Fiscal Federalism 2022

Making Decentralisation Work

Report

20 December 2021

4. Spending autonomy and public sector performance across levels of government

Abstract

Differences in spending autonomy across sectors and countries may have important consequences for intergovernmental fiscal relations. Yet the share of subnational expenditure often does not reflect the true degree of decision-making authority. This chapter first presents new measures of subnational spending autonomy across key sectors of local government service delivery, which allows the analysis of alternative arrangements on outcomes. The chapter then examines how the performance of the public sector can be strengthened through the assignment of responsibilities at the most effective level of government for their delivery, focusing on health care. Performance monitoring and evaluation systems are also illustrated that enable benchmarking of public services to promote learning about good policy practices.

Introduction

The Coronavirus pandemic has brought renewed attention to the performance of government services, both in relation to the crisis itself and the enduring fiscal and other policy challenges that will be its legacy. Public deficits and debt have risen substantially, so that budgetary resources are even more constrained. Government resources will need to be harnessed as cost-effectively as possible if consequent societal needs, in healthcare, welfare services, education and other community services are to be adequately met in coming years. The ability to reduce debt and pay services will depend on income growth and reversing the downward trend in productivity that has occurred in many countries since the global financial crisis. This will be influenced by government policies and by the productivity of the public sector itself.

The decentralisation of public service provision is a central topic in fiscal federalism. Through the well-designed devolution of public services, it may be possible to realise efficiencies, better adapt services to local preferences and improve outcomes (OECD/KIPF, 2016[1]). Typically, decentralisation implies that subnational governments assume responsibility for all or part of the provision of a particular service. This necessarily raises questions related to whether the level of government financing the various services also possesses decision-making authority over them. As such, metrics that aim to capture the true extent of subnational spending power must consider both the extent of subnational spending on services, as well as the level of subnational control over the provision of said services.

There is increased recognition that the public sector – whether at central, regional or subnational levels – has the potential to do better in many areas, whether service-related or regulatory in nature. The degree of subnational spending power is generally depicted as the subnational expenditure share as a proportion of total government expenditure. However, because of barriers and restrictions on subnational decision making, including earmarked grants, mandatory spending and national standards, simple expenditure shares can misrepresent the true level of subnational decision-making autonomy. This makes accurately comparing and measuring decentralisation across countries difficult, far beyond the purely statistical challenges that cross-country comparisons already face. To address these challenges, the OECD’s Network on Fiscal Relations has developed a set of survey-based spending power indicators which seek to reflect more accurately the extent of subnational control over spending within individual sectors, such as health care, education, housing, long-term care and transportation (Beazley et al., 2019[2]; Dougherty and Phillips, 2019[3]; James et al., 2019[4]; Kantorowicz and van Grieken, 2019[5]; Phillips, 2020[6]). The Network and its partners have also developed empirical insights from cross-country comparisons regarding the role and contribution of different levels of government and the determinants of performance (Dougherty et al., 2019[7]; Lastra-Anadón and Mukherjee, 2019[8]; Jong et al., 2021[9]; Dougherty, Renda and von Trapp, 2021[10]; Dougherty et al., 2021[11]). This chapter synthesises this work, along with related studies, and draws implications for policy making as well as for further research. Main findings include:

Measurement of public sector performance is fundamental to improving outcomes, yet it is also extremely challenging to do well. A combination of approaches will generally be called for, including improved data and methods, appropriate assignment of responsibilities across levels of government, effective benchmarking and monitoring systems, as well as institutions that are custom fit for purpose, including for in-depth diagnosis and evaluation of policy needs.
Aligning spending and financing responsibilities for public services at the most effective level for delivery is a core central government imperative. Differences in the nature of public services such as education and health have an important relation with how they can be best organised across government levels. Depending on the target sector and population, economies of scale may also vary across countries and time, with periodic adjustments needed to improve performance.
Spending power is quite balanced across the sectors explored, but is less decentralised in health and long‑term care, and more decentralised in social housing, transport and primary and secondary education (referred to as “education”) services. That is, regional and local governments, as well as service providers, tend to have more power to make decisions about how social housing, transport and education services are delivered, compared to health and long-term care.
The overall policy framework for service sectors is often centrally controlled, but varies in certain key aspects across sectors – policy autonomy in the long-term care and health sectors is the most centralised, while policy decisions are least centralised in the social housing sector. In contrast, input and budgeting decisions are more likely to be decentralised.
Deploying public sector performance monitoring and evaluation systems can enable not only the benchmarking of public expenditures across subnational regions and municipalities, but also promote learning about good policy practices and engage public stakeholders in a constructive feedback process.

The chapter first discusses the development of indicators of spending autonomy across key sectors, with a special focus on health, including a decomposition of the main components. Second, composite indicators of spending autonomy are presented. Third, public sector performance measurement is discussed along with empirical findings on how the decentralisation process can be used to improve outcomes in some sectors. Finally, performance benchmarking and monitoring systems are discussed and considered in an intergovernmental relations context.

Expenditure shares, spending power and fiscal autonomy

Measuring subnational autonomy through expenditure shares

In measuring the fiscal dimension of subnational autonomy, scholars have generally relied upon indicators such as expenditure and revenue shares to determine the level of decentralisation. Although these indicators can give a first impression of the extent to which subnational governments are decentralised or fiscally autonomous (particularly on the revenue side), these indices fall short in adequately capturing the complexity and multifaceted nature of fiscal arrangements and, by extension, the actual degree of autonomy subnational governments enjoy (Blöchliger and King, 2007[12]). This issue becomes especially apparent in the case of subnational spending autonomy.

The breakdown of subnational expenditure and subnational expenditure as a share of government expenditure are common metrics to help clarify the spending power of subnational governments. While these indicators do not capture the complexity of fiscal arrangements, they can provide a useful guide of how much fiscal power regional and local jurisdictions enjoy. These metrics are constructed using data from the OECD National Accounts Statistics, which uses the Classification of Functions of Government (COFOG). Now considered as a worldwide standard, COFOG data classifies government expenditure by the purpose for which the funds are to be used.

First-level COFOG splits expenditure data into ten functions (general public services; defence; public order and safety; housing and community amenities; economic affairs; environment protection; health; recreation, culture and religion; education; social protection). COFOG II further divides the ten expenditure functions into 69 sub‑functions. The second level of COFOG is particularly important for public finance analysis, as it allows for the breakdown of social protection into different programme areas. Although the sectors evaluated in this survey more closely align with the COFOG II classification, these COFOG data are not available in many countries, reducing the ability to accurately compare the new spending power indicators in this chapter with expenditure shares.

The following figures show the current state of spending decentralisation as measured by the composition of subnational government expenditure in OECD countries (Figure 4.1) and subnational government expenditure as a proportion of total government expenditure. Education represents the largest sector in the subnational government expenditure, averaging 24% of all subnational government expenditure across OECD countries in 2019. However, values for individual member states vary considerably. In Estonia, Iceland, Israel, Latvia, the Slovak Republic, Slovenia and the United States, spending on education exceeded 30% of subnational budgets, and in Latvia, it is 41%. Health accounts for the second highest budgetary outlay, accounting, on average, for 18% of subnational government expenditure. It exceeded 20% of subnational budgets in Australia, Austria, Denmark, Finland, Spain, Sweden and the United States, and reached 48% in Italy. The third largest subnational budget item is ‘other’ expenditure representing 15% of subnational government expenditure, which includes defence; public order and safety; housing and community amenities; recreation, culture and religion; and environment expenditure. Public order, safety and defence expenditures accounted for around 7% of subnational government expenditure. Housing and community amenities represented on average around 3% of subnational government expenditure across the OECD. General public services and social protection (which includes current and capital social expenditure) both represent around 14% of subnational government expenditure.

Figure 4.1. Composition of subnational government expenditure across economic functions

OECD weighted average across countries, 2019

While the subnational share of government expenditure (Figure 4.2) is often used to describe the degree of spending power of subnational governments, it does not necessarily indicate whether subnational governments can autonomously decide on the allocation and implementation of the financial resources that are at their disposal. An alternative approach is to consider only certain types of funding arrangements, in particular own-source revenue, as they are assumed to result in more freedom for subnational governments in the way public funds are allocated.

Figure 4.2. Subnational government expenditure as a share of total government expenditure, 2019

However, the true spending autonomy of subnational governments can only be inferred from taking into account existing political-institutional factors, such as national regulations. More specifically, ‘gauging spending power entails detailed assessments of each policy area’s regulatory environment and intergovernmental fiscal frameworks’ (OECD/KIPF, 2016[1]). For instance, one tool of central governments to tie the hands of subnational governments in terms of spending authority is through intergovernmental transfers. Grants from the central level of government to subnational governments can either be non-earmarked (unconditional) or earmarked (conditional). If the latter is the case, resources that are transferred to lower levels of government are controlled by the central government as the transferred resources are intended to be spent on pre-determined programmes (Blöchliger and King, 2007[12]). Such imposed upper-tier regulations thus reduce the actual discretion of subnational governments over expenditure items. As a consequence, large subnational expenditure ratios do not necessarily equate with true spending autonomy (Dougherty and Phillips, 2019[3]).

Developing spending autonomy indicators

Faced with these measurement issues, scholars have developed more refined indicators of the spending power of subnational governments to adequately capture to what extent sub-entities have authority over expenditure items. Developing indices of spending autonomy is, however, not as straightforward as it may seem as the spending side of the budget comprises a wide range of policy areas. In addition, whereas ‘autonomy’ in terms of taxation boils down to the ability of subnational governments to set tax rates, tax bases, or both, spending autonomy is more multifaceted. In essence, the spending autonomy of subnational governments depends on the extent to which subnational entities exert influence over rules and regulations in different policy areas and whether they are free from constraints imposed by upper-level governments.

To overcome this apparent complexity, Bach, Blöchliger and Wallau (2009[13]) split up spending autonomy into different categories of rules and regulations (Figure 4.3):

Policy autonomy: The extent to which subnational decision makers exert control over main policy objectives and main aspects of service delivery.
Budget autonomy: The extent to which subnational decision makers exert control over the budget (e.g. is budget autonomy limited by upper level regulation).
Input autonomy: The extent to which subnational decision makers exert control over the civil service (personnel management, salaries) and other input-side aspects (e.g. right to tender or contract out services).
Output and monitoring autonomy: The extent to which subnational decision makers exert control over standards such as quality and quantity of services delivered and devices to monitor and evaluate standards, such as benchmarking.

The preceding categories form the basis of the OECD’s spending autonomy indicators. Survey questions are developed within each category to assess each aspect of spending autonomy. The responses received are then combined to compute the indicators.

Figure 4.3. Classifications of spending autonomy

Defining the indicators

Levels of government

A main characteristic of a decentralised government is the existence of several governing bodies, which have political, administrative or funding power at a subnational level. Three levels of government are defined – central/federal, state/province/region, and local/municipality. In this chapter, subnational governments are defined as sub-central levels of government. Regional governments are upper-tier municipalities including states, territories or provinces. Local governments are the lowest tier of government including counties, cities, districts, municipalities, councils or shires. In the context of countries with only two levels of government, the lower level is defined as local government.

This chapter focuses on spending power in five key areas of the public sector – heath care, education, housing, long-term care and transport services.

Spending power of subnational decision makers

Spending power should be interpreted beyond budgeting decisions. It describes the level of control or authority of subnational decision makers, including deciding how services are organised, how funds are allocated, the preferred level and quality of inputs and outputs and how service delivery is measured and monitored (OECD/KIPF, 2016[1]). The spending power indicators presented in this chapter aim to provide an accurate representation of spending power, and encompass features such as subnational governments’ right to introduce new government programmes, to amend regulations, to grant subsidies and concessions, to abolish spending programmes, to decide on the ratio of recurrent to capital spending, and to allocate funding across priority areas.

Questionnaire scope and methodology

Country questionnaire responses

Indicator values for data on education, long-term care, transport services and social housing are drawn from responses to a recent OECD survey on the spending power of subnational governments. This survey was sent to countries in early 2018, with seventeen OECD countries and four partner countries responding to the survey in full. A further four OECD countries and one partner country provided partial responses to the survey. The indicator values subsequently computed as described in Dougherty and Phillips (2019[3]) and are available on the OECD Fiscal Decentralisation database.1

Indicator values for data on health care are drawn from a separate survey on performance measurement systems in the health sector and responsibilities across levels of government. This survey was sent to countries in late 2017 with responses received from 28 OECD countries and three partner countries. Differences in the design of the two questionnaires results in differences between the health data and the data collected for the other four sectors. The health survey was not specifically formulated to construct indicators, and as such, the checkbox questions on responsibilities between levels of government allowed for the following possible responses: central government; regional government; local government or ‘other’. Given the varying nature of the ‘other’ responses, this has not been included in the spending power indicator. This means there is effectively a ceiling on the indicator values for health – health cannot have an indicator value higher than seven whereas the indicator limit for all the other sectors is ten. This reduces comparability across the sectors, and sectors cannot be directly compared within countries, when looking at health and another sector. For further details see Beazley et al (2019[2]).

Indicator values for housing were taken from Phillips (2020[6]), who computed them based on responses to the OECD’s 2019 Questionnaire on Affordable and Social Housing (the QuASH). Thirty-five OECD countries and seven partner countries responded to the survey in full.

Computing spending autonomy indicators

The indicator set can be differentiated and shown as an “indicator tree” with low-level indicators (LLI), medium-level indicators (MLI) and the high-level summary indicator (HLI) (Figure 4.4). HLI’s are constructed for each of the five policy sectors.

Figure 4.4. Indicator tree for spending autonomy

In order to construct a composite indicator from individual survey questions, the country responses to each question are transformed into LLIs using the values shown in Table 4.1, which describe one specific aspect of decision making in each sector. Indicator values are scaled between 0 and 10, with a higher value associated with greater decentralisation. While indicator values are scaled between 0 and 10, the ordinal ratings are arbitrary (Table 4.1). If answers to the questionnaire indicated shared responsibilities, which was often the case, the arithmetic mean of the indicator values for the decision-making levels involved was used.

Level of decision maker	Indicator value
Central	0
Regional	3
Local	7
Provider	10
Not applicable response	Not included
Social security fund (used for aged care indicators only)	0
Other (included in health survey only)	Not included

Service providers are considered a separate decision-making level receiving the highest indicator value. This implies that a move of spending responsibilities from a state or local government to providers increases subnational autonomy. While allocating spending power to providers tends to bring services closer to citizens, it does not necessarily increase subnational government power over a specific service. This should be taken into account when interpreting the results.

An alternative approach to scoring the data can be taken, which equally distributes scores between zero (central) and one (local) across levels of government, according to whether a level of government has autonomy on a given competency. This more uniform approach allows for balanced comparisons across different sectors and between federal and unitary countries. It is described in detail by Kantorowicz and van Grieken (2019[5]) and is also available in the Fiscal Decentralisation database. This approach is also used for enhanced comparability when developing a composite indicator of spending autonomy (Box 4.1).

The results by sector and by policy dimension are shown in Figure 4.5, Panels A and B.

Figure 4.5. Spending autonomy indicators by sector and by policy dimension

Spending power in health

Overall results

There is an imbalance between the powers of different decision makers in health, with 62% of decisions in the sector controlled by the central government. This high centralisation could be due to a range of factors. There are obvious efficiencies that come from having certain aspects of health centralised, for example, data sharing requirements. Secondly, many of the positive social and economic benefits from having a high-quality health care system, maintaining minimum national standards and adequately funding preventative health flow to the country as a whole, rather than being confined to sub-jurisdictional borders. Further, due to the cost and complexity of many health systems and hospital procedures, a lack of economies of scale can discourage subnational involvement. Central and regional power is relatively diversified across the four classifications of autonomy, but, similar to other service sectors, the power of local governments and providers is more concentrated on aspects of decision making that involve inputs.

Figure 4.6 shows the allocation of responsibility for decisions in health care, across respondents. It is calculated as the number of times a country responded that a level of government was responsible for a health decision, and then shows these sub-totals as a proportion of the total ‘yes’ responses, for each country. Decision-making power across many facets of the health sector in surveyed countries is strongly skewed towards the central government. This strong centralisation of health responsibilities is despite a general trend towards decentralisation of health care over the last 20 years, which has transferred competences to the subnational level. However, some OECD countries such as Australia, Germany or Sweden, have recentralised over the last 20 years (OECD, 2018[14]). On average, central governments are nearly twice as likely to be responsible for the health decisions surveyed, compared with regional governments, and four times more likely compared with local governments. As shown in the figure below, health remains a centralised responsibility in several countries, but most strongly in Greece, Chile and Iceland. At the other end of the spectrum, the subnational government is usually responsible for health decisions in Canada, Switzerland and Spain.

Figure 4.6. Decision-making power in the health sector, across levels of government

% of decisions at each level

Reponses to ‘other’ reflect the presence of other significant decision-making power across areas of spending power. Responses to other included public and private health insurance funds, and public and private service providers, particularly hospitals.

A shared responsibility is when two or more decision makers are responsible for the same decisions and is the result of multiple levels of government or authorities being responsible for the financing or policy making of service delivery. Figure 4.7 shows the level of shared responsibilities in health care. A high number of shared decisions suggests the presence of more complex frameworks and more overlapping responsibilities. This has the potential to generate inefficiencies in intergovernmental relations, and reduce transparency and accountability of public policies and government spending. Taller columns represent countries with a greater number of shared responsibilities in health care, including Argentina, Australia, and Denmark. Interestingly, Canada, Germany and Spain have low levels of shared responsibilities despite these countries being federal, where power is shared with subnational governments.

Figure 4.7. Shared responsibilities in health care

Cumulative number of areas for which each level of government is responsible

Policy decisions in health care

The majority of survey respondents stated that the central government is responsible for key decisions about policy (Figure 4.8). Specifically, setting public health objectives was a central government responsibility and a regional government responsibility, for 91% and 38% of respondents, respectively. Setting the legal framework (e.g. a law establishing objectives for and rights and obligations of hospitals) was the responsibility of the central government for 97% of respondents, and deciding on the various forms of service provision (public vs. private provision) was the responsibility of the central government and the regional government for 75% and 28% of respondents, respectively.

Setting minimum regulations/standards in hospitals was the responsibility of the central government in many countries (88% of respondents), but not in Belgium, Canada, Norway or the United Kingdom. Explicit minimum standards for service coverage, whether social and/or geographical, promote equal access for all citizens. Belgium’s current framework of minimum standards has been in place since the ‘6th state reform’, of which the last stage was finalised in July 2014. This reform involved transferring some health care competences (mainly for elderly residential care, mental health, recognition of medical professions and hospital standardisation) from the central government to communities. However, even if competences in some fields were transferred, the ‘playing field’ for the communities is still subject to national co-ordination or framework of rules. For example, regional rules for hospital standards cannot change the rules for social security, or the exercising of medical professions, or the financing rules of hospitals.

Figure 4.8. Responsibilities for key policy decisions between central and regional governments

Proportion of respondents that stated it was the responsibility of central or regional governments

Budgeting decisions in health care

Compared to policy decisions, key budgeting decisions were more evenly split across decision makers, but central governments have considerable power (Figure 4.9). Setting the level of taxes earmarked for health care and setting the base and level of social contributions/premiums for health care was the responsibility of the central government for 91% of respondents.

The same percentage of respondents answered that the central government was responsible for designing and implementing a scale for user contributions or co-payments, as well as differentiating user contributions according to the social situation of users. User contributions cover all individual payments to service providers, including private co-payments through insurance schemes, in return for a service. User contributions for health services can potentially contain excess demand, reducing pressure on government budgets and improving the quality of public services. However, user fees may be less suited for demand management when services are not particularly price sensitive, which may be the case for acute hospital care (Blöchliger, 2008[15]). Indeed, there is considerable evidence showing that excessive user fees and other out-of-pocket payments can impede access to care and cause financial hardship (WHO, 2010[16]).

Deciding on the resource allocation between sectors of care, in terms of hospital care, outpatient care, long-term care etc. was more evenly split with 66% and 39% of respondents suggesting that it was a central and regional government responsibility, respectively.

The central government is often responsible for regulating private hospital activity and determining the level and type of public funding for private hospitals. In Belgium, the definition of ‘hospitals’ is officially regulated and private health sector providers must be not-for-profit. For-profit institutions can enter the market but do not receive direct public financing. In Denmark, if public hospitals are unable to offer a service within a given timeframe determined by the central government, public hospitals may refer the patient to a private hospital, and the public sector pays the costs. In addition, private hospitals offer treatments funded by user fees or private insurance.

Budgeting decisions concerning hospitals were more evenly shared across decision makers compared to other budgeting responsibilities (Figure 4.9). Financing new hospital buildings was a central government responsibility and a regional government responsibility, for 59% and 47% of respondents, respectively. In Italy, a specific national fund for investment in health care is used for the financing of new hospital buildings. Previously, regions used to finance new hospital buildings through public-private partnerships. Financing new high-cost equipment was the responsibility of the central government for 50% of respondents, the responsibility of regional governments for 47% of respondents, and the responsibility of the other entities, like hospitals, for 41% of respondents. Similarly, financing the maintenance of existing hospitals was a central government responsibility and a regional government responsibility, for 50% and 47% of respondents, respectively. Financing hospital current spending was a central government responsibility (50%) and a regional government responsibility (34%). As would be expected, these key financing decisions are more likely to be the joint responsibility of central and regional governments in federal countries.

Many countries responded that entities other than central, regional or local governments were responsible for budgeting decisions in hospitals. These key decisions, for example financing hospital staff’s salaries, are often made internally by the individual hospital. For example in Switzerland, most hospitals have sufficient autonomy to decide on their own investments, but regional governments are able to influence decisions through their service plans.

Figure 4.10 shows the responsibility of regional governments in key budgeting decisions in federal and unitary countries. In federal countries, regional governments have a high level of responsibility for key financing decisions especially concerning hospital decisions, such as financing new hospitals, and hospital maintenance.

Despite greater decision-making power by subnational governments, central government has much of the responsibility over key budgeting decisions. Some of these key budget decisions, like setting the level of taxes, and setting the total budget for public health care, can restrict the revenue-raising potential of regional governments. This creates a mismatch, where the central government has greater influence with regard to revenue-raising decisions, while regional governments are more often responsible for financing, especially concerning hospitals. This mismatch suggests that the traditional indicator of decentralisation, measured as the subnational expenditure share as a proportion of total expenditure, overestimates the true level of budget autonomy in some, mainly federal, countries.

Figure 4.9. Responsibilities for key budgeting decisions across levels of government

Share of respondents that stated it was the responsibility of central, regional, or local governments or other

Figure 4.10. Regional government responsibility for key budgeting decisions, by federal and unitary countries

Share of respondents that stated it was the responsibility of regional governments

Labour and input decisions in health care

Labour and input decisions include the hiring and firing of staff, determining working conditions, establishing training rules and planning of necessary hospital infrastructure. The responsibility for these decisions was more evenly shared across levels of decision makers.

The hiring and firing of staff was the responsibility of the central government for 31% of respondents, the responsibility of regional governments for 31% of respondents, and the responsibility of the other entities, like hospitals, for 59% of respondents. Determining working conditions (salary scales, pension rules, and working hours) was often a shared responsibility across decision makers, and was a central government responsibility, a regional government responsibility, and the responsibility of other entities, for 88%, 34% and 47% of respondents respectively. In Australia, the relevant employer determines working conditions but must do so in accordance with legislated conditions of the central and regional governments. In the Netherlands, health care providers are responsible for determining working conditions but must comply with collective labour agreements.

Setting remuneration methods for physicians was a central government responsibility, a regional government responsibility, and the responsibility of other entities for 78%, 28% and 31% of respondents, respectively. This shared responsibility generally involves the central government establishing an overall framework for remuneration, with joint responsibility by subnational decision makers like insurers, health care institutions or doctors’ associations. In the Netherlands, for instance, the national market authority provides the regulatory framework for remuneration, which is implemented with considerable discretionary power by private insurers. Independent physicians benefit directly from this and remuneration of employed physicians also depends on their employer’s policy. Physician remuneration is also often the responsibility of regional governments in federal countries.

Local governments have little overall power regarding health care decisions, but were most likely to be responsible for input related decisions. In particular, these decisions include the planning and provision of necessary hospital infrastructure and infrastructure maintenance, and the hiring and firing of staff.

Output and monitoring decisions in health care

Key output and monitoring decisions in health care are shown in Figure 4.11, which includes the breakdown of responsibilities across levels of government. Output decisions, especially regarding hospitals, were split across decision makers. For example, determining the opening or closing of hospital units was a central government responsibility and a regional government responsibility for 56% and 50% of respondents, respectively. Determining the allotment of hospital beds across hospitals was the responsibility of the central, regional, and local governments for 50%, 38%, and 22% of respondents, respectively, and the responsibility of other entities for 31% of respondents. Determining the size of health care districts was the responsibility of the central government for 47% of respondents, and the responsibility of regional governments for 38% of respondents.

Monitoring decisions were more likely to be the responsibility of central government. Deciding on performance measurements, indicators and targets of service providers was a central, regional and local responsibility for 78%, 34% and 31% of respondents, respectively. Monitoring of service provision (does supply meet users’ needs, and is access for users from different regions or different social groups ensured) was the responsibly of central government for 78% of respondents and 34% and 16% for regional and local governments, respectively.

Figure 4.11. Responsibilities for key output and monitoring decisions, across levels of government

Share of respondents that stated it was the responsibility of central, regional, or local governments or other

Comparing responsibilities between different aspects of health spending autonomy

Areas of spending power consist of policy, budget, input, output and monitoring. As shown in Figure 4.12, central governments still have considerable spending autonomy. However, they are most likely to be responsible for decisions regarding the policy and budgetary aspects of health care, and have less control over decisions regarding the inputs and outputs as well as monitoring of health care. Decisions for input-related matters, such as determining which services can be out-sourced and deciding on the contractual status of staff, fall more on subnational governments, especially for regional governments in federal countries.

Local governments have little decision-making power in the health care sector, but have more responsibility with regard to health inputs, namely, deciding on hospital infrastructure maintenance and planning hospital infrastructure. Financing the current spending of hospitals and financing new high-cost equipment are more likely to be the responsibility of local governments in federal countries.

Figure 4.12. Responsibilities across areas of spending autonomy in health care

Spending power in other sectors

Housing

Housing responsibilities are typically held by multiple levels of government. In the majority of countries, the governance of the housing sector is shared between national government and local governments, with national governments having a slightly larger role regarding overall policy priorities, while local governments are responsible for the implementation of social housing programmes, the allocation of social housing, sustainable urban development and spatial planning.

As shown in Figure 4.13, local governments take on more responsibility for the output decisions and infrastructure financing of social housing provision. While local authorities still have a lot of responsibility for social housing policy decisions, national governments on average are more likely to be responsible for deciding on policy metrics. When averaging across respondents, social housing autonomy is most likely be assigned to central decision makers, with 57% of decision making the responsibility of national governments. Local governments also play an important role – on average they are responsible for 45% of decision making in the social housing sector, while autonomy for regions (23%) and providers is lower.

Figure 4.13. The categorisation of social housing responsibilities across actors

Average of respondents (%)

When looking at the social housing spending power indicators, output, monitoring and input decisions are more likely to be devolved to lower level actors, than policy decisions. This is consistent with the findings for the other five policy areas examined in this chapter. The devolution of spending power follows a similar trend as for other public services. Policy decisions have a spending power index of 2.4 on average, compared to input decisions which have a spending power indicator of 3.4, on the scale of 0 and 10. Social housing is on average most decentralised in Canada, Estonia, Colombia, Iceland and the Netherlands. Decision making is more devolved to lower-level actors in federal countries (spending power of 3.6), compared to unitary ones (2.9) in particular with regard to input decisions (Figure 4.14). The constitutional underpinning of countries does, however, not seem to affect the broad trend that policy decisions are more centrally determined, and output, monitoring and input decisions are more likely to be devolved.

Figure 4.14. Spending power indicators for social housing by sub-category

Education

The education sector surveyed in the questionnaire relates specifically to the primary and secondary education sector, including the administration, inspection, operation or support of schools. In the education sector, the central government has on average the most decision-making power with it being the sole or shared decision maker in 50% of the aspects of education surveyed. Despite this, regional governments, local governments, and education providers all have important powers across the countries surveyed, with these entities being a sole or shared decision maker in 26%, 32% and 29% of decisions respectively. These results support the view that the decentralisation of education, especially education financing, has become a global feature.

This is a positive outcome, as studies show that fiscal decentralisation may raise the overall share of the budget devoted to public investment and education, thus increasing human capital. Many studies also show that fiscal decentralisation may positively affect education performance, as measured by the OECD Programme for International Student Assessment (PISA) (Kim and Dougherty, 2018[17]).

To the extent that the central government has decision-making responsibility it tends to be most concentrated in policy, and output and monitoring decision making, while local government autonomy is most pronounced in regard to budgeting decisions and education providers have the most power over input decisions (Figure 4.15 and Figure 4.16). This aligns with the general hypothesis that areas of education services like setting the curriculum and setting overall standards for schools should be and generally are, centralised, while decision making around school and teacher management is generally assigned to the subnational level.

Figure 4.15. Distribution of responsibilities in education sector across decision makers

Figure 4.16. Distribution of responsibilities in education sector across autonomy classification

Long-term care

The long-term care sector includes a wide range of services that are provided over an extended period to people with a reduced degree of functional capacity. The OECD questionnaire focused mainly on services and benefits in the form of institutional care (e.g. nursing homes and assisted living facilities), home care by professional care providers, and included informal care, to the extent that governments offer subsidies, tax-credits or income support to assist relatives or friends acting as caregivers.

In the long-term care sector, the central government has the most decision-making power, with it being the sole or shared decision maker in 51% of the aspects of long-term care surveyed. Despite this, regional governments, local governments, long-term care providers and social security funds all have some power across the countries surveyed, with these entities being a sole or shared decision maker in 37%, 27%, 21% and 7% of decisions respectively.

Based on arithmetic averages, central government power is by far the most concentrated in decision making regarding policy aspects of long-term care services, with respondents suggesting that the central government is in charge of 71% of policy autonomy decisions in long-term care. Regional governments have the most autonomy in budgeting, and output and monitoring aspects of long-term care, and the local government is also most likely to be in charge of budgeting (Figure 4.17 and Figure 4.18).

Figure 4.17. Distribution of responsibilities in long-term care across decision makers

Figure 4.18. Distribution of responsibilities in long-term care across autonomy classifications

Transport services

Transport services includes the construction, maintenance, operation and administration of water, road and railway transport systems and facilities, and does not take into account non-scheduled bus services, funiculars, cable cars, chairlifts and air transportation.

In the transport services sector, the central government again has the most decision-making power on average, but power is more balanced towards transport providers and local governments. In contrast to the long-term care sector, regional governments have the least decision-making responsibilities, on average. There are also differences across the various transport sub-sectors, with rail services more centralised, and less likely to be the responsibility of local governments compared to other transport services. This is consistent with the premise that the local level often lacks the economics of scale to address urban-rural linkages across vast geographical areas, which are best undertaken by central governments. Bus services are more likely to be decentralised than the average, with high decentralisation of budgeting autonomy for bus services.

The central government responsibilities are relatively balanced across the four classifications of autonomy, but are still more skewed towards policy decisions. As would be expected, local governments are more likely to be in charge of transport services in urban areas, including buses, urban roads and bridges. The responsibilities of transport providers are more likely to be input orientated, especially with regard to the conditions and acquisition of workers (Figure 4.19 and Figure 4.20).

Figure 4.19. Distribution of responsibilities in the transport services sector across decision makers

Figure 4.20. Distribution of responsibilities in the transport services sector across autonomy classification

Extending the indicators: Computing composite indicators with uncertainty

As described earlier, low-level indicators (LLIs) obtained from survey responses are aggregated to compute medium-level indicators (MLIs) and subsequently high-level indicators (HLIs). This process introduces uncertainty which arises from several sources. First, there is uncertainty about the relative weight that should be attached to each LLI when computing the MLIs. Second, there is uncertainty owing to the fact that only a fraction of possible policy sectors are covered by the survey data. In effect, the sectors considered in this chapter constitute a sample of all possible sectors which fall within the purview of government. Finally, uncertainty arises from missing observations. As is common with survey data, not all information for every sector was made available by each country which participated in the initial questionnaire.

Random weights (RWs) have been hitherto employed to compute confidence intervals for MLIs and HLIs. This method relaxes all assumptions about the relative weights that should be attached to lower-level indicators when computing higher-level indicators. However, given that the RWs are drawn from a uniform distribution between zero and one, the mean indicator values are asymptotically equivalent to indicators using equal weights for all LLIs.

In order to account for other sources of uncertainty, namely unobserved sectors and missing observations, Kantorowicz and van Grieken (2019[5]) employ alternative methods to compute point-estimates and confidence intervals for higher-level indicators, using multi-level indicators computed by provider level. They implement two approaches: Bayesian factor analysis and country product dummies (CPD), which yield similar results. A technical description of the latter method is provided in Box 4.1.

Box 4.1. Combining multiple sectors into a synthetic indicator

Description of the country-product-dummy (CPD) approach

The CPD approach was originally developed for assessing purchasing power parity across countries. Applied to the spending autonomy of subnational governments, its main purpose is to tackle the problem of missing observations. As was noted in the methodological part, although a large overlap exists, we deal with different sets of countries across policy sectors, leading to a shrinkage of the sample when attempting to create the “balanced sample” composite indicator of spending autonomy.

The CPD method as applied in this study works as follows. In the first stage, we define the CPD model, which regresses observed LLIs on a country-fixed effect, autonomy dimension fixed effect, policy sector fixed effects and an error term. The estimates are obtained by OLS. The model is as follows:

${L L I}_{c a i} = α_{c} D_{c} + β_{a} {A D}_{a} + δ_{i} P_{i} + ε_{c a i}$

Where c, a, and i are country, autonomy and policy indexes (the right-hand side contains only dummy variables); LLIs are the values observed for a given dimension of autonomy in a given country and policy sector. From the first-stage model, we obtain the predicted values for missing LLIs. In the second stage, we run another CPD regression by augmenting the predicted LLIs for unobserved LLIs (if the predicted values of LLIs are outside the range, the values are trimmed). The resulting alpha coefficients ( $α_{c})$ from the second-stage regression reflect the country-specific level of overall spending autonomy of subnational governments. Constructed confidence intervals allow for assessing statistical differences between countries and, for instance, vis-à-vis the OECD average.

The CPD approach completes missing observations by regressing LLIs on several fixed effect variables and an error term. By using the results obtained to generate fitted values, it becomes possible to include additional countries for which data was otherwise missing. Subsequently, confidence intervals can be constructed using the standard errors generated through the regression analysis, which enables statistical inference in cross-country comparisons.

These approaches are applied to develop a composite spending power indicator for subnational governments that combines the HLIs across the five policy sectors, generating a single spending autonomy score for each country (Figure 4.21) on a scale of 0 to 1. More precisely, the CPD approach is first applied to generate the missing observations and generate confidence intervals, and is available in the Fiscal Decentralisation database. The dataset with missing observations generated is then subsequently employed to generate confidence intervals according to the random weights method.

According to this composite measure, the average spending autonomy indicator is roughly 0.6. Countries such as Canada, Switzerland, the United Kingdom, Belgium and Australia obtain the highest scores. Among the countries with the lowest spending autonomy scores are Greece, Slovenia, Iceland, Israel, Ireland, Luxembourg and Lithuania. The Czech Republic, Poland, South Africa and Chile lie just around the average score. A comparable ranking is obtained if the scores are calculated by RW using the CPD-imputed values (Kantorowicz and van Grieken, 2019[5]). The advantage of RW over CPD confidence intervals is that RW provides variation around the point estimates, which informs to what extent spending autonomy is balanced across autonomy dimensions and policy sectors. Chile, the Netherlands and Estonia stand out as the countries with the largest variation of spending autonomy values, indicating that these countries show rather inconsistent degrees of autonomy across autonomy dimensions and policy sectors.

Figure 4.21. Composite indicators of spending autonomy

The effects of decentralisation on service provision

Decision-making power and responsibilities for public services by subnational governments vary widely across countries, though very few countries around the world rely solely on services being devised and administered by central governments alone. Greater decentralisation to subnational actors increases the need for co‑ordination in itself, but is even more necessary for policy areas where there is a lot of shared decision making. Shared responsibilities can be an issue when roles overlap, which generates inefficiencies in intergovernmental relations, and reduces transparency and accountability, as the public is unsure which level of government is responsible for the delivery of public services and government spending. The responsibility for public services are often shared across levels of government, more so in federal countries than unitary countries, either through explicit legislation or through residual policy acquisition (Bach, Blöchliger and Wallau, 2009[13]). Shared responsibilities make it more crucial to establish governance mechanisms to manage these joint responsibilities, including platforms for dialogue, fiscal councils and contractual arrangements.

In the recent volumes OECD/KIPF (2018[17]) and OECD (2019[18]), the broad effects of decentralisation are considered. The literature and recent evidence suggests that the evidence can be fairly ambiguous at the whole-of-economy level, given strong country specificities and mixes of existing public service decentralisation. Thus, the marginal effect of centralisation or decentralisation reforms may have important effects on efficiency and equity, and an in-depth analysis is necessary. Nevertheless, in particular sectors, the evidence for potential improvements in outcomes through devolution can be considerably clearer.

The non-linear impact of decentralisation in health care

Decentralisation may allow for health systems to better target regional needs, while fostering good competition (Oates, 1999[19]); at the same time, taken too far, it can lead to fragmentation and overly high costs (Rodden, 2003[20]). Dougherty et al. (2019[7]; 2021[11]) carried out a novel empirical analysis aimed at exploring the role of institutions – notably the degree of administrative decentralisation across levels of government – in health care decision making, public spending on health care and life expectancy. To examine the relationship between the measured degree of decentralisation and public spending on health care and life expectancy, the study estimated a system of simultaneous non-linear equations, based on a micro-founded model. This empirical analysis also built upon earlier OECD studies.

The administrative decentralisation indicator and measures of other institutional features of health care systems such as financial incentives to improve quality and depth of basic coverage are used in the analysis of the impact of policy changes on public spending on health care and life expectancy. In order to carry out this analysis, a novel indicator of health care decentralisation was constructed over the 2008 to 2018 period, based on which level of government was responsible for thirteen policy or service areas. The resulting indicator yields a scale of 0 to 6, with 0 a fully central government decision, and 6 is a fully local government decision; shared decisions are scored with an intermediate value of 3. Marginal effects of varying degrees of decentralisation on public expenditure and life expectancy are shown in Figure 4.22.

Figure 4.22. Marginal effect of decentralisation on public spending on health care and life expectancy

Several new findings emerge from this analysis:

The results point to a statistically significant non-linear effect of “administrative decentralisation” on public health care expenditure and life expectancy.
The sign and size of the coefficients suggest that a moderate degree of decentralisation reduces public spending on health care and increases life expectancy – saving public resources and improving the quality of outcomes – as compared to countries with very low decentralisation.
However, “excessive” decentralisation is associated with higher public spending on health care and lower life expectancy – reversing cost-saving and quality-boosting effects – as compared to a situation with an intermediate degree of decentralisation.

Additional confirmatory estimates are also made using hospital microdata for almost a dozen countries, which include controls for specific disease groups (Dougherty et al., 2019[7]).

The impact of decentralisation in other sectors

Given the essential role of schools in the education sector, outcomes can be measured using PISA scores for 15-year-olds. Recent empirical analysis finds linear improvements with decentralisation, with a stronger relationship to revenues and tax autonomy than spending, as well as fiscal-administrative interactions, depending on school autonomy (Lastra-Anadón and Mukherjee, 2019[8]), as mentioned in Chapter 1 (Figure 1.4). At the same time, maintaining economies of scale plays an important countervailing role (OECD/EC-JRC, 2021[21]), both for education and healthcare. Note that the decentralisation of municipal organisations as a whole have also been found to yield improved outcomes in terms of labour productivity, conditional on high quality of local governance and low administrative fragmentation (Jong et al., 2021[9]).

Box 4.2. Measuring public sector performance

Measurement of public sector performance is central to improving outcomes, yet it is also extremely challenging to do well. Measuring, assessing and enhancing performance is a valuable exercise across the public sector but especially in the context of countries with multiple layers of government. Within the public sector, performance improvements provide a means by which citizens can get better value for their taxes. By raising the performance of services, governments may also be able to respond to growing demand without having to resort to higher taxation or increased debt burdens – a particularly important consideration currently.

The concept of performance is linked to both the efficiency and the effectiveness of public service delivery. Service providers that lie at the production possibility frontier are seen as achieving ‘best practice’, which is linked to the concept of efficiency. There are a number of ways to define efficiency, however a distinction is often made between technical efficiency and allocative efficiency. It can also be linked to tracing technical change, identifying changes in efficiency, identifying real cost savings in production, identifying inefficiencies, and assessing standards of living.

Technical efficiency characterises a production process where the maximum possible output has been achieved, given a fixed set of inputs and a certain technology. The concept of technical efficiency is similar to the notion of productivity, defined as a ratio of a volume measure of output to one of input use. Allocative efficiency refers to the allocation of resources, either in financial terms or in labour-capital terms, and occurs when the input-output combination is cost-minimising. Efficiency more broadly is difficult to measure as it requires knowledge of the sector’s production possibility frontier – that is, which providers are performing ‘best practice’ combinations. In practice, measurement of productivity is often a primary focus, since the best possible result that is achievable does not need to be known. A common metric is labour productivity, often calculated on the basis of output produced per labour hour worked, although broader measures may seek to use a cost-weighted set of outputs and inputs.

Figure 4.23. The relationship between public sector performance, efficiency and effectiveness

Effectiveness is another critical component of performance of service delivery and is related to quality, equity and access. Effectiveness often aims to identify the intangible value attached to the service and measure its performance against that value. Integrating effectiveness can deliver a more comprehensive view of performance especially with regards to public sector services. In contrast to efficiency broadly the ratio of outputs to inputs, effectiveness is the ratio of defined outcomes to inputs, and conditional on the quality of service provision. When conceptualising the measurement of public service performance, the integral role of the consumer or citizen to the production process needs to be considered. This refers to the level of customer satisfaction but also to the access that citizens have to public services.

The concept of performance measurement is interesting to study from a lens of multi-level government. Over the past decades, many countries have delegated more authority and responsibilities to subnational governments (OECD/KIPF, 2016[1]; OECD, 2019[18]). Decentralisation can create competition which can lead to policy innovations, transparency and better performance. However subnational governments may lack the resources and expertise to tackle major projects, particularly those requiring co‑ordination among governments and those that may be better suited to central administration. The economic, social and financial effects of decentralisation are still being debated especially as the benefits and costs often depend on locality specific circumstances and way decentralisation is designed and implemented (Jong et al., 2021[9]). Governance and oversight at the right scale depends on a number of factors including the types of services being provided, local government functions, public finance and even factors like trust in government and the history of government institutions (Krawchenko, 2021[22]).

With decentralisation comes a larger number of stakeholders and service providers, and greater complications in intergovernmental fiscal relations, which can intensify information gaps between levels of government. The creation of multiple public agencies can also make co‑ordination more complex, hindering the central government’s ability to steer public policy in the desired direction. Performance measurement systems can correct co‑ordination failures that are likely to be exacerbated by decentralisation and create a common frame of reference.

There are heightened difficulties in assessing performance within the public sector given multiple objectives. Public organisations are responsible for the delivery of key services with the expectation that they will be provided efficiently, effectively, and equitably. The performance of public sector agencies and service providers drive or constrain the performance of state intervention generally, and impact the efficiency of public spending and policy making within key service sectors. While private sector services are expected to be provided cost-effectively or with a profit objective, the performance of public services is generally viewed through a broader lens of both efficiency and effectiveness including against policy objectives like equity and access. While all elements of performance are relevant, some will be more important than others depending on the service, which makes overall measurement and assessments more challenging.

Source: Dunleavy (2017[23]); Lau et al. (2017[24]); Phillips (2018[25]).

Benchmarking and monitoring of public sector performance

Indicators to measure performance

Benchmarking in different public service domains and contexts is often a useful tool in promoting ongoing improved performance (Phillips, 2018[25]). The effectiveness of a benchmarking system is in turn influenced by the selection of indicators that are used to benchmark service providers, which if not chosen carefully, can distort the incentives and policy objectives of participants. Moreover, participation in benchmarking or performance systems can improve the collection and organisation of high quality data, resulting in governments improving their information base and performance metrics even before engaging in formal comparisons. The need to settle on common service objectives and performance indicators requires that these be given explicit and systematic consideration, which may be a novel development in itself and helpful in building a culture conducive to seeking improvement.

Productivity, efficiency and effectiveness are easily confused and can be difficult to separate (Box 4.2). While output measurement is critical for productivity or efficiency measurement, however, outcomes are integral to measuring the effectiveness or quality of services and provide a more comprehensive illustration of performance. For example, in the school education sector, it is possible that productivity could rise when measured by output metrics like student numbers or preferably hours of instruction, following a policy initiative to increase class sizes. But if this was accompanied by weaker learning outcomes, as reflected say in external test results or tertiary acceptances, performance in a more fundamental sense could have declined. The OECD PISA results are based on test scores of 15-year-old pupils, and represent a well-known example of outcome measurement of the performance of the educational system. Measuring outcomes is more necessary for services as opposed to goods, due to their intangible nature. Outcomes are also arguably more useful to assess for public sector sectors rather than those provided in the private market because the purpose of the public sector to achieve a set of desirable social or economic outcomes.

Measuring costs

Benchmarking subnational or regional government service providers through a cost or price approach can provide useful information on which operators or regional governments are most cost-efficient. The dissemination of detailed regional cost information to sub-national governments may help reduce information asymmetries and reveal unjustifiable or avoidable differences in costs of services. Comparing the average costs across providers, some variations will be due to factors that are beyond the control of service providers, like geographical variations in consumer needs or differences in price and wage levels.

Compared to the private sector, the estimation of the actual costs of public sector activities is relatively complicated. Most public services are not bought and sold, and many government services are collective goods, like policing and environmental protection, which cannot be consumed individually (Lau, Lonti and Schultz, 2017[24]). Public sector accounts are typically not as detailed or disaggregated as those in the private sector, making it difficult to obtain information on all input costs. Alternatively non-monetary factors can be used, like the number of civil servants involved in a public activity, pupil to teacher ratios or construction costs per dwelling completed.

Composite indicators of performance

Composite indicators are an aggregate index comprising individual indicators and weights that represent the relative importance of each indicator (Figure 4.24). The construction of a composite indicator is not straightforward and the methodological challenges raise a series of technical issues. However composite indicators are much easier to interpret than trying to find a common trend in many separate indicators. Examples of composite indicators include ‘star ratings’ or ‘report cards’. These composite performance ratings have taken on great importance as they are often used to reward or penalise organisations. The transparent reporting and easy understanding of these can sharpen incentives for public service managers to do better, to the extent that greater public awareness about deficiencies leads to political pressure to address them.

Figure 4.24. Performance benchmarking approaches

Such indicators have proven to be useful in ranking regions or service providers in benchmarking exercises, however, they can send misleading policy messages if misinterpreted. Arguments in favour of composite indicators include focusing attention on important policy issues, offering a more rounded assessment of performance and presenting the ‘big picture’ in a way, which the public can understand. They provide an attractive option for accountability purposes, as it is easier to track progress of a single indicator over time rather than a whole package of indicators. Ideally sub-indicators should include output, process and outcome measures. Outputs are more immediate and are more representative of what is important to users. However, final outcomes are obviously important because they are a better indicator of the impact of government services.

However composites need to be published with indications of uncertainty to communicate the sensitivity of the reported measure and be supplemented with more detailed performance information or breakdown of the elements. This is because research suggests that rankings of performance indicators can be unstable and that subtle changes to construction, like weighting systems and aggregation rules, can cause large changes in rankings.

Consumer experience and satisfaction surveys

To achieve meaningful measures of service quality, central and regional governments are implementing measures of consumer experience and satisfaction, in order to identify specific domains of satisfaction or value, or to measure service performance against explicit national standards. Customer satisfaction measurement may also enable an organisation to understand the extent to which satisfaction with a service is influenced by factors outside its control and to understand what is really driving satisfaction with a service experience. Numerous researchers who focused on customer satisfaction have indicated that it is an emotional reaction influenced by the interaction of users’ pleasure, expectations of performance, assessment of consumer experience, and consumer interests (Phillips, 2018[25]). In Denmark, a national performance target relating to student well-being was introduced in 2014, which generated the implementation of a national survey on student well-being. Results from the survey are intended to help devise initiatives to enhance students’ well-being.

Consumer satisfaction survey results, such as those derived from in-patient surveys in hospitals, are a useful type of performance measurement system. Hospital performance is becoming more focused on health education, patient empowerment, comfort, complaint mechanisms and continuity of care. Patient experience measures are a complement to quantitative quality measures, as long as the information is collected using psychometrically sound instruments, employing recommended sample sizes and adjustment procedures.

With regards to health care surveys specifically, patient-reported indicators of health system performance largely relate to patient-reported experience measures (PREMs, such as whether the patient feels they were adequately involved in important decisions about their care), and patient-reported outcome measures (PROMs, such as whether the patient is free of pain after an operation care). Some OECD countries are conducting PREMs surveys and, to a lesser extent, are experimenting with PROMs (Lau, Lonti and Schultz, 2017[24]). A number have found that patient reports of better patient-centred hospital care were significantly associated with better survival rates after treatment. These findings suggest that many aspects of service delivery that are outside the usual realm of quantitative testing are important for consumer satisfaction narrowly, and important for consumer outcomes more generally.

Systems to measure and assess performance

In most countries, subnational jurisdictions have a significant role in service delivery and regulation, but central governments share the incentive to see them perform well, being either ultimately responsible in whole or part to citizens, or seen that way by electorates. As a performance assessment and reporting mechanism, benchmarking has the advantage of flexibility, in that it can accommodate a variety of indicators and be progressively expanded or adapted over time. Indeed, many countries have reported the use of central government performance measurement systems, especially in the hospital and education sectors. Benchmarking can be a means by which a central government is able to overcome informational barriers to monitoring the performance of sub-national jurisdictions in order to improve transparency and rectify underperformance. National performance and benchmarking systems may be particularly sought when regionally administered services are largely centrally funded. As economies have evolved and markets have become more integrated, a number of services or activities previously regarded as ‘local’, increasingly have national ramifications.

The typical centralisation of monitoring and policy levers is also expected from an accountability and equity viewpoint. Fiscal decentralisation can obstruct the redistributive role of the central government. With high levels of subnational spending autonomy, the central government may not have sufficient resources to reduce any large income differences across the regions of a country. Centralised monitoring may be needed to reduce inequalities and ensure a broad access to services. Centralised monitoring and policy is also beneficial in the absence of strong local democratic processes, where subnational governments do not feel accountable for their spending behaviours.

The setting of national policy objectives, minimum requirements, oversight and benchmarking may be justified on the grounds that some services have broader national, and sometimes international, implications. In many countries the central government is responsible for setting these minimum requirements, especially with regards to health and education. National service requirements may be in terms of immunisation requirements in the health sector, setting curricula, education requirements of teachers and minimum access requirements in the education sector, or regulating education or training requirements in certain professions.

Collegiate systems

Collegiate (or collaborative) benchmarking is based on learning from best practices, as opposed to using ‘naming and shaming’ techniques. It generally involves consultation and collaboration between levels of government. Collegiate benchmarking is unlikely to involve the public dissemination of ‘highstakes’ performance information meaning that sub-national governments may be more likely to participate in these forms of benchmarking and share their experiences. Collegiate benchmarking will more likely be implemented if all levels of government perceive that it will lead to new or better information channels, improve policy effectiveness, or if they can share the additional resources and political leverage. Central governments can play an important role in facilitating knowledge and data sharing across regional governments and consolidating data sources.

A performance system based on collegiality is more likely to be amenable to regional governments in a federal, decentralised system. In Australia for instance, the central and regional governments undertake a collaborative exercise to produce an annual report on the performance of subnational service delivery (APC, 2021[26]). Benchmarking may also be voluntary rather than imposed by the central government, which may be more prevalent if sub-national governments are not heavily reliant on the central government for funding. For example, performance comparisons across local authorities in Germany are largely voluntary and self‑managed. The Association of Local Government Management (KGSt), for instance, encourages voluntary participation from districts and municipalities, with the KGSt establishing voluntary benchmarking networks and a Common Assessment Framework to assess performance across governments.

Competitive systems

Competitive benchmarking aims to generate competition amongst regional governments by disseminating information and thus facilitating comparisons by citizens and voters. In fact, the performance monitoring of public services in many countries seems to have moved from using performance information to improve organisational processes towards transparency mechanisms to enhance accountability. While transparency of public services has increased, the availability of information is only useful if it actually achieves the long-term objective of improving the efficiency and effectiveness of or access to public services.

Some research suggests that this public dissemination of information is likely to improve service delivery though the best benchmarking system is debatable. Across many OECD countries, the public dissemination of performance information is seen as a useful component of benchmarking systems by central government in key service areas, including transportation, education and long-term care. If performance measurement and benchmarking is accompanied with public transparency, it can create competition between regional governments and service providers and thus lead to efficiency gains. Revenue and expenditure decentralisation creates competition among regional governments for mobile, tax-paying citizens, which have an incentive to move to regions that offer services and taxes that match their preferences (Tiebout, 1956[27]).

The role of “tailored” institutions for evaluation and oversight

In discussing the role of measurement systems and issues related to the decentralised provision of services, it was noted that their effectiveness in promoting better public sector performance can be significantly influenced by governance arrangements, institutional frameworks and other contextual matters. A variety of public institutions have been established in different countries specifically for the purpose of promoting or upholding public sector performance in key dimensions. Unlike regulatory agencies, although characterised by significant independence, they are generally limited to informational or advisory roles, and have no executive power.

This distinction is relevant to concerns expressed about the rise in the power of ‘technocratic’ or bureaucratic institutions and their potential to displace or undermine democratic structures. In the case of the institutions identified here, there is a case that by serving to promote transparency, accountability and better informed decision making by government itself, they are actually strengthening democratic governance.

The relevant institutions

The principal institutions in this category are: public sector oversight bodies, supreme audit institutions, fiscal oversight/reporting bodies, policy research and advisory institutions and regulatory oversight bodies. Such institutions have complementary roles across the policy ‘cycle’, from informing initial decisions, to monitoring implementation and administration, assessing programme performance and reviewing a policy itself and whether its objectives have been met.

There are also institutions that span jurisdictions or regions within a country, such as those responsible for advising on transfers for fiscal equalisation purposes, or co‑ordinating and monitoring national reforms or other cross-jurisdictional policy initiatives. These tend to be more prevalent within federal systems of government, with few economic, fiscal or regulatory institutions having a “vertical scope” of responsibility (Dougherty, Renda and von Trapp, 2021[10]). One important exception is Spain’s AIReF fiscal institution, which has a national mandate and provides important support at all levels of government and to regions.

A key benefit of decentralised government is the ability to address the needs and preferences of citizens more effectively, particularly where circumstances differ. A key risk in this approach is fragmentation and incoherence from diverse approaches may diminish outcomes in aggregate. A number of institutions have emerged during the COVID-19 crisis to reduce this risk and bring additional benefits from ‘joined up’ co‑ordination. One of the more notable institutional innovations around the OECD was the adaptation of Australia’s Council of Australian Governments (COAG) into a ‘National Cabinet’ (Box 4.3).

Box 4.3. Australia’s National Cabinet

Australia co-ordinated actions across levels of government in order to tackle the COVID-19 crisis through the use of executive federalism (see Chapter 9). In the Australian case, the key body that was put to use to co‑ordinate action was the Council of Australian Governments (COAG). This Council has brought together the Prime Minister and the Premiers of the States as well as the authorities in charge of sectorial policies from Ministries, territories and states. It existed prior to the COVID-19 pandemic, having been formed in the 1990s as part of the implementation of policy reforms and deregulation. In a typical year, the COAG had met once or twice in order to discuss relevant matters and to improve intergovernmental co‑operation through joint decision making and oversight of the implementation of policies. Historically, the Prime Minister set the agenda and thus has a leadership role.

In the months following the pandemic outbreak, the COAG developed such a prominent role in managing the COVID-19 crisis that it was transformed into the “National Cabinet”. Important changes were made in its governance to adapt to the severity of the crisis. The COAG organised meetings much more regularly – often on a bi-weekly basis. In addition, the scope of the discussions was widened to encompass multiple topics related to the COVID-19 crisis such as health, the economy, education, transport logistics, law and order, among others. In order to handle such a variety of topics, different councils that respond to the COAG were created, each involving authorities from specific sectors from different levels of government. These technical councils have been crucial to provide a technical position to the executive authorities, which are from both sides of the political spectrum and, thus, help to minimise political factors in decision making.

Source: de Biase and Dougherty (2021[28]).

References

[26] APC (2021), Report on Government Services 2021, Australian Productivity Commission, https://www.pc.gov.au/research/ongoing/report-on-government-services/2021.

[13] Bach, S., H. Blöchliger and D. Wallau (2009), “The Spending Power of Sub-Central Governments: A Pilot Study”, OECD Economics Department Working Papers, No. 705, https://dx.doi.org/10.1787/223123781022.

[2] Beazley, I. et al. (2019), “Decentralisation and performance measurement systems in health care”, OECD Working Papers on Fiscal Federalism, No. 28, http://oe.cd/il/FFwps.

[15] Blöchliger, H. (2008), “Market Mechanisms in Public Service Provision”, OECD Fiscal Federalism Working Papers, No. 6.

[12] Blöchliger, H. and D. King (2007), “Less than you thought: The fiscal autonomy of sub-central governments”, OECD Economic Studies 43, https://dx.doi.org/10.1787/eco_studies-v2006-art12-en.

[28] de Biase, P. and S. Dougherty (2021), “Federalism and public health decentralisation in the time of COVID-19”, OECD Working Papers on Fiscal Federalism, No. 33.

[11] Dougherty, S. et al. (2021), “The impact of decentralisation on the performance of health care systems: a non-linear relationship”, The European Journal of Health Economics, https://doi.org/10.1007/s10198-021-01390-1.

[7] Dougherty, S. et al. (2019), “The impact of decentralisation on the performance of health care systems: A non-linear relationship”, OECD Working Papers on Fiscal Federalism, No. 27, http://oe.cd/il/FFwps.

[3] Dougherty, S. and L. Phillips (2019), “The spending power of sub-national decision makers across five policy sectors”, OECD Working Papers on Fiscal Federalism, No. 25.

[10] Dougherty, S., A. Renda and L. von Trapp (2021), “Independent Oversight Bodies: Lessons from Fiscal, Productivity and Regulatory Institutions”, in Plaatjies, D. (ed.), Making Institutions Work, HSRC Press.

[23] Dunleavy, P. (2017), “Public sector productivity: Measurement challenges, performance information and prospects for improvement”, OECD Journal on Budgeting, https://dx.doi.org/10.1787/budget-17-5jfj7vb36p5c.

[4] James, C. et al. (2019), “Decentralisation in the health sector and responsibilities across levels of government: Impact on spending decisions and the budget”, OECD Journal on Budgeting, https://dx.doi.org/10.1787/c2c2058c-en.

[9] Jong, D. et al. (2021), “A comprehensive approach to understanding urban productivity effects of local governments: Local autonomy, government quality and fragmentation”, OECD Regional Development Papers, No. 11, OECD Publishing, Paris, https://dx.doi.org/10.1787/5ebd25d3-en.

[5] Kantorowicz, J. and B. van Grieken (2019), “Spending Autonomy of Sub-central Governments: Conceptualisation and Measurement”, Background paper to the Network on Fiscal Relations, https://oe.cd/KvG19.

[17] Kim, J. and S. Dougherty (eds.) (2018), Fiscal Decentralisation and Inclusive Growth, OECD Publishing and Korea Institute of Public Finance, http://oe.cd/FDIG.

[22] Krawchenko, T. (2021), “Public sector and productivity: Governance at the right scale”, OECD-EC high-level expert workshop Productivity Policy for Places, http://www.oecd.org/regional/W3-S1-Krawchenko.pdf.

[8] Lastra-Anadón, C. and S. Mukherjee (2019), “Cross-country evidence on the impact of decentralisation and school autonomy on educational performance”, OECD Working Papers on Fiscal Federalism, No. 26, https://dx.doi.org/10.1787/c3d9b314-en.

[24] Lau, E., Z. Lonti and R. Schultz (2017), “Challenges in the Measurement of Public Sector Productivity in OECD Countries”, International Productivity Monitor, Vol. 32, pp. 180-195, http://www.csls.ca/ipm/32/lau.pdf.

[19] Oates, W. (1999), “An Essay on Fiscal Federalism”, Journal of Economic Literature, Vol. 37/3, pp. 1120-1149, https://doi.org/10.1257/jel.37.3.1120.

[18] OECD (2019), Making Decentralisation Work: A Handbook for Policy-Makers, OECD Multi-level Governance Studies, OECD Publishing, https://dx.doi.org/10.1787/g2g9faa7-en.

[14] OECD (2018), OECD Regions and Cities at a Glance, OECD Publishing, Paris, https://doi.org/10.1787/reg_cit_glance-2018-en.

[21] OECD/EC-JRC (2021), Access and Cost of Education and Health Services: Preparing Regions for Demographic Change, OECD Rural Studies, OECD Publishing, Paris, https://dx.doi.org/10.1787/4ab69cf3-en.

[1] OECD/KIPF (2016), Fiscal Federalism 2016: Making Decentralisation Work, OECD Publishing, https://dx.doi.org/10.1787/9789264254053-en.

[6] Phillips, L. (2020), “Decentralisation and inter-governmental relations in the housing sector”, OECD Working Papers on Fiscal Federalism, No. 32, http://oe.cd/il/FFwps.

[25] Phillips, L. (2018), “Improving the Performance of Sub-national Governments through Benchmarking and Performance Reporting”, OECD Working Papers on Fiscal Federalism, No. 22.

[20] Rodden, J. (2003), “Reviving Leviathan: Fiscal Federalism and the Growth of Government”, International Organization, Vol. 57/4, pp. 695-729, https://doi.org/10.1017/s0020818303574021.

[27] Tiebout, C. (1956), “A Pure Theory of Local Expenditures”, Journal of Political Economy, Vol. 64/5, pp. 416-424, https://doi.org/10.1086/257839.

[16] WHO (2010), World Health Report 2010: Health System Financing, the Path to Universal Coverage.

Note

← 1. The OECD Fiscal Decentralisation database is available at http://oe.cd/FDdb.

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & Events

News & Events

Featured Events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment