Inner-city traffic congestion – Prisoners’ dilemma and social preferences Torsten Marner Institute of Transport Economics University of Munster Am Stadtgraben 9 48143 Munster Germany phone: +49 (0)251 8322818 email: torsten.marner@iv-muenster.de Inner-city traffic congestion – Prisoners’ dilemma and social preferences Torsten Marner Abstract This paper presents an analysis of inner-city traffic congestion in order to evaluate different instruments and their ability to cope with congestion. It is assumed that users do have heterogeneous preferences that are differentiated in materialistic, reciprocally motivated and altruistic preferences. Dependent on the interaction of users with different preferences, various instruments and their impacts on the model are analyzed. It is shown that the consideration of social preferences may change the user equilibrium, so that instruments that are proposed under the assumption of strictly selfish user behaviour need to be scrutinized. However, if materialistic preferences are predominant, measures such as road pricing are appropriate instruments. If social preferences are predominant, it may be better not to implement road pricing. 1. Introduction Traffic congestion is to be regarded as one of the central problems of inner cities. Traffic volume systematically exceeds road capacity because of an inefficient use of the restricted road capacity. Thus, congestion and its well known consequences such as time losses, environmental pollution and noise arise. As a consequence, significant welfare losses are originated. For instance for Germany, congestion costs are estimated in a range between 16,35 and 193,46 bill. € (Frank and Sumpf, 1997; Link, 2002; INFRAS and IWW, 2004), dependent on the method of evaluation. An overview of international congestion costs is given by Krause (2003), Kossak (2004) and INFRAS and IWW (2004). Especially in inner-cities congestion costs are significant. Based on INFRAS and IWW (2004), the proportion of innercity congestion costs amounts between 26% and 32% of the whole congestion costs in Germany. Compared with federal highways and motorways, inner-city streets suffer from significantly higher marginal congestion costs (Newbery, 1990; Sansom et al., 2001; INFRAS and IWW, 2004). An overview of investigations of marginal congestion costs is presented by Mayeres (1996) or Schrage (2005). Assuming individual rationality and selfish user behaviour inner-city traffic congestion is presentable as stable Nash-equilibrium. In spite of congestion, users do not have incentive to change their strategy “use the car” to drive into the inner-city. This causes a classical prisoners’ dilemma, because all users were better off, if they would use a cooperative strategy such as using public transport instead of using the car (Rapoport, 1967; Tucker, 1983). In chapter 2 we model the binary decision problem of the users to choose either car or public transport to arrive at the inner-city at a particular time. Referring to Schelling (1978) and McCain (2003) we use a game theoretic n-person framework, but in contrast to Schelling and McCain we additionally consider capacity limits and bus tracks and trams, as it is usual in larger inner cities. Based on Schelling (1978) and Dawes (1980) we show in chapter 3 that our model represents a prisoners’ dilemma. Under the neoclassical assumption of individual rationality and thus selfish user behaviour the users do not have any chances to break this dilemma, so that external interventions such as road pricing are essential. In economic literature road pricing is regarded as efficient instrument to mitigate congestion (Pigou, 1920; Knight, 1924; Walters, 1961; Eisenkopf, 2002). The aim of the paper is to evaluate, if an external intervention is essential to battle inner-city congestion. Fur this purpose it is reasonable to take the heterogeneity of users in account. Following this basic approach, we introduce heterogeneity of users in chapter 4. Thus, we reject the assumption of unconditional selfish user behaviour by considering social preferences such as altruism and reciprocity (Hackett, 1993; 1994; Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000). In contrast to common economic literature we try to show the impacts of the existence of social preferences on the user equilibrium. Numerous experimental investigations document the existence of social preferences in real life (Güth et al., 1982; Roth, 1995; Güth, 1995; Fehr and Fischbacher, 2002), especially by analyzing games such as the ultimatum game. Hence, it seems to be recommendable to use these thoughts for transport economy too. However, if materialistic preferences are predominant, an external intervention will be essential. Chapter 5 consequently demonstrates the impacts of different measures to battle congestion: capacity extensions, road pricing and improvements of public transport (Marner, 2004). If social preferences are predominant, it may be not reasonable to battle congestion by coercion, because in the case of existing social preferences users may be able to mitigate congestion without coercion. Chapter 6 analyzes the interrelation between social preferences and cooperative behaviour, before chapter 7 draws a conclusion. Numerous game theoretic investigations of traffic behaviour are available so far. An interesting and actually overview of these analyses is presented by Hollander and Prashkar (2006). 2. Modelling inner-city traffic congestion Inner-city congestion can be seen as a logical consequence of users’ individual rationality. Individual rationality means that users will choose those alternatives with highest expected utility. Our model shows that at restricted capacity, congestion can be understood as stable user equilibrium with actually realized traffic volume exceeding capacity (Marner, 2004). With the aid of figure 1 we analyze the users’ decision to use either the car or public transport at a particular time starting from a point A outside the inner-city to a point B inside the innercity.i Figure 1 shows the utility of a particular user on his way towards the inner-city either as car user or as public transport user given a particular allocation of traffic.ii - Insert figure 1 here The ordinate of the coordinate system documents the users’ respective utility, whereas the abscissa illustrates the proportion α of the users who choose the car to cover the distance AB at a particular period to arrive at the inner-city. The continuous line represents the utility of the public transport users. If α = b (or as well α = 0), his utility will be y. The dashed line however documents the car users’ utility. Using the car at α = b (or as well α = 0) implies a utility x > y. There is a nearly perfect correlation between α and traffic density (Bauriedl and Winkler, 2004). The lower the proportion of car users at AB, the more players uses public transport to enter the inner-city. Thus, fewer vehicles are needed to transport the fixed number of road users. Because of the higher capacities of busses or trams a particular number of cars will be substituted by a bus or a tram. An articulated bus for example is able to transport as much passengers as some 74 passenger cars (Verkehrsbetriebe STI, 2006), while the passenger car unit (PCU) of a bus is – dependent on the length – 2,5-3 (Transport Research Laboratory, 1996). According to this the net effect of the substitution of a car by a bus or a tram is a 25 to 30 fold relieve of the streets. Independent from a concrete dimension, traffic density decreases significantly, so that α can be interpreted as traffic density. Consequently, the strategy of each player “to use public transport” can be defined as cooperative strategy C, whereas the strategy “to use the car”, that tends to result in a stimulation of congestion, can be defined as defecting strategy D. b can be interpreted as the capacity limit of the considered road. For all α ≤ b road capacity is not yet exhausted and therefore road users are not confronted with time losses while driving from A to B. If α > b, the demand for utilization of the road exceeds capacity limit resulting in congestion. Thus, for all α > b, the users are faced with opportunity costs of time. The graphs of the utility functions of car users and public transport users run parallel for all α ≤ b. Nevertheless, the utility of car users exceeds the utility of public transport users in this range. This assumption is based on two groups of arguments: cognitive-rationale arguments, that make allowance for trading off perceived costs and utilities, emotional arguments, that make allowance for aspects as reputation and feelings (Steg et al., 2001). Cognitive-rationale reasons for a higher utility of choosing the alternative “car” in the case of “no congestion” can be differentiated in flexibility, independence, reliability, accessibility, speed, convenience and easier use of the car as means of transportation of goods (Hensher and Stopher, 1979; Recker and Golob, 1976; Jakubowski et al., 1998; Steg, 2003). Examples for emotional reasons can be craving for sensation, feeling of power, feeling of superiority, enjoying motoring, feeling of freedom and privacy protection (Näätänen and Summala, 1976; Lawton et al., 1997). For all α > b congestion implies different impacts on the utility functions of car users and public transport users. Although in the case of congestion the utility of the car user decreases as well as the utility of the public transport users, however we identify three reasons for a lower mitigation of utility in the case of public transport use than in the case of car use: First, car users do not have the same options to escape congestion as public transport users often do have by using non-congested bus tracks. Second, for rail-bound traffic as one part of public transport there is no congestion problem anyway.iii Third, stress-related mitigation of utility in case of congestion is assumed to be higher for car users because of their active driving role. Thus, an increasing α determines a sharper decline of the graphs of the utility functions in the car users’ case (Marner, 2004). The different slopes of the two curves imply the existence of an intersection point N. N’s position is dependent on the capacity limit b and the devolution of the utility curves that are determined by users’ preferences. Assuming users to be individual rational, N with α = c emerges as stable Nash-equilibrium (in pure strategies). A Nash-equilibrium exists if and only if “no player has incentive to deviate from his strategy given that the other players do not deviate” (Rasmusen, 1994).iv If N is therefore identified as stable user equilibrium, there will be no incentive for any user to change, neither from public transport to car nor vice versa. Thus, the proportion of car users is defined α = c, the proportion of public transport users is defined = 1-c and the resulting utility is equal for every user z. Given the above mentioned and absolutely realistic assumptions, the distribution (α,) of users on the regarded route AB converges independent from the actual proportion α towards intersection point (c, 1-c). This can be explained as follows: Assume α to be c with the actual equilibrium point (c, 1-c). Is there incentive for any player to change her strategy? If e.g. a car driver wishes to use public transport from now on, the proportion of car users α changes marginally to α’ = c - with symbolizing the proportion of exactly one new car driver of all users on the regarded route. Consequently, at the point α’ = c - , for the new public transport user the continuous line of the utility function is valid, which however at α’ = c - proceeds below the dashed line so that at α’ = c - each player has incentive to use the strategy “car”. Thus, the regarded player would do better returning to the starting point N with at α = c. Hence, there is no incentive to change the strategy. In analogy, at user equilibrium N there is no incentive for any public transport user to change her strategy. N is not only a stable Nashequilibrium but also a unique equilibrium, because all thinkable distributions converge towards N. Each initial distribution below α = c, e.g. at α’ = c - 2, α’ = c - 3 etc. a fortiori implies incentive for a change from public transport use to car use and each initial distribution above α = c, e.g. at α’ = c + 2, α’ = c + 3 etc. a fortiori implies incentive to change from car use to public transport use (Marner, 2004). Thus, in the case of n users, equilibrium point N with (α,) = (c, 1-c) is realized. 3. The prisoners’ dilemma For each user the existence of equilibrium point N implies a utility of z < y, so that in the case of mutual choice of the cooperative strategy “public transport use” the utility for each of the users would be higher. Assuming individual rationality, at least a proportion α = c of users choose the defective strategy “car use”, resulting in an undesirable outcome for each user and therefore in an inefficient equilibrium N with (α,) = (c, 1-c). This inefficiency can be characterized as an alleviated multi-person prisoners’ dilemma. The prevalent definition of a multi-person prisoners’ dilemma consists of two conditions: a) Each user benefits more from a socially defecting choice, i.e. using the car to arrive at the inner-city, than from a socially cooperating choice, i.e. using public transport, independent from other users’ strategies and b) all users are better off if all choose the cooperative strategy than if all choose the defective strategy (Dawes, 1980). The mathematical structure is as follows: Each player chooses one of two available strategies: Defecting (D) or Cooperating (C). Each players’ utility depends on the own strategy and on the number of players choosing strategy D or C. Let D (β) be the utility of the users who defect in the considered multi-person game with α being the proportion of players who defect and β being the proportion of players who cooperate. Let C (β) be the utility of the users who cooperate when β is the proportion of players who cooperate (including themselves). A prisoners’ dilemma is characterized by the following inequalities: c) D (β) > C (β+ε) und d) D (β=0) < C (β=1). Inequality c) means that in the case of parallel utility curves defecting is a dominating strategy for all users. Inequality d) indicates that collective defection confronts all players with less preferred results as in the case of collective cooperation (Dawes, 1980).v However, in our case of non-parallel curves with an interception point N, condition a) as well as inequality c) is not fulfilled, because a proportion of users is better off by choosing strategy “cooperate”. Hence, by definition, the players are not confronted with a prisoners’ dilemma, but in fact, they are. The only difference between the case of parallel lines of utility, that satisfies the conditions of a prisoners’ dilemma, and the case of intersecting curves, that does not satisfy the conditions of a prisoners’ dilemma, is that in the first case, at the equilibrium no player uses the cooperative strategy, but in the second case, at the intersection point N, some players do use the cooperative strategy. Both equilibriums are inefficient and in both cases the players’ utility would be greater than in the case of collective cooperation. The only difference is the extent of the dilemma. With intersecting curves, the player’s utility is greater than in the case of parallel curves, but by all means less than the collective cooperative utility (Schelling, 1978). So in fact, our binary decision problem can be interpreted as an alleviated multi-person prisoners’ dilemma. Based on Schelling’s definition of a prisoners’ dilemma with an inefficient potential equilibrium, our binary decision problem must be called a prisoners’ dilemma, because it exactly satisfies the following conditions: “There is a unique equilibrium with some choosing …. [D], some …. [C]. …All would be better off if all chose…[C]. The collective total would then be a maximum. The collective total would be even higher if some still choose … [D], everybody still being better off than at the equilibrium, but not equally so.” (Schelling, 1978). The existence of a prisoners’ dilemma implies an inefficiency that calls for measures to change the incentive structures of the players. Social dilemmas in general are defined in terms of social payoff structures. Thus, the simplest method for stimulating cooperative behaviour is to change those payoff structures. This is possible by providing incentive to use the cooperative strategy, in our case “use public transport”. Incentives can be differentiated in internal and external incentives to be cooperative. Internal incentives are based on social preferences of the users. Those can be differentiated in aspects such as moral, normative concerns, fairness, reciprocity, inequity aversion and altruistic concerns (Hackett, 1993; 1994; Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000). By providing internal incentives to the players to be cooperative equilibrium may convert from an inefficient to an efficient equilibrium and the prisoners’ dilemma breaks up. To ensure cooperative behaviour in dilemma situations the often used assumption of completely rational and selfish behaviour must be rejected. This is true for a) the players have to contemplate the situation and understand the complexion of the dilemma, so that possibly existing social preferences as well as external interventions can influence the players’ decisions and b) the players must have beliefs that the other players will not defect, because they are able to anticipate a greater utility in the case of collective cooperation than in the case of non-cooperation, albeit individual rationality suggests defective behaviour (Dawes, 1980). If social preferences exist in an adequate proportion, no coercion by means of political interventions will be necessary to procure an efficient equilibrium. External incentives are based on the idea, that individuals are not able to break a dilemma situation without external intervention. Thus, a dilemma could legitimate a political measure. External authorities could implement a political measure to correct existing but inefficient incentive structures by coercion.vi There are three categories of reasonable measures to break a dilemma situation: a) providing incentives not to defect, i.e. implementing measures that lower the attractiveness of using the car, b) providing incentives to cooperate, i.e. implementing measures to strengthen alternatives to using the car, e.g. improving public transport, c) capacity extensions (Marner, 2004). Both, internal and external incentives tend to result in a change of social payoff structures and therefore are appropriate measures to break up the dilemma. The efficacy and adequacy of the nature of the incentives are dependent on the characteristics of the users. So, in contrast to many theoretical models, we absolutely realistic have to assume that users are heterogeneous. 4. Heterogeneity of users The conventional assumption referring to economic behaviour is that there is no heterogeneity of users concerning their preferences. In fact, users do have different preferences, so that the assumption of collective selfish behaviour, that is caused by unconfined materialistic preferences, must be rejected (Hackett 1993;1994; Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000; Poulsen and Poulsen, 2002). Not only real life suggests evidence for the existence of social preferences. Charity or tipping the waiter in face of the knowledge, that you will never see him again, so that you cannot expect a return flow, can be interpreted as consequences of social preferences. But also scientific evidence is given by numerous experimental investigations (Güth et al., 1982; Roth, 1995; Güth, 1995; Fehr and Fischbacher, 2002), so that it is reasonable to consider heterogeneity of users concerning their preferences. We distinguish two categories of preferences (Fershtman and Weiss, 1998; Guttmann, 2000): rational, selfish preferences and social preferences. Rational preferences are to be defined as follows: a rational and selfish user chooses the alternative with highest expected utility for him, completely independent from other players’ utilities. For users with rational, selfish preferences, exclusively materialistic aspects are relevant for their utility functions, so that those users seek to maximize their monetary return. Thus, we restrict rational and selfish preferences to materialistic preferences (Poulsen and Poulsen, 2002). In the following we restrict social preferences to contain the following types of preferences: altruism and reciprocity. Altruism may be defined as unconditional niceness. Independent from the behaviour of other users, altruistic users always tend to be kind (Hirshleifer, 1977; Dawes, 1980; Becker, 1982). Altruism is to understand as the phenomenon that the welfare of other users positively influences the utility of the regarded altruistic user (Axelrod, 2000). Reciprocity may be defined as “conditional niceness” (Poulsen and Poulsen, 2002). Basically these users are kind, until they face unkind behaviour of others, so that their initially kind behaviour turns to unkind behaviour. Thus, reciprocity means to be kind towards players, who are expected to be kind and to be unkind towards players, who are expected to be unkind (Axelrod, 2000, Poulsen and Poulsen, 2002). Consequently, the possible outcomes of the existence of reciprocally motivated users are ambiguous and depend on the composition of the user group (Ehrmann et al., 2006). As shown above, we restrict the analysis to three relevant types of preferences: the materialistic preference type (m), the reciprocator preference type (r) and the altruist preference type (a) (Poulsen and Poulsen, 2002). Let a simplified prisoners’ dilemma have the following monetary payoffs that are illustrated by figure 2. - Insert figure 2 here – Let A > 1 > 0 > b be true and C and D stand for “cooperate” and “defect”. Individual rationality leads players exclusively to care about their own monetary utility, so that outcome 0 is realized, that is worse than 1. Different preference orderings generate different user behaviour. From there preference orderings, we can derive following best replies. Best replies document the decision players choose, given that opponent players choose either C or D. For players with materialistic preferences (m), D is a strictly dominant strategy. For players with altruistic preferences (a), C is strictly dominant. Reciprocally motivated players choose C, if they expect opponent players to be cooperative and D, if they expect them to be defective. Here, the outcome (C,C) is strictly preferred to other outcomes with (C,C) > (D,C) > (C,D) > (D,D). You have to keep in mind, that these simplified accomplishments with dominant strategies refer to the simplified illustration of figure 2 to simplify the understanding of the impact of social preferences. Materialistic preferences (m) and altruistic preferences (a) are non-critical for the analysis, because m-players generally use strategy “defect” and a-players generally use strategy “cooperate”, but the choice of reciprocally motivated players remains ambiguous, because it depends on the decisions of the unknown opponents. Thus, in the case of reciprocally motivated players, the interaction of players is of vital importance. There are different determinants that explain the impacts of interaction of heterogeneous players. First, the proportions of players with m-, r- and a-preferences play an important role and second, the beliefs of the reciprocally motivated users are important. If they expect to meet rather altruistic or reciprocally motivated than materialistic motivated players, they will tend to use cooperative strategies and therefore, the proportion of cooperative strategies increases and the user equilibrium tends to move towards a non-congested situation. Figure 3 illustrates the impacts of the consideration of social preferences by showing the interaction between users with heterogeneous preferences in an extended game. - Insert figure 3 here – The tree of figure 3 documents, that for selfish players D is a dominant strategy, so that they always use the defective strategy, independent on their belief of the composition of the user group. In contrast to selfish players, altruists do have a dominant strategy C. Reciprocally motivated players do not have dominant strategies. If they are confronted with a defective strategy, they will react by using a defective strategy in the next move, if they are confronted with a kind player choosing a cooperative strategy, they will be kind too and choose a cooperative strategy in the next move, too. Starting with a materialistic player, we see, that in the first move of our game, exclusively defective strategies are used (3 out of 3). The second move consists of three subgames with completely nine best replies. In the left subgame the m-player makes the choice, in the middle subgame it is the r-player and in the left subgame it is the a-player who makes the choice. In the second move, in 66, 67% (six out of 9) of the cases the respective player’s best reply is D. In the third move the proportion of D as best reply decreases to 55, 56% and in the fourth move to 51, 85%, in the fifth move to 50, 62% and so on. The limiting value, still for starting with a materialistic player, is 50%. If you start with an altruistic player at root of the tree, the left and right subgames remain unmodified and only the middle subgame changes. But even the middle subgame tends to a limiting value of 50% for n → ∞. Thus, assuming a rectangular distribution of preferences, independent from the starting point, 50% of the strategies are cooperative. This can be interpreted as an evidence for the functioning of cooperation, even without coercion. The following chapters 5 and 6 take up these thoughts and assume that a dominance of selfish or materialistic preferences causes the necessity of coercion, because in the case of a higher rate of defecting the user equilibrium tends to be rather at a congested state, while with a dominance of social preferences coercion rather should be rejected, because natural cooperation tends to be higher, so that the user equilibrium is rather near the capacity limit. 5. Rationality, selfish behaviour and instruments to minimize congestion Assuming a dominance of rational and selfish user behaviour, the efficacy of social preferences concerning users’ behaviour is negligible. Thus, internal incentives do not have any relevance for the user equilibrium. Consequently, there is a need to revert to external incentives by means of political intervention. In this context, road pricing can be seen as the most efficient instrument to mitigate welfare losses caused by congestion (Eisenkopf, 2002). Two central concepts concerning road pricing are user equilibrium and system optimum of traffic flow distribution in road networks. User equilibrium means that all users try to minimize their individual travel costs, whereas system optimum means, that users cooperate to minimize total network cost. The difference between user equilibrium and system optimum causes a welfare loss. Road pricing and other instruments to battle congestion have the aim to minimize this difference. The effectiveness of an implementation of road user charges corresponds to the fundamental principal of marginal-cost pricing, which suggests, that at congested roads users have to pay an optimal toll equal to the difference between marginal-social cost and marginal-private cost with the objective of maximizing social surplus (Dupuit, 1844; Pigou, 1920; Knight, 1924; Walters, 1961). By internalizing the external effects of congestion by means of optimal pricing, incentives for an efficient use of the scarce good road infrastructure are provided. Thus, density decreases, until capacity utilisation is welfare optimal and the welfare loss caused by the external effect “congestion” is avoided. In economic literature it is common sense, that first best solutions are not feasible, so that numerous second best solutions are proposed (May and Milne, 2000; Sumalee, 2004; Yang and Huang, 2005). Albeit those solutions do not generate efficiency like first best solutions do, they at least are more feasible than first bests because of lower transaction costs and major practicability. Furthermore, do not generate considerably higher efficiency than conveniently implemented second bests appear to do (Proost and van Dender, 1999). Empirical evidences from the implementation of the London City Congestion Charge (Marner, 2004; Eichinger and Knorr, 2004; Prud’homme and Bocajero, 2005) and Stockholm’s recently successfully completed trial (Armelius and Hultkrantz, 2005; Abboud and Clevstrom, 2006) demonstrate, that second best solutions my be successful for traffic control purposes. Already the experiences of the pioneer of road pricing, Singapur with its electronic road pricing ERP can be evaluated predominantly positive (Olszewski and Xie, 2005). Especially after these positive experiences and after the recently implemented tolling of the inner-city of Bologna it is not amazing, that numerous large cities such as Paris, Brussels, Oslo, Milan or Helsinki are concerned with concrete plans for an implementation of a pricing scheme for the congested inner-cities (de Palma et al. 2006a; de Palma et al. 2006b; Vold, 2006; de Palma/ Lindsey, 2006; Proost/ Sen, 2006; o.V., 2006a; o.V., 2006b). Adequate overviews of the concrete theory of urban road pricing are presented by for instance May and Nash (1996), Arnott (2001) or Yang and Huang (2005). In the regarded model framework of figure 1, road pricing causes the following, whether by optimal pricing or second best pricing: Users are assumed to be heterogeneous and therefore have different willingness’s to pay. Thus, a differentiated analysis of user groups is required. Those users with a willingness to pay exceeding the toll continue using the defective strategy by using the car and are confronted with two effects: first, their utility is mitigated by the increasing price of using the road infrastructure. Second, users’ utility will increase by saving opportunity costs of time, if congestion is mitigated. If net effect is positive, so that the average of the regarded category of users is better off or after implementing pricing, predominantly depends on the amount of the toll and the preferences of the users. As the aim of pricing is the mitigation of congestion, it has to be guaranteed, that the amount of the toll is high enough to burden the average of the user group. In this case, the average incentive to use the car decreases. Figure 4 illustrates the impact of pricing by shifting the utility curves of the car users downward. - Insert figure 4 here – Users with a willingness to pay that is lower than required to cover the user price – that could be interpreted as the sum of individual conventional costs and the toll – will be crowded out by either making them change their strategy to be cooperative or putting them out of the market. Exclusively considering the problem of congestion, this development is enjoyable, because hence decreases and therefore the equilibrium point N of figure 1 is shifted towards a non-congested state. A sufficient toll tends to activate a crowding out of car users and therefore enforces cooperative behaviour by thenceforward at least a few users will have incentive to use public transport. Thus, similar to the idea of internalisation of external effects by Pigou, car users are crowded out, traffic density decreases and the congestion-caused welfare loss decreases too. In Figure 4, the thick drawn, dashed line illustrates the utility of a car user after implementation of pricing, while the thin drawn, dashed line illustrates his utility before pricing. It is shown, that road pricing shifts the proportion of car users from c0 to c1 with a significant decrease of congestion, equivalent to the minimization of the distance c-b. c0-b is significantly greater than c1-b, so that the Nash-equilibrium changes from N0, that is true for the non-pricing case, to N1, that is true for the pricing case. The prisoners’ dilemma is minimized, because the difference between z1 and y is significantly minor to the difference between z0 and y. Thus, basically the implementation of a pricing system can be evaluated as an appropriate measure to solve the inner-city congestion problem. However, choosing an appropriate amount of the toll that is dependent from several determinants such as politicoeconomic aspects (Evans, 1992; Gomez-Ibanez, 1992; Lehmann, 1996; Jones, 1998) and individual urban frameworks (Marner, 2004) is of great importance. In general, for an adequate amount of toll is true, that the toll has to be exactly so, that the car users exactly from this amount on have incentive to change their strategy. Figure 5 shows the impacts of an optimal charge on the users’ strategies simplified for two players. - Insert figure 5 here – First matrix illustrates the utilities of the players in the case without charging. Assuming a) an ordinal ranking T > R and P > S and b) 2R > T+S > 2P, the Nash equilibrium is the strategy combination (car use, car use) with an utility of 2P for both users, that is lower than 2R, that could be achieved by a collective rational choice of the strategy (public transport use, public transport use). This implies a stable prisoners’ dilemma. Second matrix illustrates the case with optimal charging. A price topt causes a more expensive use of the regarded route AB for the car users with an initially mitigated utility. Consequently, for a part of users T-topt < R and P-topt < S is true. So, this is true for those users with their willingness to pay not covering the increase in prices. Those users do not only have a modified payoff structure as every user does have, for them it is furthermore individual rational to change their strategy and use public transport after pricing. They will be crowded out, so that decreases and converges to capacity limit b. For all car users, topt causes an average mitigation of utility, so that the considered utility curve at figure 4 is shifted downward. It must be pointed out, that the positive welfare effect that is caused by optimal pricing will not be overcompensated by prohibitive high transaction costs or system costs that could generate a governmental failure (Prud’homme and Bocajero, 2005; Baum et al., 2005). Further actual discussed alternatives to reduce inner-city traffic congestion are a) capacity extensions and b) improvements of public transport. a) capacity extensions Capacity extension in general causes an increase of supply of street space and tends to reduce congestion by shifting the capacity limit b of figure 1 rightwards (Small et al., 1989). However, there are several significant restrictions to capacity extensions: There are substantial urbanistic, political and spatial barriers, so that capacity extension would be either impossible or at least prohibitive expensive. You have to consider – perhaps in the context of cost-benefit-analyses – substantial investment costs. Capacity extensions imply induced traffic (DeCorla-Souza and Cohen, 1998; Marte, 2005), so that, following Downs (1962) and his fundamental law of congestion extension of infrastructure coercively generates an increase in volume of traffic. Induced demand will cause a new short-term congestion, if capacity of the road network is extended (Downs, 1962). Figure 6 illustrates the consistence of this idea with the implications of our model. - Insert figure 6 here – Capacity extension causes a shift of the capacity limit b0 to b1. Consequently, the thin drawn utility curves are shifted rightwards, so that in the case of capacity extension the thick drawn curves are true. At (c1,z1) a new stable Nash-equilibrium (in pure strategies) N1 is originated, but however, c1 and b1 are equidistant to c1 and b0, so that congestion remains constant. The new equilibrium N1 implies z1 = z0, so that capacity extension does not influence the utility of the concerned users. In our model, induced traffic (c1-c0) follows Downs’ prediction and compensates capacity extension that remains inefficient. Indeed, you have to take into consideration, that infrastructure investments generally induce significant positive welfare effects (Aschauer, 1989; Gramlich, 1994; Baum and Behnke, 1997; Baum and Kurte, 1999; Vickerman, 2000, Hartwig and Armbrecht, 2005). b) improvements of public transport We investigate two categories of an improvement of public transport that have different impacts on the utility curves of our model. First, we analyze substantial improvements of the capacities of busses by implementing new routes, higher frequencies or larger busses and by supplying an improved service, for instance by implementation of a 24 hour-service. In our model, these measures cause a parallel translation of the utility curves of the public transport users. Second, we analyze an improvement of bus lanes or a consideration of tram or subway systems. These measures cause a more even decline of the regarded utility curve. Figure 7 illustrates the impacts of both cases. - Insert figure 7 here – After consideration of an improvement of public transport, at (c2, z2) a new stable Nashequilibrium N2 is originated. Compared with N1, N2 converges to the capacity limit b, so that the dimension of the social dilemma significantly decreases. Empirical evidences, i.e. of the London case, show that improvements of public transport can be seen as appropriate accompanying measures, if the revenues of road pricing are used for public transport (Marner, 2004; Santos and Bhakar, 2006). In this context, Armelius and Hultkrantz (2005) show that in larger cities with a greater significance of public transport, such as Stockholm or London, it is more unproblematic to compensate users by using the revenues for public transport. The aggregate impact of the measure package “road pricing plus improvement of public transport” is documented by figure 8. - Insert figure 8 here – The thin drawn curves illustrate the status quo that is shown in figure 1, while the thick drawn curves show the impact of the package “road pricing plus improvement of public transport”. Equilibrium converges from c0 to c3 = b and implies – dependent on the intensity of intervention – a non-congested situation. In the case shown in figure 8 an intensive intervention breaks the dilemma. However, at least, we identify a relocation of the equilibrium towards a non-congested situation. The utility of the users increase from z0 to z3, that is however lower than x, but higher than y. The dilemma is solved, congestion decreases or eventually disappears and the congestion caused welfare loss is eliminated or at least lowered. 6. Social preferences and cooperative behaviour Assuming a dominance of social preferences, road pricing or other instruments that could be implemented by coercion, will be redundant. Given the assumption, that a significant proportion of preferences are social, the users will make the user equilibrium shift from N0 to N3, as figure 9 shows. - Insert figure 9 here – The realization of N3 implies a utility of z4 for the users and therefore a decreasing welfare loss. However, dependent on the degree of social preferences, congestion may not be completely eliminated, but following arguments may speak against an intervention: First, congestion is a temporary phenomena. The route AB is not congested permanently and furthermore you have to consider temporal and spatial relocation, so that second, a cost-based reduction of traffic demand would generate welfare losses. In economy it is common sense that traffic generates substantial welfare effects (Aschauer, 1989; Gramlich, 1994; Baum and Behnke, 1997; Baum and Kurte, 1999; Vickerman, 2000; Hartwig and Armbrecht, 2005). Thus, an optimal congestion is existent at b + γ. γ albeit tends to be small and depends on the structure of traffic flow and tends. The existence of an optimality of congestion means, that the occurrence of social preferences could be adequate for a minimization of congestion. Third, the system costs and transaction costs of an intervention have to be taken into account. Prohibitive high costs, as identified for the London case (Prud’homme and Bocajero, 2005), involve the danger of a state failure. There are two further effects that influence the impact of social preferences. First, the analyzed daily problem of intermodal choice may be seen as a typical repeated game. The regarded decision problem recurs day by day. Repeated games tend to favour cooperative behaviour, because players learn to be trustful and eventually learn that the game not only consists of materialistic players, but of a mix of materialistic players and players with social preferences. Thus, in repeated games the probability of learning that collective behaviour tends to increase the utilities of all players is much higher than in one-shot games (Axelrod, 2000). Compared with this, anonymity is one important determinant of defective strategies. The probability of defecting is much larger in an anonymously environment. On congested routes, n → ∞ players are involved in the game, so that the degree of anonymity is very high. At this you have to differentiate between nearly permanent congestion at large cities such as London or Paris and probably temporary congestion at small cities. In small cities anonymity is significantly lower, because the probability that persons know each other is substantially higher. However, anonymity tends to favour defective behaviour, because especially in small cities with n being significantly lower than in larger cities each player rather has reinforcement control over other players (Dawes, 1980). Users in smaller cities with a lower degree of anonymity rather can punish other users for defective behaviour than in large cities with a degree of anonymity near 1. If users have the chance to punish defective users, they can influence others’ choices. For this case, Rapoport (1967) shows that prisoners’ dilemma is not a dilemma anymore. 7. Conclusions The model documented, that users are captured in a prisoners’ dilemma, if all users are assumed to be completely selfish. Allowing for heterogeneity, it is shown, that you have to differentiate users in players with materialistic, altruistic and reciprocally motivated preferences. If materialistic preferences dominate the regarded user group, instruments to provide users with incentive for an efficient use of the inner-city roads are essential. It is shown that road pricing in combination with improvement of public transport then is an appropriate measure to battle congestion. Capacity extension is not able to mitigate congestion because of the expected induced traffic. If social preferences dominate, laissezfaire may be the best policy, because social preferences cause users to behave cooperative, so that the prisoners’ dilemma tends to break. If social preferences are predominant, it will be furthermore easier to strengthen the consciousness of the users for cooperative behaviour. It is shown that the interaction of users plays an important role and that it is dangerous to ignore the heterogeneity of users and commend policies that are solely based on unconditional selfish users. What remains nebulous is the actual proportion of social preferences. However, numerous studies suggest a significant proportion of social preferences in real life, especially by analyzing ultimatum games (Güth et al., 1982; Rabin, 1993). Further experimental research could investigate social preferences especially in the transport sector with its specific characteristics such as anonymity of the users and repeated games. The analyzed daily problem of intermodal choice may be seen as a typical repeated game. While repeated games tend to favour cooperative behaviour, the fact, that especially in larger cities the users do not know each other and therefore the degree of anonymity is high, tends to favour defective behaviour. Thus, it would be interesting to learn more about the actual proportions of social preferences. References Abboud, L., Clevstrom, J. (2006) “Stockholm’s traffic buster”, The Wall Street Journal, Vol. XXIV, No. 147, 29.8.2006, p. 2. Armelius, H., Hultkrantz, L. (2005) “The Politico-Economic Link Between Public Transport and Road Pricing: An Ex-Ante Study Of The Stockholm Road-Pricing Trial”, Working Paper Series, Örebro University, Working Paper No. 8, 2005. Arnott, R. (2001) “The Economic Theory of Urban Traffic Congestion: A Microscopic Research Agenda, Paper presented at the workshop on Environmental Economics and the Economics of Congestion: Coping with Externalities”, Venice International University, San Servolo, July 18-19, 2001. Aschauer, D. (1989) “Is Public Capital Expenditure Productive?”, Journal of Monetary Economics, 23, pp. 177-200. Axelrod, R. (2000) Die Evolution der Kooperation, Oldenbourg, München. Baum, H., Behnke, N.C. (1997) „Der volkswirtschaftliche Nutzen des Straßenverkehrs“, Schriftenreihe des Verbandes der Automobilindustrie, Köln. Baum, H., Geißler, T. und J. Schneider (2005) „Pkw-Maut für Deutschland? – Eine kritische Analyse“, Studie für den Verband der Automobilindustrie (VDA) e.V., Frankfurt a.M.. Baum, H., Kurte, J. (1999) Wachstums- und Beschäftigungseffekte des Verkehrs, Deutsches Verkehrsforum, Köln. Bauriedl, S., Winkler, M. (2004) „Typisierung europäischer Regionen auf ihrem Weg zu nachhaltiger Entwicklung“, NEDS Working Papers, No. 4, 08/04, Hamburg. Becker, G.S. (1982) „Altruismus, Egoismus und genetische Eignung: Ökonomie und Soziobiologie“, Becker, G.S. (ed.), Der ökonomische Ansatz zur Erklärung menschlichen Verhaltens, pp. 317-332, Tübingen. Binmore, K. (1992) Fun and Games – A Text on Game Theory, D.C. Heath and Company, Lexington, MA. Bolton, G.E., Ockenfels, A. (2000) „ERC: A Theory of Equity, Reciprocity, and Competition“, American Economic Review, 90, pp. 166-193. Dawes, R.M. (1980) “Social Dilemmas”, Annual Review of Psychology, 31, pp. 169-193. DeCorla-Souza, P., Cohen, H. (1998) “Accounting for induced travel in evaluation of urban highway expansion”, Transportation Research Board, 77th Annual Meeting, 1998. De Palma, A., Lindsey, R. (2006) “Modelling and evaluation of road pricing in Paris”, Transport Policy, 13, pp. 115-126. De Palma, A., Lindsey, R. und E. Niskanen (2006) “Policy insights from the urban road pricing case studies”, Transport Policy, 13, pp. 149-161. De Palma, A., Lindsey, R. und S. Proost (2006) “Research challenges in modelling urban road pricing: An overview”, Transport Policy, 13, pp. 97-105. Downs, A. (1962) “The Law of Peak-Hour Traffic Congestion” Traffic Quarterly, 16, pp. 393-409. Dupuit, J. (1844) “On the measurement of the utility of public works”, Annales des Ponts et Chaussees, Memoires et Documents 2nd Series, 8, pp. 332-375, translated by: Barback, R.H. (1952), in: International Economic Papers, 2, pp. 83-110. Ehrmann, T., Hartwig, K.-H., Marner, T., Schmale, H. (2006) „Investitionsanreize im Schienenverkehr – eine experimentelle Untersuchung“, Internationales Verkehrswesen, 9/2006, pp. 398-404. Eichinger, A., Knorr, A. (2004) „Congestion Charging – das Beispiel Londons. Ökonomische Grundlagen und Zwischenbilanz“, Internationales Verkehrswesen, 9/2004, pp. 366-371. Eisenkopf, A. (2002) Effiziente Straßenbenutzungsabgaben – Theoretische Grundlagen und konzeptionelle Vorschläge für ein Infrastrukturabgabensystem, Aberle, G. (ed.), Giessener Studien zur Transportwirtschaft und Kommunikation, Band 17, Universität Gießen. Evans, A.W. (1992) “Road congestion pricing: When is it a good policy?”, Journal of Transport Economics and Policy, 26, pp. 213-243. Fehr, E., Fischbacher, U. (2002) “Why Social Preferences Matter – The Impact of Nonselfish Motives on Competition, Cooperation, and Incentives”, Economic Journal, 112, C 1-33. Fehr, E., Schmidt, K.M. (1999) “A Theory of Fairness, Competition, and Cooperation”, The Quarterly Journal of Economics, 114, pp. 817-866. Fershtman, C, Weiss, Y. (1998) “Why do we care what others think about us?”, Ben-Ner, A., Putterman, L. (ed.), Economics, Values and Organization, pp. 133-151, Cambridge University Press. Frank, D., Sumpf, J. (1997) „Abschätzung der volkswirtschaftlichen Verluste durch Stau im Verkehr“, BMW-Studie, München. Gomez-Ibanez, J.A. (1992) “The political economy of highway tolls and congestion pricing”, Transportation Quarterly, 46, pp. 343-360. Gramlich, E.M. (1994) “Infrastructure Investment: A Review Essay”, Journal of Economic Literature, 32, pp. 1176-1196. Güth, W. (1995) “An Evolutionary Approach to Explaining Cooperative Behavior by Reciprocal Incentives”, International Journal of Game Theory, 24, pp. 323-344. Güth, W., Schmittberger, R., Schwarz, B. (1982) „An experimental analysis of ultimatum bargaining“, Journal of Economic Behavior and Organization, 3, pp. 367-388. Guttman, J.M. (2000) “On the evolutionary stability of preferences for reciprocity”, European Journal of Political Economy, 16, pp. 31-50. Hackett, S. (1993) “Incomplete Contracting: A Laboratory Experimental Analysis”, Economic Inquiry, 31, pp. 274-297. Hackett, S. (1994) “Is Relational Exchange Possible in the Absence of Reputations and Repeated Contact?”, Journal of Law, Economics and Organization, 10, 2, pp. 360-389. Hamburger, H. (1973) “N-person Prisoner’s Dilemma”, Journal of Mathematical Sociology, 3, pp. 27-48. Hartwig, K.-H., Armbrecht, H. (2005) „Volkswirtschaftliche Effekte unterlassener Infrastrukturinvestitionen“, Studie im Auftrag des Bundesverbandes der Deutschen Zementindustrie, des Hauptverbandes der Deutschen Bauindustrie und des Verbandes der Automobilindustrie, Berlin. Hensher, D.A., Stopher, P.R. (ed.) (1979) Behavioural travel modelling, Croom Helm, London. Hirshleifer, J. (1977) “Shakespeare vs. Becker on Altruism: The Importance of Having the Last Word”, Journal of Economic Literature, 15, pp. 500-502. Hobbes, T. (1651, 1947) Leviathan, London. Hollander, Y., Prashker, J.N. (2006) “The applicability of non-cooperative game theory in transport analysis”, Transportation, 33, pp. 481-496. INFRAS, IWW (2004) “External Costs of Transport. Update Study“, Final Report, Zürich, Karlsruhe, Oktober 2004. Jakubowski, P., Knitschky, G., Lehmann, C. und H. Tegner (1998) „Motive und Ursachen von Personenmobilität verstehen“, Münster. Jones, P. (1998) “Urban road pricing: public acceptability and barriers to implementation”, Button, K.J. und E.T. Verhoef (ed.), Road Pricing, Traffic Congestion and the Environment, Issues of Efficiency and Social Feasibility, Cheltenham, Northampton. Kossak, A. (2004) „Straßenbenutzungsgebühren“, Internationales Verkehrswesen, 56, 12/2004, pp. 536-540. Krause, T. (2003) „Dynamische Straßenbenutzungsgebühren zur Reduzierung von Staus“, Hasenkamp, G. et al. (ed.), Volkswirtschaftliche Analysen, Band 10, Frankfurt/Main u.a.O. Lawton, R., Parker, D., Manstead, A. und S. Stradling (1997) “The role of affect in predicting social behaviors: The case of road traffic violations”, Journal of Applied Social Psychology, 27, pp. 1258-1276. Lehmann, C. (1996) „Die politische Ökonomie des Road Pricing“, Ewers, H.-J. (ed.), Vorträge und Studien aus dem Institut für Verkehrswissenschaft an der Universität Münster, Heft 30, Göttingen. Link, H. (2002) “The Pilot Accounts for Germany, UNITE Deliverable 5”, Ann. 1. Marner, T. (2004) „Die Problematik innerstädtischer Staus: ein spieltheoretischer Ansatz, Armbrecht, H. u.a., Politische Weichenstellungen für den Verkehr: Verkehrsplanung, Bepreisung, Deregulierung. Vorträge und Studien aus dem Institut für Verkehrswissenschaft and der Universität Münster, Heft 41, Göttingen 2004, pp. 61-108. Marte, G. (2005) „Die Bewertung des induzierten Verkehrs – ein Verfahrensvergleich“, Internationales Verkehrswesen, 11/2005, pp. 491-494. May, A.D., Milne, D.S. (2000) „Effects of alternative road pricing systems on network performance“, Transportation Research, 34 A, pp. 407-436. May, A.D., Nash, C.A. (1996) “Urban Congestion: A European Perspective on Theory and Practice”, Annual Review of Energy Environment, 21, pp. 239-260. Mayeres, I., Ochelen, S. und S. Proost (1996) “The Marginal External Costs of Urban Transport”, Transportation Research D: Transport and Environment, 1, 2, December, pp. 111130. McCain, R. (2003) Game Theory: A Non-Technical Introduction to the Analysis of Strategy, South-Western Educational Publishing, Belmont, CA. Näätänen, R., Summala, H. (1976) Road user behaviour and traffic accidents, Elsevier, Amsterdam. Newbery, D. (1990) ”Pricing and congestion: economic principles relevant to pricing roads”, Oxford Review of Economic Policy, Vol. 6, No. 2. Olszewski, P., Xie, L. (2005) “Modelling the effects of road pricing on traffic in Singapore”, Transportation Research Part A, 39, pp. 755-772. o.V. (2006a) „Mailand steht City-Maut bevor“, DVZ Nr. 99 of 19.8.2006, p. 2. o.V. (2006b) „Transporteure machen sich Luft“, DVZ Nr. 113 of 21.9.2006, Spezial: CityMaut Mailand, p. 7. Pigou, A.C. (1920) The Economics of Welfare, first edition, Macmillan, London. Poulsen, A., Poulsen, O. (2002) “Social Preferences in the Prisoner’s Dilemma Game: An Evolutionary Analysis”, Working Paper, Department of Economics, Aarhus School of Business. Proost, S., van Dender, K. (1999) “TRENEN II Stran – Final Report for Publication”, Leuven. Proost, S., Sen, A, (2006) “Urban transport pricing reform with two levels of government: A case study of Brussels”, Transport Policy, 13, pp. 127-139. Prud'homme, R., Bocajero, J.P., (2005) “The London congestion charge: a tentative economic appraisal”, Transport Policy, 12, 3, 2005, pp. 279-287. Rabin, M. (1993) “Incorporating Fairness into Game Theory and Economics”, American Economic Review, 83, pp. 1281-1302. Rapoport, A. (1967) “Optimal policies for the prisoner’s dilemma”, Psychological Review, 74, pp. 136-148. Rasmusen, E. (1994) Games and Information. An Introduction to Game Theory, 2 nd ed., Blackwell, Malden, MA. Recker, W.W., Golob, T.F. (1976) “An attitudinal modal choice model”, Transportation Research, 10, pp. 299-310. Roth, A. (1995) “Bargaining experiments”, Kagel, J., Roth, A. (ed.), Handbook of Experimental Economics, pp. 253-348, Princeton University Press. Sansom, T., Nash, C., Mackie, P., Shires, J. und P. Watkiss (2001) “Surface Transport Costs and Charges. Great Britain 1998, Final Report for the Department of the Environment, Transport and the Regions”, Institute for Transport Studies, University Of Leeds, Leeds, July 2001. Santos, G., Bhakar, J. (2006) “The impact of the London congestion charging scheme on the generalised costs of car commuters to the city of London from a value of travel time savings perspective”, Transport Policy, 13, 2006, pp. 22-33. Schelling, T. (1978) Micromotives and Macrobehavior, Norton and Company, New York, London. Schrage, A. (2005) „Straßenmaut und Verkehrsstaus“, Sinn, H.-W., Fuest, C. (ed.), Beiträge zur Finanzwissenschaft, 19, Tübingen. Small, K.A., Winston, C. und C.A. Evans (1989) Road Work – A New Highway Pricing and Investment Policy, Washington, D.C. Steg, L. (2003) “Factors Influencing the Acceptability and Effectiveness of Transport Pricing”, Schade, J., Schlag, B. (ed.), Acceptability of Transport Pricing Strategies, Amsterdam u.a.O., pp. 187-202. Steg, L., Vleg, C. und G. Slotegraaf (2001) “Instrumental-reasoned and symbolic-affected motives for using a motor car”, Transportation Research F, 4, 2001, pp. 151-169. Sumalee, A. (2004) “Optimal road user charging cordon design: A heuristic optimization approach”, Computer-aided Civil and Infrastructure Engineering, 19, pp. 377-392. Transport Research Laboratory (1996) “The Use of Traffic Signals in Developing Cities”, Overseas Road Note, 13, Overseas Centre, Transport Research Laboratory, Crowthorne, Berkshire, UK. Tucker, A.W. (1983) “The Mathematics of Tucker: A Sampler, A Two-Person Dilemma: The Prisoners Dilemma”, The Two Year College Mathematics Journal, 14, 3, pp. 228-232. Verkehrsbetriebe STI (2006) Mobilität im Berner Oberland, Zahlen und Fakten 2005, Thun. Vickerman, R.W. (2000) “Economic Growth Effects of Transport Infrastructure”, Jahrbuch für Regionalwissenschaft, 20, pp. 99-115. Vold, A. (2006) “Phased implementation of transport pricing for Greater Oslo”, Transport Policy, 13, pp. 140-148. Walters, A.A. (1961) “The theory and measurement of private and social cost of highway congestion”, Econometrica, 29, 4, October, pp.. 676-699. Yang, H., Huang, H.-J. (2005) Mathematical and Economic Theory of Road Pricing, Elsevier, Amsterdam u.a.O. Captions to illustrations Fig.1: The model – Inner-city traffic and the prisoners’ dilemma Fig.2: Payoffs in a simplified prisoners’ dilemma game Fig.3: Interaction between users with heterogeneous preferences Fig.4: The model – Impacts of road pricing Fig.5: Impacts of an optimal charge in a simplified 2x2-game Fig.6: The model – Impacts of capacity extensions Fig.7: The model – Impacts of an improvement of public transport Fig.8: The model – Impacts of the measure package “road pricing plus improvement of public transport” Fig.9: The model – Impacts of the consideration of social preferences Figure 1: The model – Inner-city traffic and the prisoners‘ dilemma utility of the user either as car user or as public transport user x utility of the car user y utility of the public transport user N z 0% b Source: Own illustration c proportion of 100% car users Figure 2: Payoffs in a simplified prisoners‘ dilemma game C D C 1 b D a 0 Source: Poulsen/Poulsen (2002), p. 6. Figure 3: Interaction between users with heterogeneous preferences m D D m D D D Move 3: D vs. C 15:12 m r a D D D DDD C CC m r a r D D m r a m r a m D DD m r a D a C C D r a DDD C CC m r a Move 4: D vs. C 42:39 Source: Own Illustration Move1: D vs. C 3:0 m r a m r Move 2: D vs. C 6:3 C a C DDD CC C CC m r a m r a m r a Figure 4: The model – Impacts of road pricing utility of the user either as car user or as public transport user x utility of the car user y utility of the public transport user N1 z1 N0 z0 0% b c1 Source: Own illustration c0 100% proportion of car users Figure 5: Impacts of an optimal charge in a simplified 2x2-game without road pricing public transport use car use with road pricing public transport use car use public transport use R, R S, T public transport use R, R S, T-topt car use T, S P, P car use T-topt, S P-topt, P-topt Source: Own illustration Figure 6: Impacts of capacity extensions utility of the user either as car user or as public transport user x utility of the car user y utility of the public transport user N1 z0 = z1 N0 0% b0 b1 Source: Own illustration c0 c1 proportion of 100% car users Figure 7: Impacts of an improvement of public transport utility of the user either as car user or as public transport user x N2 utility of the car user z2 utility of the public transport user y N0 z0 0% b c2 Source: Own illustration c0 100% proportion of car users Figure 8: Impacts of the measure package „road pricing plus improvement of public transport“ utility of the user either as car user or as public transport user x N3 utility of the car user z3 y utility of the public transport user N0 z0 0% b = c3 Source: Own illustration c0 proportion of 100% car users Figure 9: The model – Impacts of the consideration of social preferences utility of the user either as car user or as public transport user x utility of the car user N1 z4 y utility of the public transport user z0 N0 0% b c4 Source: Own illustration c0 proportion of 100% car users To simplify the analysis we create a binary decision problem – car versus public transport - by ignoring less relevant alternatives as cycling, taxi or walking. ii The model assumes a particular allocation of traffic of a sufficient large number of vehicles because otherwise a high proportion of car users could generate equilibrium beneath capacity limit, particularly in the case of just one car on the network. This assumption is insofar non-critical as the violation of the assumption will not result in an economical problem. iii Even if the proportion of rail-bound traffic is low, it at least will tend to result in a reduction of the mitigation of utility. iv See Binmore (1992), who identifies a Nash-equilibrium by the players mutually playing best responses to the strategies of the other players. v See Tucker (1983) for the first technical formulation of the 2-person prisoners’ dilemma. See Hamburger (1973) for a technical analysis of the multi-person prisoners’ dilemma. vi There are many thinkable possibilities to provide incentives to cooperate, by coercion as well as without coercion, that are not analysed in this framework, ranging from the constitution of a “Leviathan” (Hobbes, 1651, 1947) to public appeals. i