Chapter 17: Fiscal federalism
Fiscal federalism concerns the division of revenue collection and expenditure responsibilities between different levels of government and between countries in an economic union. Most countries have a central (or federal) government, state or county governments, town councils and, at the lowest level, parish councils. The central government can usually choose whatever tax instruments it pleases and, although it has freedom in its expenditure, it usually focuses upon national defence, the provision of law and order, infrastructure and transfer payments. The taxation and expenditure powers of lower-level governments are more restricted. This raises the question of why there should be more than one level of government and how the functions of government are allocated between levels. The next section considers the rationale for multi-level government, focusing upon the availability of information. An analysis of the arguments in favour of multi-level government is then given and the optimal
structure investigated. The issues of accountability and decentralisation are then analysed. The essential elements of inter-regional risk sharing are presented next and the distinction between insurance and redistribution is discussed. The final section provides empirical evidence on the extent of decentralisation by countries and functions.
17.1 Multi-level government
The arguments for having government are founded on the existence of market failure and the wish to improve equity. To justify multi-level government, the case must be made that the objectives of efficiency and equity are better-served by a combination of local and central government.
Decisions should be taken at a national level if they involve public goods which serve the entire economy. The obvious example is national defence. The central provision of these services makes it natural to finance them through centrally-organised taxes. Local public goods which benefit only those resident within a defined geographic area could be administered at the national level but there are three arguments to support a lower-level decision. Firstly, the lower level can have more precise information on local preferences. Secondly, at the national level political pressures may prevent there from being any differentiation of provision between communities. Finally, the Tiebout hypothesis implies that efficiency requires numerous communities with different levels of public good provision.
• Multi level government has to be justified
• Preferences of the population affected by a public good
• Decisions made at an appropriate level
17.2 The costs of uniformity
Uniform provision of public goods and services by heterogeneous jurisdictions will be a compromise between the different levels of demand. It must involve some loss in welfare relative to differentiated provision.
A uniform level of provision, G 0 , causes a loss of welfare to society compared to what would be achieved if each group could be supplied with its preferred quantity. Furthermore, the loss increases the more widely dispersed are preferences and the more members there are of each group.
This analysis shows how uniformity can be costly in terms of foregone welfare. A policy of uniformity can then only be supported if the costs of differentiation exceed the benefit. Such costs could arise in the collection of information to determine the differentiation and in the administration
costs of a differentiated system.
The ability to differentiate public good provision between regions allows more accurate targeting of resources to where they are required. This is an equity argument for not having uniform provision. Decentralised decision-making allows each region to communicate its needs to the centre and permits the centre to make differential allocations to the regions.
1. Uniformity imposes a cost
2. Wider dispersion of preferences raises the cost
3. Cost minimised by achieving trade-off
17.3 The Tiebout hypothesis
Despite the clear costs of uniformity, it requires more than this to justify multi-level government. The route to doing this is to use the Tiebout hypothesis that was developed for local public goods. The same arguments apply here.
A community provides local public goods. When consumers have heterogeneous tastes, there is an advantage to different communities having different levels of provision. Each offers a tax rate, level of provision and type of provision that appeals to its population. By choosing where to live, consumers reveal their tastes for public goods. This allows communities to know what to supply and allows consumers to obtain what they want. An efficient equilibrium must ensue.
Transaction costs are relevant in practice, and the problem of optimally dividing a finite population into a limited number of jurisdictions will arise. The fact that the first-best allocation will not be achieved does not necessarily undermine the argument for decentralisation. Starting from a uniform level of services that is too little for some consumers, and too much for others, then a move away from this uniform level by some jurisdictions must benefit some of the consumers. In this way, even restricted decentralisation can be efficiency-increasing.
• Tiebout hypothesis supports heterogeneity
• Location choice reveals tastes
• Efficiency is ensured
17.5 Optimal structure: efficiency versus stability
An issue that arises in designing an optimal structure is that the division may be different for different public goods. There are many public goods provided by the government and, if each were allocated at the correct level of decentralisation, this would imply an equally large number of levels of government. This would lead to the accumulation of costs.
This argument is now illustrated in a simple location model that trades off scale economies against diversity of preference. The point of departure is that centralised decision making produces a ‘one size fits all’ outcome that does not reflect the heterogeneity of tastes. The uniform provision follows from political economy considerations preventing centralised majority voting from allocating different levels of public goods to different districts. It is only by decentralising the majority voting at the district level that it is possible to differentiate public good provision but at some cost of duplication.
• Different public goods should be supplied at different levels of
• Each level of government implies additional costs
• Optimal structure trades off costs and benefits
A government is ‘accountable’ if voters can discern whether it is acting in their interest and sanction it appropriately if it is not. The problem is then to confront politicians with a trade-off between diverting rents and losing office or doing what voters want and getting re-elected. In this view, elections can be seen as an accountability mechanism for controlling and sorting good from bad incumbents.
There are severe problems in monitoring and evaluating the incumbent’s behaviour in order to make informed decisions about whether to re-elect or not. Voters face a formidable agency problem because they are inevitably poorly informed about politicians’ behaviour and type. Moreover, the electoral sanction (pass or fail) is such a crude instrument that it can rarely induce the politicians to do what the public wants.
From this perspective, it might be reasonable to try to organise competition among politicians in order to control them. In this respect, decentralisation is an effective mechanism to control governments’ expansive tendencies. The basic argument is that competition among different decentralised governments can exercise a disciplinary force and break the monopoly power of a large central government.
he information will be revealed if there is at least one government that chooses a different policy from that of the others. Therefore, comparing the performance of their incumbent with other incumbents facing similar circumstances, voters can gain increased control over their politicians and deduce what is attributable to circumstances as opposed to government actions.
• Politicians are self-serving
• Accountability regulates their behaviour
• Comparison across governments provides a check on performance
17.6 Risk sharing
Inter-regional insurance is fundamentally about sharing risk among a group of regions so that no region bears an undue amount of risk.
There are some fundamental principles in mutual insurance. First, risk-sharing is more effective the broader the basis on which risks are pooled. Second, it is more advantageous for a region to engage in mutual insurance with other regions when risks are negatively correlated across regions. Third, there must be sufficient symmetry across regions. The reason is that with an asymmetric regional distribution of risks, some regions will systematically and persistently subsidise others. The distributional considerations will then dominate insurance aspects. Fourth, risk-sharing arrangements require reciprocal behaviour: a region with a favourable shock will help other regions if it can reasonably expect that those regions will in turn help it out in bad circumstances.
Risk-sharing agreements without commitment must be ‘self-enforcing’ in the sense that no region has an incentive to defect unilaterally from the agreement. To be self-enforcing, the risk-sharing arrangement must be such that the expected net benefits from participating is at any time larger than the one time gain from defection (by not making transfer when called upon).
Several implications can be drawn from this simple model of risk-sharing without commitment. First, the time horizon will influence the amount of mutual insurance that is sustainable. The second implication is that the level of risk-sharing that regions can achieve increases with risk aversion. The third implication is about the effect of income inequality. Intuition would suggest that mutual insurance is more likely if regions are ex-ante identical and that regional inequality limits the scope for insurance. But this is not true. The reason is that risk-sharing redistributes ex-post from the region with a positive shock to the other region, but it does not redistribute ex-ante from the rich to the poor region.
• Risk-sharing between regions
• Need for asymmetries
• The role of commitment
17.7 Evidence on decentralisation
The degree of decentralisation of government activity can be measured in several different ways. Three of the measures of fiscal decentralisation are: (i) share of total public revenue collected by the central government; (ii) share of the central government in all public expenditures (including
income redistribution payments); (iii) share of the central government in current government consumption expenditures.
First developed countries are generally more decentralised. Latin America countries decentralised mostly during the period 1980 to 1995. African countries are the most centralised and display little decrease in centralisation (with almost all government spending occurring at the central level). Looking at the world level average (involving up to 48 countries) also reveals a general trend toward greater decentralisation, with the central spending share declining from 75 per cent in 1975 to 64 per cent in 1995.
The functional decentralisation of government activity country-by-country. Housing and Community Amenities are the most decentralised, with an average of 71 per cent, followed closely by Education and Health with an average of 64 per cent each. The least decentralised are the expenditures for Social Security and Welfare with an average of 18 per cent. This is consistent with the normative view that income redistribution is better achieved at the central level.
• Measurement of decentralisation
• Developed countries more decentralised
• Activities of government have different degrees of decentralisation
17.8 Determinants of decentralisation
The positive literature on decentralisation suggests certain empirical regularities concerning the forces that promote decentralisation. Both country size and income per capita play a crucial role in explaining decentralisation. The empirical evidence suggests that for different measures of decentralisation, larger and richer countries are more decentralised. Diversity as measured by urbanisation also increases decentralisation.
Richer regions demand more autonomy because regional income inequality is such that mutual insurance becomes pure redistribution. Moreover, the demand for more autonomy is exacerbated, rightly or wrongly, by the perceptions in the rich regions that the regional transfers are very much influenced by the opportunistic behaviour of the receiving regions (i.e. some form of moral hazard problem at the regional level).
• Decentralisation related to country size and income
• The threat of separation raises decentralisation
• Mutual insurance can become redistribution
READING: Hindriks and Myles 2004, Intermediate Public Economics, MIT, 2004
Chapter 19: Fiscal Federalism
Firstly, why should there be more than one level of government? Using the logic of economic reasoning, multi-level government can only be justified if it can achieve something that a single-level cannot. Explanations of what this can be must revolve around access to information and how this can be best utilized. If this argument is accepted, and it is explored in detail below, then a second question arises. How are the functions of government allocated between the levels? A brief sketch of how this works in practice has already been given; is this outcome efficient or does it reflect some other factors? p453-454
The difficulty that arises here is that the optimal division may be different between public goods. The examples in the introduction have discussed how fire services are organized at a very local level, education at a higher level and defence at an even higher one. There are many other public goods provided by the federal government. If each were to be allocated at the correct level of decentralization, this would imply an equally large number of levels of government. p458
The point of departure is that centralized decision making produces a “one size fits all” outcome that does not reflect the heterogeneity of tastes. The uniform provision follows from political economy considerations preventing centralized majority voting from allocating different levels of public goods to different districts. It is only by decentralizing the majority voting at the district level that it is possible to differentiate public
good provision but at some cost of duplication. p459
the optimal amount of decentralization will be achieved when the benefits from further decentralization, in terms of matching the diversity of tastes, outweigh the cost of differentiating public good provision. Using this basic model we can now illustrate the tendency for majority voting
to lead to excessive decentralization. p460
This result suggests excessive decentralization under majority voting because the critical cost level under majority voting is higher than the critical cost level for optimality. In particular for any cost C ∈ (α/8, α/4), majority voting leads to decentralization (C α/8). Therefore under majority voting there is too much decentralization: voters who are located at the extreme have an incentive to support decentralized provision to get a public good closer to what they want, but the democratic process does not internalize the negative externalities imposed on voters located in the center. p461
It has been seen that the degree of decentralization differs quite substantially between countries. It is also instructive to measure decentralization of public expenditures by function to see whether this is consistent with normative advice. From a normative point of view decentralization is desirable when the need to tailor spending to local preferences dominates the possible economies of scale and cross-regional spillovers. p468