Abstract
When price-cap rules determine the structure of prices for a long period, they suffer a credibility problem and introduce an element of risk especially if a firm’s profits are “too large”. Profit sharing may be seen as a device to pre-determine price adjustments and thus to decrease regulatory risk. We analyse the effects of profit sharing on the incentives to invest, using a real option approach. Absent credibility issues, a well designed profit sharing system may be neutral relative to a pure price cap. With regulatory risk, profit sharing is preferable to a pure price-cap one, if it intervenes for high enough profit levels.
Similar content being viewed by others
Notes
The idea is that for long periods of time (which end with a price review), the regulated price should increase at a rate equal to the difference between the expected inflation rate (the retail price index, RPI) and an exogenously given component (x) which, roughly speaking, represents the expected increase of productivity the firm should attain. By making prices—at least within these periods—insensitive at the margin to firm’s choices, the RPI−x rule appears to eliminate the downward bias and the phenomenon known as “underinvestment”. As Beesley and Littlechild (1989) put it when listing the main arguments in favour of RPI−x, “Because the company has the right to keep whatever profits it can earn during the specified period (and must also absorb any losses), this preserves the incentive to productive efficiency associated with unconstrained profit maximization”.
For instance, in 1995 the UK electricity regulator—realising that previous decisions had been too mild—intervened on prices well before the price review, due for 1999. Other times, the same type of concern has lead to direct political interventions: for instance, a “windfall tax” on profits was introduced in Britain with the 1997 budget, affecting 33 privatised utilities, whose profits had been considered excessively high. See among others Vickers (1993).
Weisman (1993) shows that when price cap rules incorporate an element of profit sharing, price caps may represent a worsening relative to a pure cost based regulation (a notoriously inefficient set-up). Notice however, that the evidence in favour of pure price cap schemes is extremely weak. For instance, Gasmi et al. (1999) show that profit sharing may yield better results. Ai and Sappington (2002) show extremely mixed results in the case of the US telecoms.
Investment irreversibility is a crucial feature of many regulated sectors. For instance, Hausman and Myers (2002) claim that over the 1997–2000 period the revenues of the three major US railroads were inadequate, since the existing regulatory constraint did not take correct account of sunk costs and irreversible investment. A similar point is made by Pindyck (2004), who criticizes the US Telecommunication Act of 1996 as it “ignores the basic fact that sunk costs do matter in decision-making when those costs have yet to be sunk” [p. 12].
An idea of the empirical relevance of irreversible investments in regulated industries can be obtained looking at the so called “stranded costs” i.e at the value of assets that following liberalisation will hardly find a remuneration, but cannot be shifted to a different productive use. According to Lyon and Mayo (2005) these costs can be estimated for the US electricity sector “in the neighborhood of $200 billion”.
On this point, see Dobbs (2004).
This result can be usefully linked to some recent results of the empirical literature. For instance, Crew and Kleindorfer (1996) and the papers they review stress how the presumed superiority of RPI−x rules does not emerge so clearly from experiences in different countries and sectors. A similar point emerges from Gasmi et al. (1999).
The issue of the time consistency of the regulator is also analysed by Levine et al. (2005).
Another related paper is Teisberg (1993), who however, assumes that the regulated firm can abandon the project (a form of reversibility).
As we look at the consequences of different regulatory schemes on a firm’s decisions, assuming that information on costs is symmetric entails no loss of generality. The choice of the optimal price should instead consider asymmetric information, but this is beyond the scope of the current work.
This may be due to external factors affecting technology and input prices. Notice that mark up m t might also be thought of as a function of some endogenous input ν as well. In this way we could introduce into the model the idea that the firm might choose a reversible input (e.g., effort) to minimize its current costs. Formally this would mean that we would have m t = maxν m(ν;t). The quality of the results would remain unchanged (Dixit and Pindyck 1994, Chap.10).
This can be directly relevant in several cases, as the size of most investment projects that utilities face is by and large determined by the size of the area they want to serve. Building a new electric line connecting two nodes of a transmission system to improve its reliability, or a pipeline to sell gas to a new city, are choices that entail an expenditure that can only partially be controlled by the firm.
Notice that one could also have an intervention rule e.g. based on the level of revenues instead of profits. See Sappington and Weisman (1996) for a (qualitatively analogous) formulation.
In a discrete-time framework it would be sensible to introduce a delay between the observation of a profit level and the adjustment of the x factor. In this set-up this would introduce a very substantial analytical complication with no relevant change in the results.
It would be possible to extend the current analysis to a case where different switch points \(({\widetilde{\Uppi}}_{1}, {\widetilde{\Uppi}}_{2},\ldots) \) and increasing values of the x factor are introduced (the equivalent of a progressive taxation). Moreover, the case in which price dynamics may be adjusted downwards if profits are too low might be a straightforward extension. This point is also discussed in Panteghini and Scarpa (2003).
The term δ(x j ) must be positive in order for the net value of the firm to be bounded. For further details on this point see Dixit and Pindyck (1994, Chap. 5).
We will relax this assumption in Sect. 5.
The VMC requires that a firm’s present value, net of both the explicit cost I and the opportunity cost (which is equal to the option value O PC (Π, x l )) is nil. When this condition is met, a firm’s net profitability is no less than zero, and so the investment can be undertaken. Since the VMC is not an optimality condition for investment, we also need a SPC, that is a first-order condition for optimum. For further details on these conditions see Dixit and Pindyck (1994) and Sødal (1998).
Substituting π PC *(x l ) into the system, one easily obtains
$$ A=\frac{1}{\beta_{1}(x_{l})}\frac{\Uppsi -1}{\Uppsi \delta (x_{l})}\left(\pi_{PC}^{*}(x_{l})\right)^{1-\beta_{1}(x_{l})} > 0. $$Proposition 1 generalizes Panteghini and Scarpa (2003), who consider a start-up investment.
Another way to look at the issue is to stress that profit sharing is equivalent to equity participation by the consumers. Recall, in fact, that when \(\pi > \Uppsi {\widetilde{\Uppi}},\) a given part of the surplus is redistributed to the consumers. When instead \(\pi < \Uppsi {\widetilde{\Uppi}},\) consumers do not share the bad result. We can thus say that the profit sharing device is equivalent to a case where consumers are endowed with a put option with strike price \(\Uppsi {\widetilde{\Uppi}},\) written on the firm’s profits. If, therefore, the firm’s return drops below \(\Uppsi {\widetilde{\Uppi}}\) (bad result), consumers sell their equity participation at zero price. Then, they will re-buy (at zero price) their participation when the firm faces a good result, namely when \(\pi > \Uppsi {\widetilde{\Uppi}}.\)
An extension to N stages is straightforward.
For further details see Dixit and Pindyck (1994, Chaps. 5, 8).
See Dixit and Pindyck (1994, Chaps. 5, 6).
It is easy to ascertain that, given derivative \(\frac{\partial \beta _{1}(x)}{\partial x} > 0,\) inequality β1(x h ) > β1(x l ) holds.
See Proposition 1.
References
Ai C, Sappington D (2002) The impact of state incentive regulation on the U.S. telecommunications industry. J Regul Econ 22:133–160
Altug S, Demers F, Demers M (2000) Political uncertainty and irreversible investment: theory and an application to Quebec. CEPR Working Paper 2045
Beesley M, Littlechild S (1989) The regulation of privatized monopolies in the United Kingdom. RAND J Econ 20:454–472
Bernanke BS (1983) Irreversibility, uncertainty, and cyclical investment. Quart J Econ 98:85–103
Biglaiser G, Riordan M (2000) Dynamics of price regulation. RAND J Econ 31:744–767
Buckland R, Fraser P (2001) Political and regulatory risk: beta sensitivity in U.K. electricity distribution. J Regul Econ 19:5–25
Burns P, Turvey R, Weyman-Jones TG (1998) The behavior of the firm under alternative regulatory constraints. Scot J Polit Econ 45:133–157
Crew M, Kleindorfer P (1996) Incentive regulation in the United Kingdom and the United States: some lessons. J Regul Econ 9:211–225
Dixit A (1991) Irreversible investment with price ceilings. J Polit Econ 99:541–557
Dixit A (1995) Irreversible investment with uncertainty and scale economies. J Econ Dyn Control 19:327–350
Dixit A, Pindyck RS (1994) Investment under uncertainty. Princeton University Press, Princeton
Dobbs IM (2004) Intertemporal price cap regulation under uncertainty. Econ J 114:421–440
Gasmi F, Laffont JJ, Sharkey W (1999) Empirical evaluation of regulatory regimes in local telecommunications markets. J Econ Manage 8:61–93
Harrison JM (1985) Brownian motion and stochastic flow systems. Wiley, New York
Hassett K, Metcalf G (1999) Investment with uncertain tax policy: does random tax policy discourage investment? Econ J 109:372–393
Hausman J, Myers S (2002) Regulating the United States railroads: the effects of sunk costs and asymmetric risk. J Regul Econ 22:287–310
Laffont JJ, Tirole J (1986) Using cost observations to regulate firms. J Polit Econ 94:614–641
Levine P, Stern J, Trillas F (2005) Utility price regulation and time inconsistency: comparisons with monetary policy. Oxford Econ Pap 57:447–478
Lyon T (1996) A model of sliding-scale regulation. J Regul Econ 9:227–247
Lyon T, Mayo J (2005) Regulatory opportunism and investment behavior: evidence from the U.S. electric utility industry. RAND J Econ 36:628–644
Mayer C, Vickers J (1996) Profit-sharing regulation: an economic appraisal. Fisc Stud 17:1–18
Moretto M, Panteghini PM, Scarpa C (2003) Regulation and irreversible investments of endogenous size. CES-ifo Working Paper No.1040
Panteghini PM, Scarpa C (2003) The distributional efficiency of alternative regulatory regimes: a real option approach. Int Tax Public Finance 10:403–418
Pindyck RS (2004) Mandatory unbundling and irreversible investment in telecom networks. NBER Working Paper Series No.10287
Sandmo A (1979) A note on the neutrality of the cash flow corporation tax. Econ Lett 4:173–176
Sappington D, Weisman D (1996) Revenue sharing in incentive regulation plans. Info Econ Policy 8:229–248
Sødal S (1998) A simplified exposition of smooth pasting. Econ Lett 58:217–223
Teisberg EO (1993) Capital investment strategies under uncertain regulation. RAND J Eco 24:591–604
Vickers J (1993) Privatization and the risk of expropriation. In: Baldassarri M, Paganetto L, Phelps E (eds) Privatization processes in Eastern Europe. Macmillan, London
Weisman D (1993) Superior regulatory regimes in theory and in practice. J Regul Econ 5:355–366
Author information
Authors and Affiliations
Corresponding author
Additional information
We would like to thank Michele Moretto, the editor and one referee for useful comments on an earlier draft.
Appendix
Appendix
1.1 The profit dynamics
In order to determine the dynamics of profit, we first derive the dynamics of the demand function. Using (1), (2), and (6), and applying Itô’s Lemma one obtains
Using (3), (4), and (15) one can obtain the profits’ dynamics under price cap regulation
where α(x j ) ≡ (1−η) (RPI−x j ) + α m is the expected growth rate of per-period profits.
1.2 The derivation of (8) and (9)
Using dynamic programming, the firm’s value O PC (Π(t)) can be written as
Expanding the right-hand side and using Itô’s lemma one obtains
where \(O_{PC_{\Uppi}}=\partial O_{PC}/\partial \Uppi (t)\) and \(O_{PC_{\Uppi \Uppi}}=\partial^{2}O_{PC}/\partial \Uppi^{2}(t),\) respectively.
To compute the value function, it is assumed that O PC (0, x l ) = 0, namely when Π is very small, the project is almost worthless. Thus, Eq. (17) yields the function (8), where β1(x l ) is the positive root of the following characteristic equation \(\frac{\sigma^{2}}{2}\beta (\beta -1)+(r-\delta (x_{l}))\beta -r=0.\)
Let us next turn to the firm’s value after expansion. Using dynamic programming, the firm’s value V PC (Π(t)) can be written as
Following the same procedure, assuming that V PC (0, x) = 0 and that no speculative bubbles exist yields function (9).Footnote 27
1.3 The derivation of (11) and (12)
To derive function (11) we follow the same procedure used to compute (8) and (9). We can thus write the general solution as
where parameters β1(x h ) and β2(x h ) are the roots of the characteristic equation \(\frac{\sigma^{2}}{2}\beta (\beta -1)+(r-\delta (x_{h}))\beta -r=0,\) with β1(x h ) > 1 > 0 > β2(x h ).Footnote 28 In the \((0,{\widetilde{\Uppi}})\) region, condition O PS (0, x l , x h ) = 0 holds: this implies that C 2 = 0. In the \(({\widetilde{\Uppi}},\infty)\) region, instead, there are no boundary conditions. To compute B 1 and B 2, we thus apply the VMC and SPC to the two components of (10) at the switch point \(\Uppi ={\widetilde{\Uppi}}.\) Both parameters depend on the regulatory coefficients x l and x h . It is easy to ascertain that \(B_{1}\;{\varpropto}\;C_{1}\) and \(B_{2}\;{\varpropto}\;C_{1}. \) Function (10) is thus obtained.
To compute (12) we follow the same procedure. Notice that, after undertaking I, the firm loses any flexibility. Therefore, V PS (Π, x l , x h ) does not embody any real option. In this case, the general solution is given by the sum of a perpetual rent, with discount rate δ(x h ), and a homogeneous (exponential) part. Again, it is assumed that V PS (0, x l , x h ) = 0 and that no speculative bubbles exist. Given these conditions, it is straightforward to obtain (12). To compute V 1 and V 2, we let the two branches of (12) meet tangentially at the switch point \(\Uppi ={\widetilde{\Uppi}}.\)
1.4 Proof of Proposition 1
Substituting Eqs. (11) and (12) into the VMC and SPC one obtains the following two-equation system
in the \((0,{\widetilde{\Uppi}})\) region. Solving (17) yields the trigger point
which is equal to the pure price cap one (10). The equality π PS *(x l ) = π PC *(x l ) proves Proposition 1.□
1.5 Proof of Corollary 1
Recall Eqs. (11) and (12). When the x factor is characterised by a permanent upward shift, the switch level represents an absorbing barrier. Since no downward jump in the x factor is allowed, we have B 2 = V 2 = 0. It is worth noting that the two-equation system obtained with the VMC and SPC is equal to (19). This implies that the trigger point is equal to (20). The Corollary is thus proven.□
1.6 Proof of Proposition 2
To prove Proposition 2 let us first show that investment is never undertaken at any point π ≤ π PC *(x l,0). To do so, suppose ab absurdo that in the case at hand (without profit sharing and with regulatory risk) the optimal investment timing, π**, is less than π PC *(x l,0). Given Proposition 1, we thus would have π** ≤ π PC *(x l,0) = π PS *(x l,0).
Using the per-period benefit arising from an incremental investment, (Ψ −1)Π, and applying standard techniques one obtains the present value in the (0, π PS *) region of the net gain from investing as
where i = 0,1. When i = 0, the profit is such that the starting factor x l,0 does not change, while i = 1 means that the x factor jumps upward when \(\pi =\Uppsi {\widetilde{\Uppi}}.\) The term \(H_{i}\Uppi^{\beta_{1}(x_{l,0})}\) measures the present value of the expected loss due to the increase in x.
The x factor may or may not jump upwards. If this does not happen, the firm’s loss is nil. Therefore for NV 0 this second term is nil (i.e., H 0 = 0):
Let us next solve the following optimal investment timing problem
It is worth noting that a Brownian motion satisfies the Markov property. Namely, the probability distribution for all future values of Π depends only on its current value. Applying this property, and following Harrison (1985), one can rewrite (22) as
Solving problem (23), it is straightforward to obtain
where β1(x l,0) is the positive root of the characteristic equation \(\frac{\sigma^{2}}{2}\beta (\beta -1)+(r-\delta (x_{l,0}))\beta -r=0,\) and where \(\phi^{-1}\equiv \frac{\lambda} {\delta (x_{l,1})}+\frac{\lambda}{\delta (x_{l,0})}.\)
Remember that we assumed ab absurdo that π** ≤ π PC *(x l,0) = π PS *(x l,0). However, given x l,0 < x l,1 we have δ(x l,0) < δ(x l,1). This implies that δ(x l,1) > ϕ > δ(x l,0), and thus π** > π PC *(x l,0) = π PS *(x l,0). This inequality contradicts the starting assumption, and thus proves point (a).
To prove points (b) and (c) let us next apply Proposition 1: when π reaches π PC *(x l,0), and the x factor does not rise, then investment is undertaken. Otherwise, the firm will wait until π = π PC *(x l,1) > π PC *(x l,0). This concludes the proof. □
1.7 Proof of Proposition 3
Let us define O PS,i (Π) as the firm’s value before expansion, under the old (i = 0) and the new (i = 1) regime. Similarly, define V PS,i (Π) and W PS,i (Π) as the value functions after, respectively, undertaking investment I and F, under state i. We first compute the closed-form solutions in state 1. Then we turn to the pre-change case (state 0). Finally, we compare the two trigger points obtained and show that they are equal.
1.7.1 State 1 (the new regime)
Let us first compute the firm’s value under state 1 i.e under the new regime. Before expansion, the firm’s value can be written as
Expanding the right-hand side and using Itô’s lemma one obtains
where x = x l , x h,1. The Eq. (21) has the following solution
Let us turn to the post-investment value function V PS,1(Π), which can be written as
Expanding the right-hand side and using Itô’s lemma one obtains
where x = x l , x h,1. Solving (26) yields
Let us focus on the \({\Uppi}{\in}\left(0,{\widetilde{\Uppi}}_{1}\right)\) region. Using the VMC and SPC, we obtain the trigger pointFootnote 29
and
Let us next focus on the second-stage investment F. In this case the value function can be written as
Using the boundary condition W PS,1(0) = 0, and following the usual procedure we obtain the solution to (30),
where x = x l , x h,1. Substituting (27) and (31) into the VMC and SPC yields the trigger point
Manipulating (32) one easily obtains (14).
Finally, remember that the switch levels \({\widetilde{\Uppi}}_{0}\) and \({\widetilde{\Uppi}}_{1}\) must be sufficiently high to ensure that a tighter regulation is applied only under good states. This implies that \(\Uppi^{**}\in \left(0,\min ({\widetilde{\Uppi}}_{0},{\widetilde{\Uppi}}_{1})\right). \) If, otherwise, at least one of the two switch points were low, the profit-sharing regulation would be implemented in the bad-news region. This would cause a distortion.
1.7.2 State 0 (the old regime)
Let us start with the pre-investment value. The Bellman equation is
Expanding its right-hand side and using Itô’s lemma yields
Defining Y(Π) ≡ O PS,0(Π)−O PS,1(Π), and subtracting (24) from (33) yields
Given the condition Y(0) = 0, Y 2 is nil. Using the solution of Y(Π) and Eq. (28) we obtain the option function under regulatory risk
where the parameter Y 1 is an unknown.
After investment I, the firm’s value is
Expanding the right-hand side and using Itô’s lemma one obtains
Define next X(Π) ≡ V PS,0(Π)−V PS,1(Π) and subtract (26) from (35), so as to obtain
The function X(Π) has a standard solution
where β i (x l , λ) are the roots of the characteristic equation \(\frac{\sigma^{2}}{2}\beta (\beta -1)+(r-\delta (x_{l}))\beta -(r+\lambda)=0.\) Given the condition X(0) = 0, X 2 is nil.
Using the solution of X(Π) and Eq. (22) we obtain
where the parameter X 1 is an unknown.
Finally, given the boundary condition W PS,0(0) = 0, the post-investment function will have the following form
where Z 1 is an unknown parameter.
1.7.3 The trigger points
Let us now compute the trigger point above which investment I is profitable under policy risk. Substituting (34) and (36) into the (VMC) and (SPC) we obtain a two-equation system
Divide (39) by β1(x l ,λ) and substitute it into (38) so as to obtain
Substitute (29) into (40) and multiply it by \(\frac{\beta_{1}(x_{l})-1}{\beta_{1}(x_{l})}\frac{1}{I}.\) We thus obtain
Multiply (41) by \(\frac{\beta_{1}(x_{l},\lambda)}{\beta _{1}(x_{l},\lambda)-1}\) so as to obtain
Adding and subtracting 1 from the LHS yields
Define \(y\equiv \left(\frac{\Uppi}{\Uppi^{*}}\right)\) and \(\phi \equiv \frac{\beta_{1}(x_{l},\lambda)-\beta_{1}(x_{l})}{\left[\beta _{1}(x_{l},\lambda)-1\right] \beta_{1}(x_{l})} < 1.\) Thus, Eq. (42) can be rewritten as
Equation (43) has more than one solution. We thus compute these solutions and identify the optimal one. As can be noted, the equality y = 1 holds in Eq. (43). This entails that Π** = Π*. Substituting Π* into system (38) and (39) one thus obtains (X 1−Y 1) = 0. This is the first couple of solutions of system (38) and (39).
Define y′ as any other solution. Given inequalities β1(x l ) > 1, ϕ < 1 and β1(x l )ϕ < 1, it is easy to show that any other solution is y′ > 1. This implies that the trigger point obtained would be Π** > Π*. Substituting this new solution into system (38) and (39) yields (X 1−Y 1) > 0. Thus (Π** > Π*,(X 1−Y 1) > 0) is the second couple of solutions. However, this couple is sub-optimal. To show this, assume ab absurdo that the optimal solution is (Π** > Π*,(X 1−Y 1) > 0). Then, using the definitions of X(Π) and Y(Π), we define the pre-reform added value arising from investment (net of both the opportunity and the effective costs), as
Using (VMC) and (44) we obtain F(Π**) = 0. Rewrite (44) as
Since in Π = Π* the post-reform added value [V PS,1(Π)−O PS,1(Π)−I] is nil, we have \(F(\Uppi^{*})=\left(X_{1}-Y_{1}\right) \Uppi^{*\beta_{1}(x_{l},\lambda )} > 0.\) This means that, in the interval Π ∈ (0,Π**), there exists at least one point (Π = Π*) such that the project’s payoff is strictly positive. Thus, a rational firm facing a positive payoff in Π = Π*, would immediately invest instead of waiting until the trigger point Π** is reached. This contradicts the assessment that (Π** > Π*, (X 1−Y 1) > 0) is the optimal investment strategy. The remaining solution (Π** = Π*, (X 1−Y 1) = 0) is thus the optimal one.
Following the same procedure, we can show that function (37) has the same form as functions (34) and (36). Therefore, substituting (36) and (37) into the VMC and SPC and manipulating the results yields the following equation
which has the same form as Eq. (42). Following the same procedure, we can thus show that the post-reform trigger point coincides with Π*. This proves Proposition 3.□
Rights and permissions
About this article
Cite this article
Panteghini, P.M., Scarpa, C. Political pressures and the credibility of regulation: can profit sharing mitigate regulatory risk?. Int Rev Econ 55, 253–274 (2008). https://doi.org/10.1007/s12232-008-0040-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12232-008-0040-y