Credit availability for SMEs in the crisis: which role for the European Investment Bank Group?

Alessandro Fedele Andrea Mantovani April 16, 2012 Francesco Liucci

Abstract In this paper we consider a moral hazard problem between a bank and a creditworthy rm which needs funding. We rst study under which conditions the rm does not obtain the loan. We then determine whether and how the intervention of an external nancial institution can facilitate the access to credit. In particular, we focus on the European Investment Bank Group (EIBG), which provides (i) specic credit lines to help banks that nance small and medium-sized enterprises (SMEs) and (ii) guarantees for portfolios of SMEs loans. We show that especially during crises the EIBG intervention allows to totally overcome the credit rationing problem. Keywords: credit crunch, moral hazard, EIBG, SMEs JEL Classication: G01, D82, D21

We thank the seminar audience at 6th EUROFRAME Conference on Economic Policy Issues in the European Union, London 2009, for precious comments and insightful discussion on a previous version of the paper. The usual disclaimer applies. Department of Economics, University of Brescia, Via S. Faustino 74/b, 25122, Brescia, Italy. Email: [email protected]. Department of Economics, University of Bologna, Strada Maggiore 45, 40125, Bologna, Italy, and Institut dEconomia de Barcelona (IEB), C/ Tinent Coronel Valenzuela, 1-11, 08034, Barcelona, Spain. Email: [email protected] Department of Economics and International Business, Oxford Brookes University, OX33 1HX, Oxford, United Kingdom. Email: [email protected].

Introduction

The severe crisis of the world nancial system has been generated by a multiplicity of factors, among which, the failure of nancial regulation systems. Banks became reluctant to grant loans and many creditworthy investors were denied the access to credit due to a sudden tightening of the conditions required to obtain a loan. Since then, many economists and policy-makers have agreed on the necessity to enhance the scope of nancial regulation and improve the provision of liquidity through traditional and newly created instruments. The aim of our paper is to pinpoint which role supranational nancial institutions can play in their attempt to solve, or at least mitigate, information problems between lenders and creditworthy borrowers. This has a fundamental importance especially in periods of crisis, where trust between economic agents has to be re-established. In particular, we will target credit rationing problems faced by Small and Medium Enterprises (SMEs henceforth). Notwithstanding the relevance of the topic, relative little attention has been given to the specic analysis of external nancial institutions that reduce information asymmetries in credit markets. Our contribution lies at the crossroads between two streams of research. On the one hand, the issue of government support in credit provision with nancial informational frictions has been studied by DeMeza and Webb (1987), Innes (1991) and Hellmann and Stiglitz (2000), inter alii. Williamson (1994) and Wang and Williamson (1998) consider public intervention which eliminates information asymmetries caused by costly state verication.1 On the other hand, the positive eect of external mediators has been examined by Myerson (1986) and Forges (1986) in communication games. Fedele and Mantovani (2008) show that delegation of hidden tasks to a third agent can mitigate the informational asymmetry between a start-up entrepreneur and a bank. We adopt a simple moral hazard model in which a wealthless rm applies for a bank loan to invest in two alternative risky projects: the good project yields a positive expected value, but it requires a more intense eort by the rm. The bad project is less demanding but its expected value is lower than the good one, and in most cases negative. The bank can not verify the rms choice and this generates the information asymmetry in the form of a moral hazard (Holmstrom, 1979). In the rst part of our paper we consider two alternative scenarios: before (without) or after (with) the intervention of the external nancial institution. In the rst scenario, the rm applies for funds and it is induced to implement the good project only in exchange for a suciently high informational rent. We nd a wide interval region in which the loan is not granted and a credit crunch occurs. In the second scenario, an external nancial institution favours the lending process by an appropriate combinaArping et al. (2010) consider the support by the state in the form of credit guarantees and loan subsidies to entrepreneurs who are capital constrained and subject to moral hazard.
1

tion of two instruments: (i) co-funding of the investment project and (ii) provision of guarantees the bank receives if the project fails. We demonstrate that the intervention of the external agent drastically reduces the occurrence of the credit crunch. In the second part we evaluate the eectiveness of the tools adopted by the nancial institution. We show that the credit crunch area can be further reduced by increasing the percentage of co-funding and the amount of guarantee. Yet, only if the spread between the cost of raising capital for the bank as compared to that of the nancial institution is suciently high, the credit crunch can be completely eliminated. This is exactly what happened since the intensication of the nancial crisis in September 2009.2 Therefore, under the theoretical conditions of our model, the intervention of the external nancial institution becomes extremely eective, in that it allows the bank to always nance a creditworthy project. More generally speaking, it may help the economic recovery by stimulating the credit market. As mentioned before, we are particularly interested in interventions targeted to SMEs. In the US, the Small Business Administration (SBA) actively engages in provision of direct loans and bank loan guarantees. The situation is more ambiguous in Europe. The European Monetary Union lacks a common institution capable to ease possible shortage of liquidity in the European markets. These structural decits have been increasingly obvious in the aftermath of the nancial crisis. The creation of the European Systemic Risk Council and of the European System of Financial Supervisors was intended to improve bank regulation and supervision across borders, but we are looking for proper supranational credit institutions which may actively help rms to achieve credit. This is why we turn our attention to the European Investment Bank Group, which is particularly qualied to play the role of the external nancial institution that we have in mind. The potential impact of the European Investment Bank Group deserves additional attention if one considers that the recent nancial turmoil has raised the cost of raising capital, one of the crucial factors in our paper. This contributed to deteriorate the attractiveness of risky investments, such as those proposed by SMEs. The EIBG, which is highly rated, introduced anti-crisis measures and became a stable anchor for banks wishing to nance rms navigating in stormy waters. The remainder of the paper is organized as follows. In the next section we describe the functions of the European Investment Bank Group. The formal model is laid out in Section 3. Sections 4 and 5 consider the two scenarios described above, respectively before and after the intervention of the nancial institution. In Section 6 we characterize the importance of such an intervention and suggest how to eliminate the credit crunch. Section 7 concludes the paper.
The interest rate spreads on government bonds of many EU countries have dramatically risen (Sgherri and Zoli, 2009) due to a higher risk aversion in international nancial markets (Schuknecht, von Hagen and Wolswijk, 2009).
2

The European Investment Bank Group

The European Investment Bank Group (EIBG henceforth) was established in 2000 to bring the European Investment Bank (EIB) and the European Investment Fund (EIF) under the same umbrella. The EIB was founded under the Treaty of Rome as the EUs long-term lending institution. As dened by the Treaty on the Functioning of the European Union, pursuant to Article 309, the task of the EIB is to contribute, by having recourse to the capital market and utilizing its own resources, to the balanced and steady development of the internal market in the interest of the Union.3 The EIF, instead, was created in 1994 to support the development of SMEs. Its main operations regard venture capital and guaranteeing loans. Complementing EIBs product oering, the EIF promotes "the implementation of European Community policies, notably in the eld of entrepreneurship, technology, innovation and regional development".4 The relationship between the EIB and the EIF aims at encouraging a productive sharing of expertise in support of SMEs. The EIBG as a whole is engaged into two main activities: the disbursement or partial-disbursement (co-funding henceforth) of loans, and the provision of guarantees. Let us look at their specic functioning, as this will represent the theoretical ground that we use to model the intervention of the nancial institution. The rst activity is particularly related to the general commitment of the EIB towards the integration, development, and economic and social cohesion of the EU Member States, on a non-prot maximizing basis.5 This special feature allows the EIB to lend almost at the cost of borrowing. Since it does not distribute dividends to its shareholders, any euro earned is retained in order to cover the EIBs expenses, and to refund its operations and programs. The EIBs capital is subscribed by EU member States. The EIB raises resources through bond-issues and other debt instruments (EIB Statute, 2009). The EIBs "rm shareholder support, [. . . ] strong capital base, exceptional asset quality, conservative risk management and [. . . ] sound funding strategy" constitute the reasons for its constant triple-A credit rating, assigned by Moodys, Standard and Poors, and Fitch.6 The EIB provides dierent kinds of loan. The individual loans are addressed to projects with a total investment cost higher than 25 million euros. They are available for promoters in both public and private sectors, including banks, and nanced directly by the EIB. They are not the subject of our paper. We focus instead on intermediated loans, which are credit lines granted to intermediary banks (or other nancial institutions) to facilitate the nancing of projects proposed by SMEs.7 The rm has to
Ocial Journal of the European Union, C 115 of 9.5.2008: Consolidated Version of the Treaty on the Functioning of the European Union. 4 EIF website: "About Us". 5 EIB website: "About the EIB". 6 EIB website: "About the EIB". 7 More precisely, these credit lines are directed to SMEs with total maximum cost lower than 25
3

apply directly to the intermediary institutions to which the credit lines are oered. Requirements for application may vary according to the respective intermediary, as "the conditions of nancing (interest rate, grace period, loan period etc.) are determined by the respective EIB partner bank".8 Nonetheless, one of the mission of the EIB is to stimulate competition among the intermediaries so as to pass on to SMEs the generous credit conditions oered by the EIB. The shortage of liquidity caused by the recent crisis augmented the quest for partnership by commercial banks, thus increasing the bargaining power of the EIB and actually giving vigor to competition in the banking sector. This will be a crucial assumption of our theoretical model. The second activity, the provision of guarantee instruments, is carried out by the EIF. Two main products are worth mentioning, the Credit Enhancement-Securitisation and the Guarantees/Counter-Guarantees for portfolios of micro-credits, SME loans or leases. The former, that will be explicitly considered in our model, concerns operations conducted by using guarantees provided by the EIF, which is directly involved in the SMEs loans securitisation transaction. The latter relates to dierent European framework programmes, such as the "Competitiveness and Innovation Framework Programme" (CIP 2007-2013), managed by the EIF, on behalf of the European Commission, in supporting SMEs activities and development.9 Regarding the securitisation transaction, the guarantee support is of fundamental importance for our paper. Under the Basel II capital requirements, the banks ability to raise funds is conditional to the soundness of their credit exposures. In such a scenario, they can increase their lending capacity by transfering part of their portfolio risk through securitisation transactions. In other words, securitisation works like a loan insurance, as the credit exposure attached to every loan is transferred from a bank to an investor, by issuing notes (called asset backed-securities) on the capital market. The bank must pay a fee to the investor in exchange for such protection, but in case of default it is the investor which repays the loss. The credit rating of the securitised loans is connected with the one assigned to the nancial institution covering that specic part of the risk. In case of support by the EIF, even in presence of SMEs moral hazard, the rating of the covered loans is enhanced, due to the zero risk-weighting assigned to assets guaranteed by the EIF.10 The benecial role of the EIBG in the securitization transactions is supported by dierent studies (e.g. Robinson 2009, Janda 2008, IRManagementdienste GmbH 2007, European Commission 2006 and AMTE Final Report 2006). However, and this will be very important for the policy implications of our paper, the EIBG set limits to the provision of loans and guarantee support. As for the cofunding, the intervention covers up to 50% of the initial investment needs. It has been
million euros (small-sized enterprises) or up to 50 million euros (medium-sized enterprises). 8 EIB website: "Intermediated Loans". 9 EIF website, Guarantees & Securitisation. 10 EIF website, Credit Enhancement.

argued that the imposition of a ceiling aims at avoiding opportunistic behaviors by the partner banks.11 In order to preserve the achievement of the EIBGs objectives, in particular the support of SMEs, the EIB Corporate Operational Plan (2009) species conditions and constraints in relation with its activities. With regard to the selection of the partner bank, the EIBs evaluation concerns an overall assessment of its credit worthiness in conjunction with its historical experience in the SMEs sector. In the contractual arrangement between the EIB and the partner bank, the amount of funding granted by the EU institution is calculated on the basis of the intermediate banks capability to originate loans without the EIBs credit facilities. The EIB controls the reliability of the budgeted amount set by the partner bank and it disburses the credit lines accordingly. Moreover, the EIBG intervention creates a positive surplus through the leverage eect. Every partner bank is obliged to pass on to the nal beneciary the nancial benet generated by the EIBs involvement, "For each euro provided by the EIB the partner bank undertakes to lend at least two to SMEs, so creating a leverage eect".12 The EIB estimated the leverage factor between its nancing and the total value of the investment as a range between 2 and 10 times.13 This is consistent with Robinson (2009), who found that during the period 2004-2006 the leveraging factor generated by the EIBs lending activity in the EU ranged between 4.8 and 23.9.

The model

We consider a wealthless rm which needs funds to implement a business plan based on two alternative risky projects. Project i, i = H, L, yields a per-unit-of-investment return a with probability p (ei ) (0, 1) and 0 otherwise; ei is a nontransferable eort, whose per-unit-of-investment disutility has a monetary equivalent equal to c (ei ). We assume a linear specication for both the eort disutility and the success probability: c (ei ) = ci and p (ei ) = pi . Moreover, we let cH = c > 0 = cL and pH > pL . Project H entails a bigger eort cost than project L, but succeeds with higher probability. The nancial contracting game begins when the rm applies for a bank loan, whose amount is normalized to one. The timing of the game is as follows: at t = 0, rst the bank designs a loan proposal, then the rm decides whether to accept it or not. If the loan is granted, in the time span between t = 0 and t = 1, the rm chooses between projects H and L, i.e. it decides whether to exert an extra eort cost c cH cL = c, thereby increasing the success probability by p pH pL , or to shirk. This choice is assumed to be hidden, thus giving rise to a moral hazard problem between the lender
Lelarge et al. (2008) present evidence from the French government loan guarantee program (Sofaris). While they argue that the loan guarantees are eective in helping young French rms to nd nance and to grow, they also nd that guaranteed rms are more likely to adopt risky strategies, and that these rms le more often for bankruptcy. 12 EIB website: "EIB Loans for SMEs". 13 EIB website: "EIB Directors approve anti-crisis measures for 2009-2010".
11

and the borrower. At t = 1 returns accrue and the rm repays the bank. If the rm does not apply for the loan, or the loan is not granted, the project cannot be implemented and therefore the rms outside option is zero. An additional possibility for the bank is to monitor the behavior of the rm by paying a xed cost equal to m.14 Monitoring is assumed to make the agents behavior perfectly observable. The choice of the bad project is detected with probability one, in which case a non-monetary punishment can be imposed to the rm. Hence, the moral hazard issue disappears as the rm has no incentive to shirk if the punishment is suciently high. We study the eect of an external nancial institution on credit availability by considering two alternative scenarios. In the rst one, the rm resorts to a local bank (LB, henceforth) to obtain the unit of capital necessary to start the project. In the second one, the loan application is directed to an intermediary bank (IB) supported by an external nancial institution (FI). The IB is required to concede just a fraction [0, 1] of the unit of capital since it receives from the FI a monetary amount (1 ) which must be used for nancing the rm. Moreover, the FI provides an additional guarantee W to insure the bank against the projects failure. The players of our game are supposed to be risk-neutral. Furthermore, banks operate in a perfectly competitive environment. We opt for this assumption because one of the objectives of the EIBG, our specic case study, is to stimulate competition in the banking sector, as partner banks should be induced to pass on to the nal beneciary the favorable credit conditions that they enjoy.15 One may object that the LB bank is not subject to such a pressure. Yet, the hypothesis of the same degree of competition for both representative banks allows us to isolate the eect of the FI on credit availability without changing the market conditions in which banks operate. Moreover, we will show that relaxing the assumption of a perfectly competitive credit market for the LB does not modify our main results. We can now describe the dierent features characterizing the two scenarios. (i) When resorting to the LB, the rm is oered a standard debt contract {r}, where r [1, a] is the gross interest rate due by the rm only in case of success. The rms expected share is ui pi (a r) ci , (1) where ci is the rms nontransferable eort cost. The banks expected surplus is: vi pi r ,
14

(2)

One can think of the time spent by the banks to directly monitor the behavior of the rm, parameter m thereby representing the opportunity cost of not devoting that time to other productive activities. Hauswald and Marquez (2006) assert that the aquisition of information via monitoring implies costly screening and other losses in terms of time, eort and resources employed by the bank. 15 In general, the process of deregulation that took place in the last decades removed many restrictions on competition in the banking sector: an excellent survey is provided by Carletti (2008).

where represents the banks unitary cost of raising money, for example by issuing bonds or collecting deposits. Summing up (1) and (2) gives the expected welfare when project i is implemented: si pi a ci . (3) (ii) When the rm applies to the intermediary, the debt contract is {R}, where R [1, a] is the gross interest rate.Moreover, Ui pi (a R) ci is the rms expected share, Vi pi R + (1 pi ) W (W ) (1 ) pi (W ) is the IBs expected share and Zi (1 ) pi (1 pi ) W + (W ) (1 ) (6) (5) (4)

is the FIs share. Expressions (5) and (6) show the way we model guarantee provision and co-funding by the FI. On one hand, the IB pays a monetary amount (W ) > 0, with > 0, in exchange for the guarantee W which it receives from the FI if the rms project fails (with probability 1 pi ). On the other hand, (1 ) is the amount lent by the FI to the IB to nance the rms project; is the unitary gross remuneration the IB owes to the FI only if the project succeeds (with probability pi ) in return for capital borrowed. Finally, parameter represents the FIs unitary cost of raising money, while (W ) is the IBs unitary cost of raising money, which depends on the amount of guarantee. The functional form of (W ) will be specied in the next subsection. Summing up (4), (5) and (6), we obtain the expected welfare Si when project i is implemented, in presence of the FI: Si pi a ci (W ) (1 ). (7)

3.1

Initial conditions: the region of interest

The initial conditions of the game represent the starting point of our analysis. They are derived from the study of the activity of the EIBG. Based on the constant triple-A credit rating enjoyed by the EIB compared to a worse rating for a local bank, we rst assume that the cost of raising money is lower for the FI than for the LB: < . (8) Second, we specify the functional form for (W ), the IBs unitary cost of raising money. Recall that the guarantee W insures the IB against the credit risk when nancing the rms project. We assume that the risk-neutral IBs utility is indirectly aected by the degree of insurance: the higher W , the lower the risk perceived by risk-averse 8

where f (W ) < 0. Given condition (8) and the functional form for (W ), the FI will generate a higher expected welfare by reducing the cost of nancing: S Si si = (W ) (1 ) > 0. As for the welfare produced by the investment project: Assumption 1 a (a, a) and c (0, c), where a = implies that:16 c+ p ,a= and c = . This pH pL pL (11) (10)

bondholders and/or depositors on IBs assets. They will, in turn, require lower returns, thus reducing the cost (W ). The upper bound on (W ) is for W = 0 and equals , the same cost sustained by the LB, which receives no guarantees. By contrast, the IB, when fully or over insured (W R), incurs the same cost borne by the highly credit rated FI. As a consequence, if W = 0 (W ) = (9) < f (W ) < if 0 < W < R if W R

(i) sH > 0 > sL and (ii) SH > max {SL , 0} .

The expected total surplus is always positive when the rm behaves. On the contrary, the project fails when the rm does not behave and the LB provides the loan. Furthermore, we allow for the possibility that SL 0, i.e. if the loan is granted by the IB the project may have positive NPV even if the rm shirks because of the reduced cost of nancing. This has to do with the general mission of the EIBG, whose top operational priority is to support the investments of SMEs, though this may require a very generous commitment in sectors where the risk of failing is high. As the nancial crisis particularly contracted SMEs access to credit, we do not want to neglect the role that the FI can play in generating a positive surplus even in presence of a low eort by those players so seriously hard-hit by the liquidity constraints in the nancial markets. Finally, regarding the monitoring cost: Assumption 2 m > SH , i.e. the cost of monitoring is higher than the maximum expected welfare. Dierent sources of theoretical literature agree on the inability of lenders to monitor project outcomes at suciently low cost (see Townsend 1979, Williamson 1986, Border and Sobel 1987, and, more recently, Krasa and Villamil 2000, Lacker 2001 and Hvide and Leite 2007).
The condition sH > 0 requires a > a to hold, where a increases with c because the higher c, the lower, ceteris paribus, sH pH a c . A higher a is therefore needed to satisfy the condition.
16

The local bank case

In this section we consider the case in which the external FI does not intervene and only local banks are available in the market. Since the representative LB is competitive, it sets the gross rate r in order to maximize the rms share ui pi (a r) ci , otherwise the rm applies to a rival bank. The maximization is subject to the LBs participation constraint vi pi r 0 and the rms incentive compatibility constraint ui ui , with i = H, L (and i = L, H). This ensures that choosing project i instead of project i gives a higher share to the rm. Since ui is decreasing in r, the LB sets the gross rate at the lowest value with the eect that the IBs participation constraint becomes binding: vi = 0 if r = ri . pi (12)

There are two levels of r which result in the LB breaking even, depending on the rms project choice: rL p and rH p , with rL > rH . If the rm shirks the project L H succeeds with lower probability and the LB is forced to charge a higher rate, otherwise it would incur a loss. The rms incentive compatibility constraint is: uH uL r r a c . p (13)

The rm behaves only if R r: for relatively low values of gross interest rate the negative eect of the eort disutility is outdone by the positive eect of the increased success probability. The following three cases have to be accounted for: rH < rL r, rH r < rL and r < rH < rL . The rst two cases lead to the same result, hence they are treated together through inequality rH r a aN c + . pH p (14)

In this scenario the rm chooses project H according to the above reasoning. It follows that the break-even level is rH , which can be plugged into uH to give: uH (rH ) sH = pH a c . (15)

The rm accepts the LBs proposal rH because it yields the entire surplus sH , whose value is positive under Assumption 1. On the contrary, if rH > r, i.e. a < aN , then the rm chooses project L and the LB breaks even in rL = pL > rH . Substituting rL into uL yields sL , and the LBs proposal is not accepted by the rm given that sL < 0 under Assumption 1. The possibility for the LB to monitor the rm at cost m > 0 is ruled out by Assumption 2. We can summarize the result of our analysis as: 10

Lemma 1 When the rm applies to the local bank, (i) the loan is granted if a aN , in which case the expected welfare is sH ; (ii) the loan is not granted if a < aN and the resulting expected welfare is nil. Proof Directly follows from Assumption 1 and 2 and by inspecting (14). The rms equilibrium expected share (15) is positively aected by the expected return a and negatively by both the cost of raising fund and the eort disutility c. When a < aN the expected gain is low relative to the costs. This is more likely to occur in a period of crisis, in which case the rm does not accept the LBs loan proposal. Figure 1 illustrates the results of Lemma 1 in the parametric space (c, a). Notice that aN divides the interval region of interest a (a, a) in two parts, as its vertical intercept (when c = 0) coincides with that of a, but its slope is steeper. The area a (a, aN ] represents the credit crunch. Figure 1: The Local Bank Case a pL Loan aN No Loan a

pH

In the Appendix we study the case of a monopolistic local bank. We nd that the result of Lemma 1 still holds, the only dierence being a redistribution of the total surplus between the bank and the rm. This conrms that the hypothesis of a local competitive banking sector does not aect our main ndings. 11

The intervention by the external nancial institution

In this section the rm resorts to the IB supported by the external FI which provides co-funding and additional guarantees. Recall that we tailor the intervention of the FI based on our analysis of the EIBG, whose lending activity is ruled by the EIB on a nonprot maximizing basis. In our model this implies that the FIs unitary remuneration is determined through the break-even condition Zi = 0. From (6): Zi = 0 = (1 pi ) W (W ) + i . pi pi (1 ) (16)

There are two break-even levels of , depending on the rms project choice, with L > H . Plugging the above value into the IBs share (5), one gets: Vi |=i pi R (W ) (1 ) . (17)

We rst study the rms choice, starting from the incentive compatibility condition: UH UL R a c r. p (18)

Inequality (18) puts an upper bound on R: as explained in the previous section (see (13)), the rm nds it protable to behave only if R r. Thus, the IB must solve two problems, depending on whether the rm chooses project H or L: max UH pH (a R) c
R

(19)

s.t. VH pH R (W ) (1 ) 0 and R r ; max UL pL (a R)

(20)

s.t. VL pL R (W ) (1 ) 0 and R > r As we assume that the banking sector is perfectly competitive, the rm has full bargaining power when it applies for the loan. Hence, the bank sets R to maximize the rms share, otherwise it loses the client. The IBs participation constraint is binding: Vi |=i = 0 if Ri = (W ) + (1 ) . pi (21)

Now we consider the two cases illustrated above. First, when RH r a aB (W ) + (1 ) c + , pH p (22)

the rm chooses project H and the IB solves problem (19). Plugging RH into UH gives: UH (RH ) = SH = pH a c (W ) (1 ). (23)

The rm receives the entire welfare SH , whose value is positive under Assumption 1. 12

On the contrary, when RH > r, i.e. if a < aB , the rm is induced to select project L as the break-even condition for the bank requires RL > RH . Therefore, the IB solves problem (20). Substituting RL into UL yields: UL (RL ) = SL = pL a (W ) (1 ). The rm ends up with the entire bad welfare SL , where: SL 0 a aL (W ) + (1 ) . pL (25) (24)

If SL 0, the rm accepts the loan proposal by the bank and produces surplus SL . Monitoring is not a viable option, as one can easily verify from Assumption 2. Notice that, from (10): (i) aB represents a rightward parallel shift as compared to aN ; (ii) aL represents a downward parallel shift as compared to a. Moreover, aB aL c c and aB a c c (p)2 ( (W ) + (1 ) ) pL pH (26)

p ( (W ) (1 ) ) pL

(27)

where both c and c belong to the interval (0, c), as it can be easily ascertained. It follows that: Lemma 2 When the rm applies to the intermediary bank supported by the nancial institution, (i) the loan is granted in a aL a aB and the resulting expected welfare is either SH for a aB or SL 0 for aL a < aB ; (ii) the loan is not granted if a < aL a < aB and the expected welfare is nil. Proof Directly follows from Assumption 1 and 2 and by inspecting (22), (25) and (26).

The results of Lemma 2 in the interval region of interest are depicted in Figure 2. After comparing this gure to the local bank case (Figure 1), it becomes evident that the area of credit crunch can be dramatically reduced by the intervention of the external nancial institution. We can therefore claim that: Proposition 1 The intervention of an external nancial institution in support to the lending activity of an intermediary bank mitigates the moral hazard problem between the bank itself and a rm which needs a loan to start a creditworthy project. Proof Directly follows from comparing Lemma 1 and Lemma 2 and by our previous discussion on the respective position of aB and aL in the region of interest.

13

Figure 2: The Intermediary Bank Case a pL Loan with high eort aN aB a Loan with low eort aL No Loan a

pH

The combination of co-funding and guarantees provided by a nonprot external agent, like the FI, represents a useful device to soften the credit crunch problem in periods of recession, where informational and monitoring costs are high relative to projects returns. In the next section we will investigate the mechanism underpinning the positive role of the external nancial institution and suggest how to improve the eectiveness of its action. Turning to the welfare implications, consider the eort choice and loan arrangement in our two scenarios and take into account Figure 2. First of all, the rightward shift of aB implies an expansion of the area in which the rm behaves and the loan is granted. The expected surplus is now SH : (i) in a aN the rm behaves under both contractual environments, therefore the net welfare gain is S; (ii) in aB a < aN the rm behaves only when the FI gives support, hence the net welfare gain is SH . Second, the downward shift of aL gives rise to a new area of surplus SL , which is non-negative for suciently high values of a. In particular, in aL a < aB the net welfare gain is SL . Notwithstanding the low eort exerted by the rm, the deriving negative eect on the success probability is outweighed by the decrease in the cost of nancing induced by the FI. 14

On the elimination of the credit crunch

The initial aim of our analysis was to nd a theoretical road to model the role played by an external agent which intervenes to restore a credit market seriously hit by the crisis. We described the problem faced by a rm which asked for a loan from two dierent types of banks. In the rst case the bank was left alone and we found a relatively big area in which a credit crunch existed. In the second case such an area shrank, thanks to the measures adopted by the FI in support of the intermediary banks lending activity. However, nothing has been said yet regarding the extent of the credit crunch area remaining after the intervention of the FI. Therefore, the mechanism through which the FI intervenes must be further investigated. Consider aB and aL , whose values are both decreasing in W and increasing in .17 The respective benecial rightward shift of aB and downward shift of aL are proportional to the amount of guarantee and co-funding. Therefore, the two instruments are substitutes as they both reduce the cost of raising capital for the IB, as it appears from (17) vis vis (2). Focus, for instance, on the region where a aB , which requires RH r (see condition (22)). Given that RH is increasing in and decreasing in W , whilst r is independent on and W , RH r is more likely to be satised in precence of higher co-funding ( ) and/or higher guarantees (W ). An analogous reasoning holds for a aL , which solves SL 0. The economic intuition is as follows: on one hand the FIs intervention reduces the gross interest rate Ri charged by the IB, thereby enlarging the area where the rm chooses project H. On the other hand, it raises the value of welfare SL , thus increasing the area where the rm accepts the banks proposal. It is worth verifying whether the FIs intervention can completely eliminate the credit crunch area. This occurs as long as c c in Figure 2. We can easily prove that: Lemma 3 A necessary condition for the intervention of the nancial institution to completely eliminate the credit crunch problem is pH + p pH (28)

Proof When the FI fully commits to reduce the intermediary institutions cost of raising funds to the level , it sets = 0 and/or W = R. Consider the values c and c which appear respectively in (26) and (27). After substituting = 0 and/or W = R into c and c, inequality (28) solves c c. Lemma 3 highlights the importance of the cost of nancing. When (28) is not satised, a region of credit crunch still persists even in the presence of full support by the FI ( = 0 and/or W = R). Only when the ratio between the cost of raising
Notice that aB = aN and aL = a if = 1 and W = 0: under such conditions we would return to the LB case represented in Figure 1, in which neither co-funding nor guarantees were available.
17

15

capital with and without the intervention is suciently high, the credit crunch can be eliminated. More exactly, the equality c = c is satised by (W, ) = 1 (1 ) , (29)

In Figure 3 we represent (W ) as a function of and identify the locus of points below (resp. above) which ( c) is negative (resp. positive). The feasible set for c the FIs support is (W ) [, ] and [0, 1]. Notice that (W, )|=0 = and (W, )|=1 = / , which belongs to the interval [, ) given (28). Figure 3 The intervention of the FI (W )

(W, ) No credit crunch M N Possible credit crunch

Figure 3 deserves an additional explanation. Ceteribus paribus, higher levels of co-funding (1 ) and guarantee W are respectively captured by a leftward and a downward shift of the above locus, since (W ) < 0 for (W ) (, ). For example, point N represents a weak intervention in support of the IB and a credit crunch may occur, depending on the values of c and a, because ( c) > 0 (see Figure 2). On the c contrary, a stronger support by the FI is shown in point M , which results from the combined action of higher co-funding and guarantees. This is sucient to eliminate the credit crunch. Finally, notice that (W, ) decreases with , conrming that and

16

W act as substitutes, given that smaller guarantees ( ) are needed to overcome the credit crunch if the FI provides a higher percentage of co-funding ( ). The above discussion can be summarized in: Proposition 2 When the rm applies to the intermediary bank supported by the nancial institution, the credit crunch problem can be completely eliminated only if (i) Lemma 3 holds and (ii) the nancial institution provides a suciently high combination of co-funding and guarantees, the minimum level of which is indicated by (W, ) in (29). The results of Proposition 2 are particularly interesting for the purpose of our analysis. In our model and are the cost of raising funds respectively for the FI and the LB, with the former having the best credit rating, while the latter a substantial worse one. As the interest rate spreads on government bonds of the EU countries have risen dramatically after the intensication of the nancial crisis, the spread between and is nowadays likely to satisfy condition (28). In such a scenario, an appropriate intervention by the FI in support of the intermediary bank becomes very eective given that it completely eliminates the credit crunch problem.

6.1

Which role for the EIBG?

We now evaluate whether the analytical results obtained above apply to the concrete case study that we have in mind. Can the EIBG remove the obstacles in the credit market and unlock access to loans for SMEs? As described in Section 2, the EIBG limits the amount of co-funding to 50% of the project cost and sets an upper bound on guarantees. The aim is to avoid the occurrence of moral hazard between the EIBG and the intermediary bank. In our theoretical model, this translates into [, 1] , with 1/2, and W 0, W , with W (, ), where W denes the minimum cost of raising funds attainable under the constraint W W . Substituting into (W, ) one obtains: (W, ) = 2 , (30)

which is the minimum cost of raising funds necessary to fully eliminate the credit crunch area when co-funding is at its maximum of 50%. Notice that: (W, ) < W ( <) On the contrary: (W, ) W We can therefore state: W + . 2 (32) < W + . 2 (31)

17

Proposition 3 The eectiveness of the EIBG ultimately depends on the spread between and . In particular: (i) if (31) holds, the credit crunch problem can be eliminated only by relaxing the upper constraints on co-funding, [, 1], and/or guarantees, W 0, W ; (ii) if (32) holds instead, such constraints do not prevent the EIBG from overcoming the credit crunch. The result of Proposition 3 indicates that only when the spread is suciently high (interval 32), the EIBG can eliminate the credit crunch by respecting the constraints on co-funding and guarantee support. This reinforces the results of Proposition 2 by conrming that, the more severe the crisis, the more eective the intervention in support of intermediary banks which nance SMEs productive projects.

Conclusion

In this paper we have demonstrated how nonprot top-credit-rated nancial institutions providing additional credit and guarantees to intermediary banks can mitigate informational problems between lenders and borrowers. This is crucial in periods of crisis, where trust between economic actors has to be re-established. Focusing on the European context, we have argued that the European Investment Bank Group is a very good candidate for this task, as it can eectively support intermediary banks to nance creditworthy projects proposed by SMEs. Indeed, we have shown that the intervention of the EIBG turns out to be particularly helpful when the credit market is hit the hardest by the nancial crisis. Recent reforms and new measures taken by the EIBG reinforce the validity of the message conveyed in our contribution. After the dramatic deterioration of the situation on the nancial markets and the expansion of the economic crisis, the EIBG reinforced its skills with anti-crisis measures. In particular, the EIBG managed to provide support for some 105 000 SMEs in 2009 and 115 000 SMEs in 2010.18 Moreover, the new Loan for SMEs product, launched by the EIB in 2008, may nance up to 100% of the investment costs, instead of the usual maximum of 50%. We are convinced that a prompt and lasting recovery has to pass through a widespread feeling of trust, primarily raising the morale of those "small" and innovative entrepreneurs that represent the backbone of the newly established economic and nancial system. We have shown that the EIBG can mitigate the squeeze of the credit availability through its ability to raise funding at low costs. Nonetheless, the activity of this institution should be included in a broader and integrated European framework. In this regard, Gros and Mayer (2010) propose to set up a European Monetary Fund to deal with euro area member countries in nancial diculties, like Greece. They argue that such a fund could allow borrowing at favorable conditions, relying on a portfolio of high-rated
18

EIB website: "The EIB Group supports SMEs" (July 2011).

18

bonds of countries with strong public nances, in order to provide guarantees for a specic issuance of public debt.

Appendix
In this Appendix we consider the case of a monopolistic local bank, which sets the gross rate r in order to maximize its expected share vi pi r , provided that the rm participates and selects project i. The following rms constraints have to be satised: the participation constraint ui pi (a r) ci 0 and the incentive compatibility constraint ui ui , with i = L, H; they ensure that choosing project i gives to the rm a higher share than the outside option and the choice of project i, respectively. c uL by r we obtain r r, where recall that r a . Moreover, Solving uH p ci ui 0 when r ri = a . Since cH > 0 and cL = 0 it is always true that ri > r, pi hence the bank faces a trade-o: its share is increasing in r, but when r > r the rm chooses the bad project, thus reducing total welfare. The bank is then forced to propose a lower interest rate to have the rm selecting project H. We have now all the elements to solve the problem of the bank, i.e. to maximize vi pi r . When setting r, the maximum interest rate that induces the rm to select pL pL project H, the bank gets vH () = sH p c, whilst the rm obtains uH () = p c. r r On the other hand, when setting rL > r, the maximum interest rate that induces the rm to participate by choosing the bad project, the bank ends up with vL (rL ) = sL and the rm with zero. Alternatively, the bank has the possibility to monitor the proper implementation of the good project at total cost m. In this case we know that the moral hazard issue disappears: the banks problem is to choose r to maximize vH (m) pH rm, subject c only to the rms participation constraint uH 0. The solution is r = rM a : pH the bank gets vH (rM ) = sH m and the rm zero. The IB compares its expected share when setting either rL or r or rM . First, setting rL is not protable as vL (rL ) = sL is negative under Assumption 1. Second, Assumption 2 is sucient to rule out monitoring, as it implies that vH (rM ) is lower c than zero. Finally, vH () 0 if a aN r + . pH p We can conclude that the LB proposes the contract {} and that the rm accepts r if a aN , otherwise no loan is proposed: the result of Lemma 1 still holds, with the only dierence that the rm does not appropriate the entire welfare when the loan is granted. Modifying the relative bargaining power between the parties produces only a redistributive eect.

19

References
[1] AMTE The Association des Marchs de Taux en Euro (2006), Supporting SME Financing using Securitization Techniques, AMTE Final Report. [2] Arping, S., Loranth, G. and A. D. Morrison (2010), Public initiatives to support entrepreneurs: Credit guarantees versus co-funding, Journal of Financial Stability, 6(1), 26-35. [3] Border, K. C. and J. Sobel (1987), Samurai Accountant: A Theory of Auditing and Plunder, Review of Economic Studies, 54, 525-540. [4] Carletti, E. (2008), Competition and Regulation in Banking, in A. Boot and A. Thakor (eds.), Handbook in Financial Intermediation, Elsevier, North Holland. [5] De Meza, D. and D. Webb (1987), Too Much Investment: a Problem of Asymmetric Information, The Quarterly Journal of Economics, 102, pp. 281-292. [6] EIB (2009), EIB Statute, Luxembourg EIB. [7] EIB (2009), Corporate Operation Plan 2009-2011, Luxembourg EIB. [8] Fedele, A. and A. Mantovani (2008), Complementarity, Coordination and Credit, Journal of Institutional and Theoretical Economics, 164, pp.230-253. [9] Forges, F. (1986), An approach to communication equilibria, Econometrica, 54, pp. 1375-85. [10] Gros, D. and T. Mayer (2010), Towards a Euro(pean) Monetary Fund, CEPS Policy Brief 202, February. [11] Hausawald, R. and R. Marquez (2006), Competition and Strategic Information Acquisition in Credit Markets, The Review of Financial Studies, 19, pp. 967-1000. [12] Hellmann, T. and J. Stiglitz (2000), Credit and equity rationing in markets with adverse selection. European Economic Review, 44(2), pp. 281304. [13] Holmstrom, B. (1979), Moral Hazard and Observability, Bell Journal of Economics, 13, pp. 7491. [14] Hvide, H. K. and T. Leite (2007), Optimal Debt Contracts under Costly Enforcement, CEPR Discussion Paper, 6040. [15] Innes, R. (1991), Investment and government intervention in credit markets when there is asymmetric information. Journal of Public Economics, 46, pp. 347-381. [16] Janda, K. (2008), Which Government Interventions Are Good in Alleviating Credit Market Failures?, Working Papers IES, 2008/12, Charles University Prague, Faculty of Social Sciences, Institute of Economic Studies. [17] Krasa, S. and A. P. Villamil (2000), Optimal Contracts when Enforcement is a Decision Variable, Econometrica, 68, pp. 119-134. [18] Lacker, J. M. (2001), Collateralized Debt as the Optimal Contract, Review of Economic Dynamics, 4, 842-859. [19] Lelarge, C., D. Sraer and D. Thesmar (2008), Entrepreneurship and credit constraints: Evidence from a French loan guarantee program. NBER Chapters, in: International Dierences in Entrepreneurship, National Bureau of Economic Research, Inc.

20

[20] Myerson, R. (1986), Multistage Games with Communication, Econometrica, 54, pp. 323-58. [21] IR-Managementdienste GmbH (2007), The Leverage Eects of European Cohesion Policy under the Structural Funds, Report Commissioned for the Committee of the Regions (Vienna: IR-Managementdienste GmbH). [22] Robinson, N. (2009), The European Investment Bank: The EUs Neglected Institution, Journal of Common Market Studies, 47 (3), pp. 651-673. [23] Schuknecht, L., von Hagen, J. and G. Wolswijk (2009), European government bond spreads in the current crisis, VoxEU.org, 21 December. [24] Sgherri, S. and E. Zoli (2009), Euro-area sovereign risk during the crisis, VoxEU.org, 17 November. [25] Townsend, R. M. (1979), Optimal Contracts and Competitive Markets with Costly State Verication, Journal of Economic Theory, 21, 265-293. [26] Wang, C. and S. D. Williamson (1998), Debt contracts and nancial intermediation with costly screening, Canadian Journal of Economics, 31, pp. 573595. [27] Williamson, S. D. (1986), Costly Monitoring, Financial Intermediation and Equilibrium Credit Rationing, Journal of Monetary Economics, 18, 159-179. [28] Williamson, S. D. (1994), Do informational frictions justify federal credit programs?, Journal of Money, Credit, and Banking, 26, pp. 523-544.

21