Information about Asia and the Pacific Asia y el Pacífico

11 Basic Issues of Decentralization and Tax Assignment

Ehtisham Ahmad, Vito Tanzi, and Qiang Gao
Published Date:
September 1995
  • ShareShare
Information about Asia and the Pacific Asia y el Pacífico
Show Summary Details
Vito Tanzi*

This chapter deals with an issue normally referred to as the tax assignment problem. A government is responsible for performing various functions. It needs to raise tax revenue in order to finance the functions that require spending. It may be monolithic or unitary, in which case all responsibilities are carried out at one level, or it may be decentralized, whereby some responsibilities are carried out by the top level, that is, the national government, and some by lower levels, normally local jurisdictions. When responsibilities are decentralized, it is necessary to assign resources to the lower levels. This chapter discusses the way in which various taxes can be allocated to different jurisdictions. It also discusses briefly some international experiences and comments on Chinese tax development.

Governments must perform some basic functions, which economic literature has generally classified in three broad categories. First is resource allocation: that is, the need to allocate resources in ways that improve the welfare of a country. In a market economy, the assumption is that the private sector carries out much of the allocation of resources but, given the presence of externalities and public goods, the government is still required to perform some important allocative functions.

Second is the redistribution of income. Once again, the market allocates resources and, through its allocation of resources, generates a certain distribution of the total income among the citizens. Most societies, however, have expected the government to play some role in this area. In particular, the government, in order to support a higher standard of living of poor individuals, has been required to tax proportionately more the income of richer individuals than that of lower-income persons. These poorer individuals often belong to particular categories, such as the disabled, the young, the old, those unable to earn an income because of poor health, the unemployed, or those who live in depressed or poor regions. There has always been a concern that the redistributive role of the government should not carry with it a high cost in terms of the efficiency of the economy by reducing incentives to work, save, and invest or by creating an inefficient government bureaucracy.

The third important function of the government is stabilization of the economy. Most economies experience periods of economic instability. This instability may be reflected in the behavior of prices, of output, of the balance of payments, or in some other ways. When economic conditions diverge significantly from the desired objectives, the government is often required to promote policies aimed at stabilization.

The government tries to perform the above-mentioned functions by using various instruments such as (1) taxes; (2) spending programs; (3) regulations; and (4) other policies, such as monetary and credit policy, exchange rate policy, and so forth. The activities of the government aimed at performing the above-described functions can be organized (1) in a completely centralized way, that is, through a unitary form of government; (2) in a completely decentralized way, that is, through the activities of local jurisdictions; or (3) through a central government that coordinates and complements the activities carried out by local jurisdictions. In some cases, these local jurisdictions can be more properly called local “governments,” especially when their existence is guaranteed by the constitution of the country. In fact, it may even precede the existence of the national government, as is the case in the United States and Germany.

These three alternative organizations of the public sector have implications for public financing and the way the government performs its function. Experts who have studied these problems have generally agreed that complete centralization (whereby a powerful unitary government has responsibility for all the public finance functions, that is, for all the spending and taxing in a country) may be accompanied by some significant costs. Some of these costs may be associated with the fact that such an organization leads to little experimentation and little innovation. The way of doing things determined by the national government can perpetuate itself and become rigid and anachronistic when there is no pressure to innovate. With many smaller jurisdictions competing for resources, innovation is likely to take place even as errors are made. The benefits derived from innovating will often compensate for the errors made. Experts agree also that when tastes and attitudes differ from region to region, as would likely be the case in a large country or in a country with different ethnic and cultural groups, centralization brings about inefficiency as it forces every region to “consume” the same mixture of taxes and public spending. For example, an area might like to spend more for cultural activities, while another might like to spend more for roads. But both would be forced to adjust to the pattern imposed by the national government thus reducing the welfare of their citizens.

A third problem associated with complete centralization may arise when some taxes and some spending programs experience diseconomies of scale. Because they are carried out at a national level, some programs may become inefficiently large and thus lead to organizational problems. Also, bureaucratic inefficiencies and rent-seeking activities can become important in large institutions or programs. Furthermore, central governments may have less direct information on local needs and preferences than local jurisdictions.

By the same token, total decentralization brings with it some inefficiencies that, for particular functions, can be larger than those connected with complete centralization. For example, the pursuit of a stabilization policy for the whole country becomes very difficult because each one of the jurisdictions will not have an interest in pursuing the necessary policies. The reason is that the benefits derived from these policies will be diffused among all the jurisdictions and will thus benefit only marginally the local jurisdiction that implements the stabilization policies. In other words, the spillover effects from policy actions will be large.

Experience among countries shows that when governments become very decentralized, the government loses its ability to pursue efficient stabilization policies. Examples from Brazil, Argentina, Nigeria, and some other countries support the view that excessive decentralization is likely to make the pursuit of stabilization more difficult.

Excessive decentralization also reduces the ability of the government to pursue redistributive policy. Highly decentralized governments have little interest in pursuing redistributive policy because generous spending policies aimed at helping their poor would become very expensive because they would attract the poor from other jurisdictions. Thus, the jurisdiction that promoted these policies would end up carrying a high cost by helping the poor of other jurisdictions. By the same token, highly progressive tax policies would push the rich out of the jurisdiction and would thus reduce its tax base.

Excessive decentralization would also make it more difficult for the country to allocate the right amount of resources to the production of national (i.e., countrywide) public goods, because the national government would not have the means to do so and the local governments would not have the interest to finance these expenditures because the benefit to each of them from these policies would be small. There would be an incentive for each jurisdiction to argue that it derives little benefit from the national public goods and therefore the burden of financing those goods should be shifted to those who do benefit. This is the classic free-rider problem.

Public finance experts generally prefer to have a central government that takes over nationwide responsibilities, such as stabilization, redistribution, and generation of national public goods, but leaves local jurisdictions free to carry out government functions of particular interest to the individuals who live in those jurisdictions. The less spillover the policies generate outside the jurisdiction, the more they should be the responsibility of the jurisdiction. In this way the pattern of taxes and spending can be tailored to local preferences. This would increase the welfare of all the citizens.

Tax Assignment Policies

The basic question that arises then is the assignment of spending and taxing responsibilities between the national government and the local jurisdictions. The question has often been discussed separately for taxing and for spending assignments. In other words, which taxes should be assigned to the different levels of government and which spending responsibilities? It must be recognized, however, that some simultaneity exists in making that kind of decision, because the easier it is to assign taxes to local jurisdictions, the more justified it is to assign expenditure responsibilities to them. It is important not to create an imbalance between expenditure responsibilities and the means available to local jurisdictions to carry them out including own-tax resources, proceeds of shared taxes, and transfer from other levels of government—as discussed below.

If decisions about the delegation of spending responsibilities to the local jurisdictions have been made, the basic issue is how to provide them with the means to finance them. It is generally agreed that local jurisdictions should not be allowed to engage in deficit financing and should be required to balance their budgets annually. In other words, they should cover their expenditures out of current revenues, except for exceptional circumstances related to large and productive capital projects when some borrowing might be allowed. There must therefore be a hard budget constraint applied to local jurisdictions so that their budgets are actually balanced not only ex ante, or on a provisional basis, but also ex post. The budgets must also be balanced without recourse to gimmicks, such as buildup of arrears, anticipation of taxes, extrabudgetary accounts, and the like.

The assignment of funds to local jurisdictions can follow several options. The first, and probably least attractive, option is to assign all tax bases to the local jurisdictions and ask them to transfer upward to the national government some of the revenue so the national government can meet its spending responsibilities. This option is clearly unattractive and welfare-reducing for a variety of reasons. It is not consistent with a national policy aimed at redistributing income through the use of the tax system. It is not consistent with a policy that assigns to the public sector the role of stabilizing the economy and that depends on the use of the tax system to help achieve this objective. It may not be consistent with the lowest possible way of raising public resources, because it inevitably excessively fragments the tax system. And, finally, it may provide the wrong incentives to the local jurisdictions if they know that part of the taxes they collect will be shared with the national government. There is evidence from some countries that this policy leads to inefficient tax administration.

The second option is for the national government to collect all taxes and to transfer some of these funds to the local jurisdictions so they may finance their own spending responsibilities. The transfer of funds to the local jurisdictions can be done through earmarking and thus sharing of total tax revenue or through earmarking and sharing of specific taxes. The first alternative is clearly superior to the second because it gives the local governments a more stable revenue source and gives the national government more freedom in pursuing tax policy options.

There are several problems with the option of transferring from the national authorities all the revenue of the local jurisdictions. First of all, by breaking the nexus between decisions to collect tax revenue and decisions to spend that revenue, the concept of the tax price for public spending (that is, the idea that spending decisions carry a specific cost expressed through the taxes paid) is broken at the local level. Local officials and local taxpayers may not connect the benefits they derive from public spending with the taxes paid. They may, thus, not exercise the required restraining function on expenditure, and taxpayers will be less willing to support the tax effort. Also, no attempt will be made at the margin to bring the benefit from the last unit of money spent in line with the cost of the last unit of money collected in taxes, as would be required by an optimal size of public sector.1

A third alternative is the more customary one of assigning to the local jurisdictions some taxing power and of complementing, if necessary, the revenue raised locally with some transfers from the national government. The taxing power provided to the local jurisdictions can be given by (1) assigning to them the exclusive use of some tax bases; (2) allowing them to share some bases with the national government; and (3) allowing the local governments to piggyback on some national taxes, for example, by asking the national government to charge higher rates on sales taxes and to transfer the additional revenue to the local government.

The way in which transfers are provided is very important to ensure an efficient use of public money. This leads us to the question of which taxes can be assigned to local jurisdictions and which must remain the responsibility of the national government. By assigning specific tax bases to the local jurisdictions and by not making the grants received from the national government dependent on the taxes raised locally, the local jurisdictions would have, at the margin, the option of spending more if they could raise their own tax revenue. Their perception of the benefit cost of this action would presumably guide their decisions on spending and taxing.

The assignment of tax bases to local jurisdictions must take into account several considerations. The first is the importance of objectives (other than the basic one of raising revenue in the most neutral and efficient way) being pursued through taxation. The more important are these other objectives in connection with particular tax bases, the less advantageous it is to leave these tax bases to local jurisdictions. For example, the more weight the government assigns to the objective of income redistribution (through progressive taxation) or to stabilization (through the use of built-in stabilizers) the better it is to leave certain tax bases, such as the progressive income tax and the corporate income tax, to the national government. The reason is that progressive income taxes are the best tool to redistribute income through the tax system, and corporate income taxes (together with the personal income taxes) are the best tax tool to provide a built-in stabilizer to the economy.

Second is the mobility of the tax bases. If a tax base can escape taxation at the local level by moving to another jurisdiction, that base is not a good candidate for local taxation unless harmonization of tax rates among local jurisdictions reduces or eliminates the mobility of tax bases. Thus, the more mobile is a tax base, the greater is the presumption to keep it at the national level. Incidentally, in an integrating world economy this problem reappears within an international context, especially for small countries.

Third, the more important are economies of scale in tax administration for a given tax, the stronger is the argument for leaving the tax base for that tax to the national government. Economies of scale may depend on informational requirements, for example, on a national taxpayer identification number, on technical requirements (such as the use of computers, and so forth), or on other factors. Fourth, the greater is the need to maintain a level field for the whole country (to avoid distortions created by different tax rates), the stronger is the presumption that the tax should be left to the national government.

Given the above premises we can quickly survey the assignment of tax bases starting with the simplest and moving to the more difficult ones. Obviously, this is a complex subject that requires much more space than is allocated here to provide complete treatment.

Import and Export Taxes

Economists generally consider import and export taxes inefficient and undesirable sources of revenue. Their importance in industrial countries is limited, but they still account for a large share of revenue in developing countries. However, to the extent that countries do decide to use them, perhaps because of strong revenue needs, they should always be imposed by the national government, so as to reduce the possibility that major distortions are introduced within the country by differential foreign trade taxes imposed by different jurisdictions.

Taxes on Land and on Real Property

Land and existing structures are among the most immobile of tax bases. Land in particular cannot move because of tax factors nor can existing buildings. Where they are used, taxes on real properties are often imposed by local jurisdictions. Apart from the fact that these tax bases are immobile, it has been argued that they are guided by a benefit-received principle, because spending by local jurisdictions benefits (and thus increases the value of) local properties. Of course, while land and existing structures cannot move, new structures will not be built if the jurisdiction taxes them considerably more than other institutions.2 Also, the capitalization of property taxes in the price of the properties, that is, the reduction of the property values because of the imposition of taxes implies that, even though the base cannot move in a physical sense, it can move in a fiscal sense because property taxes that are not used to provide local services beneficial to the properties will often reduce the market value of the properties. In other words, given the revenue from this tax, the effective tax rate after the capitalization will be higher than the statutory rate before the capitalization.

Sales Taxes

Among the sales taxes, we must distinguish single-stage (excises and retail) from the multistage ones (turnover and value-added taxes). The issues that arise for excises and retail sales taxes are similar. These taxes can be given to local jurisdictions provided that jurisdictions do not levy these taxes with highly different rates. Thus, a jurisdiction that relies on general retail sales taxes or on excises on specific products will not encounter difficulty with these taxes if the rates at which it taxes the products are not so much higher than the rates imposed by neighboring jurisdictions. If the rates are higher, it may encourage its own citizens to shop in the lower-rate jurisdictions. Major factors in this case are the vicinity of the other jurisdictions, the cost of travel, and the value of the goods purchased. If jurisdictions with lower tax rates are near, if the cost of travel is low, and if the item purchased has a high value (e.g., consumer durables or jewelry), then the difference in tax rates cannot be too high or the tax base will migrate toward the low-rate jurisdictions. Generally, competition among jurisdictions limits the scope for rate differentials and thus limits the freedom of actions of local jurisdictions.

Retail sales taxes and excise taxes are generally simple taxes. However, in economies with many small sellers, retail sales taxes may be difficult to administer especially if the rates are high. High rates would promote high evasion, and evasion by retailers when a retail sales tax is in effect is more difficult to control and more costly in terms of revenue losses than, say, tax evasion by retailers under a value-added tax. The reason is that when retailers evade paying a retail tax, the whole tax is evaded while, for the value-added tax, only a share of the tax is lost.

Value-Added Taxes

Value-added taxes imposed with a credit mechanism are generally rebated on exports and imposed on imports because they follow the destination principle. This principle stipulates that these taxes should be paid by the final consumers and should thus not distort trade relations. The application of the destination principle requires border checks by the jurisdiction that imposes the tax. It is neither feasible nor desirable, however, to impose border checks on trade within a country, because this would impose excessive costs and would impede trade flows within the country. For these reasons, value-added taxes are best left as a responsibility of national governments.

The European Community (now European Union) has been considering the use of value-added taxes based on the so-called origin principle, which dispenses with border tax adjustments by taxing production rather than consumption. But, as long as the tax rates in different jurisdictions diverge and as long as the balances of trade between jurisdictions are not in equilibrium, this alternative has many serious shortcomings.3

Personal Income Taxes

Personal income taxes can be global or schedular. In other words, they can be imposed on the total income received by a taxpayer—that is, by combining wages and salaries, interest and dividends received, and income from all other activities—or they can be imposed separately on each type of income.

Schedular Income Taxes

Schedular income taxes can be used by local jurisdictions with less difficulty especially if the taxes on some incomes, such as interest and dividend incomes and wages and salaries, are withheld at the source by those who pay these incomes and the taxes withheld are considered as final taxes. In this case, the taxpayer does not need to present a tax declaration and has no further tax liability. If these taxes are not considered final taxes but are related to the particular situation of the taxpayer, and especially if the incomes received come from jurisdictions other than the one where the taxpayer resides, then administrative problems arise because the tax authorities in the jurisdiction where the taxpayer resides may not receive the necessary information to be able to tax the individual. At an early stage of development, most individuals derive only one source of income, and this income is generally earned in the same jurisdiction where the taxpayer resides. The more developed countries become, however, the greater is the proportion of individuals likely to earn different kinds of income (interest, dividends, and profits, in addition to wages and salaries). Furthermore, these incomes may be earned in different jurisdictions.

Global Income Taxes

For global income taxes to operate well in a modern economy, it is necessary that all the incomes that the taxpayer receives from different sources and different jurisdictions be combined before the tax is calculated. The tax administration of the jurisdiction where the taxpayer resides is unlikely to have information about income earned outside the jurisdiction. Therefore, tax evasion is likely to be significant. For this reason, it is better to leave this tax base to the national government, which is in a better condition to get the relevant information.

Should the local jurisdiction also wish to tax income, it would need to arrange with the national government the sharing of information. The easiest way of taxing global income locally is for the local jurisdictions to use the same statutory base for the national income tax and agree with the national government to share relevant information. This also reduces the compliance cost for the taxpayer because the same information can be used for both the payment to the national tax administration and the payment to the local jurisdiction. Tax competition among jurisdictions would ensure that no jurisdiction imposes tax rates that are much higher than those by other jurisdictions. If they did, some taxpayers, and consequently, some capital, would leave the high-taxed jurisdictions in order to establish residence in the lower-taxed ones.

Business activities may be associated with small establishments or with large enterprises. Small, family-type establishments often do not keep good records of their transactions and are thus taxed on the basis of presumptive principles. For example, they may pay a business tax that may be related to personal income often assessed on the basis of gross sales, or floor space in which the activity takes place, or on the basis of some other criteria. For this kind of activity, local jurisdictions often have as much as, or more information than, the national government. As a consequence, these taxes can be imposed as efficiently at the local level as at the national level. However, to reduce abusive or corruptive practices, careful monitoring of officials that determine the taxes is required.

For larger establishments, the situation is different. They often have branches in various parts of the country, may trade with other countries, and may buy inputs from businesses in other jurisdictions and sell their output in other jurisdictions. They are also more sensitive to tax factors. If these taxes were imposed locally, several difficulties would arise. The local jurisdiction would find it difficult to verify the information provided by the enterprises (related to the cost of the purchases from other areas and to the prices at which the produced output is sold). If the rates at which enterprises were taxed were highly different across jurisdictions, these rates would become an important factor in determining capital movements within the country and could generate serious distortions in the allocation of resources. For these and other reasons, it is good policy to leave the taxation of enterprises in the hands of the national government. Once again, the local jurisdictions can get a free ride by using the information from the national tax administration to tax also, at low rates, these enterprises using the same tax base. As long as the local rates are low, the local taxation of enterprises would not generate major difficulties.

Tax Assignments in Practice

There are many different patterns to be observed across countries in terms of tax assignments and own revenues accruing to various levels of government. To some extent these differences may be explained by varying constitutional and administrative arrangements. Some countries have unitary constitutions, whereas others are federal in nature; tax administration may be either centralized or decentralized. All these factors affect the patterns observed.

Nevertheless, some broad patterns do emerge. Property taxes are predominantly assigned to state or local governments (some Scandinavian countries, and others in the developing world, are exceptions). Income taxes are mainly, but not exclusively, assigned to the center, although there is some local “piggybacking.” Some East European countries, such as Romania, are an exception.

Chinese Tax Reform Proposals

Much progress has been and is being made toward the solution of the fiscal problems in China. A major tax reform is being planned that will modernize the tax system making it more efficient, more productive, and more equitable. I heartily endorse the guiding thought for the tax system reform given in Chapter (9): a uniform tax code, equitable tax burdens, a simplified tax system, a rational division of powers, straightening out of the apportionment of tax monies between the national government and local governments, and building a taxation system that is consistent with the needs of a socialist market economy.

The unification of the enterprise income tax for all types of enterprises is an important step, although I would have had a slight preference for a single rate rather than two. Similarly, the standardization of the individual income tax should be supported, with the slight reservation that the monthly deduction of Y 800 for living expenses may be too high, given China’s current personal income level, and may reduce excessively the revenue from this tax. However, if this deduction remains unchanged in nominal terms, the fast growth of nominal income in China will bring it to a more normal level. Limiting this tax “on people having fairly high incomes” may eliminate much of the potential tax base. Again, the standardization of proposed value-added tax (VAT) should be implemented as proposed. It will be important to limit the rates to two, and to keep the tax base as broad as possible. The proposed consumption tax is a useful complement to the VAT, although the number of taxed items should be kept low. I am unable to express a judgment with regard to the business taxes. I understand that these taxes are applied largely to activities that would not be covered by the VAT. Thus, in a way, business taxes would require or represent an erosion of the VAT base. Perhaps, at some future time, these activities could be added to the VAT base so that the business taxes could disappear.

As for the other taxes contemplated by the tax reform, in general, I would hope that (1) they be kept to a minimum, and (2) that the assignment of many of these taxes to local governments would not prevent the elimination of some in the future.

The most important feature of the tax reform will be the elimination of the contracting system, the shortcomings of which have been eloquently outlined in Chapter 2. This will, indeed, be an important change. The tax reform will be accompanied at the same time by the introduction of a tax-sharing system. Actually, the system will have several features of which proper tax sharing is just one. Some taxes, and especially the VAT, will be collected only by the central government. Others will be collected by and assigned to the local governments. Several questions will deserve close analysis before final decisions are made:

(1) Are the administrative mechanisms in place to allow each jurisdiction to administer the taxes for which it has the collection responsibility?

(2) Does the planned assignment of tax bases to different jurisdictions reflect the most rational criteria?

(3) Are the percentages agreed for the division of the revenues shared in common consistent (when added to their own taxes) with the expenditure responsibilities of the different jurisdictional levels?

(4) What formulas or criteria will be used to share the revenues subject to revenue sharing not just between the central and local governments but among the subcentral governments?

The determination of local governments’ responsibilities will, of course, play a large role in this process. The above questions are clearly in the mind of the Chinese officials dealing with these issues. The answers will require a lot of hard analytical work and difficult political decisions. It will also require a lot of work in setting up the new institutions that will (1) collect revenue, (2) plan and manage public spending, and (3) monitor and allocate the collection and the disbursement of the shared taxes. Much progress will be required in tax administration, in budgeting, and in the setting up of an administrative structure to handle cash receipts and payments for the various institutions.

One point that I would like to make is that even when the current tax reform is carried out, the government should not tie its hands in reforming the tax system again at some future date, perhaps to reduce the still large number of taxes that would remain in the tax system. It is important to retain this flexibility. In other words, the tax-sharing and tax assignment system that is put into place should not become an obstacle to future tax reform. It, itself, should be reassessed periodically.

These are exciting times for China. Many of the changes proposed are fundamental and necessary. They are also difficult. The government should persist with these reforms to achieve the various objectives that it has set to achieve such as (1) better control over the country’s fiscal policy to control inflation and other macroeconomic problems; (2) better allocation of scarce resources to better satisfy the needs of the people and to sustain a high rate of growth; and (3) to achieve a more equitable distribution of income so that all parts of China can share in the growth of the economy. The achievement of these objectives would make China a successful socialist market economy.

The experience of other countries that have carried out fundamental tax reforms or that have successfully dealt with problems created by different layers of governments within their frontiers can help by pointing to alternative options and by indicating the options with the greatest chance of success. Reform is always an endogenous product, however, and its characteristics can only be determined taking into account the special features of a country and its constraints as well as its strengths.


International Monetary Fund.


Problems of this kind developed in Italy after 1978 when a reform centralized revenue collection but decentralized expenditure responsibility.


Some countries, for example, Italy, have at times given preferential property tax treatment to new construction. This policy not only reduces revenue but also creates serious equity issues because new buildings are often owned by wealthy people.


The possibility of sharing the base for sales taxes between the national government and the local governments is a realistic one, as long as, once again, it does not lead to widely divergent rates between, especially contiguous, jurisdictions.

    Other Resources Citing This Publication