Choosing between Centralized and Decentralized Models of Tax Administration

Abstract

The fundamental questions addressed in this article are the following: what is the most appropriate approach to organizing the vertical structure of tax administration, and what are the determinant factors that may make an approach more or less optimal in any particular country? Both centralized and multi-level tax administration have certain advantages and disadvantages. However, given the general approach to fiscal decentralization in a country and, more in particular, assignment of taxes and other revenue sources to different levels of government, it is possible to identify better and worse approaches to organizing tax administration.

Keywords:

• tax administration
• local taxes
• fiscal decentralization
• tax autonomy

Notes

¹As we will see in the next sections of the article, there are country examples for these cases.

²See Rubinfeld (1985) who argues that that basic principle of tax administration design should be the lowest possible administration and enforcement costs.

³The relative advantage of different tiers of governments in achieving economies of scale and the existing empirical evidence on this naturally reflect the specific technology prevailing at that time, which can change in the future. We are grateful to an anonymous referee for pointing this out.

⁴However, entering tax returns and processing payments exhibit diseconomies of scale and, beyond certain point, create a bottleneck so that some large countries like the United States deconcentrated this operation to regional processing centers while some Latin American countries (e.g., Colombia, Bolivia, and Ecuador) have outsourced this operation to the banking sector (Ramirez Acuña, 1992).

⁵However, these points should not be interpreted to say that a decentralized approach will never be able to achieve some economies of scale in the use of information. The issue here is that this realization of economies of scale will require information exchange agreements, which can be costly, cumbersome and hard to enforce.

⁶However, Mikesell (2007) points out that the common wisdom that collection performance improves with higher remuneration of tax collectors does not have much empirical support.

⁷Specialization of duties has the added benefit of improving internal control and lowering the chances of fraud and corruption.

⁸Even, in the United States, “state and city tax departments are increasingly ‘outgunned’ in attempting to enforce [the corporate income tax]” (Tannenwald, 2001:42).

⁹The empirical evidence on the existence of size of the economies of scale is reviewed below.

¹⁰However, even in the case of a large sub-national government size, informational economies such as those related to third-party information or economic activities of the same taxpayer in different jurisdictions may not be realized.

¹¹This has been a common practice in a variety of countries such as South Africa and Ukraine and other transitional countries.

¹²The practical feasibility of the required managerial flexibility as well as associated costs and limitations can be learned from country experiences with the “New Public Management” (Fjeldstad & Moore, 2009).

¹³These elasticities were calculated based on the translog specification using the mean values of regressors, in particular the average number of 2,390 agricultural parcels, 1,814 commercial parcels, and 27,733 residential parcels per county.

¹⁴In terms of points of contact there can be some positive aspects in terms of visibility and accountability of local authorities to taxpayers. See the discussion of accountability issues below.

¹⁵In 1997 Nova Scotia, New Brunswick, and Newfoundland/Labrador) replaced their sales tax with a harmonized sales tax (HST) with a 7 percent federal rate (lowered to 6 percent in July 2006 to 5 percent in January 2008) and 8 percent provincial rate applied to the same base as the federal sales tax elsewhere. The HST is administered by Revenue Canada but changes in rate and base require unanimous agreement of the three provinces.

¹⁶Some U.S., states (e.g., Maryland) collect local income taxes as a single line on the state income tax return so that the existence and magnitude of the local tax liability can be obscured in this joint collection and enforcement process. In Canada, there is a single return for provincial and federal income taxes, although a separate calculation of some detail is required for each. And in American states in which local sales taxes are collected, the local and state taxes are collected together, without differentiation, on taxable transactions.

¹⁷Of course, the presumption here is that the central tax administration would prepare its audit plan at least partly on the basis of the “demographics” of different regions, as opposed to the individual characteristics of taxpayers (who happened to be distributed unequally across different regions.)

¹⁸However, in countries with challenging environment for tax administration (e.g., some developing countries), gains in administrative costs of private tax collection can outweigh social costs.

¹⁹See Grady (1997).

²⁰It goes without saying that these motivations may also be present at the central level.

²¹See the discussion in Martinez-Vazquez et al. (2007)

²²This type of externalities also exists when related taxes (e.g., PIT and payroll taxes) are separately administered by different central government agencies.

²³The strategic behavior in tax enforcement by sub-national units may also affect central government revenues in those cases where sub-national jurisdictions are charged with collecting (some) central government taxes. For example, the pre-1994 arrangements in China basically fell apart when the provincial tax administrations strategically put less effort in enforcing central government taxes so to favor their local enterprises leading to a drastic decline in central government revenues.

²⁴Although private tax collectors in the United States do work with multiple jurisdictions and seem to manage working out contractual arrangements with each of their clients. In Colombia, Bolivia, and Ecuador, payments from the national government to the banking sector for accepting and transcribing tax returns is determined based on the frequency of delays and number of mistakes in document processing (Ramirez Acuna, 1992).

²⁵For the information on individual country cases we draw on Vehorn and Ahmad (1997), Martinez-Vazquez and Timofeev (2005), Mikesell (2007), and other sources cited throughout this section.

²⁶This arrangement in the Nordic countries is facilitated by the importance of local taxes that piggyback on central government taxes, in particular, the personal income tax.

²⁷However, the case of Bosnia-Herzegovina with a de facto confederation of the Bosnian/Croat Federation of Bosnia and Herzegovina and the Bosnian Serb-led Republika Srpska can be considered a special case of decentralized tax administration. See, for example, Fox and Wallich (1999).

²⁸Some other countries in transition in Asia such as Vietnam have more recently lived through similar experiences.

²⁹A recent example is provided by in Tanzania, where a pilot decentralization of forestry fees collection from the district to the village level led to multifold increases in revenue (Lund, 2007). This was due to the fact that 95 percent of collections were retained by the village council as opposed to being completely remitted by a salaried district officer to the national treasury with subsequent returning a share of this money back to the district budget.

³⁰In this sense, it has been argued that, if the federal government of the United States were to eliminate its (global) income tax, the states would have to do the same, as they would lack the information necessary to adequately administer and audit their state income taxes (McLure, 1999).

³¹As of 1999, the Internal Revenue Service has written agreements with 126 state agencies representing 50 States, the District of Columbia, American Samoa, Guam, Puerto Rico, the U.S. Virgin Islands, New York City, Louisville, St. Louis, Cincinnati, Cleveland, Toledo, Philadelphia, Pittsburgh, Kansas City, and Columbus. State tax agencies may use the tax data provided for justified tax administration purposes only. Any unlawful disclosure of federal tax data subjects the violator to both criminal and civil penalties. Agency officials are required to submit a Safeguard Procedures Report within 30 days after receipt of federal tax data, and an annual Safeguard Activity Report. On-site safeguard reviews can be conducted by IRS personnel as needed to ensure confidentiality and security of federal tax data, with a minimum requirement of a review once every five years. Agencies are required to maintain a system of standardized records of requests for inspection or disclosure. For more details see Program 21.004: Exchange of Federal Tax Information with State Tax Agencies in the 2009 Catalog of Federal Domestic Assistance.

³²Note that population-weighted votes conform with the objective function of the single tax administration model producing a social optimum in Lopez Laborda et al. (2004),

³³In Spain, some of the incentive problems of upward revenue sharing of the two regional governments under the special fiscal regime, the Basque Country and Navarre, are solved by stipulating in advance a fixed sum of money that that the regional governments need to transfer to the central governments.