The only difference between Chicken and Prisoners' Dilemma in this format is the transposition of p and s for each player , but the games are quite different .sx Each of them contains a puzzle :sx in the case of Chicken a puzzle about precommitment , in the case of Prisoners' Dilemma a puzzle about rational actors arriving at a foreseeable but suboptimal outcome .sx These two games are the foundation of modern non-zero-sum game theory in social science .sx This is not the place to review the vast literature that has developed from them .sx It is the place for one warning and one encouragement .sx The warning is that what is true of the two-person one-off game is not necessarily true for repeated and/or multi-person games , so it is dangerous to make sweeping inferences from them .sx The encouragement is that nevertheless the insights to be derived from these very simple cases go a long way towards explaining puzzling features of political life in many different times and places .sx The Provision of Public Goods .sx Chicken and Prisoners' Dilemma are examples of collective action problems .sx A collective action problem is any situation in which players' individually rational actions lead to an outcome which could have been bettered if players had chosen differently .sx To illustrate this most general and hence vague formula , consider a problem which has troubled economists for a long time :sx the optimal provision of public goods .sx A public good is anything which is at least to some extent indivisible and/or non-excludable .sx A well-known case is clean air .sx You cannot take a pound of clean air away from the common stock and sell it on a market stall .sx Nor can you exclude anybody from enjoying the benefits of clean air , whether or not she has contributed to the cost of producing it .sx Public goods pose a problem for market economists because they cannot be sure that the unregulated market will produce an optimum quantity of them .sx General equilibrium theory builds on the classic insight of Adam Smith that the uncoordinated and largely selfish choices of individuals may make society , in aggregate , as well off as it is possible to be .sx The modern way of putting this is that unregulated perfect competition in all markets at once leads to a point on the Pareto frontier .sx The Pareto frontier is the set of points with the property that no further trades can be made .sx As it is assumed that nobody trades unless both ( all ) parties think they would be at least as well off , and some better off , after the trade than before , this is an another sic !sx way of saying that at the Pareto frontier nobody can be made better off without at least one person being made worse off .sx There are two problems with this result .sx One is that not all outcomes which are economically efficient are socially just .sx The other , relevant here , is that markets in which there are public goods cannot be perfectly competitive .sx As you cannot force anybody to pay for clean air , nobody will have enough incentive to produce it .sx Producing clean air is the same thing as preventing the production of dirty air .sx Therefore , as in a market equilibrium it is impossible to force the polluter to pay , too much pollution will be produced .sx Ensuring that public goods are produced is much the same as ensuring that the outcomes players desire in Chicken and Prisoners' Dilemma games are produced .sx Everybody wants clean air but left to themselves may fail to produce it .sx Since the 1920s , economists have looked for ways to ensure that public goods are produced , usually by recommending that politicians or administrators introduce incentives so as to make markets behave as if they were perfect .sx It has taken a long time for academic concern to reach the policy arena but current international interest in the 'polluter pays principle' provides an example .sx It is now commonplace for economists to argue that permits to pollute , priced at the social cost of the pollution , should be sold by governments and tradeable amongst would-be polluters , while the proceeds are used to clean up the pollution .sx But there is a catch .sx If we examine the record of any government which professes to believe in market principles , it is easy to find actions which are inconsistent with that belief .sx For instance , the Thatcher administrations in the UK rejected available moves towards perfect competition in a number of policy arenas .sx Three of them concern motor cars .sx First , a market-distorting tax concession in favour of company cars , inherited from previous administrations , was retained ( albeit on a reduced scale) .sx This imposed not only direct costs arising from the market distortion and from lost revenues but also a pollution cost .sx Motor cars damage cities by reason of noise , exhaust pollution , congestion and accidents .sx A system of permits to pollute , in the form of road pricing , is technically feasible , works in Singapore and is supported by social scientists of a wide range of ideologies , but was resolutely rejected by the Thatcher government and also by the new model Labour Party under Neil Kinnock .sx Thirdly , UK car prices are held artificially high , even within the supposed free-trade area of the European Community , by a combination of requiring 'type approval' for car imports and a 'voluntary export restraint' agreement with Japanese car makers .sx Why do politicians disappoint economists so much ?sx Not because they are irrational or stupid .sx I believe that the Thatcher administration had perfectly good reasons for rejecting these three efficiency-improving changes .sx Most of the outputs of a political system are themselves public goods .sx A law empowering local authorities to charge users for their urban roads is as indivisible as clean air .sx It is not quite non-excludable because some places could be exempted from its operation .sx But in the sense which matters , it is not practicably excludable :sx it is not practicable to restrict it to those people who have contributed to the costs of its production .sx Hence the market in government rules is imperfect , just as the market in air is imperfect .sx Another way of putting the same point is as follows .sx Standard economic analysis assumes that most actors in the economy are self-interested .sx In perfect competition , universal self-interest would maximize welfare , as if by an invisible hand , no less efficiently than universal altruism .sx But general equilibrium fails because the market delivers too few public goods .sx The same Adam Smith who praised the workings of the invisible hand observed :sx People of the same trade seldom meet together , even for merriment and diversion , but the conversation ends in a conspiracy against the public , or in some contrivance to raise prices .sx .sx So economists turn to politicians and bureaucrats in the hope of getting the market regulated .sx But it is inconsistent to suppose that politicians and bureaucrats are any less self-interested than economic actors .sx In so far as the model presupposes economic actors who maximize utility ( defined over income , wealth , entitlements and status ) , it must in consistency presuppose political actors who do the same thing .sx In a democracy politicians need enough votes to be ( re-)elected and enough funds to fight their campaigns and pay for the party HQ .sx Normally , before they can maximize their utility they must first be elected .sx That politicians depend on voters and on campaign funds is scarcely news .sx The distinctive contribution of the rational-choice school is to apply game theory to these relationships and emerge with predictions about rational behaviour .sx Rational Choice and Voting .sx In the well-known spatial model popularized by Downs , politicians who wish to be ( re-)elected move to the issue position of the median voter .sx Voters maximize their expected utility by voting for the party which maximizes their party differential , provided that their party differential discounted by the probability that they are pivotal exceeds the cost of voting .sx Both parts of this model have been much discussed .sx Downs raised an obvious problem with the 'rational' voter :sx on any plausible view , the subjective probability that one's own vote is decisive is vanishingly small .sx Therefore it drags the expected utility of voting down to fractions of a penny .sx As the cost of voting and of becoming well informed is at least a few pence , it seems that the rational voter would never vote .sx Rational-choice writers have suggested some unsatisfactory ways out .sx Downs suggested that voters might nevertheless vote to save democracy from collapse , but this was to commit " a common fallacy in social thought which most of the book turns on his avoiding " :sx namely , that what was in the interests of all was ipso facto in the interests of each .sx Another popular line has been to say that people vote for the incidental benefits ( meeting their friends at the polls and so on) .sx This is a desperate move , invalidated by all empirical evidence .sx A rationally self -interested citizen would almost never vote .sx But Down's analysis is a constructive failure because it explains why apparently trivial differences in voting conditions ( wet weather or dry , Sunday or Thursday , registration at the voter's initiative or at the state's ) are associated with great differences in turnout .sx Who then are relatively more likely to vote ?sx : those whose party differentials and/or subjective probability of decisiveness are relatively high and those for whom the cost of voting is relatively low .sx There is no reason for the first two to vary systematically with class or income but there is reason to expect the last to do so .sx Voting ( where not compulsory ) is a form of leisure activity .sx Leisure is probably a luxury good in the economist's sense that the demand for it increases more than proportionately with income .sx Therefore there is reason to expect the rich to be more likely to vote than the employed poor .sx Furthermore , the cost of acquiring political information is lower for the educated than for the uneducated .sx As there is a positive correlation between education and income , this reinforces the leisure effect .sx Thus there is good reason to expect what empirical studies across time and space confirm :sx that voting turnout rates are higher among the rich and educated than the rest of the population .sx Thus a rational politician , seeking to maximize the probability of re-election , offers not what the electors want but what the median voter wants .sx The median voter is probably richer and better educated than the median elector .sx In particular , she probably drives a car which in Britain is quite likely to be a company one ( over half of new car registrations in the UK are of company cars) .sx Thus rational choice analysts have an easy explanation for the first two of our three observed departures from economic rationality , namely the company car tax concession and the absence of road pricing .sx The other half of Downs's model deals with the rational party .sx If there is just one issue dimension , politicians who wish to be ( re-)elected will converge on the ideological position of the median voter .sx Where politics is truly unidimensional , Downs's hypothesis is robust .sx Every so often , a politician condemns the consensus of the existing parties and seeks a " choice , not an echo " ( the campaign slogan of Barry Goldwater in 1964) .sx Goldwater , George McGovern and Michael Foot each led their parties to catastrophic defeat after breaking away from the consensus .sx Margaret Thatcher is often cited as a counter-example .sx However , on a number of issues , including race relations , trade union reform and nationalistic foreign policy , Thatcherism moved towards , not away from , the median voter .sx On other issues , Thatcherism was admittedly unpopular .sx However , Conservative efforts in 1990 to temper the effects of the poll tax and to slow down the implementation of NHS reform are evidence for the robustness of the Downsian median .sx So are the strenuous efforts of Neil Kinnock to lead the Labour Party to it .sx When politics is multidimensional , however , the hypothesis gets into trouble .sx Generalizing the median voter hypothesis into multidimensional issue space may sound easy .sx Suppose that one issue dimension is 'left-right' judged by voters' attitudes to the distribution of income and wealth ; another is 'liberal-conservative' judged by issues such as free speech and the death penalty ; and a third concerns the regional distribution of resources .sx