2. Evaluating fiscal equalisation: Finding the right balance

Fiscal equalisation refers to the transfer of financial resources to and between sub-central governments (SCGs) with the aim of mitigating regional differences in fiscal capacity and expenditure needs. Fiscal equalisation systems redistribute funds from wealthier governments to governments that face higher per capita costs or lower per capita revenue capacities. Such transfers may take place vertically, from higher levels of governments to lower levels of government, or horizontally, within the same level of government. Distinct fiscal equalisation arrangements first emerged during the 1940s and 1950s in a number of federal countries. Today most OECD countries have redistributive programmes to reduce fiscal disparities (OECD, 2013[1]).

Sometimes described as an enabler of fiscal decentralisation, equalisation allows SCGs to provide their residents with similar levels of public services at similar levels of taxation by evening fiscal capacity across SCGs (OECD/KIPF, 2016[2]). Moreover, equalising transfers are typically, though not always, non-earmarked. This increases the relative fiscal autonomy of jurisdictions that would otherwise have fewer resources.

Assuming that some degree of decentralisation is welfare-enhancing, equalisation systems that correct for structural differences between jurisdictions can counteract the less desirable effects of competition between SCGs that would distort firm and household decisions (Boadway, 2003[3]; Boadway and Shah, 2009[4]; Kim and Dougherty, 2018[5]). This theoretical formulation may make the task of an equalisation system appear straightforward: to reduce disparities arising from structural differences between regions. However, in practice, significant ambiguity derives from two principal sources: first, the interaction between equalisation systems and the incentives faced by policymakers and second, the problem of accurately capturing “structural differences” between regions.

Based on the 2019 fiscal equalisation questionnaire (Dougherty and Forman, 2021[6]), this chapter presents a cross-country analysis of the mechanisms underpinning fiscal equalisation, their impacts on fiscal redistribution and macroeconomic outcomes, how equalisation systems can be evaluated and reformed and the associated policy implications. New data collected in the 2019 questionnaire on the formulas and factors that determine equalising transfers offer important insights into how countries design equalisation systems and approach the myriad challenges they pose.

Key observations from the 2019 survey results include:

  • Most equalisation systems combine elements of more than one of the three principal modalities of fiscal equalisation: cost, revenue and gap-filling. These are summarised in Figure 2.1.

  • Overall, equalisation systems have a range of impacts on inter-jurisdictional revenue inequality. Systems that have a robust cost-equalising component tend not to realise as great a reduction in inter-jurisdictional revenue inequality. In these cases, the Gini coefficient of per capita SCG revenues may remain the same or even increase after equalising transfers.

  • Among equalisation systems that reduce revenue inequality, the Gini coefficient declines by 8 percentage points on average after equalisation. This corresponds to an average reduction in fiscal disparities by nearly one-third after equalisation.

  • There is little observable cross-sectional evidence to suggest the extent of fiscal equalisation correlates with an observed reduction in regional convergence.

  • Mechanism design approaches to fiscal equalisation centre on reducing the incentive to suppress (or inflate) SCG revenues (or costs). Such approaches often entail the use of a representative tax system or standardised costs to compute equalisation entitlements. In the case of revenue equalisation, certain revenues may be entirely excluded from the assessment of SCG fiscal capacity in order to promote the development of own-source revenues.

  • Assessing the impact of cost equalisation presents unique challenges because its effects cannot be captured by straightforward measures of revenue disparity like the Gini coefficient. While cost equalisation aims to facilitate equitable access to public services across SCGs, this outcome is rarely assessed in the context of reviews of equalisation systems. Accordingly, there may be an opportunity to more closely connect cost equalisation to subnational performance benchmarking.

  • COVID-19 introduces a special set of challenges for equalisation systems which are not well adapted to responding to emerging, short-term crises (see Box 2.2). Many countries anticipate that despite the asymmetric impacts of COVID-19, equalisation payments will fall because they are tied to elastic revenue streams. In some cases, the asymmetries introduced by the pandemic will not be reflected by equalisation payments for several years due to the use of lagged variables in the underlying formulae.

Guided by these insights, several good practices emerge which provide a framework for building, reforming or evaluating equalisation systems. These build on past policy messages from OECD research on fiscal equalisation that focused on the simplification of equalisation formulas to reduce manipulation by subnational governments, the use of representative tax systems (RTS), the separation of equalisation from other grants and regular monitoring (Blöchliger and Charbit, 2008[7]; OECD, 2013[1]; OECD/KIPF, 2016[2]).

This chapter is organised as follows: first, the principal modalities of fiscal equalisation are defined and their characteristics are summarised on a cross-country basis (see Table 2.1). Second, the impacts of equalisation on revenue and cost disparities are assessed (see Table 2.3) with a special focus on cost equalisation. Third, the mechanism design approaches to policy challenges in equalisation are discussed. Fourth, approaches to the assessment and reform of equalisation systems are analysed (see Table 2.6), including with reference to OECD economic surveys (see Table 2.5). Conclusions and policy implications are summarised in Box 2.1.

There are three principal modalities of fiscal equalisation: cost, revenue and gap-filling, with most equalisation systems combining elements of each. Selected equalisation systems are classified according to these modalities in Table 2.1. Cost and revenue equalisation systems aim to compensate for differences in per capita costs or revenues across SCGs. Gap-filling equalisation systems aim to bridge the gap between per capita costs and revenues for each SCG, rather than consider either side of the ledger in isolation. Previous OECD research has identified the cost and revenue equalising modalities (OECD, 2013[1]; Blöchliger and Charbit, 2008[7]). The present paper augments this framework with the gap-filling classification.

The source of the funds for equalising transfers is typically described as either vertical (from the centre to SCGs) or horizontal (from wealthier SCGs to poorer SCGs) (OECD, 2013[1]). Few equalisation systems are fully horizontal – meaning that they are entirely funded by transfers of own-source revenues between SGCs (one notable example of full horizontality is Lithuania) (Figure 2.1). Some systems incorporate sizeable horizontal components (e.g. Germany, Sweden, Switzerland, and Ireland), but the majority are entirely vertical. In turn, vertical equalisation systems are often funded by fixed shares of central government revenue streams (e.g. Japan and Korea). In other cases, funding comes from a fixed envelope of central government revenues which grows in line with macro indicators (e.g. GDP in Canada).

Revenue equalisation relies on measurements of real or potential per capita revenues (fiscal capacity) to determine equalising grants to jurisdictions. Typically, a representative tax system (RTS) based on cross-jurisdictional average tax rates is applied to an SCG’s revenue base to determine its fiscal capacity. In Canada, an RTS encompassing five revenue categories is used to calculate the theoretical revenues that would be raised should a province apply, roughly, national average tax rates across the specified revenue categories. A system of vertical fiscal equalisation then provides transfers from the federal government to provinces whose fiscal capacity falls below the average fiscal capacity of all provinces under the RTS. Australia uses a similar approach to assess revenues but complements the system of revenue equalisation with a comprehensive assessment of costs, based on standardised expenses. In this sense, it is essentially a gap-filling system. Crucially, Australia’s equalisation system is entirely financed by GST revenues, which are earmarked for this purpose (Coppel, 2018[9]).

Some revenue equalisation systems, such as Germany’s, are notable for their horizontal element: the transfer of revenues from wealthier to poorer SCGs within the same level of government. A defining feature of horizontal revenue equalisation is skimming, which is the appropriation of SCG own-source revenues above a certain threshold for redistribution via the equalisation system. In effect, this is a taxed faced by governments on their own-source revenues (See Table 2.2 for an example of the German case).

Cost equalisation systems aim to compensate SCGs with higher per capita costs relative to other SCGs, such that they do not face an undue burden. Typically, per capita costs are measured using average or standardised costs, based on budget categories, rather than the SCG’s actual expenditure. Sometimes, indicators such as geography (e.g. forest cover), poverty rates or surface area are directly integrated into cost equalisation formulae. Such formulae are sometimes highly complex, capturing a wide range of cost factors. For instance, the cost-equalising component of Italy’s equalisation system measures thirteen different factors to determine the standardised costs of eight essential functions of municipal government (SOSE, 2018[11]). Many equalisation systems integrate both cost and revenue equalising components.

Australia1 and Sweden provide examples of comprehensive cost equalisation systems. In Australia, dozens of factors covering all aspects of state expenditure as well as the underlying drivers of cost disparity enter into the equalisation formula. This allows the equalisation system to capture much of the variation in per capita funding requirements. In addition, it helps to enhance the policy neutrality of the equalisation system by employing cost variables which generally cannot be directly affected by policy choices. Similarly, the Swedish cost equalisation systems aim to assess structural differences in the per capita costs faced by SCGs through the use of sectoral expenditure models, each relying on hundreds of variables.

Gap-filling approaches to equalisation combine cost and revenue equalisation into a single transfer designed to fill the gap between assessed costs and assessed revenues. The distribution of Japan’s local allocation tax exemplifies this approach. An assessment of each municipality’s financial need is made, along with an assessment of its fiscal capacity. The local allocation tax is then used to fill the gap between these two quantities. Similarly, Korea’s general grant to municipalities aims to fill the gap between standardised financial needs and standardised revenues. Importantly, the system relies on standardised rather than actual values of revenue and cost to avoid perverse incentives.

An interesting feature of gap-filling systems is that they generally increase the inequality of per capita revenue after the nominally equalising transfers are applied. Because jurisdictions with higher per capita costs receive higher per capita payments, SCGs that initially have above-average per capita own-source revenues may see even higher post-transfer per capita revenues if per capita costs are uncorrelated or positively correlated with per capita revenues. The rise in post-equalisation revenue inequality is also driven by the fact that among gap-filling systems, the majority of the equalising transfer compensates for cost rather than revenue disparities (see Figure 2.6).

Roughly half of the equalisation systems surveyed combine cost and revenue equalising components, with distinct transfers for each purpose (Figure 2.2). The other half are split near-evenly between gap-filling systems and those that are exclusively revenue equalising, without significant cost equalising components. Importantly, no system is exclusively cost equalising, free of any measure of fiscal capacity.

Figure 2.3 illustrates the scale of equalising transfers as a percentage of total government expenditure (all levels) across sixteen OECD economies. Among the countries featured, equalising transfers average 3.6% of government expenditure, with Australia having the largest transfers as a share of total government expenditure (9.9%), and Belgium having the smallest (0.6%).2

Table 2.3 illustrates the change in the Gini coefficient of per-capita revenues of SCGs before and after equalising transfers are applied. While for a number of OECD and emerging economies, the reduction in the Gini is substantial, for others it is not:

  • For some countries, the reduction in the Gini coefficient is considerable (e.g. Canada, Germany).

  • For several countries, the reduction in the Gini coefficient is quite modest (e.g. Brazil, India).

  • In a few cases, equalising transfers have almost no impact on the Gini coefficient (e.g. Mexico, Italy).

  • In several cases, the Gini coefficient rises (e.g. Japan, Korea).

Figure 2.4 illustrates the modal variation in the effect of equalisation on disparities in inter-regional per capita revenues. Purely revenue-equalising systems demonstrate a 44% drop in the Gini coefficient on average, corresponding to a 14 percentage point reduction. Revenue/cost systems – that provide separate transfers to equalise both revenues and costs – demonstrate an average 17% drop in the Gini coefficient or a three percentage point reduction. Conversely, gap-filling systems illustrate an average rise of 36% in the Gini coefficient or a three percentage point increase.

Such outcomes suggest that the impact of fiscal equalisation cannot be measured in terms of the effect on revenue distribution alone. In cases where countries see a rise in the post-equalisation Gini coefficient (e.g. Australia, Japan, Korea), the equalising effect of the system can only be captured by examining measures that relate fiscal capacity to expenses, like Australia’s GST relativities or Japan’s fiscal capacity index. Accordingly, the appropriate metric for the effect of a given equalisation system must be related to the design of the system itself. This is particularly true where transfers are driven by variation in cost to a higher degree than variation in per capita revenues such that measures of revenue disparity fail to capture the equalising effect.

The correlation between the extent of revenue equalisation and regional income disparities appears limited at the cross-country level, with univariate regression analysis finding no significant relationship (Figure 2.5). Though past OECD work has suggested that equalisation may reduce incentive for poor regions to catch up or for households and firms to migrate to more prosperous jurisdictions (OECD, 2006[12]; OECD, 2013[1]), there are important examples of countries that illustrate high levels of regional convergence in the presence of robust fiscal equalisation. These include Canada, Germany and Lithuania.

Where an equalisation system is purely revenue-equalising, the task of assessing regional disparities is made relatively straightforward: it is fairly easy to quantify the change in disparity of subnational per capita revenues before and after the addition of equalising transfers (Table 2.3). In contrast, the effect of cost equalisation systems cannot be evaluated via a metric that simply measures the change in revenue disparities. Cost equalisation systems aim to compensate regions for elevated costs and by their nature, introduce a high degree of complexity and special set of challenges for the policy maker (Expert Panel on Equalization and Territorial Formula Financing, 2006[13]).

Across the OECD, the extent of cost equalisation varies widely (Figure 2.6), from systems that exclude costs entirely (e.g. Canada) to systems dominated by compensation for cost disparities (e.g. Korea). Notably, among gap-filling systems (e.g. Korea, Japan, Ireland and Australia) the imputed3 cost equalising component of the transfer represents the majority of total spending on equalisation. This suggests that the cost disparities faced by SCGs in these countries exceed their revenue disparities.

Figure 2.7 depicts countries’ inter-regional per capita revenue disparities before and after equalisation, as well as the inter-regional disparity of per capita expenditure after all transfers. This reveals the central challenge of observing the effect of equalisation systems. In thirteen cases (Belgium, Brazil, Canada, Estonia, France, Germany, India, Israel, Latvia, Lithuania, Luxembourg, Switzerland, and Sweden), equalising transfers reduce per capita revenue disparities. However, in the remaining seven cases examined (Australia, Ireland, Italy, Japan, Korea, Mexico, and the Netherlands), equalising transfers either have no effect on per capita revenue disparities or increase them. In five cases (Australia, Japan, Korea, the Netherlands, and Ireland), revenue disparities rise after equalisation alone. In another seven cases, disparities in per capita final expenditure appear to partially (Brazil, Canada, Estonia, India, Latvia, Sweden) or fully (Belgium) counteract the reduction achieved by the equalisation system. Finally, in two cases (Ireland and Australia), per capita final expenditure illustrates even greater regional disparity than per capita revenue after equalisation, which was already higher than per capita revenue before equalisation.

Three observations are immediately apparent. First, equalisation systems do not always reduce disparities in per capita revenues. Second, other transfers from central government or deficit spending by subnational governments may counteract the effects of equalisation, leading to an increase in the disparity of per capita final expenditure relative to per capita revenue after equalising transfers. Naturally, such an outcome may be desirable where variation in certain expenditure needs is not captured within the scope of the equalisation system. Third, the opposing effects of different inter-governmental transfers raises important questions about the coherence of decentralisation. Coherence requires that decentralisation be balanced along the political, fiscal and administrative dimensions with each level of government illustrating broadly consistent levels of competence across policy sectors. Distinct transfers with contradictory effects on SCG fiscal capacity could therefore be indicative of incoherent fiscal relations.

These observations can easily be related to the modalities of equalisation systems (see also Figure 2.4): with one exception (the Netherlands), all the systems that demonstrate a rise in post-equalisation inter-regional revenue disparity are of the gap-filling type. Notably, these systems are characterised by robust cost-equalisation components that seek to capture variations in SCGs’ per capita expenditure needs via comprehensive measures. In contrast, systems that are exclusively or near-exclusively revenue equalising, such as Canada and Germany, demonstrate unambiguous declines in the post-equalisation inter-regional Gini coefficient.

Figure 2.8 suggests that the intended effect of cost equalisation systems cannot be observed via a measure of inter-regional revenue disparity. Indeed, the nature of cost equalisation is fundamentally different from that of revenue equalisation, which may necessitate a different approach to measuring its effect. Moreover, the importance of defining a suitable measure to quantify the effects of cost equalisation is reinforced by the fact that cost equalisation may generally be positively associated with larger equalisation systems (Figure 2.9).

Cost equalisation systems range considerably in complexity, underpinned by formulas which aim to capture the drivers of cost disparities. Figure 2.10 depicts the number of variables and factors4 that enter into these formulas as a broad indicator of their ability to capture the complexity of regional cost variation. It should be made clear, however, that this is not an indicator of their administrative complexity, which is influenced by many factors other than the structure of the formulas themselves.

Several country examples illustrate the diversity of approaches to cost equalisation. During the period of the 14th Finance Commission, India accounted for the per capita cost variation faced by states using two straightforward variables: forest cover and surface area.5 This approach has the advantage of simplicity, which may allow for timely and accurate data inputs. However, such an approach may also exclude much of the variation driven by other factors. At the other end of the spectrum, Sweden uses a complex system comprised of 13 different cost models covering diverse functions of government. Australia uses similarly sophisticated methods. Likewise, Italy has adopted an intricate system of cost measurement that employs regression models to connect the extent of service provision with the service’s standardised cost. Naturally, such approaches rely on the availability of extensive and high quality data. Latvia, Lithuania, Norway and Switzerland lie in the middle with simpler cost-equalisation frameworks based around a few key variables.

Greater cost equalisation tends to be associated both with larger equalisation systems as a share of sub-central government spending (see Figure 2.9) as well as greater post-equalisation revenue disparities (see Figure 2.8). The immense complexity and large size of some cost-equalisation systems makes it imperative that their effects be measured on a comparable, cross-country basis. However, measures of the disparity of per capita revenue fail to capture the effects of cost equalisation. Cost equalisation is deeply connected to the broader objective of enabling equitable access to public services, beyond simply equalising SCG revenues. Accordingly, any assessment of its effect should take this into account. For example, recent OECD work has shown how public sector performance benchmarking can provide a yardstick for SCGs to evaluate their relative levels of service provision (Phillips, 2018[14]).

The factors and formulas used to determine the amount of the equalising transfer paid to a given jurisdiction lie at the heart of equalisation systems. Table 2.4 identifies the revenue and cost factors used to determine equalising transfers to SCGs. Some systems rely on a broad range of cost and revenue variables to determine the size of the transfer, while others consider only one principal input. Notably, while revenue factors typically relate to a few main revenue streams flowing to SCGs, cost factors can be numerous and diverse. Some countries (e.g. Australia, Italy, the Netherlands) consider dozens of cost factors based on demographic and geographic variables that are known to drive significant variation in SCG per capita expenses.

A reccurent concern is that fiscal equalisation systems that reward revenue losses may discourage SCGs from pursuing own revenue collection efforts (and therefore from applying the necessary tax rates to the appropriate base). Policymakers tend to respond by using a representative tax system (RTS) for fiscal equalisation, that is, an assessment of fiscal capacity based on the hypothetical revenues that would accrue to a jurisdiction were it to apply a certain (often cross-jurisdictional average) tax rate to an assumed tax base. In this way, jurisdictions are compensated for any decline in their capacity to raise own-source revenues rather than for direct falls in revenue itself. For example, Sweden uses an RTS based on municipal tax rates in 2003 as in principle these cannot be influenced by the policy choices of current local governments.

While an RTS can help to reduce the impact of an SCG’s tax rate and base choices on its equalisation entitlements, certain revenues have a tax base which may be particularly sensitive to factors beyond tax policy (e.g. zoning for property taxes) (de Joode, 2017[15]). One solution is to use a simplified set of taxes within the RTS in order to reduce the SCG’s incentive to make policy choices based on equalisation entitlements. Canada’s 2007 reform of equalisation sought to do this by vastly simplifying the revenues used to determine SCG fiscal capacity: a set of 34 taxes was reduced to a set of five, which aimed to enhance the policy neutrality of equalisation. Some systems contain more explicit disincentives to depress own-source revenues. Luxembourg’s approach to the horizontal aspect of its equalisation system replaces standardised revenues with yield: the ratio of revenues to the tax rate, which acts as a proxy measure of the breadth of the tax base. A municipality’s yield determines its contribution to fiscal equalisation, so cutting tax rates without a proportional drop in revenues drives up yield and hence the contribution requirement.

Conceptually, yield is similar to other measures of tax effort, such as that employed by India’s 15th Finance Commission. Yield is the ratio of tax revenues to the tax rate whereas tax effort is often quantified as the ratio of tax revenues to the tax base. Both measures depend positively on revenue and negatively on the tax rate or the tax base, respectively. Accordingly, both measures capture the effect of factors such as compliance with taxation, tax avoidance, or the administrative efficiency of tax collection to the extent that these factors affect revenues without affecting the measured tax base or tax rate.

Some jurisdictions (e.g. Belgium, Germany) use an approach that acknowledges the interaction between revenue policy choices and equalising grants by restricting equalisation to a level below 100%. This may be operationalised via equalisation coefficients that set explicit limits on the extent of equalisation. For example, Belgium compensates regions for 80% of the gap between actual per capita revenues and the mean per capita revenue. Lithuania fixes this amount at 90%. In contrast, other countries (particularly Gap-filling countries, e.g. Japan, Australia) aim for a complete equalisation of fiscal capacity relative to expenditure needs. In these cases, the design of the equalisation mechanism typically aims to mitigate any incentive to suppress own-source revenues.

Cost equalisation systems pose a more complex challenge, as evidenced by the range of factors that enter into the determination of SCG costs. Again, policymakers typically aim to quantify expenditure needs rather than direct costs. As explained above, sophisticated approaches are sometimes employed to assess need, based on economic or financial data adjusted to reflect certain considerations (see Figure 2.10). For example, Australia’s Commonwealth Grants Commission considers factors ranging from kilometres of road to pensioners. In the Netherlands, a standardised abatement is applied to a diverse set of cost factors.

Regression-based approaches to the determination of standardised costs have been implemented in some jurisdictions. At the state/local level, a model proposed for Massachusetts regressed per capita local spending on nine variables such as population density, age of housing stock and pupils per capita. The coefficients could be interpreted as the dollar change in expenditure requirement when a given cost factor increases by one-unit, with the error term capturing policy choices across jurisdictions. Moreover, only those coefficients with statistically significant values would be included when using the model to estimate standardised costs, with insignificant variables being discarded (Bradbury et al., 1984[16]). Similar, albeit more sophisticated, approaches can be observed in the Italian case, where regression-based analysis models local costs as a function of the optimal quantity of services offered, input prices and external or environmental factors (e.g. surface area) (SOSE – Soluzioni per il Sistema Economico S.p.a., 2014[17]). In effect, the Italian approach augments the Massachusetts model with measures of service output and input costs. Regression models of expenditure need are also employed in the Swedish equalisation system and during reviews of the Australian equalisation system (Commonwealth Grants Commission, 2015[18]).

Equalisation systems are not static entities, with regular reviews often built-in to their underlying legislative frameworks. Moreover, such reviews may lead to reforms of equalisation systems, sometimes concomitant with broader reforms of fiscal federalism. Japan reviews its equalisation system annually, while Canada adheres to a system of five-year reviews. Australia updates its GST allocations on an annual basis and thoroughly reviews the associated methodology every five years (Commonwealth Grants Commission, 2015[18]). India’s Finance Commission convenes every five years, setting the equalising formula for distributing tax revenues to states via a highly consultative process (Finance Commission of India, 2014[19]). In this context, it is critical for the policy maker to consider the design of the review process and the nature of possible reforms.

Several key factors have been identified that motivate the reform of equalisation systems (Figure 2.12).

Figure 2.13 summarises the questions faced by the policy maker when reviewing and reforming fiscal equalisation systems in order to systematically address the concerns enumerated above.

As illustrated in Figure 2.13, questions of system design are salient during the early stages of an equalisation review. However, the answers vary widely. Countries that choose to quantify fiscal capacity in terms of SCG own-source revenues alone will tend towards a system of pure revenue equalisation (e.g. Canada and Germany). On the other hand, countries that choose to incorporate variation in per capita cost must then choose between a system of distinct cost and revenue equalising grants (e.g. The Netherlands, Sweden and Switzerland), or an integrated gap-filling transfer (e.g. Australia, Japan and Korea).

The determination of fiscal capacity, and in turn equalisation modality, tends to hinge on two foundational questions:

  1. 1. Do SCGs have sufficient fiscal autonomy such that, were each to have equivalent per capita own-source revenues, each could make the choices necessary to deliver the desired bundle of services for its residents?

  2. 2. If not, which other drivers of inter-regional inequality, such as costs, need to be taken into account?

In response to the first question, Canada’s last review of fiscal equalisation returned with a firm “yes,” precluding the adoption of a system of cost equalisation which was seen as needless in the presence of very high provincial fiscal autonomy (Expert Panel on Equalization and Territorial Formula Financing, 2006[13]). Conversely, in the Dutch case, low degrees of municipal tax autonomy were described as a “bottleneck” with respect to equalisation reform (de Joode, 2017[15]), necessitating a compensatory increase in municipal tax autonomy concomitant with any reduction in fiscal equalisation. Moreover, the Netherlands’ equalisation system remains heavily cost-oriented (see Figure 2.6).

In contrast to Canada, Australia’s equalisation system takes place within the context of an extensive vertical fiscal gap. This means that, in principle, own-source revenues are insufficient to cover state-level costs. Accordingly, as part of each five-year review of the equalisation system, the Commonwealth Grants Commission must undertake a detailed analysis of the structural factors underlying cost inequality between Australian states. This enables the development of policy-neutral cost indicators, which allow for the measurement of fiscal capacity.

Data availability is a key consideration when designing equalisation formulas. Australia’s Commonwealth Grants Commission has identified the contemporaneity of data as important to ensuring the functionality of the equalisation system (Commonwealth Grants Commission, 2015[18]). This, combined with the vast number of variables that factor into its estimation of expenditure need, makes for a system that is highly dependent on a wide array of timely data and the infrastructure required to produce them. However, as evidenced by the broad variation in the complexity of cost-equalisation models (see Figure 2.10), this is by no means the only approach. In the guidance of the 14th and 15th Finance Commissions, India’s assessment of costs depended on only two variables: state surface area and forest cover.6 Latvia and Lithuania propose a middle ground, with four and nine cost variables, respectively, entering into their cost-equalisation formulas.

Once the equalisation modality has been determined, the question of the rate of equalisation arises. The rate of equalisation denotes the extent to which an equalisation system rectifies inter-regional disparities. In practice, this rate refers to a coefficient that quantifies how close equalising transfers bring a given SCG to mean per capita revenues (in the case of revenue-equalising systems) or expenditure needs (in the case of systems with a cost equalising component). It is important to remember that, in the absence of offsetting policy measures, this coefficient has two equivalent interpretations: first, it defines the rate at which a jurisdiction is compensated for each dollar lost when its fiscal capacity falls relative to the mean or its expenditure need. Second, it defines the rate at which a jurisdiction’s equalising transfers are clawed-back for each dollar gained when fiscal capacity rises.

Among countries that specify an explicit rate of equalisation, this rate typically varies between 60% and 100% (see Table 2.1). In fact, there is evidence from German municipalities that higher equalisation rates may promote more efficient taxation choices by local government when they face greater clawbacks in the presence of rising own-source fiscal capacity (Holm-Hadulla, 2020[20]). In the case of gap-filling systems, the difference between standardised own-source revenues and standardised costs is generally filled completely.

The rate of equalisation is closely connected to the funding source of equalising transfers, particularly where equalisation is fully (e.g. Lithuania and Ireland) or partially horizontal (e.g. Germany and Sweden). In the presence of horizontality, the rate of equalisation reflects not only the extent of compensation to poorer SCGs, but also the contribution rate faced by wealthier SCGs. Generally, wealthier SCGs are allowed to keep some of their additional revenues above the mean in order to promote tax effort and local economic development. For example, Swedish municipalities only contribute to equalisation once their per capita revenues exceed 115% of the standardised mean. In cases where equalisation is purely vertical (e.g. Japan and Korea), it is sometimes funded through an earmarked portion of central revenues. In Japan, equalisation is funded via a dedicated revenue stream consisting of 33.1% of income tax and corporate tax revenues, 50% of liquor tax revenues, and 20.8% of consumption tax and local corporation tax revenues.

The extent of equalisation often prompts a lively debate because it is seen as having direct implications for subnational tax effort, and in some cases, economic growth itself. However, it is important to note that such effects are challenging to observe directly.7 Moreover, some national reviews have concluded that equalisation systems are unlikely to have an adverse impact on economic development. Both Australia and Sweden’s last major reviews of fiscal equalisation concluded that there was no discernible evidence of such effects. While Sweden pursued a reform in 2014 that was designed to allow wealthy municipalities to retain more of their own-source revenues, this change was reversed in 2016. Subsequent reviews have focussed on increasing redistribution in favour of rural and remote regions and those with lower socio-economic status.

Equalisation is not always well adapted to replacing all other forms of inter-governmental transfers. Where equalising transfers are dependent on buoyant revenue streams they may illustrate the pro-cyclicality of government revenue more generally. Moreover, the frequent use of lagged variables in their underlying formulas may make them less responsive to emerging crises. For these reasons, sudden events such as the COVID-19 pandemic may require different instruments to mitigate asymmetric impacts (see Box 2.2). In addition, where an equalisation system replaces other transfers, it may need to adopt indicators that capture the criteria upon which the erstwhile transfers were based. As a result, equalisation formulas may become increasingly complex and difficult to interpret.

In view of this, some countries have taken the approach that equalisation systems should not supplant all other grants. For example, the 13th Finance Commission of India highlighted the important role of “grants-in-aid” noting that that they allow for the compensation of “cost disabilities faced by many states which are possible to address only to a limited extent in any devolution formula” (Finance Commission of India, 2009[22]). Similarly, Australia’s Commonwealth Grants Commission noted that equalisation alone “cannot overcome the disadvantage experienced by some indigenous communities. Where additional measures are required, they would best be undertaken outside the HFE (i.e. horizontal fiscal equalisation) system and excluded from it” (Australian Government, 2012[23]). Other examples of financing programmes designed to operate outside of the fiscal equalisation system include Canada’s Territorial Financing Formula and the grants received by the Brussels capital region outside of Belgium’s National Solidarity Mechanism.

Adjustments to equalisation systems typically create winners and losers. This often necessitates a period of transition as changes are phased-in. Such transition periods are typically characterised by concessions along the following two dimensions: (a) compensation to losing jurisdictions as a share of their loss arising from the change; and (b) a transition period of a fixed duration. The first dimension is conceptually equivalent to the determination of the equalisation rate itself. For example, if the equalisation rate is increased from 0.9 to 0.95, a jurisdiction that sees a rise in measured fiscal capacity in the following period will see a greater absolute loss in equalising transfers under the new equalisation rate.8 However, if the jurisdiction receives compensation to mitigate the difference, the equalisation rate is effectively reduced. The second dimension, the duration of transition, allows jurisdictions to increase own-source revenues to make up for any adjustment to the equalisation they receive.

Some recent transition measures have lasted a decade or more in duration and provided substantial protection to losing jurisdictions. In Switzerland, transitional grants arising from the 2008 reform amounted to about 7% of total equalising transfers in 2016. These transfers will be reduced by 5% a year until 2036 when they will be eliminated and they are not paid to cantons whose measured fiscal capacity exceeds the mean (Administration fédérale des finances, 2012[24]; 2019[25]). Sweden paid similar transitional grants to jurisdictions that saw losses arising from its 2014 reform. However, because this reform was reversed in 2016, the sums paid recently have been fairly minor as a portion of total equalising transfers, constituting about 0.1% of total equalisation funds in 2018. Italy’s approach to transition saw the old and the new equalisation formulas co-existing for a period of several years, with the percentage of local government funding distributed by the new system increasing every year, rising from 20% in 2015 to 45% in 2018 to reach 60% in 2022. It will then rise by 5% per year, reaching 100% in 2030. Finally, Canada introduced a transitional measure known as Total Transfer Protection (TTP) in the fiscal year 2010-11. It followed the move in 2009 to a fixed funding envelope for equalisation payments which grows at the rate of nominal GDP. The TTP served as a backstop ensuring that no province received less in combined federal transfer payments than in the previous year, regardless of any changes that had taken place. The TPP was in place until 2013-14 (Feehan, 2014[26]; Nadeau, 2014[27]), and amounted to roughly 5% of equalisation payments over the period.

Where equalisation systems have been examined within the scope of OECD country surveys, several measures have been recommended to respond to the challenges identified previously (Table 2.5).

The challenges of cost equalisation are reflected in the recommendations of country surveys. France and Japan, which share robust cost equalisation systems, were advised to separate measures of expenditure need from actual spending in order to enhance policy neutrality. This could include the use of standard unit costs. Additionally, surveys have urged the simplification of equalisation formulas as a way of achieving greater transparency (e.g. Sweden, Italy and Japan). Finally, surveys have noted the need to ensure that cost equalisation formulas capture relevant variables as drivers of expenditure evolve in the face of changing demographics (e.g. Switzerland and Sweden).

In some cases, country surveys have noted that equalisation systems are not fulfilling their intended purpose and may therefore be exacerbating disparities or undermining SCG fiscal autonomy. For example, in France 97.5% of municipalities received at least one type of vertical equalisation transfer in 2018 (OECD, 2019[29]) suggesting that the scope of the system had gone beyond the traditional purview of equalisation and subsumed other types of transfers. Moreover, the use of lump-sum payments and indicators based on past tax-raising ability may have been perpetuating inequalities. In response, improved targeting and the use of horizontal equalisation was recommended, based partially on past OECD work that suggests horizontal equalisation is more equalising per dollar spent (Blöchliger and Charbit, 2008[7]). Similarly, the use of targeted support to poorer regions concomitantly with the introduction of horizontal equalisation was recommended in Austria. In the interest of improving sub-central fiscal autonomy, Austria was also encouraged to allow SCGs to retain more own-source revenues, as were Sweden and Japan.

Table 2.5 also reflects the principal outcomes of major reviews and reforms of equalisation systems. It is important to illustrate how these measures connect to the challenges identified above that motivate their implementation (see Figure 2.12), by linking motivations for equalisation reform to typical policy responses. In general, reforms centre on defining a clear set of rules in order to enhance clarity and predictability for SCGs.

The Italian case is illustrative of a comprehensive equalisation reform that covers many of the elements observed in other countries. Introduced in 2011, the equalisation system was born out of a broader 2009 reform of fiscal federalism and encompassed a set of measures that aimed to address the opacity that had emerged within local government finance. To deal with high levels of regional disparity, the new equalisation system was designed to replace other transfers which were not targeted based on local fiscal capacity. This simultaneously enhanced subnational fiscal autonomy as the equalising transfers were non-earmarked. To avoid perverse incentives, the system employed standard unit costs rather than past spending when determining expenditure need. In a similar vein, the use of a formula-based system aimed to increase transparency from the perspective of local governments. In contrast, the former system of transfers tended to be negotiated between levels of government based on historical costs rather than objective allocation criteria (OECD, 2012[30]).

As in Italy, other countries’ reforms of equalisation have tended towards rules-based systems with fewer discretionary elements, meaning that allocations are based on clearly articulated formulas and set criteria. In comparison with ad-hoc negotiations between levels of government, this can ensure predictability for SCGs (Expert Panel on Equalization and Territorial Formula Financing, 2006[13]). A current reform proposal in Switzerland aims to guarantee a particular equalisation rate to cantons, with funding determined formulaically as opposed to on a periodic basis (Weber, 2019[31]). Likewise, the last major reviews of the Australian and Canadian equalisation systems emphasised the importance of rules-based systems (Australian Government, 2012[23]; Expert Panel on Equalization and Territorial Formula Financing, 2006[13]). Such rules-based approaches, however, may be accompanied by a concomitant reduction in the influence and power of ministries that were formerly responsible for advancing their policy agendas through the disbursement of local government grants. This was the case in Japan where concerns were raised that the reduction in earmarked grants in favour of the equalising local allocation tax would reduce the ability of ministries to pursue policy objectives (OECD, 2016[32]).

The Italian reform’s objective of enhancing local fiscal autonomy can be observed across reviews and reforms of other equalisation systems. For example, Switzerland’s 2008 reform aimed to enhance local fiscal autonomy through reducing the use of earmarked transfers in favour of non-earmarked, equalising transfers (Administration fédérale des finances, 2012[24]). Similarly, two recent reviews of Ireland’s equalisation system recommended allowing local governments to retain 100% of the property tax revenue they generate locally (Thornhill, 2015[33]; Department of Finance, 2019[34]). While this would enhance local fiscal autonomy, it would also eliminate the horizontal component of the equalisation system, which is presently funded through a 20% share of local authority property tax revenues. Consequently, this would need to be replaced by vertical transfers. Conversely, the Swiss equalisation reform included a renewed horizontal revenue equalisation component amounting to between two-thirds and four-fifths of the vertical component. Finally, Sweden’s 2014 reform of fiscal equalisation aimed to allow municipalities that would pay into the horizontal component of the system to retain more of their own-source revenues, in the interest of encouraging local economic development. However, this change was reversed two years later.

Japan’s “trinity reforms” provide a further study of equalisation reform within the broader context of enhancements to subnational fiscal autonomy, including the use of equalising transfers as a replacement for earmarked grants. In 2004, a revenue transfer of JPY 3 trillion to the local level replaced over one hundred earmarked subsidies. A further JPY 512 billion in earmarked subsidies was bundled into a general subsidy in 2011 (OECD, 2016[32]). Importantly, the principal mechanism by which non-earmarked funds are allocated to local governments, the local allocation tax (LAT), has an explicit equalising function. While policymakers aimed to base the LAT on equalising, policy-neutral criteria that are less influenced by the extent of past infrastructure spending (e.g. standardised unit costs), critics have noted that the revenues transferred to the local level were not sufficient to compensate for the reduction in grants. Accordingly, some subsequent reforms have focussed on raising subnational fiscal capacity through increases in local consumption taxes (OECD, 2017[35]; Tokyo Metropolitan Government, 2018[36]).

The high levels of complexity associated with some equalisation systems, and the formulas underlying them, have at times motivated simplification efforts. In Luxembourg, the principal outcome of the 2017 equalisation reform was the unification of the two former equalising transfers into a single transfer: the Fonds de dotation globale des communes (FDGC). The FDGC adheres to a single set of criteria, in contrast with the differing criteria of the two previous transfers. This drove a reduction in post-equalisation disparities since the new criteria were more equalising, and applied more broadly to funds transferred to the local level (Laurent, 2018[37]). Similarly, simplicity was one of the objectives of Japan’s trinity reforms, which aimed to favour “objective” cost equalisation criteria applied broadly to a single larger grant as opposed to many, smaller earmarked grants (OECD, 2016[32]). On the revenue measurement side, Canada’s experience simplifying its RTS is instructive when it comes to finding a compromise between accurately capturing true fiscal capacity and the excess complexity created by attempts to measure all potential revenue sources. In 2007, Canada’s RTS, which had formerly aimed to estimate provincial fiscal capacity based on thirty-three revenue sources, was scaled back to five revenue sources (Feehan, 2014[26]).

Complex cost equalisation formulas that depend on dozens or hundreds of variables demand extensive data inputs. While such systems are often the product of years of careful efforts to capture the structural drivers of inter-regional cost disparities, they require continuous updating to ensure that they remain relevant and do not inadvertently perpetuate inequalities. In the Netherlands, this is handled through Period Maintenance Reports (Periodiek Onderhoudsrapport) which examine annually whether the amounts municipalities actually spend in designated areas corresponds to the amount that they would be expected to spend for the purposes of distributing the Municipal Fund (de Joode, 2017[15]). In Sweden, a recent review of the equalisation systems suggested that it was failing to address the costs faced by rural municipalities, though it continued to take into account some criteria that were no longer relevant. In response, the review recommended that the equalisation formula be adjusted to better account for geography and demographics, extended to include library operation costs, and perhaps further extended to take account of the costs of voluntary services (Riksrevisionen, 2019[38]). Other ongoing work in the Swedish equalisation system focuses on developing inputs to the formula that would be automatically updated with the most recent data.

Australia’s equalisation methodology is updated annually and reviewed every five years to ensure that the determination of state-level cost disabilities adheres to four key principles: accurately reflecting what States collectively do, policy neutrality, practicality, and contemporaneity, or ensuring that relative GST shares are appropriate to the year in which they are applied. For example, this may involve modifications to indices that measure remoteness to better capture how remote a city is vis-à-vis the costs it faces. On the revenue side, this could involve phasing in the measured increases in tax-raising capacity derived from natural resource extraction, with the aim of ensuring that those states that have seen revenue increases from natural resources are not unfairly penalised by a concomitant drop in GST share (Commonwealth Grants Commission, 2015[18]).


[25] Administration fédérale des finances (2019), Rapport technique sur la péréquation financière: Bases pour la période 2020-2023, Confédération suisse.

[24] Administration fédérale des finances (2012), Péréquation financière nationale: Renforcement du fédéralisme, Confédération suisse.

[23] Australian Government (2012), GST Distribution Review--Final Report.

[7] Blöchliger, H. and C. Charbit (2008), “Fiscal equalisation”, OECD Journal: Economic Studies, Vol. 2008/1, https://dx.doi.org/10.1787/eco_studies-v2008-art8-en.

[3] Boadway, R. (2003), The Theory and Practise of Equalization, Queen’s Economics Department Working Paper No. 1016.

[4] Boadway, R. and A. Shah (2009), Fiscal Federalism: Principles and Practice of Multiorder Governance, Cambridge University Press, New York City.

[16] Bradbury, K. et al. (1984), “State aid to offset fiscal disparities across communities”, National Tax Journal, Vol. 37/2, pp. 151-170.

[18] Commonwealth Grants Commission (2015), Report on GST sharing relativities 2015 review, Australian Government.

[9] Coppel, J. (2018), The economic impacts of HFE: lessons from Australia, 14th Annual Meeting of the Network on Fiscal Relations across Levels of Government.

[17] Danilo Ballanti, F. (ed.) (2014), The Italian way towards standard expenditure needs.

[15] de Joode, G. (2017), Considering Differences, Ministry of the Interior and Kingdom Relations (the Netherlands), http://www.rijksoverheid.nl/documenten/rapporten/2017/06/01/rekening-houden-met-verschil.

[34] Department of Finance (2019), Review of Local Property Tax: The report of the Interdepartmental Group, An Roinn Airgeadais-Department of Finance Ireland.

[6] Dougherty, S. and K. Forman (2021), “Evaluating fiscal equalisation: Finding the right balance”, OECD Working Papers on Fiscal Federalism, No. 36, OECD Publishing, Paris, https://dx.doi.org/10.1787/253da2b8-en.

[8] Dougherty, S., A. Renda and L. von Trapp (2021), “Independent oversight bodies: lessons from fiscal, productivity and regulatory institutions”, in Making Institutions Work, HSRC Press.

[21] Dougherty, S. et al. (2020), “COVID-19 and Fiscal Relations across Levels of Government”, OECD Policy Responses to Coronavirus (COVID-19), OECD Publishing, Paris, https://doi.org/10.1787/ab438b9f-en.

[13] Expert Panel on Equalization and Territorial Formula Financing (2006), Achieving a national purpose: putting equalization back on track., Department of Finance Canada.

[10] Federal Ministry of Finance (2018), The Federal Financial Equalisation System in Germany, Bundesministerium der Finanzen [Federal Ministry of Finance (Germany)].

[26] Feehan, J. (2014), Canada’s equalisation formula: peering into the black box... and beyond., The School of Public Policy, University of Calgary, https://www.policyschool.ca/wp-content/uploads/2016/03/feehan-equalization.pdf.

[39] Finance Commission of India (2021), Finance Commission in COVID Times, Fifteenth Finance Commission of India.

[19] Finance Commission of India (2014), Fourteenth Finance Commission Report.

[22] Finance Commission of India (2009), Thirteenth Finance Commission Report.

[28] Forman, K., S. Dougherty and H. Blöchliger (2020), “Synthesising good practices in fiscal federalism: Key recommendations from 15 years of country surveys”, OECD Economic Policy Papers, No. 28, OECD Publishing, Paris, https://dx.doi.org/10.1787/89cd0319-en.

[20] Holm-Hadulla, F. (2020), “Fiscal equalization and the tax structure”, Regional Science and Urban Economics, Vol. 81, p. 103519, https://doi.org/10.1016/j.regsciurbeco.2020.103519.

[5] Kim, J. and S. Dougherty (eds.) (2018), Fiscal Decentralisation and Inclusive Growth, OECD Fiscal Federalism Studies, OECD Publishing, Paris/Korea Institute of Public Finance, Seoul, https://dx.doi.org/10.1787/9789264302488-en.

[37] Laurent, E. (2018), “La réforme des finances communales de 2017”, in Bulletin BCL 2018-1, Banque Centrale du Luxembourg.

[27] Nadeau, J. (2014), 2014-2015 Federal Transfers to Provinces and Territories, http://www.pbo-dpb.gc.ca/web/default/files/files/files/TransferPayments_EN.pdf.

[29] OECD (2019), OECD Economic Surveys: France 2019, OECD Publishing, Paris, https://dx.doi.org/10.1787/a0eee144-en.

[35] OECD (2017), Multi-level Governance Reforms: Overview of OECD Country Experiences, OECD Multi-level Governance Studies, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264272866-en.

[32] OECD (2016), OECD Territorial Reviews: Japan 2016, OECD Territorial Reviews, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264250543-en.

[1] OECD (2013), “Fiscal equalisation: A key to decentralised public finances”, in Fiscal Federalism 2014: Making Decentralisation Work, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264204577-7-en.

[30] OECD (2012), “Italy: Law 42 on fiscal federalism”, in Reforming Fiscal Federalism and Local Government: Beyond the Zero-Sum Game, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264119970-10-en.

[12] OECD (2006), OECD Economic Surveys: Germany 2006, OECD Publishing, Paris, https://dx.doi.org/10.1787/eco_surveys-deu-2006-en.

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

[14] Phillips, L. (2018), “Improving the Performance of Sub-national Governments through Benchmarking and Performance Reporting”, OECD Working Papers on Fiscal Federalism, No. 22, OECD Publishing, Paris, https://dx.doi.org/10.1787/ffff92c6-en.

[38] Riksrevisionen (2019), Det kommunala utjämningssystemet [The Municipal equalisation system], Riksrevisionen [Swedish National Audit Office].

[11] SOSE (2018), The Case of Italian Municipalities (an example of a fiscal equalisation system), https://www.opencivitas.it/sites/default/files/Allegati/Presentation%20Lithuania_II.pdf.

[33] Thornhill, D. (2015), Review of the Local Property Tax (LPT).

[36] Tokyo Metropolitan Government (2018), Tokyo’s Financial System, http://www.metro.tokyo.lg.jp/ENGLISH/ABOUT/FINANCIAL/financial01.htm.

[31] Weber, W. (2019), Recent Developments in Intergovernmental Relations, Proceedings of the 2019 annual meeting of the OECD Network on Fiscal Relations, http://oe.cd/fiscalnetwork.


← 1. Australia’s equalisation system is classified as gap-filling. Gap-filling systems require that both SCG revenues and costs be assessed to determine the size of the “gap” to be filled by equalisation.

← 2. Note this figure includes equalising transfers to regions only.

← 3. In order to disentangle the cost and revenue components of gap-filling systems, the hypothetic horizontal transfer necessary to bring to all jurisdictions to average per capita revenue was computed as the revenue equalising component of the transfer. Then, the hypothetic transfer required to bring all jurisdictions to their pre-determined level of financial need was computed as the cost equalising component. The magnitude of these two components was compared to determine the relative contributions of cost and revenue equalisation to the total transfer.

← 4. A factor is a driver of cost disparity, such as a public service or a geographic feature, rather than a variable which enters directly into a formula. Factor are quantified by one or more variables.

← 5. These two criteria were maintained in the formula employed by the 15th Finance Commission. There is some discussion as to whether the forest cover criterion should be interpreted as compensation for the cost disability imposed by forest or an incentive to reward the provision of ecological services (Finance Commission of India, 2021[39]).

← 6. The Indian Finance Commissions sometimes describe population as a cost criterion, however for the purposes of this cross-country analysis, transfers with an equal per capita impact are not considered equalising since they do not compensate for per capita cost or revenue disparities between jurisdictions.

← 7. For example, the Australian Government’s GST distribution review concluded “the current system creates perverse theoretical incentives in some instances, but there is little evidence that they have any effect in the real world. In particular, there is no evidence that HFE acts as a material disincentive to State tax reform” (2012[23]).

← 8. For example, under a theoretical revenue equalisation scheme, an SCG at 80% of mean fiscal capacity will be compensated to 98% of mean fiscal capacity at an equalisation rate of 0.9 and 99% of mean fiscal capacity at an equalisation rate of 0.95. Should its fiscal capacity increase to 90%, it will be compensated to 99% of mean fiscal capacity at an equalisation rate of 0.9 and 99.5% of fiscal capacity at an equalisation rate of 0.95. The loss in equalising transfers when fiscal capacity increases amounts to 9 percentage points in the presence of an equalisation rate of 0.9 and 9.5 percentage points in the presence of an equalisation rate of 0.95. In the latter case, the SCG must raise more own-source revenue to compensate for the measured increase in its fiscal capacity.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2021

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at https://www.oecd.org/termsandconditions.