Essex Finance Centre Working Paper Series

Transcription

1 Esse Finane Centre Working Paper Series Working Paper No 6: Publi-Private Partnerships as Collaborative Projets: testing the theor on ases from EU and Russia Dmitri Vinogradov and Elena Shadrina Esse Business Shool, Universit of Esse, Wivenhoe Park, Colhester, CO4 3SQ Web site:

2 Publi-Private Partnerships as Collaborative Projets: testing the theor on ases from EU and Russia Dmitri Vinogradov *, Universit of Esse, UK Elena Shadrina, National Researh Universit Higher Shool of Eonomis, Russia Januar 206 ABSTRACT Publi-private partnerships (PPP) allow involvement of private parties in the provision of publi goods. How does this differ from traditional publi prourement? We view PPPs as ollaborative projets with information fritions. Tpial publi prourement ontrats takle the problems of asmmetri information. However, not all projets are ontratible; some are not profitable enough to ensure partiipation of the private partner. This is due, in part, to ostl information verifiation, and in part to the profitabilit requirements of the private part. We demonstrate what speifi features of a partnership an improve feasibilit of projets, and thus both provide a justifiation of PPP as a form of publi good provision, and demonstrate how and whether it differs from prourement. We then analse real life eamples of PPP projets from the perspetive of optimal hoie of ontrats, involvement of both partners, and the features that make these PPP arrangements superior to publi prourement. Kewords: publi-private partnerships, publi finane, ontrating * Corresponding author. dvinog@esse.a.uk 2

3 . Introdution Publi-private partnerships (PPP) are a popular form of provision of publi goods jointl b the state (muniipal) bodies and private entrepreneurs. Man definitions are in use, whose fous ranges from legal ovenants and regulator framework to business risks and soial benefits (see, e.g., a review b Hodge and Greve, 2007). Opinions on the soietal role of PPP range from etremel positive (Lattemann et al., 2009) through moderatel ritial (Regan et al., 20) to epliitl negative ones (Coulson, 2005; Siemiatki, 20; da Cruz et al., 203). From the business organization perspetives, PPPs are onsidered along publi provision of servies, outsouring (ontrating out) and privatization (Joha and Janssen, 200, onsider either publi part, private part or jointl publi and private parties arring the responsibilit for the publi good provision; Stiglitz (2002) views PPP along privatization and ontrating out.). As emphasized b Stiglitz (2002), the main question is what role should be left to the government. In this paper we epliitl stud the role of the publi part in a PPP, and analse under whih onditions a PPP is soiall benefiial, whih provides a uniform framework able to aommodate the variet of views above. Formal analsis of PPP usuall resorts to ontrat theor. One of the issues, rather well studied in the literature, is whether some or all tasks within one, tpiall infrastruture, projet should be delegated to a private partner (Bennett and Iossa, 2006; Martimort and Pouet, 2008; Maskin and Tirole, 2008; Chen and Chiu, 200). Suh an "unbundling" view neglets ontributions b the partners that annot be defined as a part of the projet, for eample, improvements in effiien due to knowledge spillovers through ollaboration. Differentl, this paper approahes PPP from the perspetive of the theor of ontrating for ollaborative projets (Roels et al., 200; Kim and Netessine, 203; Roels, 204) to identif the speifi advantages 3

4 brought b a partnership as ompared to ontrating out and other forms of ollaboration between the publi and the private setor. We view PPP as a form of business organization with two speifi features: onsolidation of private and publi resoures (onsolidated enterprise), and failitation of o-prodution through improved ommuniation and other unontatable benefits. Although we derive the latter propert endogenousl as a ondition for an improvement in the feasibilit of soiall desirable projets, it onvenientl relates to the "partnership" element of the PPP. The interpretation of the term "partnership" varies in the literature (and aross disiplines), still most authors would agree that a partnership involves "the losest possible ooperation" (this definition of a "partnership priniple" is etensivel used in the EU integration poli douments, see e.g. Allen et al., 2005, p.28; Dahl et al., 204), agreement on objetives (Brinkerhoff, 2002) and different degrees of formalit, "ranging from informal, oral understandings to formal agreements" (Erhard and Brigham, 2006, p.4). These three properties suffie to improve feasibilit of projets, as ompared to tpial ontrats in presene of information fritions. Information fritions impl welfare losses due to verifiation osts. Roels et al. (200) stud the problem of ontrating for ollaborative servies under asmmetri information. The derive optimal tpes of ontrats depending on the elastiit of projet outomes to ontributions of the ollaborating parties. An important impliation of their model is that there eists a set of projets that annot be implemented beause the net profit from them does not over the reservation utilit of the servie supplier. Although ontrat design in publi-private projets differs from that in ollaborative servies, partiularl in terms of inentives and liabilities of the parties, the main impliations are the same. Yet, a lose ooperation with aligned partners' objetives, ma redue information ost and lower the reservation utilit, whih makes more 4

5 projets implementable through a onsolidated publi-private enterprise if the partnership priniple holds. PPP are thus justified onl if () a publi good annot be provided b the publi setor alone, (2) the sstem of publi prourement fails to ensure the provision of the publi good b ommissioning it to a private business, and (3) both the publi and the private setors possess omparative advantages in the provision of some resoures indispensable for the deliver of the publi good. The partnership element ensures a redution in the reservation utilit of the private partner and removal of the (part of) verifiation osts. To larif and test the theor, we investigate several ases of real-life PPP projets, with the fous on the ombination of the publi and private resoures, identifiation of omparative advantages of the partners, and the tpes of ontrats hosen to enable the partnership in eah of the ases. We also speif the real-life arrangements that orrespond to the partnership priniple above. Based on the theoretial and ase-stud analsis, our main onlusion is that the "partnership" is indeed an indispensable element that justifies PPP from a soial welfare perspetive. The forms, in whih the partnership priniple is implemented, an var. Institutional arrangements that failitate the projets that private businesses run with the publi setor, eemplif impliit provisions that embod the partnership element. Epliit provisions would inlude speial lauses in the ontrat. Tpiall, suh a lause would speif the role of the publi partner after the prourement stage, for eample in what refers to risk sharing or possible modifiation/variation of the projet speifiation. It follows that in ountries with a developed institutional struture that supports private-publi ooperation, the benefits of the partnership are available to an onsolidated private-publi enterprise, whih is then rightfull seen as a PPP. In others, speial provisions are needed to ensure effiien gains and to provide a lear distintion 5

6 between PPP and other forms of publi-private interation. 2. Consolidation of resoures and partnership Traditionall, publi goods are provided b the state, et this is often ritiized for ineffiienies. The state an eploit the effiien benefits of private business when outsouring the provision of publi goods. With appliation to the publi setor, outsouring is the delegation of publi good provision b the state to an eternal produer (private business). This gives rise to various outsouring strategies, depending on how muh ontrol would the state eerise over the outsoured business. On the one etreme there is outsouring to free ompetitive businesses subjet to no more than usual business regulation as applied to all other non-publi goods produers. Among other forms, this inludes ontrating and privatization, whih an be seen as, respetivel, a sporadi or permanent delegation of the publi good provision to the private setor (note that so far this view does not impl anthing for the terms and onditions of suh a deal between the state and the private business). Another etreme would involve speial regulator provisions suh as prie (tariff) regulation or even regular monitoring and audit of the outsoured businesses. Prie regulation is sometimes seen as a distintive feature that allows distinguishing between publi and private setors (Broadbent and Laughin 2003). However these forms might be ineffiient as well as regulation an limit effiien, whereas monitoring and audit are ostl. A onsolidated publi-private enterprise an onl be justified if it overomes the ineffiienies generated b the above-mentioned forms. In order to define a publi-private partnership, we need to formulate the following partnership priniple: eah part is interested in the overall suess of the enterprise as well as in the suess of the other ounterparts. This 6

7 priniple is onsonant with the definition of partnership b Brinkerhoff (2002) who stresses that partners mutuall agree on their objetives and rationall divide labour on the respetive omparative advantages of eah partner. Our fous on partnership is important beause (a) surprisingl, there is no lear definition of partnership in the PPP literature, although some definitions of PPP epliitl mention true partnership as a riterion, see definition iv, (b) partnership is often onfused with ooperation or ollaboration (partnership priniple is ommonl used in the poli douments on EU integration to denote the losest possible ollaboration), and () onsolidation of resoures does not neessaril impl partnership as defined above. In fat, the eisting PPP literature does not emphasize the term partnership per se, with rare eeptions (Roumboutsos and Chiara 200). A onsolidated publi-private enterprise without a partnership priniple an be justified, for eample, if the state wishes to enjo the benefits that would otherwise be delivered through privatization (i.e. the benefits of a free ompetitive business environment) but for politial or strategi reasons does not wish to privatize. In suh a ase a onsolidation of resoures is possible through a [subsidized] lease agreement but otherwise there is no distintion between suh a onsolidated enterprise and a In orporate finane a partnership is known as an assoiation "to ondut a nonorporate business", as in Erhardt and Brigham (2006, p.4). The elaborate on this definition further to stress that "partnerships ma operate under different degrees of formalit, ranging from informal, oral understandings to formal agreements filed with the seretar of the state in whih the partnership was formed. The major advantage of a partnership is its low ost and ease of formation. The disadvantages are... () unlimited liabilit, (2) limited life of the organization, (3) diffiult transferring ownership, and (4) diffiult raising large amounts of apital... under partnership law, eah partner is liable for the business's debts". This is in line with our "partnership priniple" as unlimited liabilit of both partners reates inentives to are about the abilit of the other partner to meet obligations. Yet when the "other partner" is the state, "unlimited liabilit" does not seem to work the same wa. Therefore we epliitl require that both partners are onerned about the overall suess of the projet. 7

8 usual private business. B ruling out these ases we arrive at the following definition of a PPP: a publi-private partnership is a onsolidated publi-private enterprise that adheres to the partnership priniple. This definition ields most known properties of PPP. Soial importane of PPP follows from the fat that publi funds are engaged, whih an onl be justified if the are used to produe publi goods. A formal legal agreement is needed to establish suh an enterprise and to speif the wa in whih resoures (phsial assets, knowledge or labor) are onsolidated. The partnership priniple ensures risk sharing: all ounterparts have inentives to take on risks whih the an most effetivel manage and delegate the management of other risks to the other parties who have their omparative risk management advantages. Compared to the properties of PPP formulated in the introdution, the above definition onl fails to diretl inorporate the ompetitive hoie of PPP partners b the state and the size and the length of the projet. Whereas the former seems rather an important tehnial proedure that provides the best math of partners to ensure the partnership priniple, the latter does not arise as a pre-requisite for a PPP. We address this issue in the following setion. The above definition of PPP from the business organization perspetive provides a sought preise boundar surrounding PPP and in partiular supports the ommonl aepted view that PPPs are the main alternative to ontrating out and privatization (Hodge and Greve 2007). It shares with other studies the lassifiation aording to the etent that the tasks, risks and responsibilities of former publi servie provision are transferred to the private partner (Koppenjan and Enserink 2009), but this lassifiation is now a natural onsequene of a more general view on the forms of business organization suitable for the provision of publi goods. It also follows that a PPP is desirable when prodution osts are high (this rules out autarh), 8

9 market barriers prevent free entr (this rules out ompetitive private business), and monitoring and audit is ostl or ineffiient (this rules out publi and ontrolled enterprises). PPPs are distinguished from other forms of private setor partiipation in the provision of publi goods b two riteria: onsolidation of resoures and partnership priniple, as summarized in Table 2. This distintion is operational and useful for empirial studies of PPP. B fousing on finanial resoures onl as one dimension and organizational relationship as the other dimension, Hodge and Greve (2007) provide a further distintion between different PPPs. The also refer to Brinkerhoff (2002) to note that the mutualit, i.e. interdependene and equalit in deision-making as well as equal benefits to parties onstitutes another important dimension. The latter broadl orresponds to our partnership priniple. Table 2. A tpolog of forms of publi goods provision based on onsolidation of resoures and partnership priniple. Consolidation of resoures Partnership priniple + + PPP Outsouring on lease, Contrating Contrating Privatization Publi enterprise In the remainder of the paper we fous on the forms of business organization presented in Table 2 in the following order. We first formall show when a ollaboration between the private and the publi setor is desirable, thus ruling out privatization and nationalization. Subsequentl, we fous on the onsolidation of resoures to stud the feasibilit of the publi-private projets under tpial ontratual agreements. Then we turn to the partnership priniple on top of the onsolidation of resoures. This allows us to stud the benefits that an 9

10 arise from the partnership, and more importantl, to formall speif the elements of the partnership priniple and to derive onditions when these are soiall benefiial. We further disuss whether a projet that adheres to the partnership priniple but does not involve a onsolidation of resoures, makes sense from the perspetive of soial welfare. 3. Optimal ontrat The above disussion ontrasts PPP and other tpes of publi good provision that involve private setor, along two dimensions: onsolidation of resoures and partnership. Consolidation of resoures redues the overall ost of their provision, if eah part supplies resoures whih it an obtain at a lower ost than an other ounterpart. On the other hand, onsolidation of resoures requires ollaboration, whih ma be ostl, for eample due to the need to oordinate ations, or to monitor the ations/ontribution of the partner. If the ost of ollaboration does not eeed the savings ahieved through the redution in the overall ost of resoures, onsolidation and ollaboration offer ost-effiien. In this setion we assume that the publi and the private partner ollaborate on a joint projet. Depending on the information and other osts, suh a ollaboration will not alwas be optimal, and hene some projets will be unfeasible. We will further demonstrate the benefits of the partnership that an improve the feasibilit of the projets. Our setup is similar to the basi model of ontrating for ollaborative servies suggested b Roels et al. (200). Their fous is on the o-prodution between the vendor and the buer of the servie. Involvement of the buer improves the value of the servie, whih ats as an inentive for the buer to o-produe. The main issues that arise with regards to servies, are the verifiabilit of the vendor's and the buer's inputs, as well as the unertaint of the value the buer will derive from the servie. With regards to tpial publi goods, their value is rather 0

11 deterministi; in most ases the are linked to material objets (buildings, infrastruture, failities et.). For this reason our fous is mainl on the struture of inentives, whih differs from that implementable in servies. In partiular, inputs are usuall verifiable, et the osts of their verifiation ma differ, impling different tpes of optimal ontratual arrangements for different projets. 3.. Consolidation of resoures Consider a projet that requires two [sets of] inputs denoted b for the ontribution of the publi part and for that of the private part, and generates a (soial) value of V,. The osts of inputs are and respetivel, inreasing in their argument. Denote with U the reservation utilit, i.e. the lowest paoff that the private partner prefers to the outside option of not entering the projet. If the projet is run b the private partner onl, it does not generate enough profit to over U, otherwise we would deem the projet implementable through the private setor. The involvement of the publi part is justified b its abilit to provide resoure at a lower ost than faed b the private partner. Theoretiall, both partners an provide an resoures, although literature usuall underlines effiien gains epeted from a ollaboration of a publi bod with a private partner, suggesting managerial skills as one of the sought private resoures. To provide an eample that informs our model, the ost advantage of the private partner ould refer to labour and material ost, as well as to the ost of intelletual apital and effiient management, all of whih are denoted as a single set of resoures in our setup. One an think of as a single pakage of inputs, and as the total prie of it; in a similar wa, a tpial lot in a publi prourement aution or tender would speif the whole pakage to be purhased, leaving the deision with regards to the resoures atuall needed for its provision, to the bidders. The total ost of

12 providing resoures is distint from the total paoff t to the private partner; the latter also needs to over the reservation utilit and in general depends on the tpe of ontrat hosen. The ost advantage of the publi partner would refer, for eample, to apital resoures (suh as land), knowledge and information, as well as the speifi abilit of the publi partner to redue the "red tape" (ost of ompliane and bureaurati proedures assoiated with the projet). For more disussion, eamples and relevant referenes see Vinogradov et al. (204) Contrat design Parties ontratuall agree to deliver inputs and as optimall hosen for the partiular projet. The publi part an verif the input of the private partner at ost. Similarl, the publi part ma be required to report on the deliver of the agreed input, whih inurs information (reporting) ost. This latter ost ma be assoiated with a need to provide hard evidene in the ourt of law upon a request b the private partner. This setup leaves all information osts with the publi part. Redistributing them between the partners will not affet the final result, as whenever the private partner bears information osts, these are inluded in the remuneration t. If an of the parties provides less resoures than ontratuall agreed, the are subjet to penalties: R for the private partner and G for the publi part. Penalt an be non-peuniar, assoiated with the loss of reputation and foregone future profit opportunities. We assume that R is limited b the value of phsial assets that belong to the private partner (bankrupt value), and G is also bounded from above, G G. The harshest penalt that an be imposed on a publi partner is the ehaustion of their reserves and the dismissal of the relevant publi managers. A reent eample of a penalt paid b a publi partner as a monetar transfer to the private part would be 30m paid b the Norfolk Count Counil to Cor 2

13 Wheelabrator Consortium for abandoning the the Willows ininerator projet at King's Lnn in the UK. 2 In terms of our model, this orresponds to the event "projet not delivered in full", with the publi part delivering less resoures than agreed. The speifi "resoure" in question here is the approval of the projet going ahead and seuring the state funding for it (the main reason for abandoning the projet were 92% loal itizens voting against the projet and the subsequent withdrawal of the 69m government grant). The penalt for the ounil splits into 9m from urrent reserves and m from future savings, impling a longer lasting burden for other loal projet and a reputation damage. The ontrat is designed b hoosing publi and private inputs and, remuneration t, and the verifiation mode. For the latter we introdue binar variables and whih take a value of if it is the publi or the private partner's input respetivel that the parties agree to verif. Insuffiient deliver of the verifiable input triggers penalt santions, hene information dislosure oupled with the penalt is refleted in the liabilities of the parties in a tpial ontrat. It is usuall ostless to inlude liabilities of both parties in a paper doument. Given this, the hoie of and in our model an be seen as establishing firm beliefs of the parties with regards to who of them will atuall be monitored or should redibl report on inputs provided. It is further assumed that the atual value of the projet V, is freel observable b both parties. Three tpes of ontrat are onievable: if either and 0, or 0 2 See, e.g. the BBC report 0f 3 Marh 204: 3

14 and, onl publi or private input respetivel is verified; with 0 neither input is verfied. A ontrat with is suboptimal sine knowing V, and one of the inputs reveals the other input at no ost. The latter emphasises, in partiular, that mistrust between the parties, that results in their beliefs that there is a need to monitor ations of eah other, leads to a suboptimal outome, as in equilibrium less resoures will be provided in total, ielding a lower value of the projet. The paoff to the private part thus depends on the inputs, the output and the verifiation mode: t t,, V,, following: ma,,, V. The publi partner then solves the, t,, V,,,,, () s.t. t,, V,, U, (IR) R G,,, arg ma,,, (IC-P) (IC-G) Here the individual rationalit onstraint (IR) ensures that the private partner is ompensated enough to over the ost of resoures and the reservation utilit, and inentive ompatibilit (IC) onstraints (IC-P and IC-G) ensure that neither of the parties have inentives to deviate from the agreed ontributions: penalties eeed an potential savings from deviations. Proposition. If there eists a mehanism that reveals t U then an optimal ontrat with verifiation of the private input ( 0, ) sets the remuneration to the 4

15 private partner at t,, V,0, 0 t if if,. PROOF: Setting t U ensures IR is met (binding) in the optimum. IC-P is met if R ma 0 sine is known, inreases in, and 0. Sine is unknown we an use inequalit U 0 to set R t to meet IC-P. The resulting paoff, if the ontrat is violated, is then t t t 0. QED Even if is verifable, the ost funtion and the reservation utilit U are not. For this reason, the above proposition assumes that t U is known. This requires a prie revelation mehanism when designing the ontrat, for eample a bidding/aution or a tendering phase. The penalt on the publi partner in this ase does not pla an role, as it is the input of the private part that is verified. This tpe of ontrats is input-ontingent (IC-ontrat) and enompasses various agreements where the private part is paid either on the hourl basis or eatl for the speified ontribution to the overall projet, et not onditioned on the suess of the projet as a whole. Proposition 2. If there eists a mehanism that reveals t U then an optimal ontrat that obliges the publi partner to dislose and verif information about its inputs (, 0), sets the remuneration to the private partner at t, V 0 t t G if if if and V and V, V V,,,, and penalt G on the publi partner at G G 0. 5

16 PROOF: If V V, then neessaril and V V, then neessaril. If and. For these two ases the proof is the same as for proposition. As the private partner has no inentives to provide more than, the equilibrium provision is. In equilibrium. For this reason setting V t then neessaril V as the publi part has inentives to redue osts t, if V, as the private partner provides onl and V V, is optimal. If, in equilibrium. The input of the private part annot be verified but transferring penalt to the private part is the heapest mehanism to ensure the private part is onvined the publi part has been penalised. For the same reasons as in the proof of proposition, the level of the penalt should be set at G 0, but the ost funtion annot be redibl reported to the private part, therefore G is set at the highest possible level to ensure that IC-G is met. QED In our model, there is no speifi mehanism to determine the optimal penalt on the publi partner. Paing the private partner t ompensates for foregone profits, independent of how muh resoures were atuall delivered, however this ma be insuffiient to meet IC-G. The transfer of G to the private part is the heapest available mehanism to ensure that the private part is onvined the penalt was atuall imposed. This tpe of ontrats is output-ontingent (OC-ontrat), and as follows from the proposition, it is indispensable that if the publi part onditions paoffs to the private side on the suess of the overall projet, the publi part itself should be aountable and provide redible information about its inputs. This would also require that all inputs of the publi part are learl speified in the ontrat. In both ases with input verifiation, the ontrat presumes a fee that is fied at the moment of signing the ontrat: the 6

17 amount t U is the agreed ompensation for the inputs, plus the premium to over the reservation utilit. Proposition 3. A ontrat that does not require verifiation of inputs speifies the paoff to the private partner b a bonus rate b 0 and a fied fee of s U : t bv s. PROOF: If inputs are not verified, parties fae double moral hazard, in whih ase the optimal (seond-best) ontrat is as above, see, e.g., Bhattahara and Lafontaine (995). It onl remains to establish that s U. Neither s, nor U are known to the publi part, both are a part of remuneration to the private part, determined at the negotiation stage. In the limit, with zero private ontribution and zero outome, the paoff to the private part is eatl the surharge (premium) it requires for taking part in the ollaborative projet, even if no further effort is required. In the fied-fee ontrats suh a surharge equals to U (see propositions and 2) and in the PB-ontrat it equals to s. As both denote the same partiipation premium, whih does not depend on the tpe of the ontrat, we obtain s U. QED The ontrat in the last proposition is performane-based (PB-ontrat). Roels et al. (200) disuss properties of the optimal bonus rate b in the ontet of o-prodution in servies. In the ontet of publi-private projets, the emphasis is on the possibilit to link the remuneration to the private part to the soial value generated b the projet. Conession is one eample: the private part operates an infrastruture objet and derives profit from paments b the end users. In this ase, the higher the soial value of the objet, the more intensivel it is used, and the higher is the paoff to the private part. 7

18 3.3. Choie of ontrat and feasibilit A projet is feasible if parties agree to implement it b entering one of the above ontratual agreements. This requires that the soial value V generated eeeds osts (inluding verifiation osts) and the reservation utilit of the private partner. The hoie of the optimal ontrat is based on the omparison of the total surplus the generate. To do this, from this point on assume that V,,, idential to Roels et al., 200) and that and (this assumption is (this assumes that the ost of input verifiation is linear in the input, and it does not ost a penn to reveal that one of the parties did not ontribute anthing). With this in mind, for fied parameters and that determine the tpe of the ontrat, re-write (), substituting for t, V,, U to obtain,, ma, U (2) This holds for both tpes of fied-fee ontrats. Maimisation of this Cobb-Douglas-tpe funtion is a standard eerise, with first-order onditions ensuring V and V for the optimal ombination of outputs and FF. This allows one to re-write the objetive funtion as,, U,, where, and, are optimall hosen for the given tpe of ontrat, (asterisk dropped to improve readabilit). From here one straightforwardl obtains that,0 0,,0/ 0, 0, /, 0. In this notation,0 if and onl if FF FF is the optimal publi input in the OC ontrat, and 0, in the IC-ontrat (similarl for 0, and,0 ). Ratio 8

19 9 0, /,0 measures the effet of the ontrat hoie on the amount of publi resoure ontrated (a requirement to verif input makes it more epensive than without verifiation, for whih reason, eteris paribus, 0,,0 ); similarl for input. It follows that if resoures of both parties have idential provision and verifiation osts, it is the sensitivit of the projet to these resoures that determines the tpe of the ontrat: OC-ontrat is preferred to IC-ontrat if and onl if. This is beause, for eample, hoosing a ontrat that inurs verifiation ost for resoure redues the optimal amount of this resoure that will be ontrated for the projet (due to higher ost), whih affets the final value of the projet, et the impat is smaller if the projet has lower sensitivit to this resoure. The following proposition establishes the result more generall. Proposition 4. The fied-fee ontrat ontingent on output (OC) dominates the fied-fee ontrat ontingent on the private part's inputs (IC) if and onl if or, equivalentl, ln ln. PROOF: Standard maimisation of the profit generated b the Cobb-Douglas prodution tehnolog with fator osts as above, ields fator demands, and., As disussed above, omparing the values of the objetive funtion for ontrats OC

20 and IC is equivalent to the omparison of,0/ 0, and 0, /,0. Substitution ields the first result. B taking logarithms of both sides and re-arranging, we obtain the equivalent representation. QED It follows that the deisive riteria for the hoie between OC and IC ontrats are the fator elastiities and, and the ost of information verifiation relative to the fator osts, / and /. For eample, a omple bridge would inur higher ost of resoures (per unit) than a road beause a bridge would involve man rather unique solutions, while onstruting a road involves a repetition of rather standard approahes. At the same time, it is easier to verif the qualit of the bridge and the resoures atuall used b parties, as it is one loalised objet, while it is more epensive to monitor the atual provision of resoures at eah kilometer of the road. For these reasons, the / ratio is epeted to be lower for the bridge and higher for the road. The higher the sensitivit of the projet to the private input, or the higher is the relative verifiation ost for the private input, the more likel is the OC-ontrat to dominate the IC-arrangement. The following proposition ompares the benefits from fied-fee ontrats with those from the PB-ontrat. Proposition 4. Performane-based ontrat dominates fied-fee ontrats if and onl if b b ma, PROOF: The optimisation task differs from that under fied-fee ontrats as the 20

21 2 objetive funtion of the publi part s b does not depend on the osts of provision of, and therefore the publi part hooses for an given level of private input. The private part separatel maimises profit U s b for an given level of publi input. The first-order onditions thus turn into b and b respetivel. The value of the objetive funtion in the optimum is thus s b PB,0 0, and the fator demands are:, PB b b and. PB b b This ields, 0,0 s b b b PB whih is to be ompared with., U FF Sine U s b proposition 3, for the OC-ontrat we obtain,0,0 0 FF PB if and onl if, b b b whih delivers the result. Similarl for the IC-ontrat. QED

22 B the above proposition, performane-based ontrats with the bonus rate b offer maimum advantages as this maimises the term b b on the left-hand side. In Roels et al. (200), PB-ontrats are found optimal for projets with, for whih the bonus rate should be optimall set at around 50%. For, the two results oinide, et our result is not based on the optimal bonus rate. Instead, it laims that the PB-ontrat is more likel to dominate IC and OC ontrats, if the optimal bonus rate is lose to b ; other rates an be optimal for projets with different and, et these would be less likel to dominate IC and OC ontrats. The onession eample above stresses that the bonus rate in publi-private projets is effetivel determined b the stream of inome generated b the projet over ears. In this ase our result stresses that the private part's share in this stream should depend on the sensitivit of the projet outome. The effet of a variation in b on the optimalit of performane-based ontrat is highlighted in figure. In partiular, an inrease in the bonus rate, although improving inentives for the private partner, does not neessaril impl a better overall performane of the projet as ompared to the same projet run under a fied-fee ontrat. Figure. Comparative statis for optimal PB-ontrat: bonus rate effet. For projets 22

23 ,, the dark area is the set of parameters V,, for whih a performane-based ontrat with b 0.5,0. 3 and 0. 7 dominates fied-fee ontrats if prodution osts are 0. and information osts are 0.. Figure 2. Comparative statis for optimal PB-ontrat: osts effet. For projets V,, the dark area is the set of parameters,, for whih a performane-based ontrat with b 0. 5 dominates fied-fee ontrats if prodution osts are 0.,0.0 and 0. 2, and information osts are 0.. Figure 3. Comparative statis for optimal PB-ontrat: information osts effet. For projets V,, the dark area is the set of parameters,, for whih a performane-based ontrat with b 0. 5 dominates fied-fee ontrats if prodution osts are 0. and information osts are 0., and Similarl, figures 2 and 3 show the effets of prodution and information osts on the 23

24 optimalit of the performane-based ontrat. With higher information osts, the performane-based ontrat is more likel to dominate the fied-fee ontrats, whih is intuitivel lear beause the ost of verifiation/reporting is a welfare loss that an be avoided with a performane-based ontrat. Interestingl, the heaper the prodution, the larger the set of projets for whih a performane-based ontrat dominates fied-fee ontrats (for 0, the performane-based ontrat alwas dominates). This is due to the effet of the relative information ost: a derease in the prodution ost makes verifiation/reporting relativel more epensive, and hene fied-fee ontrats less attrative Role for partnership Not all projets are feasible even if ontrats are optimall designed. If the total surplus from the optimal ontrat is below U then it is impossible to ontrat the private part, and thus the projet is deemed infeasible. The set of these projets is shown in Fig. Fig. Infeasible projets. Note that these projets an still be soiall desirable. Figure 4. Infeasible projets. For projets V,, the dark area is the set of parameters,, for whih the maimum profit generated through optimal ontrats does not suffie to over the reservation utilit of the private partner U. 24

25 Two properties are remarkable in figure 4: () a redution in the reservation utilit shrinks the set of infeasible projets, and (2) ost redution has a similar effet. In the figure, the bonus rate is set at b 0. 5, onsistent with the optimal level for performane-based ontrats in Roels et al. (200). As this is lose to the optimal rate, the feasibilit of the projets via ontrating annot be improved b hanging the bonus. If partnerships offer advantages in terms of a redution of information osts, prodution osts, or the reservation utilit, this will improve feasibilit of projets. For eample, the publi partner an provide guarantees to improve the hanes of the private partner to obtain finane from the finanial setor. These guarantees ma be impliit, as the private partner is paid b the publi one, et the ma improve the reditworthiness of the private partner, who would be able to obtain funding at a lower ost. This an, however, work in the opposite diretion, if the publi partner itself suffers from an insuffiient funding and is thus regarded as unreliable. As another eample, in 2000, in the UK HM Treasur and the private businesses (with the publi partner owning 5% of the enterprise) established jointl Partnerships UK, the PPP aimed at a provision of onsulting to the publi and private parties that onsider forming a PPP themselves. Partnerships UK dissolved in 20, with its ativities replaed in part b the Treasur (through its infrastruture projets department, Infrastruture UK) and another PPP, Loal Partnerships, aimed at assisting the projets with loal governments. A help of this tpe in the formation of new businesses, eemplifies et another ontribution of the publi setor that does not epliitl enter ontratual agreements, et has an effet on the reservation utilit, and hene on the feasibilit of the projets. 25

26 The omparative statis in figure 4 an be formalized b onsidering a redution of the reporting/verifiation ost, PPP, or and a derease in the reservation utilit of PPP the private partner, U PPP U arising through an intensive ollaboration with the publi partner. Denote with j the profit resulting from a projet ontrated through a fied-fee ( j FF ) or a performane-based ( j PB ) ontrat, and PPP j profit from a similar projet within a similar ontrat but with speial provisions ensuring the above redution in osts and reservation utilit. Proposition 5. If the partnership ensures that either of the following holds: () U PPP PPP PPP U, (2), or (3), then there eist projets, unfeasible under standard ontratual arrangements but feasible under the partnership: PPP, : ma U, : ma U. j j PPP PROOF: It is eas to hek that and U PPP U ensure PPP ma ma for an and. The rest follows from ma ma U : all PPP j j j j elements of the first set are also elements of the seond one, but not vie versa.the proof follows from ma ma PPP U PPP U, whih holds for an and. j j The reservation utilit an be viewed as the sunk ost plus a premium (the differene between the atual remuneration and the atual osts to provide the required input) that the private part requires to take part in the projet. The sunk ost, for eample, would inlude the startup ost, the appliation ost, the ost of overoming bureaurati obstales, et. (not to mention briber). The premium is a ompensation for potential inonvenienies of dealing with the publi bod (e.g. due to internal ineffiienies, delas in deisions, need to adhere to multiple requirements and odes of best pratie, et., all of whih is usuall referred to as red tape). In a 26

27 partnership, the sunk ost dereases beause of the ost advantage of the publi partner, and the premium dereases beause of the improved inentives of the publi partner. Not the least, one of the reasons for the premium to derease is in the potential reputation gains for the private partner. The redution in information ost an be due to o-prodution: the loser the ollaboration, the more obvious is the input of the ounterpart. Co-prodution is often assoiated with knowledge spillovers, and observed, for eample, in knowledge-intensive servies (Doroshenko and Vinogradov, 20), et parties should be willing to losel o-produe and be open to ehange knowledge. These two onditions are less unlikel to be ontratible. Besides, publi bodies are known to be involved in the o-prodution to a lesser degree than private firms, at least in what regards knowledge-intensive servies (Doroshenko et al., 203). In the above we onsidered three tpes of ontrats: () two with a fied-fee to the private partner, onditioned either on the suessful implementation of the projet, or on the atual input if verifiable, and (2) with a performane-based remuneration to the private partner. All of them eist in the private setor and therefore are not speifi to publi-private ooperation. Moreover, the tpe of ontrats alone annot help distinguish publi-private partnership from outsouring (part of) publi good provision through publi prourement. Yet, omplemented with provisions that ensure a redution in the reservation utilit and information and prodution osts, whih are distintive features of partnerships, the are able improve the alloation of resoures. The latter justifies publi-private partnerships from the perspetive of improving soial welfare. 4. Eamples The above three tpes of ollaborative ontrats between the publi and the private 27

28 partners are quite ommon in pratie. This framework eplains eamples of government buildings in the Netherlands (IC-tpe PPP ontrat) and roads onstrution in Finland (onession, PB-tpe ontrat) from the Introdution. Note that due to eonomies of sale, in large projets osts per unit are tpiall lower than in smaller projets. A building eemplifies a rather unique projet of a smaller size than, for eample, a highwa. The input of the private partner an be relativel easil verified: the onstrution site is aessible, usuall in an urban loation, for whih reason performing regular monitoring is not problemati. Arhiteturall, large buildings, espeiall in a entral loation, are rather unique, whih implies relativel high prodution osts per unit of input. Therefore verifiation osts relative to the provision osts are rather low, and PB-ontrats are less likel to be optimal. An IC-ontrat is more likel to be optimal for this tpe of projets, as publi input is less speifi, and the sensitivit of the outome to it is rather low. In ontrast, large-sale projets like motorwas offer opportunities for a redution in osts per unit through the eonom of sale argument, and in partiular through a rather routinized repetition of standard ations at eah segment of the road. However the length of the projet site makes private partner's inputs less observable, suggesting higher input verifiation osts. As predited b the omparative statis in the previous setion, onession is more likel to be optimal in this ase. The same logi works in the ases below. 28

29 4.. Performane-based ontrats in waste management 3 Case : Energ from waste projet in Suffolk, UK Desription: 25-ear, -billion private finane initiative (PFI) waste management ontrat between Suffolk Count Counil and SITA UK. Inludes 85-million ininerator in Great Blakenham (Suffolk, UK), a ombined heat and power (CHP) plant, with the apait to burn up to 269,000 tonnes of residual waste a ear and generate enough eletriit for 30,000 homes. [] Publi part ontribution: The Great Blakenham site (in the ownership of the Suffolk Counil) is on lease to the private partner for a pepperorn rent for the whole duration of the projet, and will revert to the Counil on epir of the ontrat or earlier termination [3]. Minimum guaranteed tonnage (70,000 tonnes a ear). Government grant of about 200 million over 25 ears, to ontribute towards the running osts [2] Private part ontribution: finane, tehnolog, building failities, management. 80 million Ininerator funded b SITA UK; at the end of the ontrat, the failit will be owned b Suffolk Count Counil [2]. Private part outsoured ativities to other subontrators, for eample, arhiteture and design of the building to Grimshaw. Liabilities of the publi part: paments for insuffiient waste deliver; penalt if projet plans withdrawn. Credibilit of penalties is evidened b a similar projet in nearb 3 Soures used: [] Kane A. (204) Great Blakenham ininerator omes online, Resoure, 4 Deember 204, available online at [2] Suffolk Energ From Waste website, [3] Suffolk Count Counil (200) Final Business Case: Appliation for the private finane initiative redits, Department for Environment, Food and Rural Affairs Waste Infrastruture Deliver Programme, Suffolk Counil Cabinet Meeting of 7 September 200, Appendi A, available at [4] Suffolk Count Counil (200) Energ from waste projet, Sutin ommittee meeting of 8 September 203, Evidene Set, available at [5] SITA (202) Eperiene with EfW projet prourement in UK, available at 29

30 Norfolk, initiated under the same onditions (and subjet to the same government regulation), that has been withdrawn b the publi part and resulted in penalties paid to the private partner (see eample in Setion 2.2). Minutes from the ounil s meeting evidene disussions of the logisti shemes to enable meeting targets on waste deliver. The risk of not meeting the waste volume target is onfirmed b estimates that UK s residual waste treatment apait will eeed suppl in 207/8 []. Liabilities of the private part: penalties for broken deadlines, failure to meet standards; business risks, eept for the risk of insuffiient waste deliver. Sensitivit to ontributions: First phase of the projet (plant onstrution) is more sensitive to the ontribution of the private part. The seond phase (waste proessing) is equall sensitive to ontributions of both parties, as energ generation depends on the waste olletion, as well as on the tehnolog implemented and the qualit of materials used in the onstrution of the waste proessing failit. Optimal ontrat: PB-ontrat, beause the sensitivit to both ontributions is rather equal. Information osts are high (e.g. it is impossible to verif tehnolog emploed without speialized knowledge). If the first phase (plant onstrution) was unbundled from the seond phase, a fied-fee output-ontingent ontrat would be optimal for the first phase as it is highl sensitive to the ontribution of the private partner, and the information ost (verifiation of the qualit of the plant) is high. Due to the high information ost at the first stage, bundling it with the seond stage is optimal, as it removes at least one information verifiation stage, and introdues inentives for the operator to ensure the neessar qualit of the failit. Atual ontrat: PB (onession). The private part derives profits from waste proessing fees, depending on the amount of waste proessed, as well as from energ generation. 30

31 The publi part shares profit from energ generation (if eeeds target). Partnership element: lose o-operation, as evidened b minutes of the Counil s Srutin ommittee [4]. Assistane evidened b the reord-breaking short projet approval time. The Counil and the Communit Liaison Group established b the Private Partner, jointl hold meetings with loal ommunit to update on progress and address issues arising. Although identifiation of the onstrution site is usuall on the private partner [5], in this ase the site was suggested b the Counil from the beginning (and aepted b all four bidders at the prourement stage) [3]. The latter redues risk of proposal being rejeted on the basis of suboptimal hoie of site (translates in the redued reservation utilit in the model). Evidene of suess: CHP opened on time and on budget []. Arhitetural award for the building. Civil Engineering award for the overall projet. [2] Suessful operation sheme allowing to aommodate etra waste delivered from Norfolk after losure of several sites there. Average eletriit generation 570 MWh per da (dail data available from this eeeds predited energ generation of 225 MWh per da (onverted from a figure of MWh a ear, [6]). Other remarks: the ownership of the failities is transferred to the publi part after 25 ears; assoiated risks (worn off failit, amortization osts, need in modernization) are on the publi part. Budget foreasting does not go beond 25 ears, and does not onsider these osts. Case 2: Waste proessing in Krasnokamsk, Perm Krai, Russia 4 4 Soures used: [] Musorne Voin: epizod vtoroi ( Waste wars: episode 2 ), Krasnokamskaa zvezda, 29/06/20, available online: [2] Bumatika offiial webpage [3] V Permi zarabotal kompleks po avtomatiheskomu razdeleniu musora ( Waste separation failit launhed in Perm, e-newspaper NeSekretno, 5//203; available online [4] Glushkov A. (203) Est takaa professia musor sortirovat ( The profession to sort waste ), e-newspaper 3

32 Desription: Solid household waste (SHW) proessing at Bekratinsk site (Krasnokamsk, Perm Krai, Russia). From 2008, the Bekratinsk landfill site is under the management of the private partner «Bumatika», seleted through a ompetitive prourement proedure. The initial ontrat is for 0 ears. The waste management sheme inludes a waste sorting failit launhed in 2008 (the onl one in Perm Krai at that time) and full operating to da, with a apait of 30 tonnes of SHW, apable of separating 8 frations of SHW. Estimated investment value RUR 38,5m (approimatel m in 2008). Total planned investment RUR 00m. The size of the site 6,7ha. Apart from the sorting failit, the operator introdued modern tehnologies and equipment of waste ompation and ground densifiation, reduing water pollutions and simplifing future re-ultivation [, 2]. In November 203, Bumatika launhed the seond sorting failit, whih is available to loal universit researhers for eperimental studies in waste proessing. Altogether, the projet inludes olletion, transportation, storage, sorting and proessing of SHW [3, 4]. Publi part ontribution: land (pepperorn rent), monopol waste proessing right in the Krasnokamsk area (where the site is loated), red-tape redution, monitoring of the ProPerm.ru, 06//203, available online: ; [5] Bumatika: kompleksni podhod v ekologii ( Bumatika: a omple approah in eolog, Business Class, 27/06/20, available online: [6] Innovatsionnaa Bumatika ( Innovative Bumatika, Business Class, 0/0/20, available online: [7] Romanova A. (202) «Musorni Veter ( Wind of waste ), Krasnokamskaia Zvezda, available online: ; [8] Bkova A. (204) «Musornoe delo» («The Waste Business»), Zvezda Newspaper, 7/06/204, available online: [9] Shishkina O. (204) Nedohodne othod ( Unprofitable waste ), Metragi, 0(43), 52-55; [0] Pusk mosorosortirovohnogo zavoda v Kransokamske otkladvaetsia ( The launh of the waste sorting failit in Kranokamsk delaed, Business Class, 9/06/2008, available online: [] Efremova (20) Prokurature ne udalos osporit ( The proseutor s offie failed to hallenge ), Kommersant-Prikamie, /0/20, available online: [2] Zusman E. (203) Sudebnaa Praktika po konessiam v regionah ( The ourt pratie on onessions in the regions ), presentation, VEGAS-LEX, Mosow, available online [3] Suhanov V, and M. Strugov (205) «Moskva poshla v othod» («Mosow is going to waste»), Kommersant-Prikamie, 07/03/205, available online: 32