medium of FDI, but the investor is precluded from providing any expertise. In this case, the optimal level of investment is given by p I = (1 + r) + 0 2() [1 F ()] f() (43) where 0 2() = p I p p II (1 + r) p I p I. (44) Comparing equation (43) with equation (22), it is straightforward that investment incentives under FDI and debt nance would coincide if there were no moral hazard incentives on the part of the investor. Thus, the distortion term 1 () and the last element of 2 () (which is given by p E p II p EE pe pi 2 (1 + r)) on the right-hand side of equation (36) represent the e ect of FDI (or allowing the investor to exercise operational control over the project). We will call 1 () the bilateral moral hazard e ect of FDI because it captures the incentive loss that occurs on the part of investor whenever the agent is induced to increase investment a little bit. Since this term is strictly positive, equation (36) leads to the conclusion that the investor s intervention beyond providing liquidity may actually increase the cost of achieving any given level of investment, thereby worsening the agency problem. This moral hazard e ect is, however, tempered by p E p II p EE pe pi 2 (1+r), which we will call the adverse selection e ect. Because this term is strictly positive for any positive I and E, equation (36) appears to suggests that the investor s intervention in the project will reduce the marginal cost of undertaking investment that occurs in term of informational rent that must be ceded to the agent. To see this, suppose rst that E=0. In this case, in order to limit the rents earned by the agent, the investor induces distortion in investment by o ering low powered incentives. Accordingly, the informational cost of implementing any given level of investment includes the rents that must be paid to all types higher than. Now suppose that E > 0 so the investor must also be motivated 21

22 to pay attention to the project. In this case, reducing the agent s incentive in a rational attempt to limit information rents will entail an o setting incentive gain on the part of the investor. Consequently, the extraction of informational rents becomes less costly. In sum, whether FDI leads to more investment relative to debt nance depends upon the relative magnitude of the bilateral moral hazard e ect and adverse selection e ect. Only in the special case where the adverse selection e ect is su ciently large vis-a-vis the moral hazard e ect will the agent invest more under FDI than under debt nance. 5 Conclusion This paper developed a framework to illustrate simply the interaction between nancing arrangements and the optimal investment in clean technology. The main point of the analysis is that the media through which nancing is obtained has implications for agency costs and the nature of investment decisions. Relative to debt nance, FDI entails less intense adverse selection but an elevated level of moral hazard problem. Adverse selection is less pronounced here because reducing the agent s incentive in a rational attempt to capture some of the information rents induces an o setting incentive gain on the part of the investor. The moral hazard problem is greater because providing incentives to the agent must necessarily come at the expense of a diminution of the investor s incentives. Hence, if the adverse selection consequences are su ciently signi cant vis-à-vis the moral hazard e ects, FDI will outperform debt nance in terms of the level of technological investment that is optimally induced. These results are obtained through a number of simplifying assumptions, which could be extended in a variety of ways. The model assumes that both the investor and the rm are risk neutral. Risk neutrality for the agent is questionable in this context, however, because agents are unlikely to possess a portfolio of similar projects spread across the developing world. Thus, a deeper exploration of the impact of nancing mechanism might involve an examination of the e ects of risk preferences. A second possible objection to this model is that it assumes that debt contract precludes the investor from exercising control over the rm. In reality, an investor may intervene even under a debt arrangement by way of a commitment to terminate funding if the rm s performance is not satisfactory. These lines of thought are beyond the scope of this study and are left for further research. 22

23 6 Appendix 6.1 Proof of Lemma 1 Equation (7) implies that for any pair of values 0 and in, p(i(); )[ + ()] + L R() + (1 + r)(b() I()) and p(i( 0 ); )[ + ( 0 )] + L R( 0 ) + (1 + r)(w + b( 0 ) I( 0 )) (45) p(i( 0 ); 0 )[ + ( 0 )] + L R( 0 ) + (1 + r)(w + b( 0 ) I( 0 )) p(i(); 0 )[ + ()] + L R() + (1 + r)(w + b() I()). (46) Adding up (45) and (46) yields [() ( 0 )][p(i; ) p(i; 0 )] 0: (47) Therefore, if > 0, () ( 0 ) or equivalently 0 () 0. (48a) Thus, incentive compatibility requires that (:) be a nondecreasing function. This implies that (:) is di erentiable a.e. If a rm type reports type ^ to the nancier, she will get the expected utility as described in (7). If ^ = is the optimal response for type rm, then the following rst-order condition will be satis ed; p(i(); ) 0 () + p (I(); )I 0 ()[ + ()] together with the second-order condition R 0 () + (1 + r)[b 0 () I 0 ()] = 0 (49) p (I(); ) 0 ()I 0 () + p(i(); ) 00 () + p (I(); )I 0 () 0 () +p (I(); )[I 0 ()] 2 [ + ()] R 00 () + (1 + r)[b 00 () I 00 ()] 0 (50) Using the fact that (49) holds identically and taking its second derivative yields p (I(); ) 0 () + p (I(); )I 0 ()[ + ()] 23