The (Mis-) Measurement of Fiscal Decentralization in Developing and Transition Countries

Abstract

There is an extensive literature on the impact of fiscal decentralization on economic growth, development, and public sector effectiveness. However, the empirical literature on fiscal decentralization has exclusively focused on measuring the finances of elected or “devolved” local governments. Other types of decentralized expenditures, including deconcentrated and delegated expenditures, have been systematically excluded from measurement and analysis in the public finance and development literature. Our analysis considers the extent to which using devolved expenditures as a proxy for all devolved expenditures may have impacted the findings of the empirical literature. We collect comprehensive vertical expenditure profiles for health and education services in twenty-nine developing and transition countries and find that by exclusively focusing on devolution, previous analyses have overlooked two-thirds of local public sector expenditures. By excluding these “nondevolved” decentralized expenditures, the previous (often inconclusive) empirical analyses are likely to have suffered from omitted-variable bias.

Keywords

fiscal decentralization devolution deconcentration developing countries

Decentralization is widely regarded as an important public sector reform in the public administration and development literature, and the study of fiscal decentralization is widely acknowledged to be an important aspect of public sector economics. While decentralization is pursued in many countries for many reasons, including administrative and political aims, the notion that decentralization can increase the efficiency of public sector spending is part of the bedrock of modern public finance (e.g., Tiebout 1956; Oates 1972). Over the past half century, a robust empirical literature has studied local government expenditure decisions in the United States and in other developed economies.¹

Similarly, a considerable empirical literature has evolved that attempts to determine the impact of fiscal decentralization on economic growth, development, and public sector effectiveness in the developing world. Yet, while decentralization has traditionally been divided into three types—devolution, deconcentration, and delegation (Rondinelli 1986)—the empirical literature on local public finance has not paid equal attention to each of these three types of decentralization. Although the empirical literature on fiscal decentralization has focused almost exclusively on the finances of elected or “devolved” state and local governments, many developing and transition countries rely considerably on deconcentration and delegation as mechanisms to provide and fund local public services.

If we are serious about determining the impact of fiscal decentralization on development outcomes in developing and transition countries, then a rigorous analysis of subnational public finances and intergovernmental fiscal systems cannot continue to ignore the measurement and analysis of nondevolved local public sector expenditures. The current study makes an important contribution in filling this void in the empirical literature by measuring the size and composition of local public sector expenditures in health and education for an initial set of twenty-nine developing and transition economies and by considering the implications that this newly available data on nondevolved local expenditures has for the existing body of empirical literature on fiscal decentralization.

This article is organized as follows. First, we briefly review the state of knowledge and practice pertaining to the measurement of fiscal decentralization. Second, we argue that the failure to properly measure and account for nondevolved expenditures in the existing empirical literature may have resulted in the misspecification of empirical models that seek to analyze the impact of fiscal decentralization on development outcomes. Third, we describe a comprehensive set of definitions and metrics to measure devolved as well as nondevolved local public sector expenditures in countries around the world. Fourth, we present and analyze the results from an effort to measure local public sector expenditures in health and education in twenty-nine developing and transition countries around the world. Finally, we conclude with lessons and next steps, not only for the research community studying decentralization and local public finance but also for the global development community at large.

Literature Review: The Measurement of Fiscal Decentralization

Rondinelli (1986, 2) defines decentralization as the “transfer of responsibility [and authority] for planning, management, resource-raising and -allocation and other functions from the central government and its agencies to field units of central government ministries or agencies, subordinate units or levels of government, semi-autonomous public authorities or corporations, or non-governmental or voluntary organizations” (emphasis added). It is fair to say that this definition, in some form, is the de facto consensus definition of decentralization among international policy practitioners (Litvack and Seddon 1999). This definition results in broad understanding of fiscal decentralization, whereby the concept and measurement of fiscal decentralization not only cover the transfer of responsibility and authority over fiscal resources to “subordinate units or levels of government” (i.e., devolution) but also include the transfer of responsibility and authority for the management of financial resources to “field units of central government ministries” (i.e., deconcentration).² This Rondinellian definition of decentralization further suggests that delegated expenditures (the degree to which responsibility and authority for the management of resources has been transferred to “semiautonomous public authorities or corporations, or nongovernmental or voluntary organizations”) ought to be included in a measure of fiscal decentralization (i.e., delegation).³

There is a considerable divide, however, between this broader understanding of fiscal decentralization—which is widely acknowledged to include devolution, deconcentration, and delegation—and the empirical analysis of the phenomenon. Informed by its roots in the study of state and local finance in the United States and other developed federal countries, the public finance and fiscal federalism literature have often taken a narrower view, often defining fiscal decentralization as the devolution of fiscal responsibilities of the federal or national government to state and local governments (Xie, Zou, and Davoodi 1999; Dafflon 2015).

There are important differences between the various types of local government structures, and it may a priori be reasonable to assume that in the absence of locally elected political leaders, deconcentrated public sector spending is not as effective and responsive as spending by elected (devolved) local governments. Indeed, the traditional fiscal decentralization literature has—both implicitly and, in some cases, explicitly—emphasized the role of devolution in part because of its greater potential to improve the efficiency of public resource allocations. Other proponents of decentralization have emphasized devolution because it enhances democratization by increasing the competitive political space, enhances political participation, and introduces an element of vertical balance of power and accountability.

However, the exclusive focus on only one form of fiscal decentralization in the public finance literature conflates fiscal decentralization (the decentralization of responsibility and authority for raising revenue and executing expenditures) with the higher degree of political decentralization (the transfer of political authority and responsibility) commonly associated with devolved local government systems. While the two forms of decentralization can (and often do) proceed in parallel, there are many cases where national governments have elected to decentralize political authority without devolving corresponding fiscal authority, and likewise, many examples of countries that have chosen to transfer fiscal resources to elected local governments without transferring meaningful political (and/or administrative) authority (Eaton and Schroeder 2010).

While the often narrow focus on devolved local government spending is understandable given the evolution of the local public finance literature (and while devolution may potentially be an appropriate proxy for decentralization if one’s aim is to study the impact of fiscal decentralization in the United States), this narrow definition excludes nondevolved (e.g., deconcentrated or delegated) local public sector finances, which likely account for a considerable share of decentralized finances in developing and transition countries.⁴

The limited attention paid by the public finance literature to nondevolved decentralization approaches was recently reflected by the Handbook of Multilevel Finance (Ahmad and Brosio 2015), which only mentioned the word “deconcentration” twice in passing, despite its otherwise comprehensive review of the latest literature on fiscal federalism. At the same time, it is increasingly recognized that granular measurement of the political, administrative, and fiscal dimensions of a decentralized, multilevel public sector is needed in order to empirically connect the levels of decentralized public sector spending observed in different countries with the potential benefits of a decentralized public sector, whether in terms of greater citizen satisfaction, faster economic growth, or improved delivery of local public services (Treisman 2002; Vo 2008; Martinez-Vazquez and Timofeev 2010). Yet, practicing the measurement of different aspects of decentralization is severely constrained by a dearth of necessary information on local governance institutions and finances, with the International Monetary Fund’s (IMF) Government Finance Statistics Yearbook being the “only global database with fiscal data for several levels of government” (Dziobek, Mangas, and Kufa 2011, 2).

Many empirical studies of fiscal decentralization admit that “[t]he most serious difficulty we face in the cross-country study of fiscal decentralization is how to properly measure the extent of decentralization” (Martinez-Vazquez and McNab 2006, 30). The lack of comparative data on local public sector finances has led virtually all empirical analyses of fiscal decentralization to rely on either the share of subnational government expenditures to general government expenditures or the share of subnational government revenues to general government revenues as measures of fiscal decentralization, based on available IMF’s Government Finance Statistics.⁵

Although the existing measures of fiscal decentralization are often acknowledged to be imperfect, the debate regarding the measurement of fiscal decentralization largely focuses on the question of whether the available measures adequately capture different degrees of discretion and accountability implied by various types of devolved local government finance (Ebel and Yilmaz 2002). In other words, rather than seeking to measure nondevolved fiscal decentralization, the existing public finance literature is mainly concerned with achieving a more detailed understanding or classification of devolved local government finances (Blöchliger 2015).

In contrast, there is little or no discussion in the existing literature about the inclusion or measurement of deconcentrated and/or delegated public expenditures as part of the proper measurement of fiscal decentralization. Torrisi et al. (2011, 13) summarize the research on nondevolved local public sector spending in a single sentence when concluding that “as for the measurement of deconcentration and delegation, in [the] literature, to the best of our knowledge, there are no attempts to measure these forms of decentralization …” Our own review of the empirical literature in search of any systematic efforts to measure nondevolved local public sector expenditures came up similarly empty-handed.

The narrow measurement of fiscal decentralization in the existing empirical literature—which essentially equates decentralization with devolution—can be especially problematic outside the United States and the relatively small set of democratic, industrialized countries that rely on elected local governments to deliver a preponderance of local public services. While the reliance on deconcentration and delegation as mechanisms for funding local public services is probably quite limited in countries that are relatively more devolved, these alternative approaches to delivering and funding localized public services may be relied on more extensively in countries where elected local governments play a smaller role. As such, the practice in decentralization and development literatures of using devolved expenditures as a proxy for all decentralized spending may mean that we are overlooking a considerable swath of public sector spending that takes place at the local level. No real evidence is available, however, about the degree to which the existing measures of fiscal decentralization fail to capture nondevolved local public expenditures.

A recent review of the considerable empirical literature on the impact of devolution on development outcomes finds that the available research is inconclusive about the impact of decentralization on a variety of development outcomes (Smoke 2013). Reflecting on the overall state of measurement of fiscal decentralization and the state of knowledge on the topic, Smoke (2013, 8) concludes with regard to the measurement of decentralization that “[t]he use of such simplistic measures for what is often the main explanatory variable in empirical analyses raises serious questions about the interpretation of statistical results and their implications for policy making.”

Theoretical Model

The total amount of devolved spending in a country is not only a simplistic measure of fiscal decentralization as argued by Smoke (2013); moreover, a critical question is whether the amount of devolved expenditures is an appropriate proxy for the total amount of decentralized expenditures. Although a proxy variable is not a direct measure of the desired quantity, a good proxy is strongly related to the unobserved variable of interest (Clinton 2004). Therefore, in order for devolved spending to be a good proxy for all decentralized spending, it must have a close correlation with the total level of local public sector spending. If it is found that devolved local expenditures do not form a good proxy for total decentralized expenditures, then we must determine what implications—if any—this may have for the existing empirical literature.

In line with this observation, we develop a rudimentary theoretical model to consider what happens when fiscal decentralization is measured using a proxy that is likely to capture only one particular type of fiscal decentralization (i.e., devolution). Although the theoretical model that we develop to answer this question is basic, its implications for the decentralization literature are substantial. Let us consider the following linear model:

y = X β + Z δ + ∊ .

In this model, the independent variable y _i is a measure of the (potential) benefits of fiscal decentralization in country i, where y is the vector of independent variables, y ₁, …, y _n .⁶ Next, X is a n × p data matrix containing values for p independent variables (which includes, among others, a variable that measures the degree to which expenditures are devolved in each of the n countries), while β is a p × 1 vector of the underlying (unobservable) parameters to be estimated.

In turn, Z is a n × q data matrix containing q measures of the degree of nondevolved decentralization in the n countries (q ≥ 1), where δ is the vector of the underlying (unobservable) parameters (the response coefficients of the dependent variable to the measures of nondevolved decentralization contained in Z) to be estimated. It should be noted that in accordance with our literature review, none of the variables contained in matrix Z has actually ever been measured in a systematic manner. Finally, ∊ is the vector of unobservable error terms, having an expected value of 0 and subject to the usual assumptions regarding the error term (Greene 1993, 171–72).

In accordance with Greene (1993, 245–46), if Z is omitted from the regression, then the estimated values of the response parameters of the other independent variables will be given as follows:

E [\hat{β} | X] = β + (X^{'} X) X^{'} Z δ

The second term after the equal sign in equation (2) reflects potential omitted-variable bias, which is nonzero if one or more of the omitted variables in Z are correlated with any of the variables included in the matrix X. In our case, this suggests that if the observed devolved local government spending is correlated (positively or negatively) with the degree of unmeasured or omitted nondevolved local public sector spending, then the empirical results of the previous literature would be biased and inconsistent.

The extent and direction of the possible bias depend on the estimators as well as the covariance between the regressors and the omitted variables. In the case of our theoretical model, the direction of the potential bias would depend on whether highly (or weakly) devolved countries proportionately tend to rely more or less on other forms of decentralized spending. Let us assume for a moment that devolved spending produces greater benefits than other, nondevolved (deconcentrated/delegated/direct) types of decentralized spending. Under this assumption, if low-devolution countries rely more heavily on nondevolved localized spending, then the omission of nondevolved local expenditures in the regression model would result in an underestimate of the impact of devolution on the independent variable.⁷

This theoretical discussion suggests that the measurement of fiscal decentralization should be driven neither by ideological arguments over the true nature of decentralization nor by the availability of data. After all, if we find a negative correlation between the share of devolved expenditures and the share of nondevolved local expenditures, then the misspecification of previous empirical models is likely to have contributed—to a greater or lesser extent—to the inconclusive nature of the existing stock of empirical research on the beneficial impact of fiscal decentralization.

Empirical Strategy and Data Collection Efforts

Until recently, no comprehensive, consistent metrics were available in the decentralization literature to measure the full extent of decentralized expenditures in different countries, as no standardized measures of nondevolved local public spending were available. In order to fill this knowledge gap, Boex (2012) developed a set of comparative metrics that define a country’s local public sector finances as the total of the country’s devolved local government finances plus the country’s nondevolved local public sector finances.

Table 1 presents the five basic types of public sector spending that are considered in constructing a “vertical” expenditure profile (an analysis of the distribution of all public sector expenditures across levels and tiers of local government and local administration) or a local public sector expenditure profile. The four types of local public sector expenditures considered as part of a vertical expenditure profile include: (1) devolved local government spending; (2) deconcentrated local administrative expenditures; (3) delegated local public expenditures; and (4) direct central government support for the delivery of localized public services.⁸ Categories 2 through 4 are jointly referred to here as nondevolved local public sector expenditures. Finally, in this classification of public expenditures, (5) central line ministry expenditures that do not contribute to the delivery of public services in a direct and localized manner are considered “truly central” (nonlocal) government expenditures.

Table 1.

Types of Public Sector Expenditures.

Public spending	Definition	Example
1. Devolved local government spending	Spending contained in the budget of elected local government entities	Spending on a local government’s education department, including local education administration, teacher salaries, and other local education spending contained in a local government’s budget
2. Deconcentrated local spending	Spending contained in the central government budget for a deconcentrated (e.g., regional or local) line department	Spending on provincial health services (in country where the province is a deconcentrated level and where the provincial health department has its own deconcentrated budget)
3. Delegated local spending	Financial resources provided by central line ministries to delegated entities/local service-delivery providers	Transfers to regional authorities, parastatal organizations (e.g., public hospitals), or nongovernmental organization/faith-based service providers
4. Central spending on local services	Central line ministry spending that is directly visible to the end user of a public service, for instance, through a “vertical program” or an “autonomous project unit”	Central line ministry spending on front line staff salaries; local service delivery infrastructure; or (medical/school) supplies procured and provided by line ministry
5. True central government spending	Central line ministry spending that does not directly support localized public services	Spending on central line ministry activities (such as policy development, central administration, inspection and oversight, etc.)

Note: Boex (2012) makes a distinction between budgetary deconcentration (counted as part of deconcentrated local spending, which requires the deconcentrated entity to be a secondary budget unit in the Chart of Accounts) and administrative (nonbudgetary) deconcentration, which is considered part of central spending on local services (e.g., spending on field administration staff contained in the central government budget).

According to the more recent second generation fiscal federalism literature, it is likely that political economy factors—rather than technical factors—play a critical role in determining the level and composition of local public sector expenditures (Weingast 2013; Ahmad and Brosio 2015). As such, it is important to recognize that different types of local public sector expenditures are not likely to be perfect substitutes in the production of localized services: depending on how these funding streams are governed, different types of local public sector spending—and even different funding flows within the same type of public sector spending—could have considerably different impacts on local service delivery outcomes. Furthermore, given the prevailing importance of political forces, there is no reason to believe that the vertical composition of public sector expenditures in a country necessarily reflects a long-term equilibrium or an economically optimal or efficient allocation of public resources.

If it were relatively easy to collect data on these different types of devolved and nondevolved local public expenditures in developing and transition economies, we might have fully replicated some of the earlier multivariate empirical studies in the literature using data for numerous years and a large number of developing and transition countries. In reality, however, collecting such data not only requires a solid grasp of the methodology but also necessitates an in-depth understanding of a country’s central and local budget processes and public financial management systems. This means that collecting accurate measures of total—devolved and nondevolved—local public sector expenditures for different countries is a tedious, time-consuming, and costly process.

In order to develop rigorous evidence on the impact of different kinds of decentralized and localized public spending on development outcomes, a number of international development agencies—united under the umbrella of the Development Partner Working Group on Decentralization and Local Governance (DeLoG)—supported the collection of vertical expenditure profiles for the health and education sectors in developing and transition countries around the world. The initial ambition was to collect data for sixty countries and for two years (2000 and 2010). The health and education sectors were chosen for this analysis because they are critical to the attainment of the global millennium development goals (MDGs; covering four of eight MDGs) and because in many countries, these two sectors represent a major share of all local public sector expenditures.⁹ Due to limitations on data availability and the expense of collecting the necessary data, however, data were ultimately collected for only twenty-nine countries and for only a single year (FY 2010) for each country.

Since this was the first attempt to consistently measure devolved as well as nondevolved local public expenditures across a wide range of countries, it is appropriate to acknowledge potential concerns regarding data quality. While the measurement of devolved local government expenditures on health and education typically requires the aggregation and analysis of sectoral expenditures across all local governments in a country, the measurement of nondevolved local public expenditures required a careful and detailed analysis of central line ministry expenditures in each of the countries covered by the study. As such, the measurement of nondevolved local public expenditures in some countries was constrained by the availability of central government budget data in a way that allowed central sectoral spending to be disaggregated into spending by deconcentrated line departments; spending by central-level line directorates that directly support the localized delivery of public services; and “truly central” public spending.¹⁰

Empirical Results: The Size and Composition of the Local Public Sector in Health and Education

Table 2 presents the descriptive statistics for the relative size of decentralized spending in the health and education sectors, respectively, in the twenty-nine developing and transition countries for which data were collected by the DeLoG partners. For each sector, the table shows the share of sectoral spending executed by devolved local governments as well as the share of nondevolved sectoral spending at the local level (i.e., the sum of deconcentrated local spending, delegated spending, and direct central funding of localized services). The table further presents the total (devolved plus nondevolved) amount of decentralized local public sector spending in health and education as well as the share of sectoral expenditures devoted to “truly central” activities. For completeness, this table also provides descriptive statistics for the most important development outcome measures in the health and education sectors, namely the under-five mortality rate (per 1,000 births) and the literacy rate of persons aged fifteen to twenty-four.

Table 2.

Descriptive Statistics for Decentralized Spending in Health and Education (as Percentage of Sectoral Spending).

Variable	Average	Standard deviation	Coefficient of variation	Minimum	Maximum
Health sector
Devolved health expenditures	26.3	29.9	1.138	0.0	98.5
Nondevolved health expenditures	52.7	29.6	0.563	0.6	98.1
Total decentralized health expenditures	79.0	16.2	0.206	25.8	99.1
Central (nonlocal) health expenditure	21.0	16.2	0.772	0.9	74.2
U5MR (MDG 4.1)	72.6	46.3	0.638	2.3	188.8
Education sector
Devolved education expenditures	30.7	31.2	1.018	0.0	90.6
Nondevolved education expenditures	54.0	30.8	0.571	1.9	96.1
Total decentralized education expenditures	84.7	12.6	0.149	52.5	99.2
Central (nonlocal) education expenditure	15.3	12.6	0.825	0.8	47.5
Literacy rate fifteen to twenty-four (MDG 2.3)	79.7	17.9	0.224	39.3	99.5

Note: “Nondevolved health expenditures” and “nondevolved education expenditures” are computed as the sum of deconcentrated expenditures, delegated expenditures, and direct central expenditures on local services. MDG = millennium development goal.

Source: Computed by authors based on data provided by the Local Public Sector Initiative.

To complement the descriptive statistics contained in table 2, figures 1 and 2 show the vertical expenditure patterns in the health and education sectors for each of the twenty-nine developing and transition countries contained in our sample. The share of devolved expenditures in each country is indicated by the dark segment of each bar, while nondevolved local public sector expenditures are reflected by the incrementally lighter segments. It should again be noted that prior to the current study, the level of nondevolved local public sector spending had simply not been measured in any of these countries, and as such, had essentially gone unobserved.

Figure 1.

Health sector expenditure patterns in selected countries, 2010.

Figure 2.

Education sector expenditure patterns in selected countries, 2010.

The table and graphs confirm that local governments often play an important role for service delivery in these sectors: in health and education, respectively, an average of 26.3 percent and 30.7 percent of total public expenditures are channeled to elected local and regional governments for the delivery of local services. This type of spending conforms to the traditional measurement of “decentralized” public finance in the empirical literature. When all types of local public sector expenditures are considered under our expanded approach, however, countries are seen to spend a much greater share of sectoral resources—an average of 79.0 percent and 84.7 percent—at the local level for health and education services.

What does this basic finding mean for the previous measurement of fiscal decentralization, which relied exclusively on devolved expenditures as a proxy for fiscal decentralization? When all—devolved and nondevolved—local public sector expenditures are taken into account, regional and local government expenditures account for only about one-third of local public sector expenditures in health and education.¹¹ In other words, to the extent that the previous empirical literature equated decentralized expenditures with devolved local government expenditures in developing and transition countries, the previous empirical literature on fiscal decentralization overlooked two-thirds of decentralized expenditures.

The direct implications of this finding are substantial. For the community of policy practice on DeLoG, it is clear that we have failed to “measure what we treasure.” To the extent that our interest is in improving local development outcomes, those working within the “public sector finance stovepipe” or within the “local governance pillar” should work much more closely with the respective sectoral line ministries to understanding the vertical aspects of public finance and governance within each sector. For the research community focused on decentralization and fiscal federalism, it should be self-evident that devolved expenditures can in no way be used as a reasonable proxy for decentralized expenditures in cross-country studies of decentralization if in so doing we disregard two-thirds of decentralized public expenditures.¹² Going forward, our primary finding suggests that there is a major need for researchers, policy practitioners, and development organizations to focus much greater attention and resources on measuring and analyzing deconcentration, delegation, and other approaches to localization in a multilevel governance context.

In addition to the direct implications of our primary finding, there is an important indirect implication of this finding on the state of knowledge with regard to decentralization (or devolution, as the case may be). As noted above, if countries that rely on devolution weakly or not at all instead rely more heavily on nondevolved localized spending, then the omission of nondevolved local expenditures in the empirical literature will have resulted in omitted-variable bias that—most likely—has led to an underestimate of the beneficial impact of devolution on the independent variable.

Although estimating the impact of different types of localized sectoral spending on development outcomes in health and education falls beyond the scope of the current study, our new ability to measure the previously unmeasured levels of nondevolved local public sector spending allows us to gain important insights into the extent to which omitted-variable bias is a likely to have resulted in misspecification. Table 3 indicates that there is a powerful negative correlation between the relative share of devolved expenditures and the share of nondevolved local public expenditures. The relevant correlation coefficients (−0.85 and −0.92, for health and education, respectively) suggest that devolved and nondevolved expenditures have been treated by central government policy makers in developing and transition countries as near-perfect substitutes: a country that spends a smaller share of its total sectoral budget in health or education through local governments (whether for technical or political economy reasons) tends to offset this through greater deconcentrated, delegated, or localized central spending in the same sector.¹³ Given the high degree of correlation, substantial concern is raised that omitted-variable bias has played a considerable role in influencing the inconclusive findings of the existing decentralization literature.

Table 3.

Correlation Matrix: Decentralized Education and Health Spending (as Percentage of Sector Spending).

Health	Devolved	Nondevolved	Total decentralized	U5MR
Devolved health expenditures	1.0000
Nondevolved health expenditures	−0.8512	1.0000
Total decentralized health expenditures	0.2854	0.2601	1.0000
U5MR	−0.2749	0.0850	−0.3505	1.0000
Education	Devolved	Nondevolved	Total decentralized	Literacy
Devolved education expenditures	1.0000
Nondevolved education expenditures	−0.9177	1.0000
Total decentralized educ. expenditures	0.2344	0.1711	1.0000
Literacy rate (age fifteen to twenty-four)	0.4478	−0.2780	0.4302	1.0000

Concluding Remarks

In the course of studying the impact of fiscal decentralization on the potential benefits of decentralization (such as greater citizen satisfaction, faster economic growth, or improved delivery of local public services), the existing empirical literature has taken an important shortcut: whether due to lack of available data or due to historical biases, researchers have almost universally used the level of devolved (local government) expenditures in a country as a proxy for all decentralized expenditures. While using local government finances may have been a reasonable proxy for decentralized expenditures when studying decentralization in the United States or in other highly devolved countries (where there is likely to be little deconcentration or delegated expenditures), our analysis suggests that this proxy cannot be defended as a reasonable approach in measuring fiscal decentralization in developing and transition economies. Our results show that two-thirds of local public sector spending in our sample is done through nondevolved mechanisms that were not measured by the previous literature.

This study’s main conclusion is obvious to the extent that the global development community and the associated research community are interested in enhancing the efficiency of public sector spending by understanding the impact of fiscal decentralization on public sector performance and development, considerably greater time and effort should be focused on measuring and analyzing the impact of nondevolved local public expenditures. Rather than proxying decentralization based on a single measure of devolved local expenditures, it would be much more appropriate for the research community to pursue the measurement and analysis of decentralized expenditures as a vector of different types of decentralized or localized expenditures. The policy corollary to this conclusion is equally plain: if a central government (or line ministry) cannot identify what share of its sectoral resources is going down to the local level for the delivery of front line public services through various different mechanisms, then it is unlikely that the government’s spending patterns are driven by well-informed policy decisions about how to best improve public service delivery outcomes at the local level.

Of course, measuring the level and composition of decentralized funding flows in a more comprehensive manner is only a starting point in achieving a better understanding of the role that local-level expenditures play in achieving improved service delivery outcomes. Future research needs to be pursued in parallel in two different directions. First, based on our more appropriate and nuanced measurement of decentralized public sector expenditures, what can we learn—if anything—about the impact of local public sector spending on service delivery outcomes? In other words, does the level of local public sector expenditures have an impact on development results (in education and health), and if so, does this impact vary by different types of decentralized expenditure? It is likely that further data will need to be gathered on devolved and nondevolved local public sector expenditures in developing and transition countries before these relationships can be explored in an empirically prudent manner.

Second, promoting a more nuanced understanding of fiscal decentralization—by explicitly recognizing not only devolution but also deconcentration and other nondevolved localized expenditures under the umbrella of fiscal decentralization—requires us to more explicitly recognize the interaction between the fiscal decentralization and the other dimensions of decentralization (i.e., political and administrative). For instance, beyond the vertical expenditure patterns themselves, the study’s data collection process further revealed that most developing and transition countries included in our sample impose considerable conditionalities on nominally devolved expenditures. In the absence of meaningful political or administrative authority, autonomy, and discretion, we may find that the impact of such notionally devolved expenditures differs little from deconcentrated expenditures. Going forward, therefore, conceptualizing, measuring, and testing the impact of different levels of political and administrative power and authority over localized expenditures may further illuminate the connection between fiscal decentralization and service delivery outcomes.

Finally, our main conclusion and recommendation—that there is a need to expand the focus of the empirical decentralization literature to include both devolved and nondevolved local public sector expenditures—should not be misunderstood to mean that devolution is any less important than other forms of decentralization in achieving effective public sector performance or positive development outcomes. If we are interested in accurately measuring the impact of devolved expenditures on development outcomes, then we need to account more carefully for nondevolved local public sector spending level to prevent omitted-variable bias. In fact, it is quite possible that doing a better job measuring and accounting for the impact nondevolved decentralized expenditures could reveal stronger empirical evidence in support of the beneficial impact of devolved local government spending.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article draws on vertical expenditure profiles for the health and education sectors in twenty-nine developing and transition countries, which were collected by the Urban Institute in collaboration with the Development Partner Working Group on Decentralization and Local Governance (DeLoG). The preparation and collection of these vertical expenditure profiles were supported and funded by the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), along with the Swiss Agency for Development and Cooperation; the French Ministry for Foreign Affairs; the UN Development Program(UNDP); the UN Capital Development Fund (UNCDF); and the World Bank. The support of these organizations is gratefully acknowledged.

Notes

References

Abdelhak

Faigy

Chung

Jihyun

Jingqiang

Stevens

Valerie

. 2012. “Measuring Decentralization and the Local Public Sector: A Survey of Current Methodologies.” IDG Working Paper No. 2012-01, The Urban Institute, Washington, DC.

Ahmad

Ehtisham

Brosio

Giorgio

, eds. 2015. Handbook of Multilevel Finance. Northampton, MA: Edward Elgar Publishing.

Bergstrom

Theodore C.

Goodman

Robert P.

. 1973. “Private Demands for Public Goods.” American Economic Review 63:280–96.

Blöchliger

Hansjörg

. 2015. “The Challenge of Measuring Fiscal Decentralization.” In Handbook of Multilevel Finance, edited by Ahmad

Ehtisham

Brosio

Giorgio

, Chapter 24, 617–32. Northampton, MA: Edward Elgar Publishing.

Boex

Jamie

. 2012. Measuring the Local Public Sector: A Conceptual and Methodological Framework. Washington, DC: The Urban Institute.

Borcherding

Thomas E.

Deacon

Robert T.

. 1972. “The Demand for the Services of Non-federal Governments.” The American Economic Review 62:891–901.

Clinton

Joshua D.

2004. “Proxy Variable.” In The SAGE Encyclopedia of Social Science Research Methods, edited by Lewis-Beck

Michael S.

Bryman

Alan

Liao

Tim Futing

, 879. Thousand Oaks, CA: Sage.

Dafflon

Bernard

. 2015. “The Assignment of Functions to Decentralized Government: From Theory to Practice.” In Handbook of Multilevel Finance, edited by Ahmad

Ehtisham

Brosio

Giorgio

, Chapter 8, 163–199. Northampton, MA: Edward Elgar Publishing.

Dziobek

Claudia

Gutierrez Mangas

Kufa

. 2011. “Measuring Fiscal Decentralization—Exploring the IMF’s Databases.” IMF Working Paper WP/11/126, The International Monetary Fund, Washington, DC.

10.

Eaton

Kent

Schroeder

Larry

. 2010. “Measuring Decentralization.” In Making Decentralization Work: Democracy, Development, and Security, edited by Eaton

Kent

Connerley

Edwin

Smoke

Paul

, 167–90. Boulder, CO: Lynne Rienner Publishers.

11.

Ebel

Robert D.

Yilmaz

Serdar

. 2002. “On the Measurement and Impact of Fiscal Decentralization.” Policy Research Working Paper 2809, The World Bank, Washington, DC.

12.

Fisher

Glenn W.

1964. “Interstate Variation in State and Local Government Expenditure.” National Tax Journal 17:57–74.

13.

Greene

William H.

1993. Econometric Analysis. 2nd ed. New York: Macmillan.

14.

Litvack

Jennie

Seddon

Jessica

. 1999. “Decentralization Briefing Notes.” World Bank Institute Working Paper 19683, The World Bank, Washington, DC.

15.

Martinez-Vazquez

Jorge

McNab

Robert M.

. 2006. “Fiscal Decentralization, Macrostability, and Growth.” Hacienda Pública Española/Revista de Economia Publica 179:25–49.

16.

Martinez-Vazquez

Jorge

Timofeev

Andrey

. 2010. “Decentralization Measures Revisited.” Public Finance and Management 10:13–47.

17.

Oates

Wallace E.

1972. Fiscal Federalism. New York: Harcourt Brace Jovanovich.

18.

Rondinelli

Dennis A.

1986. Briefing Notes: Overview of Decentralization in Development Countries. Washington, DC: Academy for Education Development.

19.

Smoke

Paul

(with Loffler

Gundula

Bosi

Giuliano

). 2013. The Role of Decentralisation/Devolution in Improving Development Outcomes at the Local Level: Review of the Literature and Selected Cases. New York: Local Development International.

20.

Tiebout

Charles

. 1956. “A Pure Theory of Local Expenditures.” The Journal of Political Economy 64:416–24.

21.

Torrisi

Gianpiero

Pike

Andy

Tomaney

John

Tselios

Vassilis

. 2011. “Defining and Measuring Decentralisation: A Critical Review.” Munich Personal RePEc Archive (MPRA) Paper No. 51441, University Library of Munich, Germany.

22.

Treisman

Daniel

. 2002. Defining and Measuring Decentralization: A Global Perspective. Los Angeles: Department of Political Science, University of California.

23.

Duc Hong

. 2008. “The Economics of Measuring Fiscal Decentralisation.” Dissertation, The University of Western Australia, Crawley, WA.

24.

Weingast

Barry R.

2013. “Second Generation Fiscal Federalism: Political Aspects of Decentralization and Economic Development.” World Development. https://dx-doi-org.web.bisu.edu.cn/10.1016/j.worlddev.2013.01.003.

25.

Xie

Danyang

Zou

Heng-fu

Davoodi

Hamid

. 1999. “Fiscal Decentralization and Economic Growth in the United States.” Journal of Urban Economics 45:228–39. Accessed May 19, 2015.