Geoffrey Brennan and James Buchanan first developed the Leviathan model of government, in their 1980 book The Power to Tax. In this early formulation government is assumed to act as a monopolist that maximizes tax revenues. Such representation is driven by the other fundamental assumption that the authors make, namely, that rational ignorance, fiscal illusion and outright collusion among elected officials deprive taxpayers-voters of any control they may have over government. Hence the reference to Hobbes’ Leviathan. In The Power to Tax, Brennan and Buchanan develop the fiscal consequences of the revenue maximizing government and point out the constitutional provisions that may hinder the Leviathan in its drive to appropriate taxpayers’ resources.
The sheer originality and the intuitive plausibility of the Leviathan model can be appreciated by contrasting its predictions to those of the traditional Musgravian public finance. To begin with, the fundamental problem of taxation is quite different in the two conceptualisations of the state. In the traditional vision, the government is a benevolent public good provider, which seeks to raise a given amount of revenues subject to certain efficiency and equity constraints.
With a revenue-maximising Leviathan, instead, citizens must impose constraints on the government that limit its ability to raise revenues to a given amount. Also the standard analysis of excess burden is turned on its head. With the amount of revenues to be raised by taxation fixed, the optimal tax base is the one that induces the minimum amount of distortion of taxpayers’ choices, namely, the most inelastic sources of revenues. With the amount of revenues to be raised the maxi-mand, the citizens seek to limit the government to the more elastic tax bases, so to shelter parts of their income and wealth from the Leviathan. The Leviathan model provides a justification of the normative public finance principle of horizontal equity, but for quite different reasons than the standard one, to treat equals equally. A citizen writing a tax constitution to constrain the Leviathan would require that the government impose uniform tax schedules across persons to limit the government’s capacity to engage in tax discrimination as a means of expanding revenues. A similar logic leads to a preference for progressive over regressive taxes, since fewer revenues can typically be raised by tax schedules imposing high marginal tax rates than by schedules imposing low ones. Moreover, the Leviathan model provides an additional justification for Wicksell’s (1896) prescription that expenditure proposals be tied to the taxes that would finance them. Wicksell puts forth this argument as a means to ensure informed choices by citizens as to the benefits and costs of public projects. With a revenue-maximising Leviathan around, the proposal is motivated by the need to ensure a balanced budget and to force the government to provide some public benefits in exchange of the revenues raised. Also debt and money creation are for the Leviathan additional, shrouded means to raise revenues. Puviani (1903) provides a classical treatment of this point. Balanced-budget provisions as well as limitations on the government capacity to print money — to the possible extent of a complete denial of such power (Brennan and Buchanan, 1980, p. 130) — are the best constitutional response. Finally, Brennan and Buchanan argue that, as long as some individuals and firms are mobile, fiscal decentralization forces governments to engage in tax competition, thereby destroying Leviathan’s monopoly on taxation and bringing government spending closer to the preferences of citizens. Hence the empirical restriction that "total government intrusion into the economy should be smaller, ceteris paribus, the greater the extent to which taxes and expenditures are decentralized" (Brennan and Buchanan, 1980, p. 15).
The empirical testing of the Leviathan model has begun in the 1980s and is still an active line of research. The results that can be evaluated as "mixed" at best. Early studies concentrate on data sets drawn from American and Canadian states. Nelson (1986) finds those states that tax personal income have significantly larger government sectors as implied by the Leviathan model; however, a Granger causality test on the estimates suggests that causality may run other way. Nelson also found that the relative size of the government sector varies inversely with the number of local government units. If one assumes that having more local government units signifies a stronger federalist structure and more intensive constraints on government through intergovernmental competition, then this result also supports the Leviathan model. Marlow (1988) and Zax (1989) find that total government size varies inversely with the relative importance of a local government, another result consistent with the Leviathan hypothesis.
At the cross-national level, several studies have also found that federalist structures inhibit government growth.However, Oates (1985) and Forbes and Zampelli (1989) conclude that having a federalist constitutional structure has a negative, but statistically insignificant effect, on the growth rates of public sectors in developed countries. Oates also found that the degree of centralization of tax revenue, a source of fiscal power emphasized by Brennan and Buchanan (1980, p. 185), is either not statistically significantly related to government growth or inversely related at both cross-national and cross-state levels of growth. More recently, Stein (1999) demonstrates that fiscal decentralization is actually associated with larger government in Latin America. Joulfaian and Marlow (1990) find similar results for a cross section of OECD countries, but rationalize them in a way consistent with the Leviathan model. They suggest that greater decentralization enhances citizens’ trust in government, which then allows them to demand more public goods. In reviewing all these contrasting results Oates (1989) resolves most discrepancies among the empirical studies and concludes that, at the local level where citizens are assumed to have the greatest mobility, the evidence is supportive of the Leviathan hypothesis; in all other situations, the Leviathan is a "mythical beast" (Oates, 1989).
Oates (1989) review essay is certainly not the end of the story. In a recent paper Rodden (2002) persuasively argues that existing cross-national studies are insufficient to dispel the myth of Leviathan for two reasons. First, they employ cross-section averages or single year snapshots. They thus shed little light on the dynamic nature of decentralization and the growth of government, both of which are processes that unfold over time. In particular, governance in many countries around the world is undergoing a major transformation since the 1980s. Cross-national empirical analyses (such as Panizza, 1999) demonstrate that a pronounced trend towards fiscal decentralisation is strongly linked with transitions to democracy, especially in large, formerly centralised countries. For these countries, average state and local expenditure as a share of the total government sector have almost doubled from the 1980s till now. Thus it may be inappropriate to conduct empirical analysis as if all countries have reached a long-term equilibrium. Second, until very recently, insufficient attention has been given to the precise institutional incentives created by different forms of decentralisation. If decentralisation is to have a constraining effect on the growth of government, it must occur on both the expenditure and revenue sides. In the vast majority of countries, however, increased state and local expenditures are funded increasingly by grants, shared revenues, or other revenue sources that are controlled and regulated by the central government. Expenditure decentralisation without corresponding local tax powers will not engender the tax competition that drives the Leviathan model, nor will it strengthen the agency relationship between local citizens and their representatives. On the contrary, decentralisation funded by "common pool" resources like grants and revenue sharing might have the opposite effect. By breaking the link between taxes and benefits, mere expenditure decentralisation might turn the public sector’s resources into a common pool that competing local governments will attempt to over-fish, with the result to intensify the growth of government. Once these effects are controlled for, Rodden (2002) finds results that are in line with the Leviathan hypothesis in a pooled sample of 70 countries.
Other critiques to the Leviathan hypothesis are moved at a theoretical level. In their public choice approach to taxation, Hettich and Winer (1999), among others, question the Leviathan model as it assumes, but does not explain the source and the stability of the dictatorial power of the Leviathan. However, political economics models of political accountability, such as Persson et al. (1997), provide an explanation of why and to what extent rational taxpayers-voters allow the government to secure rents for itself at the expense of taxpayers-voters welfare. These models can be considered as a theoretical development of the early Leviathan model. The Brennan and Buchanan (1980) formulation can be reinterpreted in the logic of the political accountability models, as a government that is not accountable at all. Moreover, models of political accountability provide and explanation of the source and stability of a government behaviour in Western-type democracies that is characterized by a significant degree of Leviathan-style rent appropriation. As such, they provide the reply to the Hettich and Winer (1999) critique.
Interestingly, these models are biased against providing Leviathan-style results, because they suppose rational, rather than fiscally illuded, voters and see the constitutional rules as a weak, rather than the only, device to constrain the behaviour of elected officials. Constitutions are incomplete contracts, which do not offer to elected politicians an explicit incentive scheme that associates well-defined payoffs with actions in all states of the world. This makes it hard to tie specific rewards or punishments to the decisions a politician takes. Basically, constitutions reward elected politicians simply by allowing them to remain in office and sanction them simply by throwing them out of office. The mechanism that distributes such rewards and sanctions is the electoral process. Thus, unlike Brennan and Buchanan (1980), the models of political accountability study whether ordinary politics, as opposed to constitutional politics, is able to tame the Leviathan. The base of the argument is that the electoral process must carry out an adverse selection process of searching for potential candidates that are less likely to behave discretionally and divert rents from the public budget at the expense of the welfare of taxpayers-voters. Two are the sources of this discretional power: (1) an asymmetry in action. When citizens elect their representatives, they temporarily delegate the exclusive decision making authority over policy making to the holders of public office. This creates room for abuse of power between elections. (2) An asymmetry in information. In many cases, politicians have access to much better information on the relative merits and precise consequences of alternatives policies than the population at large. This also creates room for potential abuse by the holders of public offices.
Clearly, any abuse of power reduces the electoral accountability of the system and the utility of voters. To minimize the welfare loss, voters coordinate on a voting strategy that makes the elected officials as accountable as possible, given the incomplete nature of the contract embedded in the constitution. The strategy is based on an intertemporal trade off: if the elected official diverts too many resources today, he is not elected again, i.e., he is removed from office and will not be able to appropriate resources tomorrow. The amount of diversion depends on the severeness of the asymmetries in action and information of above and on the institutional framework.
The simplest possible case is when voters share the same information as politicians and the institutional setting is a "pure" presidential democracy, i.e., one where political decisions are taken by a single branch, which for convenience we call the executive, elected by the voters. The elected officials derive utility at the expense of voters only because of the asymmetry in action, created by time interval between two elections. For simplicity all elected officials are the same, so the results are driven from the institutional framework, rather than by the competence of officials. The executive decides over the budget, i.e., it supplies a certain amount of public good, from which voters derive utility, using tax revenues. The executive can also appropriate these revenues, entirely or in part. To the extent that the executive diverts revenues from the production of the public good, the executive’s utility increases and that of the voters decreases. Since voters know the "technology" available to the executive for transforming revenues into the public good, they can also infer the amount of resources the executive has appropriated for personal use. Based on this information, they decide whether to reelect the executive or not.
Persson et al. (1997) show that the immediate conclusion that voters decide not to reelect the executive if it has appropriated any nonzero amount of revenues is not in voters’ best interest. If voters adopt such a voting strategy, the executive anticipates that it will be thrown out of office at the end of its mandate, and appropriates as many resources as possible. Such strategy is clearly inefficient for voters. The efficient strategy is to allow the executive to appropriate the amount of resources that makes it just indifferent between being reelected for an iteration of future terms, or appropriate everything now and not be reelected. Essentially, democracy, because of the incompleteness of the contract that the constitution establishes between voters and their delegates, has a cost in terms of resources that citizens must transfer to elected officials to motivate them to hold office. This is an endogenous Leviathan equilibrium.
Voters’ welfare further decreases when there is not only an asymmetry in action, but also an asymmetry in information. Suppose that the technology with which the executive transforms revenues into public good changes randomly, possibly because of the dynamics of the economy; suppose also that voters cannot observe the technology currently available, while the executive can. In this case voters cannot tell whether the, say, small amount of public good they receive is due to the executive having appropriated a large amount of resources for personal use or to a poor realization of the technology. In this case they coordinate on a voting strategy centered on a minimum amount of public good consumption. If they get that amount or more, they reelect the executive; if they get less, they throw it out of office. Faced with this voting strategy the executive seeks reap-pointment only if the available technology is sufficiently favorable; in this case it gives the voters the minimum amount of public goods they want. Otherwise the executive knows that, even if it diverts no resources from those it is to administer, it will not be reappointed. Hence, it has a clear incentive to appropriate as many revenues as possible. Persson et al. (1997) show that voters are hurt by incomplete information, while the executive captures some of the informational rents. In this setting, the executive diverts more resources than under fully informed voters and does so relatively more when the economic conditions are so poor as to make reelection unlikely.
Persson et al. (1997) explore the robustness of these results under different institutional regimes (parliamentary or congressional systems, different budget approbation procedures and the like) and find that in all these forms of democracies it is always rational for voters to allow holders of office to appropriate a nonzero amount of rents. Modern democratic governments are neither a monopolistic Leviathan as in Brennan and Buchanan (1980) nor the political counterpart of the perfect competition model (Becker, 1958), but something in between. Padovano (1995) argues that, to place the equilibrium government behaviour as far away from the Leviathan endpoint as possible, one must shorten and fix the length of the legislatures, adopt majori-tarian electoral systems and expand the number of political positions assigned through direct electoral processes rather than through indirect designation. Furthermore, Padovano et al. (2002) find that an independent judiciary may reduce the amount of discretion of the Leviathan.
In conclusion, the Leviathan model of government has started as a logical and fearsome possibility, has gone through the rogue waters of empirical tests and theoretical questioning of the plausibility of the underlying hypothesis, only to be reformulated in what seems to be a less extreme, but more general and solid vision of modern democracies.
