2.2.1 Consumers versus Voters

Voters, like consumers, are rational, consistently fitting means to ends. But when deciding on which party to support they are less instrumental in the sense that voting is not based only on how electoral promises translate into individual economic welfare (Reference Becker, Mulligan and HallBecker and Mulligan 2017). This reduced instrumentality has several origins, including:

1. (1) Rational ignorance: Because an individual vote has little impact on policy choice and information is costly to acquire, individuals do not find it worthwhile to acquire all of the information necessary to assess the consequences of alternative platforms (Reference DownsDowns 1957).Footnote ¹⁰

2. (2) Expressive voting: Since the link between the act of voting and what governments do is weak compared to the connection between market choices and individual welfare, voting has an expressive component that is similar to the enjoyment of the solidarity that comes from cheering for a football team (Reference Brennan and LomaskyBrennan and Lomasky 1993).

3. (3) Ideological and other nonmarket objectives: In political life, where it is less costly, people act in part to satisfy their loyalty to an ideology or other nonconsumption-related goals. This is not a matter of a lack of information (Reference CaplanCaplan 2007).

4. (4) Behavioral biases: In the electoral setting, voters have a reduced incentive to deal with cognitive biases. While also present in market decision-making, the consequences are more directly costly to individuals there (e.g., Reference SchumpeterSchumpeter 1950, 261–62; Reference Frey, Stutzer, McCaffery and SlemrodFrey and Stutzer 2006).Footnote ¹¹

As a bridge to the next pair of elements, we note that parties are surely more instrumental than are voters because parties can benefit from strategizing in the conduct of their activities.

2.2.2 Firms versus Parties

The textbook private firm continually uses its resources to produce and market goods in order to maximize the profit of its residual claimants – its shareholders. Consequently, the output of a successful firm is always in line with what its customers want and, to achieve that, the firm will follow its customers’ tastes closely. In contrast, political parties have no residual claimants who bear most of the consequences of the party’s success or failure (Reference DemsetzDemsetz 1982, Reference Demsetz and Demsetz2008). As a result, and unlike firms, a party consists of a coalition of people with differing objectives, some of whom seek to maximize expected votes – a political analogue to profits – while others pursue ideological or policy-oriented goals even at the expense of winning (Reference WittmanWittman 1983; Reference AldrichAldrich 1995; Reference RoemerRoemer 2001).

Like firms, parties cultivate a brand or, in political terms, an ideology, in part because this reduces information costs for its “customer base.” Branding is also important in politics because it attracts the ideologically committed activists and donors who help with the problems of maintaining a political organization.

Firms and parties also differ in the credibility of their commitments. Compared to a political platform, the smaller dimensionality and greater measurability of a firm’s products allows enforcement of broken promises through the courts. In cases where promises are not justiciable, the firm develops credibility by investing in a reputation for quality, an asset permitting a price premium that allows its shareholders to be favorably compensated when the firm does not cheat (Reference Klein and LefflerKlein and Leffler 1981). The unavailability of third-party enforcement for election promises and the lack of a similar compensation mechanism make it more difficult for political parties to establish credibility in their commitment to a proposed platform, though formation of a party coalition will, to some extent, require it to honor the bargains made between coalition members (Reference LevyLevy 2004). An independent press in a liberal democracy may also play a role in maintaining party credibility by exposing false and broken promises, though this too will not work as surely as adjudication of a private-market contract.

2.2.3 Markets versus Elections

A market is a decentralized allocation mechanism in which relative prices channel privately owned resources toward their highest value uses. Underlying this mechanism is the fact that in a market, the transaction value of purchases and sales are closely tied – “you get what you pay for” – and consequently differences in the intensity of individual preferences play a role in shaping the equilibrium allocation. In contrast, an election is a collective choice process in which, absent unanimity, everyone must compromise and there will invariably be a difference between what each person wants for the taxes they pay and what they actually receive.

The matching of costs and benefits is made even less precise by the presence of public goods for which the levying of individual tax-prices that mimic market pricing is impractical, and by redistribution, in which case it is counterproductive, while the role of intensity of preference is further clouded by a franchise that assigns one vote to each person regardless of income and tastes.Footnote ¹²

Personal agency is constrained in a collective choice process compared to a market because of the need to compromise with others. It is also reduced for a second reason. While consequential market exchange can be carried out alone, altering public policy requires cooperation within a political organization, an activity, like the provision of a public good, which is subject to problems of free riding.

Finally, we recall that unlike the rights granted to private agents in a market, the party winning an election has the benefit of the state’s monopoly on the legitimate use of force. This monopoly, which reduces the cost of financing the public sector, is constrained by the institutions of liberal democracy with more or less success in various circumstances.

3.2.1 A Representation Theorem

A probabilistic spatial voting model may have an equilibrium that can be represented by the optimization of a political support function that corresponds to a social welfare function even though no active social planning is involved, a result first shown by Reference Coughlin and NitzanCoughlin and Nitzan (1981).Footnote ¹⁷ We shall refer to this result as a representation theorem. Presentation of such a framework is the second step of our welfare analysis.

The particular spatial voting model that we adopt for this analysis is significantly simplified. Constraints on collective choice that are part of the standard of reference are only implicit in the model (analogously, transactions costs are only implicit in the canonical model of perfect competition in a market). The set of policy instruments from which parties fashion their platforms is assumed known, fixed, and well-behaved, as are the parties and candidates.Footnote ¹⁸ Finally, we note that the set of issues decided by any election is big, perhaps too big for the long-run stability of democratic institutions, including complete determination of the distribution of individual welfare.

The demonstration in this section that under certain conditions electoral competition may result in an economically efficient allocation of resources is more mathematical in nature than are arguments in the Element as a whole. Readers less interested in mathematical detail may wish to read the text for a general sense of the argument, and then turn to the discussion following in which some of the implications of treating policy choices as part of a competitive political equilibrium are drawn out.

Our setting is an electoral contest between two parties, each of which chooses its platform to maximize the total vote they can expect to receive in the next election.Footnote ¹⁹ Voters know who they will support – the party that promises them the highest utility – and everyone votes sincerely. But parties have only probabilistic (common) knowledge of how the voters will behave.

An incumbent party $i$ and an opposition party o compete by offering and delivering a multidimensional policy platform or program $s^p,p=\left\{i,o\right\}$ , (e.g., a vector of tax rates, tax bases, special provisions, and public expenditures and transfers) to an electorate of J voters with indirect utility functions $v_j$ (s). Usually the number of instruments in these programs will be much less than the number of voters. (This structure is discussed in the Section 5). Voters cast their ballots sincerely for the party that promises them a higher utility, so the probability a voter will support the incumbent is

${\pi }_j^i=\{\begin{array}{l}1if{v}_j\left(s^i\right)-v_j\left(s^o\right)>0\hfill \\ 0\mathrm{otherwise}.\hfill \end{array}$(3.1)

But we suppose that, as seen by the parties, the probability a citizen votes for the incumbent is known by both parties to be described by a density function f, which is increasing in the difference in the utilities the voter derives from each platform:

$f_j\left(v_j\left(s^i\right)-v_j\left(s^o\right)\right);\partial f_j/\partial (v_j^i–v_j^o)>0.$(3.2)

To fashion a competitive policy platform, the incumbent chooses a platform that maximizes its expected vote $E{V}_i$ given the platform of its challenger, subject to the structure of the economy which we write generally as R(s)=0:

$Ma{x}_{\{s^i\}}E{V}_i={\text{Σ}}_j^J{f}_j\left(v_j\left(s^i\right)-v_j\left(s^o\right)\right)\text{subject to}R\left(s\right)=0$(3.3)

If everyone votes, $E{V}_o=J-E{V}_i$ .Footnote ²⁰ In that case the solution to the opposition’s problem is equivalent to that of the incumbent and party platforms will be identical in the electoral equilibrium if one exists. (Divergence of platforms can be introduced if first order conditions for the optimal policy platforms for each party are not the same for some reason.)

If each party’s expected vote is globally concave over the issue space as well as continuous in s, so that there is a positive probability, no matter how small, of each voter supporting either party, a Nash equilibrium in pure strategies in this contest will exist. If the concavity of EV is strict, the equilibrium will be unique. Further discussion of the conditions underlying the existence of an equilibrium along with a demonstration of how the model can be adapted to expose determinants of the degree of electoral competition are provided in the Appendix.

For a policy set $s=\{s_l,l=1,2,\dots ,L\}$ , each element of which may in general affect many different voters, the equilibrium platform is defined by the following first order conditions where party subscripts are dropped because platforms converge, and λ is the Lagrange multiplier on the constraint:

(3.4)${{{\displaystyle \sum }}^{\text{}}}_j^J\frac{\partial f_j}{\partial v_j}.\frac{\partial v_j}{\partial s_l}+\lambda .\frac{\partial R}{\partial s_l}=0;l=1,2,\dots ,L$

The representation theorem is essentially the observation that the competitive equilibrium just described can be represented by the maximization of a synthetic political support function subject to the same constraint(s) facing the parties. As a proof, it is sufficient to show that the equilibrium can be replicated by the constrained optimization of a (political support) function S that is a weighted sum of voter utilities. Assuming that S is concave in s after substitution of the constraint, the problem that can be used to replicate the Nash equilibrium is:

$Ma{x}_{\{s\}}S\left(s\right)={\text{Σ}}_j^J{\theta }_j\mathrm{.}{v}_j\left(s\right)\mathrm{s}\mathrm{u}\mathrm{b}\mathrm{j}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{t}\mathrm{o}R\left(s\right)=0,$(3.5)

where ${\theta }_j=\partial f_j/\partial v_j,$ the sensitivity of each voter to a change in their welfare evaluated at the Nash equilibrium. These influence weights are particular numbers, one for each voter. This problem can be used to represent the Nash electoral outcome because the first order conditions for its solution are exactly those in (3.4) defining each party’s politically optimal policy choices in the equilibrium.

It is important to recognize that the support function S in (3.5) is not a social welfare function since the weights determining the relative importance of individuals in the equilibrium are determined by the interplay of contesting parties and not on the basis of some normative theory.

Other forms of the support function will arise depending on the nature of the density function describing voting behavior. For example, Coughlin and Nitzan and Coughlin (1992, chp4) show that if the probability of voting has the Luce form, depending on a ratio of utilities, the support function that is optimized in an equilibrium is a weighted sum of the logs of utilities. In each case, since a weighted sum of utilities is maximized, it follows that equilibrium policy choices are Pareto efficient while being situated along the Pareto frontier at a point that takes the weights into account. In this probabilistic voting framework, there is an invisible hand in the political process.

We can compare political decision-making in the spatial voting framework we are describing to the use of the Hicks–Kaldor criterion in benefit–cost analysis. Here voting converts ordinal, noncomparable preference rankings into cardinal, interpersonally comparable numbers – a probability of voting for each party – allowing the interests of different voters to be traded off. But unlike most benefit–cost analyses, ‘distributional’ weights from the equilibrium are applied to each citizen’s welfare. Moreover, the outcome may also depend on how voters view the social implications of their situation, a factor that benefit–cost analysis does not consider.

As an example of the use of the theorem, suppose s consists of a large set of tax rates t_j, one for each voter, levied on some base in order to finance a public good G. For convenience, we also assume each tax instrument only affects the welfare of the taxpayer indicated. If total revenue is $R\left(t_1,t_2,\dots ,t_J,G\right)$ the first order conditions, analogous to (3.4), characterizing the equilibrium fiscal structure are:

$\frac{ {\theta }_j . \frac{\partial v_j}{\partial t_j}}{\frac{\partial R}{\partial t_j}}=-\lambda ,j=1,2,\dots ,J ,$(3.6)

and

$\underset{j}{{{\displaystyle \sum }}^{\text{}}}{\theta }_j.\frac{\partial v_j}{\partial \text{G}}=\lambda \left(1-\frac{\partial R}{\partial \text{G}}\right),$(3.7)

where $\lambda$ is the Lagrange multiplier associated with the government budget constraint $R\left(t_1,t_2,\dots ,t_J,G\right)=G$ .Footnote ²¹ In this equilibrium, platforms are chosen so as to equalize the marginal political cost or loss of support per dollar of revenue raised across tax instruments, and each of these are in turn set equal to the marginal political benefit of another dollar of public spending net of the effects of any additional tax revenue that arises when G is increased.Footnote ²² Here a voter’s political sensitivity to a change in their welfare, indexed by ${\theta }_j$ , determines how well they are treated in the equilibrium. As (3.6) indicates, if marginal utility diminishes, a relatively influential or sensitive voter will be taxed less heavily. We should also note that in this equilibrium, there will never be an exact matching for any citizen of the taxes they pay and benefits they would like to receive in exchange.

3.2.2 Some Implications of an Equilibrium Framework for Policy Analysis

Before turning to an analysis of why outcomes often differ from those described by this fully competitive, well-functioning political system, we explore some of the implications for policy analysis that arise when public policy is regarded as part of an electoral equilibrium.

Political Feasibility in Policy Design and Advocacy

Even when it is accepted that alternative policy designs can produce superior outcomes, advisors recognize that a proposal must be supportable as a political equilibrium – that is, politically feasible – or it will not be adopted. As we noted earlier, social planning explicitly refrains from such a consideration. Feasibility is an issue of longstanding concern in the public administration literature (e.g., Reference DrorDror 1969; Reference MeltsnerMeltsner 1972). Dror argues that the infeasibility of an alternative should sometimes be taken as a challenge rather than accepted as a constraint, though it is not clear when one ought to do so. It is less common to see feasibility considered in the public economics literature, though it does arise from time to time; see Reference Besley and CoateBesley and Coate (2003), Reference Acemoglu, Golosov and TsyvinskiAcemoglu, Golosov and Tsyvinski (2008), Reference Acemoglu and RobinsonAcemoglu and Robinson (2013), Reference OhOh (2017), Reference Bierbrauer, Boyer and PeichlBierbrauer, Boyer and Peichl (2020), and Reference Kangas, Kangas, Jauhiainen, Simanainen and YikannoKangas (2021).Footnote ²⁵ The issue is given further attention in section 3.2.3. Here we use examples to illustrate a general problem of policy design that may arise when political feasibility is ignored.

As deputy assistant secretary of the Treasury from 1983 to 1985, Charles McLure directed a team charged with drawing up a US tax reform known as Treasury I (1984). This document proposed a broad base federal income tax for the United States, despite economic theory that at the time had swung in favor of an expenditure tax. The rationale for the Treasury proposal, according to the account in Reference McLure and ZodrowMcLure and Zodrow (2007), was that the designers in the Treasury knew that one component of the expenditure tax would never be accepted by the president and, as a result, the outcome would be worse than if the Treasury proposed a broad-based income tax. The component they were sure would be missing was the taxation of inheritances and bequests, which is required to preserve tax equity when an expenditure tax replaces a personal income tax

A second example involves the Mirrlees Review’s (2011) advocacy of reduced corporate taxation. In the review, capital income is more heavily sheltered from direct taxation and inheritance taxes are advocated as a means of maintaining overall tax progressivity.Footnote ²⁶ Inheritance taxes are also recommended as a way to improve equality of opportunity. The problem facing those who want to put more emphasis on wealth transfer taxation is that this form of taxation has been declining in most Western countries for several decades, perhaps because of opposition from aging electorates that want to pass on hard-earned wealth to their children, perhaps because of successful lobbying by monied interests.Footnote ²⁷ Whatever the reason, if the decline in wealth transfer taxation is regarded as a constraint to be factored into tax design, the proposals in the Mirrlees Review might look somewhat different. A Nordic style dual income tax that involves taxation of capital income at a rate lower than labor income could be a preferred alternative.

In both cases, the advocacy of a multipart social plan was likely to result in only part of the overall proposal being accepted, producing an outcome that is no longer optimal when judged from the original standpoint. This is an important point.

3.2.3 Political Feasibility, Tectonic Policies and Schumpeterian Competition

It is worthwhile to explore the nature of political feasibility somewhat further. We do so here in a manner that leads to introduction of a form of political competition that we have not so far identified.Footnote ²⁸

Suppose we consider a proposed policy in period 1, $s_1$ , that will affect individual welfare both now $\left(v_j^1\right)$ and in the future $\left(v_j^2\right)$ . Suppose also that the policy implemented in period 1 might, by changing the distribution of income or other factors, also change the political weight of some individuals or groups in a second period. Then (leaving discounting and any constraints imposed by the structure of the economy aside), given the support function

$S\left(s_1\right)={\displaystyle {\sum }_j{\theta }_j}{v}_j^1\left(s_1\right)+{\displaystyle {\sum }_j{\theta }_j}\left(s_1\right)v_j^2\left(s_1\right),$(3.8)

the governing party competing for support would adopt $s_1$ only if

$\frac{dS}{d{s}_1}={\displaystyle {\sum }_j{\theta }_j}\frac{d{v}_j^1}{d{s}_1}+{\displaystyle {\sum }_j\left[\frac{d{\theta }_j\left(s_1\right)}{d{s}_1}{v}_j^2+{\theta }_j\frac{d{v}_j^2}{d{s}_1}\right]}>0.$(3.9)

This is a minimal condition for the political feasibility of the policy. An ideal policy will make S as high as possible.

The decision made will differ from the advice given by a benefit–cost analyst (who avoids politics) for at least two reasons. This analyst would not consider either the heterogeneity of political influence, nor the impact of the policy on the political equilibrium. That is, the benefit–cost analyst considers whether or not

$\frac{dS}{d{s}_1}={\displaystyle {\sum }_j\frac{d{v}_j^1}{d{s}_1}}+{\displaystyle {\sum }_j\frac{d{v}_j^2}{d{s}_1}>}0.$(3.10)

Attempts have been made to incorporate normatively defined versions of the ${\theta }_j$ into benefit–cost studies (see Reference Boardman, Greenberg, Vining and WeimerBoardman et al. 2020 for a discussion). But most such studies weight gains and losses across individuals equally. Induced changes in the ${\theta }_j$ are simply ignored. The political strategist will be interested in the economist’s evaluation of the economic consequences of policy choices – the terms in $d{v}_j^k/d{s}_1$ – and will fill in their own estimates of the ${\theta }_j$ and of the $d{\theta }_j\left(s_1\right)/d{s}_1$ indicating how the adoption of the policy alters current and future support.

The feasibility of policies will evolve over time as political influence does. Demographic aging, for example, will alter the relative influence of young and old and hence change the political profitability attached to social welfare programs directed at these groups (Reference ProfetaProfeta 2002), perhaps leading to the crowding out of social spending as the pension demands of an aging electorate rise (Reference LindertLindert 2019).

Sometimes policy is designed to pro-actively alter future political behavior. For example, the policy of Margaret Thatcher, when prime minister of the United Kingdom, of selling public housing to their occupants was surely intended to change the voting behavior of new homeowners in a Conservative direction. Immigration policy is another strategy that may be used to shift the distributions of electoral support (Reference Razin, Sadka and SuwankiriRazin, Sadka, and Suwankiri 2011). We might think of the European Union as a long-term project aimed at creating shifts in attitudes toward political integration across member states. These are all examples of what Reference Young, Breton, Galeotti, Salmon and WintrobeYoung (1991) refers to as tectonic policy, which he thought of as yet another dimension of electoral competition, one that is analogous to Schumpeterian creative destruction in a market.

4.3.1 International Competition

As soon as we extend our analysis of electoral competition to an international setting, it becomes obvious that international competition creates problems for allocation through the political process. Consumers can purchase goods in foreign as well as in domestic markets, and private firms can operate in international markets as well as at home. But the same extension does not apply to the corresponding political elements because voting, the domain of political parties and of an electoral system are all defined strictly on a national basis. Consequently, parties will not take the implications of their domestic decisions for global social welfare fully into account. Widespread concern with some global issues and the working of international organizations may blunt the force of this observation. But it is an open question as to how well they will serve as a substitute for democratic world government.Footnote ³⁹

Harmonization is one response to the harmful consequences of international fiscal competition. The approach has a long history going back at least to Seligman in the 1920s (Reference Musgrave and EdenMusgrave 1988). The idea is to create conditions that allow distinct national policy choices to coexist without distorting international flows of goods and factors, while at the same time attenuating the incentives countries have to engage in discriminatory fiscal practices, thereby preventing “social dumping” of bad jobs and a “race to the bottom” in tax rates. This sort of policy harmonization requires adoption by international agreement of certain principles of taxation, such as the destination rather than the origin principle for commodity taxation, and the residence rather than source principle for income taxation. It also includes the single tax principle – that a taxpayer should be taxed only once, and at least once – leading to international cooperation to prevent tax evasion, such as the recent (2021) agreement to impose a global minimum corporate income tax.

Harmonization is complicated by the many avenues that states have for engaging in predatory practices. Reference SinnSinn (2003, chapter 2) suggests that tax harmonization will just lead to more use of other, nontax policy instruments, such as infrastructure spending, as countries try to maintain an advantage over their trading partners. Others point to the complications that follow from the insistence by nation states on taxation of resource and other rents at their source. These problems heighten the challenges posed by unregulated international competition among sovereign governments, and further prejudice the case against such competition as a beneficial process.

An alternative, less discussed, approach to international competition emphasizes the dangers of using fiscal coordination to shelter countries from international competition. From this perspective harmonization shelters domestic rents, whereas international competition forces tax prices down toward the marginal cost of providing public services to mobile users. Reference MuellerMueller (2000, 182) observes that citizens move toward jurisdictions with high taxes that fund high-quality public services. He argues that the typical advocate of harmonization “knows what the proper level of taxation for a country should be and how this money should be spent”, and fears that any loss in tax revenue due to international tax competition will harm these programs. He concludes, “Such fears are unfounded, if governments provide the goods and services that their citizens want and use benefit taxation.” Reference VanbergVanberg (2000, 100) adds that in this context, using benefit taxation means that tax contributions should be regarded as a price willingly paid for benefits, which in the judgment of taxpayers offset the costs, “in the light of available alternative options.”Footnote ⁴⁰

Vanberg does not rule out the possibility that some rules about how international competition is to be conducted may be desirable. Nonetheless, while international arbitrage is commonly regarded as a significant constraint on domestic policy choices, the two normative approaches embody quite different perspectives on the social utility of intergovernmental competition.

4.3.2 Special Interest Politics

We have pointed out that the scale of organization required to influence public policy is substantial when compared to that of altering personal product choice in a market. This characteristic of the polity makes it possible for small, organized groups to direct resources toward themselves despite the socially harmful consequences of the policies they seek. The classic case involves a relatively easily organized group of producers who profit from a policy-induced increase in product prices at the expense of many consumers who each suffer to an extent that is too small to provoke them into active opposition. An example of this kind, which even appears in Pareto’s “Manual” (Reference Dahl1971, 379), involves a tariff on an imported consumer good. The literature on the political economy of taxation is full of cases in which individual welfare losses for consumers aggregate to an amount that exceeds the benefit to producers (see, for example, Reference ShugartShugart 1997).

Regulations on entry into an industry that protect rents enjoyed by existing firms and industrial subsidies for them are further examples of policies with concentrated benefits and diffuse costs (Reference StiglerStigler 1971; Reference PeltzmanPeltzman 1976). Reference MiguéMigué (1977) explains the choice between entry controls and industry subsidies: the first order of business is always to reduce the elasticity of supply, followed by subsidies once entry barriers ensure that they will not be competed away by new entrants. In declining industries where entry is not a threat, subsidies are preferred at the outset, though they may not completely offset the source of their troubles (Reference HillmanHillman 1982). Reference OlsonOlson (1982) goes further to argue that as a democracy matures and special interests proliferate, economic growth declines because interest groups resist sectoral reallocations that diminish their policy-generated rents.

The existence of inefficient policies in response to special interest lobbies in the framework of the model in Section 3 stems from (1) the use of political resources C(s) to influence policy actions, and (2) a way for these resources to enhance political support even though the government does not take the full economic consequences of the policies supplied into account, a situation depicted earlier in (4.1). These resources may be transferred to political parties or used in various ways by lobbyists to influence legislative decisions; may rise to equal the size of the rents created on behalf of special interests; and are considered by some to be an additional source of social welfare loss (Reference TullockTullock 1967; Reference KruegerKreuger 1974). Reference Dixit, Grossman and HelpmanDixit, Grossman, and Helpman (1997) conclude that if there are many interests that can be played off against each other, and the cost of doing so is low, much of the rent will be transferred to the government.

In a challenge to the conclusion that special interests distort public policy, Reference WittmanWittman (1995) admonishes readers not to confuse interests with influence. He points out that even small losses of support from citizens who are part of a large, diffuse majority can, when totaled, easily outweigh the gain in support from granting a preferential policy to a small interest group.

Does electoral competition squeeze out inefficient policies of the sort we are discussing? An affirmative answer is provided by Becker who, we recall, regards political competition as a veil for rivalry among interest groups. In Becker’s view, the equilibrium outcome depends on who stands to gain the most. Competition among interest groups then leads to the survival of policies that raise output and welfare because favorably affected groups find it in their interest to lobby more for these policies than unfavorably affected groups will lobby against. The outcome replicates a social benefit–cost analysis in which the gains and losses for different groups are weighted according to their political influence.Footnote ⁴¹ Opposition by organized groups may not even be necessary for efficiency if individual voters are aware of the consequences of the policies that special interests seek, and care about their own welfare (Reference Hinich and MungerHinich and Munger 1994, 182). Both of these conditions hold in the model of the well-functioning polity presented in Section 3.

Businesses that try to influence public policy often hire professional lobbyists. Reference de Figueiredo and Richterde Figueiredo and Richter (2014, 163) inform us that in 2012, $3.5 billion was spent on lobbying politicians in the United States, more than five times the total amount spent on campaign contributions.Footnote ⁴² There is no doubt that lobbying serves the interests of the lobbyists’ clients. What is more difficult to ascertain is whether it worsens the allocation of resources (Reference Bombardini and TrebbiBombardini and Trebbi 2020). Lobbying may be an essential means of information transmission.Footnote ⁴³ Reference Achen and BartelsAchen and Bartels (2016, 321) suggest that if voters are to have their interests represented, interest groups and parties, which (we add) are more instrumental than individual voters, must do much of the work. Reference StiglerStigler (1971, 97) suggests that these agents will become more important, the more rapidly and unpredictably the interests of the electorate change. If information transmission is the essence of this activity, and/or if lobbying works as in Becker’s framework, it may improve overall welfare. A critical issue is whether there is an asymmetry of knowledge between voters and lobbyists that can be manipulated for selfish reasons. This is one reason we have discussed the information question at length in Section 4.1.

4.3.3 The Common Pool Problem

A public good problem arises when the benefits from private contributions to its supply are nonexcludable and so received by all. Private suppliers in a market cannot then be adequately compensated and there is a need for public action. In contrast, a common pool problem arises when benefits derived from exploitation of a resource are private to the user, while the associated costs are public and so imposed on everyone. In this case, there is a need for inaction, that is, for greater restraint in resource exploitation (Reference StromOstrom 1990; Reference Sandler and ArceSandler and Arce 2003).

The common pool problem we are concerned with arises when politicians spend money from a general revenue fund fed by taxes levied on the country as a whole. A common tax pool allows “some people [to] spend other people’s money” without fully accounting for the consequences of the taxation of private incomes (Reference von Hagen, Weingast and Wittmanvon Hagen 2006, 464). The immediate cause is the diffusion of authority within political parties and governments, which makes the coordination needed to match spending with taxation difficult. In the competitive political economy of Section 3, the governing party is a single responsible authority. Taxes paid by every taxpayer are politically salient, and the successful party presents a set of policies that reconciles both sides of the budget, at least as far as possible given the difficulties of levying individualized tax-prices. In reality, however, the government is a collection of semiautonomous agents or parties. Under proportional representation (PR), a governing coalition typically involves several parties, each with its own core supporters. In a majoritarian system where one party usually governs alone, there are many elected representatives who form the caucus of the party and who each have some agency despite their party membership.

Reference Persson, Roland and TabelliniPersson, Roland, and Tabellini (2007) show how the common pool problem may be modeled using a spatial probabilistic voting framework. Here we illustrate it by adapting a diagram used by Reference Inman and RubinfeldInman and Rubinfeld (2020, 144). Figure 1 is drawn for various levels of public outlay G arising within one of many electoral districts, n = 1, 2, …, N, in a majoritarian electoral system or, alternatively, for a group of core supporters of party n that is part of a governing coalition under PR. The marginal benefit of public outlay curve slopes downward to reflect diminishing marginal returns to consumption. The marginal social cost of G is assumed to be constant for convenience.

Figure 1

The common pool problem in public finance

Note: n denotes a particular electoral district in a majoritarian electoral system, or a particular party in a governing coalition under PR. Marginal social cost (n) is the full marginal social cost of supplying good G to the nth district, or to supporters of the nth member of a governing coalition. λ∙ Marginal social cost (n), 0 < λ < 1, is the marginal cost of benefits to voters in district n, or to supporters of the nth coalition member, when the full cost is spread over all taxpayers through the national tax system.

The actual marginal tax cost to taxpayers in district n, or to supporters of coalition member n, is a fraction λ of the full cost since that cost is spread through the tax system across taxpayers in the whole country. If all political agents achieve their politically optimal level G_n′, which is greater than the socially optimal level G_n* in each case, the resulting total social loss from overexploitation of the national tax base will be equal to N times area A.

With each member of the governing party, or each part of the governing coalition, treating everyone else’s supporters like fish in international waters, the question arises as to how the resulting struggle is resolved. One answer by Reference WeingastWeingast (1979), Reference Shepsle and WeingastShepsle and Weingast (1981), and others is that each risk-averse legislator or coalition partner prefers an outcome that gives to every member their politically optimal choice G_n′. They prefer this solution to the risky alternative under which each of them tries to shift more of the tax burden onto the supporters of their colleagues by engineering a smaller winning coalition; when decisions are subject to majority-rule approval, small winning coalitions are inherently unstable.

The solution in which everyone gets a piece of the action is referred to as the norm of universalism. When this outcome characterizes actual budgetary outcomes, we should expect higher levels of public spending in situations where there are larger numbers of key decision makers. There is, in fact, evidence that expenditure grows with the size of the legislature in US states (Reference Gilligan and MatsusakaGilligan and Matsusaka 1995), with the council size in US city governments (Reference BaqirBaqir 2002), and with legislature size in the world (Reference Bradbury and CrainBradbury and Crain 2001). Related work discussed by Reference Inman and RubinfeldInman and Rubinfeld (2020, chapter 5) considers how smaller sized US state senates can constrain spending by larger elected legislatures. There is also evidence that coalition governments spend more than single-party majority governments (Reference Persson, Roland and TabelliniPersson, Roland and Tabellini 2007). Consequently, the fact that PR leads to party fragmentation and thus more coalition government, as Reference DuvergerDuverger (1954) predicts, makes it unsurprising that PR governments are often observed to spend more than majoritarian ones do (Reference Persson, Tabellini, Weingast and WittmanPersson and Tabellini 2008).Footnote ⁴⁴

Since universalism leads to overuse of the tax pool and a loss of social welfare, we should also expect that interparty competition will give rise to arrangements that allow governing parties to profit from the support that can be realized by taming the problem. Strategies that have been used include delegation to, or control by, strong ministers of finance, usually found in electoral systems with single-party governments, and coordination or contract devices more appropriate for situations involving coalition government, since delegation within a coalition is difficult to organize and enforce.

Party leaders may try to get their colleagues to internalize the spending externality using high-powered incentive schemes, such as offering their presence and financial support in re-election campaigns in return for moderating spending demands, or appointment to a higher position within the party structure. These measures aim to increase λ in the thinking of individual decision makers. In terms of Figure 1, this involves shifting the lower horizontal line upward. It should be easier to do this if the party enjoys political support from all segments of the electorate, because this heightens the political saliency of all the welfare losses depicted in Figure 1.Footnote ⁴⁵

Less ambitiously, the tendency toward overexpansion may be kept in check by simply bounding the loss. This alternative strategy is represented in Figure 1 by a leftward shift of the vertical line through the point G_n′. For example, in a majoritarian Westminster system, control over the budget is given to the minister of finance who, making use of the doctrines of cabinet solidarity and budget secrecy, says “No” to the ministers of spending departments (Reference BretonBreton 1996, chapter 4). Cabinet solidarity prevents disappointed ministers from going to supporters outside of the cabinet after a budget is decided upon, while budget secrecy gives the upper hand to the minister of finance in his or her negotiations with spending departments that do not know the whole plan.Footnote ⁴⁶ According to Breton, these conventions along with a line-item veto allows a government in a Westminster system to match spending with taxing more effectively that the president in the US congressional system can. Reference de Figueiredode Figueiredo (2003) extends this argument to show how strengthening of the executive’s use of the line-item veto responds to expected changes in the scale of the common pool problem.

Since coalition government often arises in PR systems, the contract approach is more likely to emerge there. This may include budgetary agreements among coalition members on ex ante fiscal targets. Such arrangements may require monitoring by the minister of finance, in which case the distinction between delegation and contracting is less meaningful.

Empirical evidence suggests that, to an extent that varies across the cases considered, arrangements involving delegation and contracting both do have some of their intended effect, as shown by Reference Hallerberg, Strauch and von HagenHallerberg, Strauch, and von Hagen (2009), Reference de Haan, Jong-A-Pin and Mieraude Haan, Jong-A-Pin, and Mierau (2013), and others. This judgment also applies to balanced budget laws and other ex ante fiscal rules that have been adopted in both majoritarian and PR systems (e.g., Reference von Hagen, Weingast and Wittmanvon Hagen 2006; Reference Hou and SmithHou and Smith 2010; Reference Atkinson, Mou and BruceAtkinson and Mou 2016).

6.1.1 Competitiveness as Unpredictability or Closeness

The idea that a highly competitive election is one that is “too close to call” has a long tradition in the scholarly literature and in popular discourse, for a good reason: there is no meaningful contest if the outcome is always known with certainty. When deciding upon a measure of the closeness or unpredictability of an election, it is important to recognize that the underlying logic is ex ante: the expectation of outcomes must be formed prior to the outcome of the contest. If in operationalizing “too close to call” we simply replace the ex ante concept with its ex post result (plus a random term), we are implicitly invoking a strong form of rational expectations. No matter what information set is used in making predictions, all attempts at measurement must deal with this ex ante–ex post distinction.

The oldest and most well used measure of competitiveness as unpredictability is the observed first versus second place vote share margin, v₁ – v₂. In this case an election is seen as highly competitive if this vote share margin is “small” (e.g., Reference MayhewMayhew 1974; Reference JacobsonJacobsen 2004). In Table 1 a related statistic, the (absolute value of the) deviation of the Republican share of the two-party vote in US Senate elections from one-half, averaged over twelve-year periods, is used to illustrate the distributional characteristics of the vote margins of the dominant two parties for open Senate elections (that did not involve an incumbent).Footnote ⁵⁶ The table suggests that open Senate elections became considerably more competitive following the Second World War, especially in the 1960s. The post-2006 period exhibits a strong reversal of this trend, with outcomes concentrated in the upper tail of the distribution, where winning margins are relatively large for both Republican and Democratic candidates. An important take-away from the table is that competitiveness fluctuates substantially across time and, as we would see if the table were to be suitably disaggregated, across space, even though the United States is a mature democracy.

It is tempting to attribute the rise of competitiveness in the 1960s illustrated in the table to changes in voter registration and the decline after 2004 to party polarization. But caution should be exercised in drawing conclusions at least because a simple vote margin, however calculated, is not a good measure of closeness for at least three reasons. First, the significance of any vote share margin will be a function of how much variability, or volatility, there is in the proportion of the electorate that is likely to switch its vote from one election to the next (Reference Przeworski and SpraguePrzeworski and Sprague 1971; Reference ElkinsElkins 1974). Even a relatively small margin can be associated with an election that is not close if the expected volatility is, say, an order of magnitude smaller still. Volatility is usually measured by aggregating changes in the share of the vote experienced by each competing party over adjacent elections, taking care to avoid double counting. (On measuring volatility, see Reference Bartolini and MairBartolini and Mair 2007). Volatility may be incorporated by dividing each vote share margin by its associated vote volatility to derive a volatility-adjusted index.

A second problem, also emphasized by Przeworski and Sprague, is that even a volatility-adjusted first versus second place vote margin can be misleading if the winner faces competition from more than just the second-place finisher. The third problem stems from the difference between electoral turnover, which may not be affected even though margins fluctuate, and closeness. This last observation leads to measurement of the probability of loss of office and to consideration of the contestability of elections.

To deal with the second issue – accommodating all of the parties involved in a contest – we may define the margin or “distance to go” of each party (k) as the winner’s vote share less the share of that party, v_{1 −} v_k, and then deflate that margin by the volatility in the district as a whole. A highly competitive district-level election in this context is one in which, for every party k, the “distance to go” is less than volatility. The resulting party-specific, volatility-adjusted margins may then be aggregated across parties for a given electoral district using party vote shares as weights, and then similarly aggregated up to the level of the legislature. Further detail about the construction of this index is provided in Reference Winer, Congleton, Grofman and VoigtDash, Ferris, and Winer (2019).

In Figure 2, which is based on election outcomes in fourteen major Indian states between 1967 and 2009, we compare the first versus second-place vote margin, this margin after it has been deflated by volatility, and the multiparty volatility-adjusted margin just defined. (As the world’s largest democracy, India is an interesting case.) It is apparent from the figure that these indexes for Indian states do not reveal similar patterns. The unadjusted vote margin suggests that competitiveness within Indian state constituencies exhibited trend growth over time, whereas both the volatility-adjusted margin and the Przeworski/Sprague multiparty volatility-adjusted measure of competitiveness exhibit greater, but not the same, variability, and no trend. The same indexes constructed for Canadian national elections from 1867, shown in Reference Ferris, Winer, Grofman, Gallego and SchofieldFerris, Winer, and Grofman (2016), also exhibit differences in their patterns over the history of the modern Canadian state. In measuring competitiveness as uncertainty, incorporating volatility and allowing for more than two parties appears to be warranted.

Figure 2

Alternative measures of the closeness of Indian state elections 1967–2009

Source: Dash, Ferris, and Winer (2019). The fourteen major states constitute about 85 percent of the Indian population. The first versus second volatility-adjusted margin = (v1 − v2)/volatility. The multiparty volatility-adjusted index is normalized to equal 1 at the constituency level if all party-specific vote share margins included are smaller than the corresponding vote volatility. Indexes are interpolated linearly between elections.

The discussion so far focuses on majoritarian electoral systems. Reference Blais and LagoBlais and Lago (2009), and Reference Grofman and SelbGrofman and Selb (2009) extend the measurement of vote share margins to proportional representation (PR). The PR system usually involves many candidates competing under a formula that apportions seats across parties depending on relative vote shares. As a competitiveness measure that they think may be applied to all electoral systems, Blais and Lago compute the minimal number of additional votes required under existing rules for any party to win an additional seat, divided by the number of ballots per seat to allow for the fact that it becomes more difficult to change the outcome of an election as the number of seats increases.Footnote ⁵⁷ Competitiveness increases as this number falls. This measure resolves into the first versus second, party-level vote share margin in a majoritarian, single member plurality rule (SMP) system.Footnote ⁵⁸ Grofman and Selb compute a similar number, but for each party, and then aggregate across parties using vote shares. Their measure too resolves into a vote share margin in an SMP system. These measures also vary across PR systems and, in Grofman and Selb’s study, is also shown to vary over time.

6.1.2 Competitiveness and the Asymmetry of Safe Seats

An alternative approach to measuring closeness focuses on the riskiness of the contest for political parties in the polity as a whole, rather than on unpredictability at the district level. It is well known that parties that compete across multiple districts in a majoritarian system will target marginal constituencies.Footnote ⁵⁹ The strategy of targeting districts where the previous outcome can be most easily overturned suggests that the proportion of marginal seats (MS) is a useful measure of the degree of competitiveness. This measure can be improved by accounting for the fact that the party with relatively fewer marginal seats as a proportion of its total seat count, and hence with more safe seats, has an advantage in knowing where to target its resources. To do so, an adjustment in MS can be made that accounts for any asymmetry in the distribution of safe electoral districts among parties.

To classify seats as safe, we can follow Reference BodetBodet (2014) and define a safe seat as one that lies in the upper tail of the distribution of the volatility-adjusted vote margins of incumbents. The proportion of marginal seats then is equal to $MS=1-\psi$ , where $\psi$ is the proportion of safe seats in a given election. Next we compute the Euclidean deviation from, say, a three-party equal sharing of safe seats as a measure of asymmetry (where the third party is “all other” than the two leaders), and normalize this to equal 1 if one party has all the safe seats, yielding ${\varphi }_3.$ Finally an asymmetry-adjusted marginal seats measure for the election can be defined as $AMS=1-\left\{\psi {\varphi }_3\right\}$ . (See Reference Winer, Congleton, Grofman and VoigtDash, Ferris, and Winer 2019 for more detail.) This index can be computed for any majoritarian polity over time.

To illustrate the pattern of competitiveness that can be revealed by such a measure, we show in Figure 3 the change in AMS averaged across the fourteen Indian states used in Figure 2, while contrasting this index to changes in the unadjusted marginal seat index MS. The diagram indicates a considerable amount of variation in the marginal seat dimension of competitiveness, particularly in the later 1980s and early 1990s. Moreover, in the case of these Indian states, accounting for asymmetry in the distribution of safe seats among major parties reveals a smaller degree of variation in competitiveness than a focus purely on marginal seats would have recognized.

Figure 3

Change in marginal and asymmetry-adjusted marginal seats in fourteen Indian states, 1972–2009

Source: Dash, Ferris, and Winer (2019). A safe or non-marginal seat is one for which an incumbent’s volatility-adjusted margin of victory over the nearest rival is one standard deviation above the mean of a rolling three past-election sample of such margins. Indexes are interpolated linearly between elections.

The same indexes over the history of Canada from 1867 are found in Reference Ferris, Winer, Grofman, Gallego and SchofieldFerris, Winer, and Grofman (2016). What marginal seat indexes look like for the US states referred to in Table 1 has not yet been investigated.

6.1.3 Competitiveness as the Incumbent’s Probability of Loss

We conclude our overview of measures of competitiveness as unpredictability by considering closeness from the perspective of the incumbent party. Reference Kayser and LindstadtKayser and Lindstadt (2015) have constructed an index of the incumbent’s electoral risk, whatever the electoral system, by estimating the probability that the governing party will lose the next election. Kayser and Lindstadt’s measure depends upon: (1) estimation of the elasticity of seats with respect to votes (the swing ratio) under SMP, assumed to equal 1 under PR; and (2) estimation of the distribution of vote swings based on the entire history of such swings. These two are combined to estimate the probability of a seat swing large enough to close the seat gap between the incumbent and its main challenger, thus opening the door to a turnover of power. They avoid problems associated with third parties that might win, or with coalition formation, by assuming (in accordance with the record) that only two major parties or coalitions will ever form a government.

In Figure 4, we present from Reference SplinterSplinter (2019) the Kayser and Lindstadt probability of loss measures of competitiveness for Canada from 1896 to 2011 together with Springer’s recalculation of this measure using historical rather than estimated swing ratios and rolling estimates of seat swings based on several previous elections. While the calculated probabilities do differ in their levels, the methods agree in the pattern across time. Reference Kayser and LindstadtKayser and Lindstadt’s (2015, 247) table illustrates the variation in loss probabilities that exists across a large sample of democracies with different electoral systems.

Figure 4

Loss Probabilities for Canada: 1896–2011

Source: Splinter (2019). Thesis data spreadsheet: columns AL and AM. Historical data refers to the use of actual rather than estimated swing ratios and, in this case, estimated seat swings are based on rolling estimates that use eight to eleven previous elections.

Reference Cronert and NymanCronert and Nyman (2021, 339) extend Kayser and Lindstadt’s loss probability measure to allow for the possibility that in PR systems especially, but also in others, success in an election need not lead to participation in government. Their generalization requires estimation of the probability that a party enters government (or loses office) through the formation of different coalitions, which can be a complicated calculation.Footnote ⁶⁰

6.1.4 Contestability

Contestability refers to the existence of, or potential for, credible challengers to the governing party. This dimension of competitiveness is not simply another aspect of closeness since the opposition may be credible even though the government is in no danger of defeat in the next election. It is not clear how this dimension should be measured.

Sometimes reasoning about competition in private markets is carried over to the electoral context with the conclusion that contestability in a given jurisdiction rises with the number of parties. Such reasoning in the political context is, we think, wrong. In private goods markets, the entry of new firms signals the loss of market power for existing firms and allows firm rivalry to benefit consumers by either increasing their dimensions of choice, lowering market price, or both. On the other hand, the winner-takes-all nature of a majoritarian election (we consider proportional elections later) means that the entry of additional parties will decrease the likelihood that any one challenger will be a credible rival to the incumbent because entry usually fragments the vote. Moreover, the larger set of policy choices that comes from the enlarged set of parties is less meaningful than in a market because in the end, only one platform or menu of policies can be implemented. The situation is different in a market where there is room for many “restaurants” serving food.

This reasoning implies that in majoritarian electoral systems, fewer rather than more parties lead to more contestability by increasing the likelihood of there being a credible challenger. Contestability may be high even when there are only two parties, or even if there is just a credible potential challenger, as Reference DemsetzDemsetz (1968) argues is the case in a market even under conditions of natural monopoly, though the high fixed cost of party entry suggests that high contestability in an electoral system requires an actual rather than a potential rival. In this Duverger–Demsetz view of contestability, as Reference Ferris, Winer, Grofman, Gallego and SchofieldFerris, Winer, and Grofman (2016) refer to it, the pressure on voters to desert small parties with little chance of winning, which tends to reduce the number of parties, is a factor that helps to maintain contestability over time.

Thus, for majoritarian systems, the market-based conclusion that more parties necessarily mean more competition is inverted. One can argue that the situation in proportional electoral systems varies only in degree. Proportional representation allows parties to form around more distinctive ideological positions by allowing a larger number of parties to survive. Because only the winning coalition decides on the policies to be implemented, however, unlike a market where each firm can supply a different menu of products, voters are still concerned with the fact that minor party proposals are less likely to be implemented. Hence despite the greater persistence of minority interests in the legislature under PR, the same type of Duvergerian pressure winnows the political positions that can survive in equilibrium.

In Figure 5, we illustrate how contestability, measured imperfectly here as the effective number of political parties (ENP), has varied over the 1962–2017 period for five different countries, chosen to represent a variety of different electoral systems.Footnote ⁶¹ ENP weights parties by their relative size to capture both their number and their size distribution (Reference Laakso and TaageperaLaakso and Taagepera 1979): $ENP={\displaystyle \mathit{1}/\mathit{\Sigma }{v}_i^2}$ where $v_i$ is the vote share or seat share of party i. Economists will recognize this index as the inverse of the Hirschman–Herfindahl index (HHI) of market power with vote or seat shares of parties used in place of market shares of firms. The Duverger–Demsetz view suggests that contestability is at a maximum in SMP systems when two parties share the vote equally, in which case ENP = 2.

Figure 5

Effective number of political parties: 1962–2017

Source: Gallagher and Mitchell (2009) Election indices dataset online: www.tcd.ie/Political_Science/people/michael_gallagher/ElSystems/index.php, Accessed 11/1/2021.

Both the United States and Canada have single-member plurality elections, with Canada having a parliamentary system and the United States a presidential one. Duverger’s Law – that under SMP the party structure will converge on 2 – is most strongly apparent for the United States while Canada exhibits more persistent departures from 2. Israel, with its PR system, has a much larger number of parties. France and Germany are intermediate cases: France has a mixed presidential and parliamentary system based on a two-round voting procedure, while Germany uses a mixed member proportional voting system (MMP), where both geographic and polity-wide party interests are reflected in proportional seating.

In addition to illustrating the variation in ENP that arises across electoral systems and over time, Figure 5 illustrates the well-known characteristic that electoral systems using PR or MMP typically have a larger number of parties surviving over time in comparison to majoritarian, plurality-rule systems. This observation is consistent with the view that PR systems produce greater party heterogeneity at the cost of contestability, while parliamentary and presidential systems lead to more contestable elections and, hence, greater accountability of elected governments.

However, Reference Gehl and PorterGehl and Porter (2020) suggest caution in interpreting small party numbers as evidence of greater competitiveness and accountability. They note that while smaller numbers may heighten contestability, the same Duvergerian logic that delivers small numbers also allows for greater collusion among parties to restrict the entry of challengers. The need to incorporate this trade-off into a more complete measure of contestability remains an outstanding challenge. A partial solution is provided by the indexes suggested by Reference Gerring, Hicken, Weitzel and CojocaruGerring and colleagues (2018) and Reference Skilling and ZeckhauserSkilling and Zeckhauser (2002). The former combine closeness and turnover in an index based on the difference in vote or seat shares between the incumbent and its largest challenger, while the latter use the inverse of ENP, that is, the HHI index, with the length of time a party or coalition is in office as its basic element. The potential for collusion is, however, only indirectly captured in these alternatives.

Reference BuchlerBuchler’s (2011, chapter 6) argument that relatively homogeneous electoral districts in which one party dominates are better because they minimize the need for a successful party to balance conflicting demands, which is at odds with Gehl and Porter’s view, raises a second problem for the measurement of contestability. When electoral conditions favor a single party, contestability depends on the extent to which intraparty competition acts as a substitute for interparty competition. How intraparty competition is to be recognized in a contestability index is a second challenge.

6.1.5 Responsiveness

The association of competitiveness with the degree to which policy choices are responsive to the demands of citizens is one of the important ideas we pointed to in our initial discussion of the meaning of political competition. This view has a direct counterpart in a large body of empirical work that investigates how changes in the factors underlying citizen demands are reflected in changes in public policies of various kinds.

For example, Reference Besley and BurgessBesley and Burgess (2002) consider how Indian state governments respond to falls in food production and to floods in Indian states, showing the response is greater when newspaper circulation is high and political competition, measured by differences in seat shares of major parties, is intense. Reference Hobolt and KlemmensenHobolt and Klemmensen (2008) use the relationship between promises in political speeches and changes in public expenditures to conclude that “higher levels of political contestation are associated with more responsive executives” (309). (Competitiveness is measured here by the proportion of the electorate that says it would vote for the governing party in the next election, a survey based measure of competitiveness that can be added to our list of indexes.) Reference Soroka and WlezienSoroka and Wlezien (2010) show that policy choices respond well to changes in what citizens want (as measured by opinion surveys) in the case of defense and social expenditures in the United States, the United Kingdom, and Canada. Reference Kayser and PeressKayser and Peress (2012) find that vote shares for incumbents respond positively to high levels of domestic growth and to low levels of domestic unemployment, whereas global levels of these variables have little effect, while Reference Creedy and MoslehiCreedy and Moslehi (2011) use calibration of a median voter model to conclude that variation in preferences for public versus private goods plays a substantial role in explaining cross-country variation in the composition of public expenditure. Reference FujiwaraFujiwara (2015) finds that the introduction of electronic voting in Brazil effectively enfranchised less educated voters, leading to more left-wing legislators and increased spending on public health care.

This is just a small sample from an extensive literature that documents the responsiveness of policy choices to what citizens demand. An index that can be used to compare the degree of responsiveness across studies has not yet emerged from this literature.Footnote ⁶²

Finding that public policy is responsive to changes in the factors underlying citizen demands does not necessarily imply that the ideological positions expressed by parties or elected representatives are the same as those of the electorate (Reference Romer and RosenthalRomer and Rosenthal 1979). Politicians compromise their own interests and balance those of different voters in the pursuit of elected office. Accordingly, a branch of the responsiveness literature is concerned with measuring how well the ideological positions of political parties correspond to those of the electorate, sometimes in the context of a (still unresolved) contest between alternative electoral systems. Reference Blais and BodetBlais and Bodet (2006), Reference Golder and StramskiGolder and Stramski (2010), and Reference Rosset and SteckerRosset and Stecker (2019) are examples of studies of this kind.

6.1.6 Measuring Competitiveness between Elections

Interparty competition also arises between elections in the legislature. Two measures of competition between elections that have been well used in the empirical literature are the size of the seat majority held by the governing party or coalition and a related, discrete indicator of the minority/majority state of a parliamentary government (e.g., Reference Ferris, Park and WinerFerris, Park, and Winer 2008; Reference Gamm and KousserGamm and Kousser 2021; Reference PotrafkePotrafke 2021). Competitiveness in the legislature will be higher when seat shares of major parties or coalitions are similar since outcomes will then be harder to manipulate and so less predictable. It will likely also be higher when a governing party does not have a simple majority of seats.

6.2.1 Swing Voters and Economic Growth

An incumbent is less likely to win an upcoming election the larger the proportion of swing voters in the electorate is, a result shown in the Appendix. Reference Besley, Persson and SturmBesley, Persson, and Sturm (2010) test the hypothesis that the US Voting Rights Act of 1965 and other reforms of the period led to more growth in southern US states by increasing the proportion of swing voters who, they assume, favor policies conducive to growth relative to other policies from which they did not benefit. Using a standard growth regression with a political competitiveness index added, they show that the rise in competition after the mid-1960s is associated with more rapid growth of personal incomes. Competitiveness is measured by a variable like that recorded in Table 1 averaged over all state-wide contests for a variety of elected offices.

It is important to note that the policies favored by swing voters need not always favor growth over redistribution. For example, while Reference Padovano and RicciutiPadovano and Ricciuti (2009) find that the 1995 institutional reforms in Italy that permitted greater electoral freedom in the choice of regional governors reduced political rents and increased regional incomes, the same authors (2008) find evidence at the national level in OECD countries that greater competition, measured by the first versus second party seat share margin, is inversely associated with growth, possibly because more competition leads to short-term, redistributive policy choices aimed at buying votes.

There is some evidence concerning the effects of contestability on economic growth. For example, Reference Alfano and BaraldiAlfano and Baraldi (2011) consider the relationship between an HHI index of vote shares for twenty Italian regions and find an inverted U-shaped relationship with growth peaking at an HHI value of 0.3 or an ENP of 3.33. Analyzing Indian states, Reference Chhibber and NooruddinChhibber and Nooruddin (2004) find that elections involving more than two parties, in which case, we have argued, contestability is relatively low, result in the diversion of government funds away from development activities that stimulate growth. The unresolved issue of exactly how to measure contestability hangs over these results.

6.2.2 Implications for Government Size and Procurement Efficiency

We have discussed how special interest politics and the working of the common pool problem may both lead to the growth of the public sector. Does more intense electoral competition reduce it by, for example, generating higher-quality monitoring of government programs and more effective transmission of information to voters? Or, instead, does more competition lead to larger government by exacerbating these problems? A number of empirical studies indicate that competitiveness tends to reduce public expenditure, though it is not obvious why this happens. See, for example, Reference Rogers and RogersRogers and Rogers (2000), Reference Skilling and ZeckhauserSkilling and Zeckhauser (2002), Reference Ferris, Park and WinerFerris, Park, and Winer (2008), Reference Aidt, Dutta and LoukoianovaAidt et al. (2006), Reference Aidt and JensenAidt and Jensen (2009), Reference Eterovic and EterovicEterovic and Eterovic (2012), and Reference Harrison, Lin and XuHarrison, Lin, and Xu (2014). These authors each adopt a different measure of competitiveness including: the vote share received by the incumbent; a Hirschman-Herfindahl index of the proportion of time different parties were in power; the proportion of seats won by the wining party; the Polity IV index of political competitiveness across countries; and the number of years that a ruling party has been in power.

At the municipal level, Ashworth and colleagues (Reference Ashworth, Geys, Heyndels and Wille2014) use the number of participating parties in Flanders and vote volatility as a measure of competitiveness and Reference Broms, Dahlstrom and FazekasBroms, Dahlstrom, and Fazekas (2019) use the propensity to use single source contract bidding in Sweden to show that the absence of competition leads to the abuse of power through the manipulation of public procurement practices. In the case of Italy, Reference 70Coviello and GagliarducciCoviello and Gagliarducci (2017) show that worse procurement outcomes at the local level are associated with longer times in office, interpreting their result as evidence that low turnover allows time for collusion with local bidders to emerge.

6.2.3 Political Competition and the Business Cycle

The claim is often made that because information about the motivation for pre-election spending and other policies is costly for voters to acquire, governing parties can use such policies opportunistically without considering their full social costs. Evidence consistent with the role of political competition in countering this decision externality arises in a number of ways. Reference Shi and SvenssonShi and Svensson (2006) find that government budget deficits are significantly higher in developing than developed countries, a finding they attribute to the larger share of uniformed voters and lack of institutional constraints in developing countries. Even more strongly, Reference Drazen and EslavaDrazen and Eslava (2010), using a large data panel of countries find that “well informed voters not only are hard to ‘buy’ through spending increases but also are actually ‘fiscal conservatives,’ that is, they are averse to high overall government spending and deficits” (39). Within developing economies, on the other hand, Reference SchuknechtSchuknecht (2000) finds evidence of expansionary fiscal policy around elections with cycles in public investment particularly prominent. Reference Alpanda and HonigAlpanda and Honig (2009) and, more recently, Aidt and colleagues (Reference Aidt, Asatryan, Badalyan and Heinemann2020) look for evidence of a short-run monetary electoral cycle and find evidence within low-and middle-income countries, but “no cycle in … countries with strong political institutions” (Reference Aidt, Asatryan, Badalyan and HeinemannAidt et al. 2020, 409).

6.2.4 Competitiveness and the Private/Public Composition of the Public Expenditure

Using the multiparty, volatility-adjusted measure of closeness within constituencies discussed in Section 6.1.1, Winer and colleagues (Reference Winer, Ferris, Dash and Chakraborty2021) find that an increase in electoral competition was associated with changes in the composition of Indian state government budgets. In richer Indian states, more swing voters and thus greater electoral competitiveness resulted in a reallocation of state budgets from targetable private goods, such as transfers toward public goods including infrastructure, whereas in the poorest states, spending on targetable private goods grew at the expense of public goods.Footnote ⁶³ This reinforces the view that emerges from our earlier discussion of economic growth, that while greater competition leads to policies favored by swing voters, the policies swing voters favor can differ across distinct parts of a polity.

Reference Gamm and KousserGamm and Kousser (2021) show that states with more equal party representation spend more on education, health, and transportation, leading to better social outcomes and higher incomes. On the other hand, Reference Bracco, Porcelli and RedoanoBracco, Porcelli, and Redoano (2013) find that in Italian municipalities, elected officials substitute salient taxes for those that are less salient (but not necessarily economically preferable) when political competition is tighter.

In work that includes the experience of many partially or nondemocratic regimes, Reference Keefer and KnackKeefer and Knack (2007) show that public investment is dramatically higher in countries with low-quality governance, limited political checks and balances and no competitive elections. They interpret this as governments using public investment as a vehicle to increase their rent seeking when the quality of governance is low. (There can be too much of a good thing!) This paper is just one of many interesting investigations of the economic consequences of alternative forms of governance in the world as a whole, a topic deserving of its own Element.