WELFARE ORIENTED PRICING VS. PROFIT ORIENTED PRICING AND WELFARE GAIN OF PPP Alain Bonnafous* Laboratoire d'Economie des Transports (CNRS, Université Lyon 2, ENTPE) 9 March 2016 Ground address: LET, ISH, 14 av Berthelot, 69363 Lyon, France. * Professor (CNRS-LET), email: alain.bonnafous@let.ish-lyon.cnrs.fr ABSTRACT This paper deals with the alternative between the socio-economic or the financial point of view and is focussed on the issue of pricing for PPP infrastructures. With a profit pricing there is a welfare loss for each new project but the need of subsidy is lower and the same public budget could allow a more important investment programme and thus a greater welfare gain than with a welfare pricing. This conjecture is illustrated by empirical simulations. This paradoxical result means that the optimal pricing theory cannot be generalised from one single project to a programme of several projects. Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP 1. INTRODUCTION This paper deals with a research programme from which a previous communication presented during the 10th WCTR (Bonnafous and Jensen, 2005) already stemmed. We need to start with a reminder of the framework and of the main results of this paper because the present communication remains situated in the field of consequences of some mathematical properties of the same relationships between the need of subsidies and some economic parameters. One of the main issues of this reminder, presented in the section 2, concerns the “tyranny” of the financial internal rate of return (IRR) and suggests to consider the question of optimal pricing. Indeed, in the previous communication, both the financial and the socio-economic evaluation of each investment project were based on the same hypothesis of pricing. From the moment the financial IRR appears as a better ranking criterion for a welfare purpose than the socio-economic internal rate of return (ERR), we can wonder whether the optimal pricing with respect to an optimised welfare for each project remains the optimal pricing when we consider a programme of new projects planned under a public financing constraint. More precisely, this paper examines if, for this kind of programme, a profit oriented pricing, with which public funding can be reduced, provides a better welfare level than a welfare oriented pricing. The difference between these two sorts of pricing will be emphasised in the section 3. Some answers to our central question will be stated in section 4, these result being based on numerical simulations. 2 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP 2. REMINDER ON THE OPTIMAL RANKING OF INVESTMENT The previous paper concerned the choice of infrastructure investments and more specifically the optimal ranking of the projects that are to be implemented. The framework was that of Public Private Partnerships, or more generally, that of projects financed both by subsidies and by theirs own revenues such as tolled highways. The optimal investment programme (i.e. which provides the maximum of socioeconomic net present value or NPV) must be defined under a constraint of annual subsidies. This previous communication demonstrated that the optimal ranking was that of decreasing ratio NPV/subsidies. It also proved that the ranking of decreasing socio-economic internal rate of return (IRR) does not provide a better welfare gain than the ranking of decreasing financial IRR. To the contrary, when the financial IRR is higher, the need of subsidy is lower and the same public budget allows for a more important investment programme. The lower the level of this budget, the higher the interest to follow the financial IRR ranking of the project. To confirm these points, we used a set of 17 toll highway projects for which homogeneous economical data were available (Cf. Appendix 1). These 17 French projects were in competition in the early 90s. Most of them have been carried out with a small contribution from public money owing to the former French financing system, that of crossing subsidies between tolled highways (before an european reform made that impossible)). In this exercise our working hypothesis is a mechanism of PPP in which for each project a subsidy can help the operator to reach a target IRR. Subsidizing rates have been calculated from the relationships of our PPP financing model (presented in the appendix 2), taking 8% as the target IRR for the operator. 3 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP We assume that there exists a budget constraint F the first year, this constraint increasing by 2,5 % yearly. To make clear the role of the budget constraint, we have changed its value between 0<F> and 2<F>, <F> being the mean subsidy required by all the projects (<F>= 472 Meuros). For a given order of the projects, the value of F determines the rhythm of completion of projects, since each project needs to draw on this budget an amount depending on available public subsidies each year. In figure 1, we compare the total socioeconomic NPV (NPVse) returned by the 17 highway projects for different rankings, as a function of F. More precisely, we plot the % of gain obtained by choosing the specified ranking criterion compared to the total NPV returned by using the ERR as the ranking criterion. We use two other criteria: the financial IRR and the « output », defined as O(i) = NPV(i)/sub(i), where sub(i) represents the amount of public subsidies required by this project to obtain the targeted IRR (8%, as assumed above). Note that the curve « output » is very close to the curve labeled "Optimum" (triangles) which was obtained by a numerical algorithm, inspired from the Monte Carlo method and providing the maximum of total NPV. 4 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Figure 1: Comparison of the total socioeconomic NPV returned by the 17 highway projects for different rankings (We plot the % increase of the program NVP compared to that of the ERR ranking) Nevertheless, what is important for the present paper is the result which confirms in Figure 1 that the financial IRR is a better ranking criterion than the ERR, and this is the truer the tighter the budget constraint. When this constraint is lower than some value (close to one third of the average subsidy needed for a project), the NPV output of the program obtained with the ERR ranking is frankly disastrous when compared with the pure financial IRR order. This paradoxical result is attributable to the financial constraint. Indeed, when the public financing capacity is very low, an investment programme which takes little 5 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP account of the financial profitability of the project will very quickly consume the public funds available, resulting in a slower rate of entry into service. Thus, this "budget effect" explains the paradox in terms of the length of network brought into service according to the order of programme implementation. The present paper deals with the same kind of alternative between the socioeconomic or the financial point of view but focussed on the issue of pricing. On a tolled road, for instance, the pricing depends on whether its objective is that of maximising the profit of the operator or that of maximising the welfare. On the basis of the previous results on the optimal ranking, we can wonder what is the optimal pricing for a programme of new projects. With a profit oriented pricing there is a welfare loss for each new project but the need of subsidy is lower. We have thus to consider the budget effect: the profit pricing improves the financial IRR and, under the same public budget constraint, could allow for a more important investment programme and thus a greater welfare gain than under the welfare pricing. In order to explore this issue we need to formalise the relationships between the selected price, on one side, and both the welfare gain and the need for subsidies, on the other side. 3. WELFARE PRICING VS. PROFIT PRICING The contrast between welfare and profit is a classical economic problem (for instance Mills, 1995). The optimal charge of a transport infrastructure use is based on the social marginal cost (Boiteux, 1956 1960). More precisely, the « price-relevant marginal cost » equals the social marginal cost minus the private marginal cost. This pricing is usually opposed to the optimal pricing for the operator revenue. Facing to 6 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP these two facets, the Ramsey-Boiteux pricing reconciles each other. In the present paper we do not use exactly the same distinction: to the profit oriented pricing we will oppose a welfare oriented pricing based on the scarcity of public financing capacities. In order to facilitate the analysis, we consider a standard case in which the demand d is a linear function of the toll p and takes the form: d d0 p (1) Were d0 and are fixed parameters. The users surplus S is therefore given by: S d0 ( p)2 2 ( 2) And the gross revenue for the operator is given by: R d 0 p p2 (3) Thus the classical distinction proposed by Jules Dupuit (1844 and 1949)) between two pricing strategies appears clearly in Figure 2: - either the public authority decides to favour the interest of the users and consequently to choose a free use of the infrastructure, - or the public authority decides to use the road pricing in order to minimise the public expenses. 7 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP In the firs case the user surplus reaches its maximum value (d02/2). In the second, the revenue reaches its maximum value (d02/4). Note that the optimal price for the operator provides the internalisation of only half of the maximal user surplus because we are in the case of a single price. Figure 2 : Stylised demand function, users surplus and revenue d S R d0 Smax = d02/2 Rmax = d02/4 O PRmax = d0/2 Pmax= d0/ P d = d0 – .p Users Surplus S Operator benefit R In order to simplify the calculations, we shall assume that variable costs are negligible and that only the constant costs are to be covered by the operator. This hypothesis means that the maximum of the gross revenue and the maximum of the net revenue are procured by the same value of p (in our stylised case this value is equal to d0/2). Thus, regarding any optimal pricing problem, we shall only consider the variable elements of the objective functions. These functions are therefore R(p) 8 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP for the operator and U=R(p)+S(p) for the public authority. We abstract from the function U the existence of a constant of derivation as it is not relevant for the subsequent analysis. Nevertheless, when the projects are financed both by subsidies and by theirs users, any additional unit of revenue is also one saved unit of public expense and consequently one saved unit of tax. From the moment it is an accepted fact that the tax system is not neutral and not optimal, one unit of saved public expense has a value higher than one. Thus, as a measure of an element of the collective wealth, the relevant amount of the revenue is not R but .R were is a scarcity coefficient of public funds. Thus, the welfare objective function becomes: U R S ( 4) or alternatively: U d 02 1 ( 1) d 0 p p 2 ( ) 2 2 (5) In the very particular case in which the tax system is perfectly neutral (such that = 1) equation (5) becomes (6): U d 02 2 p 2b 2 (6) 9 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP When = 1, it is quite natural for the function U to be a decreasing function of p because for any additional value op p there is a loss of S from which only a part is internalized under the form of additional R which the operator must receive. Thus, in this particular case, the welfare function is maximized for p = 0. In the more general case, the derivative dU/dp can be written as : 1 U ' ( 1) d 0 2 ( ) p 2 ( 7) Thus the welfare function (5) reaches its maximum for: pU max 1 1 2 d0 2 (8) It may be easily seen that this value of pUmax varies from 0 to pRmax, the value of p for which the gross revenue for the operator is maximum. Equation (8) shows furthermore that the higher the scarcity coefficient of public funds, the nearer of pRmax.the value of pUmax. When tends to infinity pUmax tends to pRmax i.e. to d0/2. 10 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Figure 3: The utility function in case of scarcity coefficient for public funds S, R U O PUmax PRmax P Users Surplus (S) U=S+.R ( = 1) Gross Revenue (R) U=S+.R ( > 1) If we consider the linear demand function as an approximate figure of the real function and if we retain for the value used by the French Administration (presently: 1.3), we obtain: pUmax = 0,375.pRmax These relative positions of the two optimal prices will be useful later on for our exercise of simulation. 11 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP 4. THE OPTIMAL PRICING OF A PROJECT IS NOT ALWAYS OPTIMAL FOR A PROGRAMME OF PROJECTS Let us go back to the initial question, that of whether the optimal pricing in the sense of the welfare maximisation for each project remains optimal once we consider the pricing of a programme of new projects planned under the public financing constraint. In order to prove that this proposition is wrong we only need to find one case of the programme for which a profit oriented pricing provides a higher overall socioeconomic NPV than a welfare oriented pricing. For a perfect evidence, we should compare programmes with the genuine values of p optimising respectively the welfare gain and the revenue for each project. Today we do not know these values. Nevertheless, we know that for the set of 17 projects of toll highway projects the level of p used for the ex ante evaluation was equal to the mean toll on the French network during the early 90’. Thus, in order to select a reasonable rough estimate of values for the two sort of pricing we can assume that this mean toll was somewhere between pUmax and pRmax and even that it was on a median position pm. In this case, from equations (3) and (7) it follows that, ift R was boosted by 16 %, it would lead to a loss of U of 8 %. Symmetrically, a reduction of tolls R of 16 %, would increase U of 8 %. Simulations were made of investment programmes depending on what we call the welfare oriented pricing (pm - pm) or what we call the profit oriented pricing (pm + pm). We used the set of projects already well-tested. For each pricing hypothesis, subsidy rates and socio-economic NPV’s were calculated owing to equations of our PPP financing model (Cf. Appendix 2). 12 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Thus, for each pricing alternative we know the ratio NPV/Subsidy of each project and it is easy to establish the optimal ranking based on this ratio. It was assumed that the projects were subject to a budget constraint under which the annual public financing was restricted respectively to 150, 300, 450 and 600 Meuro during the first year of the programme. Subsequently the financing was assumed to increase by 2.5 per cent a year. Each of these programmes was assumed to run for 15 years. The results are given in the table 1 below. The paradoxical facet of these results is that the use of a profit oriented pricing improve significantly the welfare gain of the programme even though each project provides a lower NPV than with a welfare oriented pricing. That is of course a consequence of the contents of the two programmes, the most effective being composed of a greater number of projects. Our intuition is thus confirmed: we find the same kind of results than those which were reminded in section 2. There is also the same kind of explanation in the sense that the rate of subsidy of a PPP project is strictly depending of the financial IRR which is itself depending of the pricing system. Nevertheless, the more radical explanation derives of the properties of the function between the need for subsidy and the additional IRR which the operator must receive (Cf. Appendix 2): the concavity of this function means that the first differences between the targeted IRR and the IRR of the operation can be extremely costly, particularly in the case of projects with a low intrinsic IRR. Thus any improvement of this intrinsic IRR by a profit oriented pricing decreases the need of subsidy and increases the number of implemented projects. 13 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Table 1: Comparison of the total NPV returned by the optimal programme regarding two different pricing systems Budget Subsidies and Welfare oriented Profit oriented constraint NPV pricing pricing 150 Total programme 20227 34093 15,0 22,0 26199 39168 8,5 14,2 27865 43184 5,9 8,7 30829 45832 4,5 6,9 NPV Total NPV / Subsidy 300 Total programme NPV Total NPV / Subsidy 450 Total programme NPV Total NPV / Subsidy 600 Total programme NPV Total NPV / Subsidy There is nevertheless a slight difference with the previous results. We found that the pure financial IRR is a better ranking criterion than the ERR and that this is the truer the tighter the budget constraint. We do not observe on the results represented in 14 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Figure 4 the same role of the tightness of the constraint: the pricing effect becomes significant as soon as the budget constraint is active. Figure 4:Comparison of the total socio-economic NPV returned by highways for alternative pricing systems 25,0 20,0 15,0 NPV/Subsidy 10,0 5,0 0,0 150 300 450 600 Budget constraint Welfare oriented pricing Profit oriented pricing Several points still remain to be examined. For instance, because our results concern only a particular case, it would be useful to specify the general conditions under which an optimal programme based on the profit oriented pricing could provide a more important welfare gain than an optimal programme based on welfare pricing. 5. A POLICY ORIENTED CONCLUSION Intuitively, the optimal welfare pricing of each project would be the one that provides the highest welfare gain when the project is integrated in a larger programme, that of several projects. From the moment there is at least one case in which this rule does not work, this case is an evidence that the intuitive rule is wrong. From an academic point of view, it is clear that the results above open some 15 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP new questions for both the optimal pricing theory and the PPP’s programmes assessment. From a policy oriented point of view we can summarise our result in a few words: In a programme of PPPs under the public budget constraint, it could be of the general interest to allow the operator to determine the infrastructure pricing. REFERENCES Boiteux, M. (1956), Sur la gestion des monopoles publics astreints à l’équilibre budgétaire, Econometrica, Vol. 24. Boiteux, M. (1960), Peak-Load Pricing, Journal of Business, April, Vol. 33, n°2, pp. 157-179. Bonnafous, A. (2002), Les infrastructures de transport et la logique financière du partenariat public - privés : quelques paradoxes Revue Française d’Economie, vol. 17, n°1. Bonnafous, A., Jensen, P. (2005), Ranking Transport Projects by their Socioeconomic Value or Financial Interest rate of return ?, Transport policy. Vol.12, Issue2. pp. 131-136 (Award of the best communication of the 10th WCTR, Istambul, July.2004). Dupuit, J. (1844), De la mesure de l’utilité des travaux publics, Annales des Ponts et Chaussées, n° 116. Dupuit, J. (1849). De l’influence des péages sur l’utilité des voies de communication, Annales des Ponts et Chaussées, n°207. Hayashi Y. & Morisugi H. (2000), « International comparison of background concept and methodology of transportation project appraisal », Transport Policy, 7(1), pp.73-88. Mills, G. Welfare and Profit Divergence for a Tolled Link in a Road Network, Journal of Transport Economics and Policy, 1995, vol.29:2, pp. 137-46. 16 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Pons , D., Johenson, F. (2006), “L’influence du tarif des infrastructures sur la rentabilité sociale d’un programme”, document de travail, available on: http://halshs.archives-ouvertes.fr/halshs-00117036 APPENDIX 1: MAIN CHARACTERISTICS OF THE 17 PROJECTS Figure 5 : ERR and IRR for 17 projects used for the simulations 12,00% 10,00% TRI 8,00% 6,00% 4,00% 2,00% 0,00% 0% 10% 20% 30% 40% 50% 60% 70% 80% TRE APPENDIX 2: THE PPP’s FINANCING MODEL Suppose a standard case (Bonnafous, 2002) in which the capital cost C is incurred at an annual rate c = C/d between the dates –d and 0. When the project comes into use at time 0, the annual profit rate (revenues less operating costs) takes the form (a + bt). We now introduce this further notation: is a discount rate which may be used to calculate the NPV, 0 is that value of which makes the NPV (of the unsubsidised project) equal to zero – in other words, 0 is the internal rate of return, 17 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP is the subsidy rate, expressed as a percentage of c (and hence also of C), is the increase in the IRR that results from the subsidy rate – that is to say, for a subsidy rate , the IRR becomes 0 + . If we recognise all benefits and costs up to a terminal date T, then: 0 T NPV - c.e .dt (a b.t).e αt .dt -αt d ( 9) 0 To simplify the analysis – with little empirical effect – now set the terminal date to infinity. Then (9) becomes NPV 1 b c(1 - ed ) a (10) The IRR 0 of the unsubsidised project is therefore given by: c(1 - e 0 d ) a b 0 0 (11) When the subsidy is applied, equation (3) becomes (1 - )c(1 - e ( 0 ) d ) a b 0 0 (12) If we think of the situation as one in which we want to find the subsidy rate that yields a specified IRR of ( 0 + ), then (12) may be written as τ 1 a( 0 ) b c( 0 )(e ( 0 ) d 1) (13) Here, is expressed as a function of six variables. However, these are not all independent, because 0 depends on a, b, c and d. Furthermore, equation (12) implies that if = 0, then = 0. The function (13), as a function of 0 (or ), is concave. This property is illustrated by the family of curves shown in Figure 6, which is drawn for the numerical case given by c = 100, b = 1, d = 5 together with alternative values of a chosen so that as a increases from one value to the next, 0 increases by 0.4 percentage points. Each curve corresponds to a particular value for a, and shows how the required subsidy τ increases as we increase the IRR from 0 to the target IRR, ( 0 + ). 18 Bonnafous, Welfare Oriented Pricing vs Profit Oriented Pricing and Welfare Gain of PPP Figure 6 : The relationship between the subsidy rate and the IRR 19