Ideological party and regulation theory Massimo Di Domenico IEFE Università Bocconi September, 3 Abstract The aim of the paper is to integrate the general regulation theory with a political dimension derived from public choice analysis. It is within this area that the paper makes some contributions. In particular departing from a standard Laffont-Tirole model of regulation in which the government is taken as a benign despot, I will consider a party in power with ideological political motivation. This will depict a variety of possible analysis of the general economic model of regulation: one-party in power and asymmetric information; two parties competing structure; parties strategic behaviour; party's and citizens/firm single relationships. The attempt and novelty of the paper is then to review the regulation of public utilities theory introducing such a political dimension setting, in this way, a more complex but complete model of regulation. Keywords: ideological parties, regulation, political economy, asymmetric information, parties competition. JEL classification: H, L5 This work is partially drawn from the PhD thesis: M. Di Domenico, Political Economic Analysis of regulation: theory and application. University of Southampton, Department of Economics (). Massimo Di Domenico, IEFE-Bocconi University, Viale Filippetti, 9, Milan-Italy. Tel.39583638, fax.3958369. E-mail: massimo.didomenico@uni-bocconi.it
. Introduction The analysis attempted in this paper concerns the influence on the regulation process by groups with different motivations. The New Regulatory Economics is particularly devoted to treat the regulation problem when asymmetric information is introduced into the model. The theory investigates the asymmetric information between the firm, that benefits from lack of information, and the agency, that have to regulate it. The asymmetric information problem and the principalagent relationship introduces some inefficiencies into the model. The necessary incentives introduced into the system force to seek second-best equilibrium determined by inefficient allocation of resources. Focussing only on the economic dimension of regulation is necessary but not sufficient because important is also the political and normative dimension of regulation. The necessary evolution of a political economy/public choice approach to regulation derives from the early works by the Virginia and Chicago schools. The Virginia schools 3 focuses mainly on the rent extraction by competing groups such as bureaucrats and politicians. The rents earned are important resources that bureaucrats and politicians devote to interest groups in order to attract their favour and attention. The full social cost for the society has to be increased, then, by the deadweight losses associated with such rent granters activities. The Chicago school, instead, focuses mainly on the role of interest groups in the public policy process. The works by Stigler (97), Posner (974), Peltzman (976) and Becker (983), are positive analysis devoted to consider the contribution of interest groups on such policy formation. Both the Chicago and Virginia schools have considered the regulation problem from "demand side" without considering the role of political agents in such a process. In particular it becomes necessary to develop a "supply side" approach to the political regulation in order to investigate how political behaviours by agents, such as voters, parties, pressure groups, translate into a mechanism that influences the regulation process. Recent works by Baron (989), Spiller (99), and Laffont () specifically have developed early work on this matter opening more interesting theoretical discussions. The literature on the argument is vast. Let us mention Armstrong, Cowan and Vickers (994), Berg and Tshirhart (988), Bishop and Kay (988), Foreman-Peck and Millward (994), Laffont and Tirole (993) as reference books both for basic and advanced analysis on regulation. For example Barzel (989), Estache, A. and D. Martimort (998), Fiorina (98), Kikeri, Nellis and Shirley (99), Levy and Spiller (994), North (99), Romer and Rosenthal (986), Spiller (99), Williamson (988), Willig (994). 3 Buchanan (98), Tollison (98), Tullock (967) but also Bhagwati (98), Coate and Morris (995), Delorme and Snow (99), Krueger (974).
In order to describe the important political process of regulation the public choice/political economy theory gives a broad variety of arguments to be discussed and applied. In particular such theory focuses on the electioneering approach, i.e. political agents motivations are driven by winning the election and gaining office 4. This is explained as parties seeking a certain "surplus" available to the party in power or supply in terms of a desire for office per se. The theoretical approach I will follow in this analysis relates to the idea of ideological parties' motivations. Example of ideological parties are treated, for example, by Alesina 5 where a single two parties competition is analysed in terms of parties having fixed policy positions over a political spectrum. The idea of ideological party preferences is interestingly extended by Beesley and Coate (997) where, in this case, not the party but each person, having given political preferences, is a potential party/candidate in the political arena. The theoretical novelty introduced in this analysis relates possibly to the integration of the regulation theory model, such as the Laffont and Tirole one, with a number of political dimensions. In particular I am interested to consider a regulation process where the Benevolent dictator assumption is replaced by a party political structure having a certain fixed policy concerning the regulation process. In this case it will be possible to investigate the probable relationships between the party in power and the other agents in the system. I assume that in this way there will be the possibility to investigate on one hand, the relationship between government, regulatory agency, firm and consumers -economic relationship. On the other hand, the possible political relationship between citizens, government, firm and agency. The above relationships will be investigated in the following sections as follows. Section will briefly review the basic Laffont-Tirole model. Such a model will be modified in section 3 where a single ideological party will be introduced and the principal-agent relationship, with full and asymmetric information will be developed. Section 4 will analyse a more complex model with a two-parties competition, still in a regulatory framework. Finally section 5 will consider single relationships in more details. In particular section 5. will analyse the party citizen/consumer relations while section 5. the party and the firm one. Section 5.3 will conclude with a simple analysis of an infinitely repeated game given asymmetric information. 4 Again the literature on this argument is very vast. For the purpose of this article I focus principally on the following authors: Brennan and Buchanan (78,8,85), Downs (957), Feenstra and Bhagwati (98), Oates (98),Shumpeter (95), Wilson (98). 5 See Alesina (988) and Alesina and Rosenthal (995). 3
. The economics of regulation: the Laffont-Tirole model In this section a basic review of the economic model of governmental regulation of industries is introduced. In particular I will consider the L-T model, which includes the possibility of regulatory capture 6. The Laffont-Tirole model takes the agency problem as central. Agency models introduce an important additional element of inefficiency into the analysis. And this form of inefficiency is relevant to both economic and political relationships (Laffont ()). This inefficiency is due to the loss of control by the principal and the fact that the agent can extract an informational rent. The L-T model is set in this context. In particular, it considers the regulation of a natural monopoly subjected to control by a regulator. The firm knows the level of cost reducing activity and the technological parameters of production and this represents private informational advantage over the regulator. In this case the firm will generally extract an informational rent. The model also considers other agents as the congress or government (which acts as the principal) and the regulatory agent that has to supervise the firm s activity on behalf of the government. In this framework the Social Welfare Function (SWF) is maximised by the Congress- the benevolent dictator agent of the model- defined in terms of the sum of all surpluses in society, subject only to the constraints imposed by technology and the availability of information... General Structure of the Model I now introduce the main characteristics of the L-T model in order to build the model with ideological party in section 3. Specifically the model considers the relationship between the firm, the regulator and the congress. For each of these agents, I specify the following definitions, which also serve to introduce relevant notation, for each of the agents considered in the model. The firm The marketable output q is produced by the firm at cost: C = ( β e) q where β is a technological parameter and e is effort, or cost reducing activity. For simplicity I will assume that β can take either of two values: high ( β ) with probability (-v) and low ( β) with probability v. It is assumed that: 6 The presentation, and notation, therefore draws heavily on Laffont and Tirole (993) particularly sections.-.3. I do not explicitly include the analysis of regulatory capture in this introductory review of the model - see Laffont and Tirole (993) section.4. See also Laffont (). 4
C β > ; if β is high the technology is inefficient in the sense of more expensive; C e < and C ee ; so that effort reduces cost at a decreasing rate. Given the firm's effort e it is also possible to introduce the firm manager s disutility of such effort expressed in monetary term Ψ(e), where Ψ (e) > (effort is costly to managers), and Ψ (e) > (convex function). The cost C, the output q and the price P( are all assumed to be verifiable. The key issue in this model of regulation is the problem of asymmetric information in the form of moral-hazard and adverse-selection. The regulator does not know the technology of the firm or the cost reducing effort of the managers and this represents an asymmetric information problem. The parameter β is private information to the firm, the adverse-selection issue. At the same time, the level of effort e is also unverifiable, setting up perverse incentives for the firm. This aspect of the problem represents the moral-hazard issue. Revenues, costs and transfers. The government receives all of the firm s net revenue R( but pays a transfer t to the firm. The managers utility function becomes then: () U = t Ψ() e To accept the regulation constraint, the effort costs must be covered by the monetary transfer t. Of course such a transfer must assure the firm at least the level of utility that is available in any outside option. For simplicity I normalise this firm s outside opportunity or reservation utility to. Consequently the firm s individual rationality (IR) constraint is: ( ) t Ψ( e) U represents the firm s rent or surplus. It is assumed that both firm and regulator are risk-neutral with respect to income. The regulator An important agent is represented by the regulator. It is an intermediary in charge of controlling the firm s activity. It receives rewards in the form of income, s, from the congress. The regulator s utility function is then: ( 3 ) V(s) = s - s* where s* represents its reservation income. For simplicity I assume that the regulator s activity is necessary to regulate the firm s price and cost. That is why the government pays at least s* to the agency in each state of nature. The L-T model specifies the complex activity of the regulatory agency. In particular the regulator obtains information about the firm s technology. A signal σ carries that information and the following probability system is considered: ξ is the probability that 5
the regulator receives the true information (σ = β) and (-ξ) is the regulator s probability of learning nothing (σ = ). It is also important to specify the report function r {σ, }. At the same time the regulatory agency must report to the congress r = if it has learned nothing (σ = ). If the regulator learns the truth it may either report the truth, r = β, or report, falsely, that it has learnt nothing, r =. The model considers a situation without collusion between the regulator and the congress. In the model set in section 3 I consider the regulator as a central agent reporting to the Congress the relevant information. The consumer The utility of the consumer is simply defined as ( 4 ) S( - P( q where S( represents the benefit or surplus derived from the consumption for the good q, and P( q the expenditures for that consumption. For simplicity and giving the development of the model with a two parties system, I now suppose two types of consumers (type and type ) with a proportion α of the population of the first type, and (-α) of the second type. Type consumers derive a benefit θ ( S( from consuming the good q produced by the firm, while type consumers derive the benefit θ ( S( with θ θ >. In particular the parameters θ ( and θ ( are function of the quantity q with θ > and θ. This specification allows to consider two groups of citizens as having different needs or preferences concerning the good produced by the firm; type citizens having a stronger preference for q than type. This will represents an important specification in order to consider how these differential preferences across the two groups of consumers will influence their political activity 7. At the beginning of time α is a random variable in the interval [,]. Consumers surplus is then specified for each group of consumers as: (5) V = α θ ( (S(-P(-(λ)[T- α θ (P(q] V = (-α) θ ( [S(-P(q]-(λ)[T-θ ( (-α) P(q] This specification of consumers into two groups allows a number of ways in which I may specify the motivations of the political party in power. The congress 7 See Laffont (996) who introduces this way of modelling the division between the groups of consumer/citizens. 6
The congress role is to maximise social welfare which is given by the sum of consumer, firm and regulatory agency surpluses. That is: (6) W = U V [ S( P( q ( λ)( s t ( β e) q P( q] or W = [ S( λp( q] - ( λ)( s * ( β - e) q Ψ( e)) - λu - λv where the terms in the squared bracket in the second expression represents the generalised consumer s surplus, whereas the rest of the formula introduces the cost of the project given the distortion term λ. In particular the term λu and λv are the social costs associated with the rents that firm and agency earn and that the Congress would like to reduce. The congress only observable variables are the cost C, the output q (or the price p = P() and it also receives the report from the regulatory agency, r. It has lack of information concerning the firm s technology parameters β or the agency s signal σ. The congress maximises the expected social welfare W by designing the incentive schemes s(c, q, r) and t(c, q, r) for the agency and the firm respectively. The process could be timed in the following way: initially the relevant information is learnt, simultaneously, by all the agents in the system. That is, the congress, and the consumers learn that β belongs to { β β } ; ; the regulatory agency learns σ and the firm learns β. The probability distributions are common knowledge 8. The congress then designs the incentive schemes for the firm and the agency. Next, the regulatory agency makes a report while the firm chooses effort and price. The story finishes with the transfer to the firm, as specified in the regulatory contracts. The full information case (σ = β) This case depicts the situation in which the congress drives the firm s rent to zero because the parameter β is known accurately (for all β { β β } ; ) In the full information environment, of course, the maximisation of social welfare depends on effort ( ee. ;) In this case, even with the inefficient firm's effort, there is not distortion (e = e *) and the SWF satisfies: 8 The cumulative distribution function is F(β) (with F ( β) =, F( β) = ) and F(β) has a continuous and strictly positive density f(β) on [ β, β]. The firm then, assumes only the two values β and β respectively. 7 with probability (-v) and v
(7) W FI W FI (e *) The asymmetric information case (σ = ) In this case, the basic regulatory aim is to prevent efficient firms from claiming to be inefficient, and so extracting a rent. To prevent such advantage, it is necessary to modify the full information rationality constraint by adding an incentive constraint. Φ() e = Ψ() e Ψ( e β ). More generally, L-T demonstrate that fixing exogenously the inefficient firm s effort at e, the expected welfare under asymmetric information becomes: AI F I F I ( 8) W ( e) = max{ W ( e) λ vφ( e)} = W ( e) λvφ( e) e (the strict concavity of W AI depends on the strict concavity of W FI ). In this situation the corresponding market price is higher than under full information. The general idea expressed in this model is that in the absence of asymmetric information, the firm cannot extract rent and so the firm s output can be socially optimal. In effect, in this case, the benevolent dictator can simply run the firm. With asymmetric information, the firm has an informational advantage and the problem becomes one of designing a regulatory regime to motivate the firm to use its informational advantage for general benefits instead of increasing its monopoly rents. Given the main L-T model assumptions it is interesting to introduce some relevant modifications that take account of the more political aspect of regulation. In particular, in the next section I will consider a new actor, the political party, having an ideological political motivation. 3. The politics of regulation: political motivations The L-T model gives an important framework to consider asymmetric information between various agents. The intent of the following is to modifying the economic model of regulation to introduce an explicitly political component. The idea is to focus on the political relationship between the citizen/consumer and the congress, and the impact of this relationship on the economic relationship between the congress, the regulator and the firm. Generally the idea is to consider the firm as producing goods or services for consumers needs. This is not the only effect. In such a system the consumer is also a citizen reacting to the firm - and its regulation - politically. The citizen/consumer s satisfaction or dissatisfaction may also represent a political matter. The citizens could support a certain political activity given the services they receive. This political preference may then influence the relationship between the congress, the regulatory agency and the firm. 8
For example, if the consumer is not satisfied by the current system because she does not approve of the firm s production, she could try to effect change by expressing this disapproval politically. It is true, there are many forms of political expression, and the models discussed here will consider only a simple case. In particular, it will only consider a very simple relationship between political and economic activity. This represents only a first step toward more complex models. In the first analytical point I will consider only one political party having an ideological motivation given the interests of citizens. This might be thought of as viewing the degree to which government is responsive to pressure from public opinion, without actually modelling the mechanism by which public opinion impacts on government. In this way, the intention is to begin to modify the standard L-T model, investigating how the political relationships interact with the economic relationships, and how this could influence the economic result. 9 In the remainder of this section I provide an account of political and economic interaction considering the ideological party s motivation in a simple one-party structure of politics extending the L-T model. In this way, I seek to explore one dimension of the political structure - the benevolence-ideology dimension. Clearly, this involves abstracting from the analysis of inter-party competition and, particularly, the analysis of elections. These topics, and the interaction between the electioneering dimension of political parties and the ideology dimension, is developed in paragraph 5. 3.. The ideological party In the analysis that follows, I will maintain the form of the L-T model. In particular, the specification of the firm and the regulatory agency will be essentially identical to that reviewed above. The main differences between the models relate to the specification of consumers and the congress/political party. In the first part of the model I do not consider political elections - there is only one political party - so the ruling party is not concerned with the threat of the loss of office. In this political structure, the political party receives from the citizens a political support, and the nature of this support will depend on its policy which in turn directly influences the welfare of citizens. The Leviathan model of Brennan and Buchanan (985), where the assumption of a benevolent dictator is replaced by an assumption of a self-interested dictator is well representative of a purely 9 I have already analysed a model with one-two parties but with different relationships between the motivation of that party/parties and the interests of citizens. See the above mentioned PhD thesis and the forthcoming article, A political economy approach to regulation theory. See Laffont (996) who introduces this way of modelling the division between the groups of consumer/citizens. See also Calvert (985), Ferejohn (986) for discussion of the ideological party assumption. 9
ideological party. Obviously a purely ideological party, without constraints imposed by elections or other institutions, will simply follow whatever policy coincides with its own ideology, regardless of the implications for consumer/citizens. In such a model, the ideological party imposes its preferred policy to the other agents. The main implication of an ideological motivation is that in this case the model does not edogenise how the party forms its decisions. This means that is not specified the direct links between party s behaviour and citizens or/and other institutions. The party, here, is a monolithic and self interested actor that pursues its optimal policy/ideology while the relationships with the other agents is not investigated. Clearly, then, a model of such ideologically motivated parties will be of great interest in the setting of a one-party model, when the party is in power and has to regulate the firm. Moreover it is appropriate to consider it when multi-party politics with elections is considered. In the model to be discussed, politicians have as main aim the regulation, through the regulatory agency, of the firm. In order to view the different cases I must allow for a political decision making. For this reason, I introduce the following politician s surplus function: (9) U = d(q,µ (e ()) ( q c) where q is the firm s production, e is political effort and µ is the monetary disutility function > of that effort (with du /dq and du < ; du /de < and du < ). < The main difference between the L-T model, and the models introduced here is that the political party/congress is no longer necessarily a monolithic and neutral actor. This important change will have an effect on the optimal form of regulation under asymmetric information. It is necessary to analyse in details eqn.9 The first term of the equation is represented by the factor d. The implication of such a factor is that in order to enforce the optimal level of output q*, the politicians in power would have to expend some level of political effort, e *. However, with asymmetric information, this efficient level is not obtained. It will be demonstrated, below, that with complete information the party in power will behave efficiently because observed by the other agents. Production is then socially efficient because the firm s managers cannot gain a rent, and the regulatory agency will report all the relevant information about the firm s cost to the government. At the same time, politicians cannot appropriate any rent and so their activity is also efficient (i.e. d= ). With asymmetric information both productive and political inefficiency is introduced. The basic L-T model accounts for the rent that the firms could earn because of its informational advantage. The same happens for the politicians in power that could benefit of a rent because of asymmetric
information. Specifically the politicians can gain a rent by varying their effort with respect to the socially optimal level of effort. This could be the case, for example, of an inefficient firm and a party that should improve its control activity over the firm if it wants to have the support of the citizens. By contrast if the firm s activity is efficient, the party in power could behave inefficiently. The party in power may decrease the political effort (e ) gaining a rent from the consumer. In particular the citizens believe that the party s activity is efficient because of the firm s performance. The novelty, here, relates to the efficient-inefficient party s effort with respect to the optimal policy q*. Very important is also the squared term of equation (9) that characterises the ideological party. The party with ideological motivations has its own ideology/political preferences and it wants to implement it given the citizens and firms preferences on the outcome. The term q represents the policy outcome and c i identifies the party s ideal point in the policy space. The particular quadratic functional form represents the idea that, as political outcomes depart from the party s most preferred policy, utility decreases at an increasing rate, but this is not essential for the basic results of the standard model. More importantly, it is assumed that the party has a symmetric and single-peaked utility function (as here) because, in this case, the party always prefers the policy closest to her own ideal point. Of course, this symmetry assumption is assumed for simplicity and it is not always valid. At the same time the concavity of the utility function implies that party is a risk-averse agent. 3. The ideological one-party model As specified in the previous section the ideological party case is characterised by a party having its own political preference together to a certain non-neutral activity. With an ideological party in power the Congress maximises the following SWF: () maxw q s. t. = αθ [ S( P( q] ( λ)[ s t ( β e) q αθ P( q] θ ( α)[ S( λp( q] ( λ)[ s t ( α) θ P( q] PB U V U U with U = -½ (q -c) d(q,µ (e ()) Specifically the Congress maximises () taking into account the individual rationality constraint for the portion of citizens not belonging to party supporters (those of type ). It is then possible to obtain: (i) αθ p ( β e) p αθ ( q * c) λ = αθ' [ S( q*) λp( q*) q*] λ η ( λ) p ( λ) p
with θ >. (ii) Ψ (e) = q e=e*; (iii) U=; (iv) V - α ; (v) Do the project iff α θ [S(λP(q]-(λ)[Ψ(e*)(β-e*)q] ; (vi) d(q,µ (e ()) = ; (vii) du \de =d ( ) µ (e ) e =e *; (viii) du \dq = -(q*-c) Many of these results are broadly similar to those in the basic L-T case with an important difference concerning the party s motivation. Result (i) considers the price-cost margin as a function only of the portion of citizens that supported the party ( parameter α θ or (-α) θ ). The maximisation process considers the party s political support that differs depending on which group of citizens is considered. The citizens preferences θ s contribute to determine the specific second best equilibrium (i). (ii) considers the optimal effort for the firm and (iii) sets utility equal to zero because it is costly to leave a rent to the firm. (iv) analyses the IR constraints, while (v) specifies the condition to start the project. Point (vi) specifies, in this full information setting, the necessary condition to avoid party s discretion while (vii) shows the optimal political effort. Important for this analysis is also point (viii). The absolute value of the term q*-c captures the impact that ideological party has on the efficient second best equilibrium. Considering the party as an important agent, having a specified utility function, together to the consumer and the firm implies a second best equilibrium that is affected by the policy implemented. This result opposes the result obtained for the Partially Benevolent Party. In that case the party forms its policy according to consumer and firm utilities. At the same time it has not an own preferred policy to be implemented. The party s activity is considered as discretional and constrained by a fully informed Congress. With ideological party, instead, the ideology is an important and exogenous factor which is considered by the welfare maximiser. The results obtained, then, are affected by such a political preference and the second best equilibrium is also function of such a preference. Giving the above definitions and results it is important to analyse a model with asymmetric information and ideological party. 3.3. The cases compared: Complete information As before, the firm cannot gain any rent because the firm s technological parameters is known in this full information environment. The same story applies to the citizen that knows the party s
activity (the effort variable), so that the party will not be able to extract a rent. Formally this is shown below. I have already introduced the utility function U for the party in power that includes the function d related to the party s political effort. Given result (vii) above the party earns a rent when its effort is lower than the optimal one i.e. e ( e *(. To better analyse the asymmetric information environment it is necessary to consider the possible probability functions for every state of nature. For an efficient firm s production (, the party s behaviour could be efficient or inefficient with the following probability: y: the probability that the party undertakes the efficient effort e ; -y: the probability that the party undertakes the inefficient effort e The above probabilities system implies a complex firm s and party s joint behaviour. For example, one of the possible states of the nature is represented by a firm s efficient production that satisfies the consumer\citizen. The party in power, knowing the citizens satisfaction by the efficient production q, could reduce its effort to e, thereby earning a rent. A necessary incentive has to be introduced, then, in order to reduce such an inefficient rent. For the purpose of this analysis, I consider only the inefficient ideological party s behaviour when the firm is efficient and vice versa. Ideological party and full information It is possible to indicate the SWF with an ideological party as it follows: {[ () W = v(- y) αθ( ][ S( λp( q] - ( λ) s * Ψ(e) ( β e) ( q c) d(q, µ (e ()) - λu} ( { ( v) y [ αθ ( ][ S( λp( q] ( λ) ( s * Ψ( e) ( β e) ( q c) d( q, µ ( e ( )) λu Equation () reproduces the SWT for two states of nature i.e. v(-y) and (-v)y. Differentiating () gives: Ψ'(e) = q for e, and Ψ'(e) = q for e ; } λv 3
d'( ) µ '( e ( ) =, d'( ) µ '( e ( ) = for e and e respectively (corresponding to the efficient levels e * and e *). Differentiating () with respect to q I obtain: () p[ αθ (] - ( β - e) p [ αθ( λ ( α) θ ( λ] = λ η [ αθ q S q P q q ( q * c) ' ( )][ ( *) λ ( *) *] ( λ) p ( λ) p where p = P(q ) (it is possible to find an analogous result for q ). The result is similar to (), but takes into account the different states of nature of the system. In this situation neither the firm nor the party gain any rent because of complete information. In particular d= and U=. Given this structure, it will be easier to verify in which cases inefficiency arises. As for both the fully benevolent and for the partially benevolent party, in the ideological party case the structure of full information is sufficient to imply efficiency. There is no great surprise in this, but it is worth noting that partial benevolence is all that is required - as long as government is responsive to the interests of at least some of the people it will be forced to be efficient. For the ideological party we have a similar situation and the party is basically efficient with respect to the group of citizens it is responsive to. The only and substantial difference, here, is introduced by the party s preference that determines a distortion with respect to the consumer/citizen and firm s preference. 3.4. Ideological party and asymmetric information The asymmetric information situation is analysed considering the following two cases: ) σ = the agency does not know the firm s technological parameter β; ) γ = the consumer does not know the party s activity or effort. For the purpose of this analysis, condition ) is very important. The citizen does not observe directly and have not information concerning the party efficient or inefficient activity because of lack of information 3. It is then necessary to introduce in the model an incentive function for the political party in order to avoid it claiming to be efficient when this is not true, as illustrated by Laffont and Tirole (993). In particular, incentive compatibility is now required for both the firm and the party. The optimal situation at the level of the firm is represented by: See agin the PhD thesis for a more detailed PBP and FBP cases. 3 See Estache, A. and D. Martimort (998) for more comprehensive analysis on this point. 4
U = Φ(e) where Φ(e) = Ψ(e) - Ψ(e - β ) At the level of the party, it is necessary to consider an incentive function Φ e ) (related to the inefficient effort e ). In this case an additional constraint of the form d ( q, µ ( e( ) Φe = d( q, µ ( e( ) d = Φe d is necessary with respect to the full information case. The modified SWF is then rearranged as (3) W= v(- y)[ { αθ (][S( λp(q]- ( λ)(s* Ψ(e) ( β - e) - λφ( e) - (q - c) d( ) Φe ( λ )(s * Ψ(e) ( β - e) - (q - Differentiating (3), I obtain the usual results (4) Ψ'(e) = q that is e = e * c) for e; d( ) }- λv ( (5) } (- v) y{ [ αθ ( ][S( - λp(q] - λφ' (e) v (- y) Ψ'(e) = q *(e) - ( λ)( v) y for e giving the results e < e * and q < q *. The reduced effort variable and the quantity with respect to the optimal condition is a necessary condition in order to reduce the firm s rent avoiding that the managers affirm to be inefficient when it is not true. To the firm s inefficient behaviour it is now important to add the political new dimension to this model. The following proposition introduces it: Proposition : With asymmetric information the party earns a rent by varying its effort e from the efficient level e *. This increases the divergence between the price and the marginal cost in the Ramsey formula. The proposition reintroduce the idea of political body having discretion expressed in term of effort. The effort of the political body enters in the model formalising the idea of inefficiency associated to costly political effort. Differentiating with respect to the party s effort it is possible to obtain: (6) d'( ) µ ' (e() = Φe ' for e (with Φe ' > ), and (7) d'( ) µ ' (e) = for e 5
Equation (6) shows that the party s utility variation depends on the incentive function Φe for inefficient party s effort. Similarly to inefficient firm, the effort is different from the optimal equilibrium, that is e > e *. Equation (7), instead, shows a result equivalent to the complete information case e = e *. The idea of an incentive function Φe, for inefficient effort e, is then introduced in order to force the party toward more efficient behaviour. For example, this incentive could be the result of the action of an institutional body having the power to judge and constraining the party toward a more efficient behaviour 4, limiting the party s ability to extract rent. In particular equation (6) shows that the inefficient party s utility is reduced by the component Φe, forcing the party to increase its effort. Maximising equation (3) with respect to the quantity variables q and q it is possible to determine: (8) [ p( αθ( ) ( β e)] λ[ αθ( ] = p λ η p( λ) d( ) d( ) µ ( e q µ ( e ) q ) dφe dq [ αθ q S q P q q ( q * c) '( )][ ( *) λ ( *) *] ( λ) p ( λ) p (9) [ p ( αθ( ) ( β e)] λ[ αθ( ] = p λ η p( λ) d( ) d( ) µ ( e ) q µ ( e ) q [ αθ q S q P q q ( q * c) '( ) ][ ( *) λ ( *) *] ( λ) p ( λ) p The above equations (8) and (9) define price-marginal cost relations also taking into account the party s effort in an asymmetric information environment. The results show the party s behaviour once an incentive is introduced. In particular equation (8) relates to the case where the firm is efficient, in this case it is possible to obtain: (8a) d( ) ( i) q d( ) µ > ; ( ii) µ q > ; ( iii) dφe dq >. 4 I do not consider explicitly this important aspect here. It is necessary to add that also this judicial body could not be benevolent- see Laffont () for more specifications. 6
Equation (i) shows that for increasing firm s efficient production q, the party behave inefficiently, earning an hidden benefit. The lack of information give to the party the opportunity to indicate itself as efficient declaring it to the other agents in the system, when this is not true. Equation (ii) considers the marginal variation of the effort due to the variation of the efficient quantity. As it has been already stated, in this situation, the party decreases its effort in response to an increase in the level of output The incentive effect on the party s behaviour is illustrated by equation (iii). Specifically the equation shows that the party incentive increases with the efficient quantity. The difference between equation ()- without party s incentive, and equation (8) is due to the hidden party s socially inefficient activity. Holding conditions (i), (ii), and (iii), the price-marginal cost differential of equation (8) diverges, moving further away from the efficient equilibrium. For inefficient firm the reverse is true. Equation (9) represents such an alternative situation: (9a) d( ) d( ) µ ( e ) ( iv) < ; and ( v) <. q µ q In this case the party faces a greater incentive to act efficiently when the regulated firm s production is inefficient. The main result concern the fact that the party decreases the hidden benefit (condition (iv)) and, at the same time, it is constrained to be more active in order to limit the firm s inefficient behaviour (condition (v)). The convergence toward a more social optimal equilibrium is, for this reason, attained. Another important aspect that has to be included into the analysis concerns the policy implemented by the party once in power. I will return later on this analysis, but it is interesting to distinguish and to introduce the pre-commitment and no pre-commitment cases concerning the announced and expected by the voters party policy. In particular if pre-committed the voters believe in the party s announced policy and q a =q e otherwise the voters expected policy becomes q e t =E(q t I t- ) q a t. In the no pre-committed case, instead, the party follows its preferred policy and it sets q=c. It is trivial to think about the pre-committed case for an asymmetric information environment. The ideological party would benefit by implementing a policy according to its optimal choice c. The party, then, once in power, and with asymmetric information, tries to implement its optimal policy moving from the optimal SWF policy. Analytically: q* = arg max W FI represents the optimal policy from the Social Welfare point of view. For other than q* policy (for example for policy q c) the SWF is trivially specified as W AI W FI. Also considering party and firm s inefficient efforts it has already been demonstrated that: 7
W AI FI FI { W ( e, e ) λ v Φ( e) vφ( e )} = W ( e, e ) vφ( e) vφ( e ) ( e, e ) = max λ e, e Generally speaking, and according to previous results 5 political motivation matters. The benevolent party determines a social price that takes into consideration all the citizens in the system, while the partially benevolent party determines the relative social price taking into account only the citizens that have supported it. The ideological party represent a more complete case with respect to the benevolent party case. The results obtained depends on such motivations Adding asymmetric information into the model it is possible to illustrate more complex cases, giving more opportunities to consider and integrate the political dimension with general regulation models. 4. Regulation and competitive politics In this section I am concerned to analyse a two-party political system with ideological motivation, faced with a policy issue of regulating a firm via a regulatory agency. In this context the citizens is considered both as the consumers of the product of the regulated firm and voters in the political elections. Departing from the parties ideological motivation of the previous sections I will concentrate on a more complex political environment where the parties compete for office having, as main aim, the regulation of a specific firm. In this case the model considers the parties as constrained by the voter s expectation about the parties policies that will influence the result of the election, and the opponent party s strategy. The major elements to be added therefore relate to the strategic behaviour of parties when faced with electoral competition, and the voting behaviour of individual citizens 6. The standard economic approach to two-party competition begins by considering a one dimensional issue space, simple majority voting, and by specifying citizen-voters preferences over policy outcomes 7. The analysis developed in this model considers an electorate composed of a continuum of voters giving the distribution of voters' ideal points. I will generally assume that this distribution is uniform over the issue space, unless specified otherwise. Assuming a single majority voting system the following step is to consider the two ideological parties, and the policy platforms chosen by the parties in the Nash equilibrium of the strategic game 5 PhD thesis. 6 The analysis offered in this paper considers a two parties system trying to describe pretty well system like that of the US, but less well more complex political system where more parties are involved in the political arena. See Congleton and Steunenberg (998) for a more extensive analysis. 7 See for example Mueller (3). 8
that represents the election. Introducing the two parties system I shall assume, for the moment, that parties are committed to their announced platforms and voters can confidently expect the winning party to act in accordance with their announced policy platform - the issue of credibility of commitments and the role of repeated elections etc. is not relevant at this stage. In this static, or one-shot context the standard starting point is to assume that parties are motivated purely to win the election. Again the standard median voter theorem applies. The parties, then, converge until they each offer a policy platform that is identical to the ideal point of the median voter. Under simple majority voting with just two parties, the ideal point of the median citizen is an unbeatable platform 8. An alternative model proposed here, though, specifies the party s preferences as being rather more similar to the citizens, so that parties are ideological rather than being motivated purely to win the election. In this case, each party has an ideal policy, but also recognises the value of winning the election, and so faces a trade-off between sticking to their own preferred policy and moving away from this policy in order to increase the probability of wining the election (see, for example Alesina, 988, Alesina and Rosenthal, 995). The Nash equilibrium that is determined in the game with two parties of this sort is unique (Calvert (985)), and does not generally involve the complete convergence of the two parties. The platform adopted by each party lies between their own ideal policy and the ideal policy of the median voter. The models to be described in the following sections build on these foundations, in order to view the extent of policy convergence in two-party systems under ideological motivations. 4.. One-shot Nash-equilibrium with full information I need to introduce some further notation to fully consider parties interaction. To be more specific, I consider two parties and a continuum of citizens located on a one dimensional issue space: Assumption : The issue space I is represented by the unit interval [,]. The voter s belong to the continuum set T, and have preferences that are symmetric and single-peaked over I. Each voter will therefore choose to support the party that offers the platform closest to the voters ideal point. Given the issue space I and the continuum set T, it is possible to represent the distribution of voter s ideal point, by the cumulative distribution function F. 8 This result derives from the seminal works by Hotelling (99), Black (958) and Downs (957). Authors such as Hinich (977), Enelow and Hinich (984), Coughlin and Nitzan (98), Ledyard (984) and Tovey (99), instead, have considered that convergence might not be coincident with that of the median voter (for discussion see Mueller (3)). Congleton (989), on the same line of reasoning, has considered the extent to which the campaign contribution of politically active groups may draw candidates away from the position of the median voter in order to rise funds. 9
Assumption : F : I [, ] is continuously differentiable and strictly increasing with F() = and F() =. The continuous density function f, generated by F, f : I [, ], is single peaked or strictly quasi-concave, i.e. there is a unique x I such that f( ) is strictly increasing on the interval [, x ] and is strictly decreasing on the interval [x, ]. I assume that Assumptions and hold unless stated otherwise in the model that follows. Furthermore, I will denote by Pb(x, x ) the probability that party wins the election (so that - Pb(x, x ) will denote the probability that party wins the election). The probability function Pb(x, x ) is assumed continuous and twice differentiable in each argument. It is now possible to analyse the one-shot Nash equilibrium when the political system is characterised by two ideological parties. For this reason I introduce the Alesina model for ideological parties motivation. 4.. The ideological party case The analytical starting point of this section is represented by Alesina s work (988). Here I apply that general model of two party competition to the issue of regulation policy. The results obtained determines an equilibrium regulation policy followed within the two-party electoral system with ideological parties. As considered before, the parties do not maximise the prospect of winning power. They aim to serve their own ideological, or serve their respective constituencies. The utility functions are then: () U =U (s,t )d for party, and, for party : () U =U (s,t ) where d again represents the positive utility payoff (if any) from being in office per se, s is the sum paid to the regulatory agency, and T is the tax levied on the consumer/voter 9. The party s utility function depends on the tax T and the payment s because, in this model, I want to consider two relevant effects: a direct and an indirect ones. The direct effect is relative to taxation. Of course the consumer will support the party if ceteris paribus, she agrees with the party s tax proposal in particular if it offers a lower tax holding other variables constant. The indirect effect instead, refers to the fact that the transfer to the agency affects the firm s activity (the quantity and hence the utility of the consumer. The party s probability to be re-elected also depends on the 9 In this case I have considered the transfer to the agency and the tax T on consumers instead of considering the argument q as in section 3. The results are similar.
level of output enjoyed by the consumer, and so there may be a trade off between lower taxes and improved service provision. Each party proposes alternative policies with respect to the optimal situation T*, s* where the firm produce at MC=P and the agency receives the exact reward required for efficiency. Party s ( s) policy is indicated by s,t (s, T ). Following Alesina s reasoning (988), the parties objective functions could be expressed in quadratic form: () U ( T ) t = q u( T t) = q t= t ( T t c ) c > and < q < for party, and for party : ( 3) t t U ( T ) = q u( T t) = q T t [where, as before, q t = discount factor (equal for the two parties); and c = bliss point controlled by the party in office. I assume it equals for party ]. In equations () and (3) I have omitted the reward s to the agency. This is because I have supposed the condition T=s. The tax level determination then becomes the crucial variable manipulated by the party to obtain political consent. The voting decisions are then based upon the citizen s rational expectations of policies that the two parties would follow if elected. These policies are T and T, for party and party. At the end of time t-, the parties announce their policies for time t: T a a t and T t. If voters believe in it then T e t = T a t and T e a t = T t otherwise T e a t = E(T t I t- ) T it (I t- = information set available to the voter at time t-). The electoral outcome is uncertain. For party victory is associated the probability P=P(T e, T e ) (giving the condition T=s). It is also assumed that ) P is time invariant and common knowledge;