OECD Employment Outlook 2026

Geographic Disparities in Jobs and Incomes

Report

OECD Employment Outlook

7 July 2026

Available in:

English
français

Download PDF

2. Workers, households, regions: Geographic disparities in labour market outcomes and the distribution of income

Copy link to 2. Workers, households, regions: Geographic disparities in labour market outcomes and the distribution of income

Sebastian Königs

Sérgio Pinto

Javier Terrero-Dávila

Juan Andrés Álvarez Mejía

Abstract

Where people live shapes their employment opportunities and living standards. This chapter examines geographic inequalities in labour market outcomes and, in turn, disposable household incomes and income mobility across small regions in OECD countries. It analyses the extent and drivers of regional disparities in employment rates, relating them to the geographic concentration of economic activity and the composition and characteristics of the local population, and discussing the role of labour mobility across regions. It examines regional disparities in median disposable household incomes, their relationship to regional GDP per capita, and their links to regional labour market outcomes – both in the cross-section and over time. It quantifies the contribution of between-region inequalities to overall income inequality across households and examines what regional disparities imply for the prospects of individuals to move up the income ladder over time.

In Brief

Copy link to In Brief

Key findings

Where people grow up and live shapes their employment opportunities, career progression, and ultimately their living standards. Large inequalities across regions within countries carry real costs for economic growth, as the potential of lagging regions remains underexploited. They also pose challenges to social cohesion if significant parts of the population feel – and in many cases are – falling further behind relative to their peers in more prosperous places.

This chapter examines geographic inequalities in labour market outcomes and, in turn, the distribution of disposable household incomes and income mobility across small regions in OECD countries – at “Territorial Level 3” (TL3) in the OECD classification (e.g, Census divisions in Canada, Départements in France). It complements statistics from existing OECD databases with newly collected data drawn from both administrative and survey sources, offering – for the first time – a granular picture of the interplay between disparities in labour market outcomes, the distribution of household incomes and people’s income mobility at this geographic scale.

The key findings are as follows:

People’s employment prospects are shaped by where they live. In over half of OECD countries, employment rates across small regions vary by more than 20 percentage points (p.p.). Regional disparities in unemployment are equally pronounced: on average, regions in the worst‑performing quintile have unemployment rates more than twice as high as those in the best‑performing quintile – a ratio that exceeds four in Italy, Belgium, Canada and the Slovak Republic.
Regional disparities in employment rates remain substantial even though there has been a period of moderate convergence. In most countries, low-employment regions have had stronger employment growth than high-employment regions since the early 2010s. This has narrowed regional dispersion in employment rates by around 50% in Slovenia, Czechia, Estonia, Ireland, the Slovak Republic and Costa Rica, and more modestly in most other countries. Only 8 out of 29 countries with available data exhibit widening dispersion in employment rates.
These disparities do not simply reflect differences in who lives where, but also the opportunities that regions have to offer. Differences in local population composition – by age, gender, educational attainment and household structure – account for between one‑third and one‑half of the employment rate gap between high- and low-employment regions. Regional economic performance is a key driver of people’s employment opportunities, with jobs in high‑value‑added services concentrated in a small number of metropolitan regions, often the capital-city region.
Labour mobility does little to narrow disparities in labour market outcomes. People move from low- to high-employment regions, but at a pace much too slow to make labour mobility alone a solution for meaningfully reducing labour market disparities. As regions with weaker labour markets are often geographically close to each other, people seeking better job prospects often need to relocate a considerable distance. Since people who leave low-employment regions often share characteristics associated with stronger employment prospects, labour mobility risks reinforcing existing labour market disparities.
Regional disparities in disposable incomes are substantially narrower than disparities in GDP per capita. Median disposable household incomes vary most across regions in large countries with many regions, such as Australia and Canada, and countries where overall income inequality is high, such as Chile and Israel. In those countries, the highest and lowest-income regions differ by a factor of two or more, while in several Central European and Nordic countries, this ratio is as low as 1.1‑1.2. Across all countries, regional disparities in disposable incomes are substantially narrower than disparities in GDP per capita, and in some countries the two measures are only weakly correlated. This underlines that GDP per capita, while commonly used as a proxy for income, is a poor measure of regional living standards.
Regional disparities in incomes are closely linked to those in employment. In most countries, regional median disposable household incomes and employment rates are positively correlated. Trends tell the same story: 10 out of the 14 countries where regional employment gaps narrowed since the early 2010s also saw regional income dispersion decline; all three countries where regional employment gaps widened also saw an increase in regional income dispersion. This underlines the role of labour earnings as the primary channel through which regional economic conditions impact household incomes. It also suggests that policies targeting regional labour market disparities can be effective at reducing regional disparities in living standards.
Incomes in capital regions are higher and more unequally distributed than elsewhere, as these regions attract a disproportionate share of high-income households. Other metropolitan regions beyond the capital record on average median incomes and inequality levels broadly in line with the national values; non-metropolitan regions have on average both lower median incomes and inequality.
However, most household income inequality occurs within regions, not between them. Even in countries where regional income disparities are wide, between-region differences account for only a small fraction of overall inequality, typically around 5%. Within any given region, disposable household incomes vary widely, and it is these differences across households within regions that strongly dominate, representing the remaining 95% of overall income inequality.
People in low-income regions have weaker prospects for upward mobility. Even though regional income disparities represent only a modest fraction of total income inequality, compounding disadvantages mean that people living in low-income regions have lower chances of climbing the income ladder over time, suffer a greater risk of income stagnation, and face a higher likelihood of remaining stuck at the bottom of the distribution.

Introduction

Copy link to Introduction

Where one lives shapes job opportunities and living standards. Firms are drawn to the better infrastructure, connectivity, services and human capital that prosperous regions offer (OECD, 2023[1]; OECD, 2025[2]). This creates a feedback loop in which these regions attract more and higher-productivity employers, while other regions fall behind (Moretti, 2012[3]; Bilal, 2023[4]). Since employment is the primary source of income for most households, and the main channel for people’s upward mobility, regional disparities in labour market outcomes can translate into unequal opportunities for income progression over the life course, and ultimately into persistent disparities in living standards. While people can move to take up opportunities elsewhere, financial and non-financial barriers, such as housing costs, the desire to preserve social networks, and high transportation costs constrain geographic mobility in practice (Ganong and Shoag, 2017[5]; Spring, Gillespie and Mulder, 2023[6]; OECD, 2025[7]). And since it is often the young and more educated who do leave economically lagging areas (De la Roca, 2017[8]; Causa and Pichelmann, 2020[9]), geographic mobility can reinforce regional inequalities.

Persistent regional inequalities carry large economic and societal costs. If economic opportunities are concentrated in few locations, this makes economies less resilient and leaves growth potential in lagging regions untapped (OECD, 2012[10]; Iammarino, Rodriguez-Pose and Storper, 2018[11]; OECD, 2023[1]). With the right regional development policies (OECD, 2023[12]; OECD, 2025[13]), these regions could contribute far more to national output. Among residents of lagging regions, the perception of being economically left behind can undermine trust in democratic and economic institutions and erode social cohesion (McKay, Jennings and Stoker, 2021[14]; Mitsch, Lee and Ralph Morrow, 2021[15]; Stroppe, 2023[16]).

This chapter studies regional disparities in labour market outcomes and, in turn, disposable household incomes and income mobility in OECD countries. It focusses on disparities across small regions – “Territorial Level 3” (TL3) in the OECD classification – a geographical level below that available from cross‑country microdata, such as the Eurostat distributions of the EU‑LFS and the EU‑SILC, in many national labour force surveys (LFS) and most household surveys across OECD countries. It draws on a major data collection effort that complements statistics from existing OECD databases with a range of newly collected indicators drawn from both administrative and survey microdata, offering – for the first time – a granular picture of the interplay between disparities in labour market outcomes, the distribution of household incomes and people’s income mobility at this geographic scale (for a detailed overview of the data sources, see Box 2.1 and Annex 2.A to Annex 2.C). This level of granularity matters: smaller geographical units reveal disparities that larger units may mask by averaging out local variation. The high granularity of the data also makes it possible to map urban-rural disparities in labour market outcomes and household income distributions, using the OECD typology classifying TL3 regions by their access to cities (Fadic et al., 2019[17]). The chapter covers the period from the early 2010s onwards, tracing regional disparities from shortly after the peak of the Global Financial Crisis (GFC) to the present.

This is the first Employment Outlook chapter on the geography of labour markets in two decades – since a chapter in the 2005 edition examined the persistence of regional disparities in employment and the role of geographic mobility (OECD, 2005[18]). It directly draws on the results from several streams of OECD work studying geographic disparities in labour market outcomes (OECD, forthcoming[19]), the distribution of household incomes (OECD, forthcoming[20]) and regional income convergence (European Commission, 2024[21]), and the accessibility of essential services (Almeida et al., 2024[22]). The chapter builds also on a range of earlier OECD work on labour markets at TL3 and local levels (Cho and Jeon, 2025[23]; OECD, 2025[7]; Ahrend et al., 2026[24]; Ahrend et al., 2026[25]); on how megatrends have been shaping the geography of labour markets (OECD, 2018[26]; OECD, 2023[27]; OECD, 2024[28]); and on regional disparities in GDP per capita and disposable household income per capita using regional accounts data (OECD, 2023[1]; OECD, 2024[29]).1 The chapter contributes to broader OECD efforts to integrate administrative employment and income data into cross-country analysis at subnational levels (OECD, 2023[30]; OECD, 2025[2]; OECD, 2025[7]; Ahrend et al., 2026[24]).

The remainder of the chapter is structured as follows. Section 2.1 documents large and persistent differences in labour market outcomes, showing that these reflect not only people’s characteristics but also disparities in the opportunities that places offer. It also highlights the limits of geographic mobility as a solution to eliminate these disparities. Section 2.2 documents disparities in the distribution of disposable household incomes, examining both income levels and inequality across household within regions. It shows that regional income disparities are substantially narrower than corresponding disparities in GDP per capita, and closely linked to regional disparities in employment. Section 2.3 shows that lower-income regions offer their residents fewer opportunities for upward income mobility. Section 2.4 provides some concluding remarks.

Box 2.1. Overview of data sources

Copy link to Box 2.1. Overview of data sources

The analysis in this chapter builds on a major data collection effort that has complemented regional labour market statistics from the OECD Database on Regions, Cities and Local Areas with additional indicators on labour market outcomes from national sources, and compiled new indicators on the distribution of household incomes and income mobility over people’s lives – all at the level of small TL3 regions. Specifically, the different parts of this chapter use data from the following sources:

The analysis of regional disparities in labour market outcomes, presented in Section 2.1, draws on statistical indicators from the OECD Database on Regions, Cities and Local Areas (https://www.oecd.org/en/topics/regions-cities-and-local-statistics.html; for 18 countries). These data were complemented with comparable indicators provided by national statistical offices (for 13 countries) or calculated directly from national microdata (for Chile);
The analysis of regional disparities in disposable household incomes, presented in Section 2.2, builds entirely on newly collected data on the level and distribution of disposable household incomes across regions, and across households within regions. For 15 countries, these data were provided by national statistical offices, drawn from survey or administrative microdata; for 9 countries, results were calculated by the OECD from microdata from administrative sources (for 4 countries), the Luxembourg Income Study (LIS) database (for 4 countries) and national survey microdata (for Chile);
The analysis of regional disparities in people’s income mobility, presented in Section 2.3, presents calculations based on administrative income microdata for a selection of OECD countries, carried out by the OECD (for 3 countries) and provided by national authorities (for 3 countries).

Certain pieces of analysis on labour market outcomes in Section 2.1 are equally based on calculations carried out by the OECD using microdata from administrative sources or the LIS database.

Since the results presented in this chapter are based on data from various different sources – primarily administrative registers (tax, social security, population and public-employment-service records); in some cases population censuses or large surveys – they are somewhat less harmonised than official OECD labour market and income statistics derived from highly standardised labour force and household surveys. Considerable care has been taken in the preparation of this chapter to identify sources of data that are sufficiently comparable, to harmonise the microdata as far as possible, and to document any remaining differences in coverage and definitions. For a detailed overview of the data sources used in the various parts of the analysis, see Annex 2.A to Annex 2.C.

A general challenge when comparing regional outcomes across countries is that the size and population of TL3 regions can vary substantially – even between countries with similar populations and territorial extension. This can affect the results in cross-country comparisons as, ceteris paribus, measured regional inequalities are usually larger for smaller geographical units of analysis. There is no obvious solution to this issue. No alternative territorial classification exists that would offer more homogeneous geographical units at a comparable level of granularity. Commuting zones are an obvious alternative, but they are unavailable in many OECD countries and, where they exist, they also vary in size and population across countries due to country-specific methodologies. Functional Urban Areas (FUAs), while constructed using a harmonised methodology, are limited to urban areas and do not cover the entire territory.

A methodological point to keep in mind is that most labour market indicators presented in this chapter are computed at people’s place of residence, not their place of work. Commuting across regional boundaries, and across national borders, can distort regional labour market statistics as workers contribute to economic activity in a region that is different from their region of residence. Earlier OECD analysis has addressed within-country commuting by aggregating metropolitan TL3 regions into a single unit in cases where at least 50% of the population of each constituent region lived in the same FUA of at least 250 000 inhabitants (OECD, 2024[29]). This chapter has not adopted a comparable approach: while the labour market statistics presented in Section 2.1 could have indeed been easily aggregated across metropolitan regions, the same does not hold for the regional median incomes and Gini coefficients presented in Section 2.2. For consistency, the chapter therefore considers metropolitan TL3 regions separately even if they host the same FUA, flagging the role of commuting in the discussion of results where relevant.

2.1. Regional disparities in labour market outcomes

Copy link to 2.1. Regional disparities in labour market outcomes

Employment represents the primary source of income for most households and the most important channel for upward mobility. Yet, in most OECD countries, high-productivity firms (Bilal, 2023[4]; Menon and Vermeulen, 2026[31]), innovation (Usai, 2011[32]) and economic activity remain unevenly distributed across regions, with significant implications for the regional distribution of labour market opportunities (OECD, 2023[1]; OECD, 2025[2]). Drawing on aggregate, TL3‑level labour market statistics and microdata analysis for selected countries, this section documents regional disparities in labour market outcomes and examines the mechanisms behind them. The analysis suggests that regional economic performance plays a central role in shaping people’s employment opportunities. It also illustrates that labour mobility alone is not sufficient to narrow regional disparities, underlining the need for place‑based policies.

2.1.1. People’s employment prospects are shaped by where they live

Despite favourable labour market conditions across most OECD countries, people experience very different employment outcomes depending on their region of residence. While employment and labour force participation have reached record levels in most OECD countries (see Chapter 1), these national averages mask substantial disparities in employment and unemployment rates across regions within countries. Across OECD economies with comparable data, the average gap in employment rates between high- and low-employment regions – defined as those in the top and bottom quintiles of regions within each country – stands at 11.4 p.p. (Figure 2.1, Panel A). Regional gaps in employment rates differ significantly across countries: they are largest in Italy (24.8 p.p.), Canada (17.4 p.p.) and Israel (17.3 p.p.), but much smaller in Czechia (3.2 p.p.), Sweden (5.6 p.p.) and the Netherlands (5.8 p.p.). Still, in over half of OECD economies, the range across regions exceeds 20 p.p. Regional disparities in unemployment rates are similarly pronounced. On average across the OECD, the unemployment rate in the lowest-performing quintile of regions is more than twice that in the highest-performing quintile; the ratio exceeds four in Italy, Belgium, Canada and the Slovak Republic (Figure 2.1, Panel B). Overall, these regional disparities are larger than those identified in recent OECD analyses for large TL2 regions (OECD, 2023[27]; OECD, 2024[28]; OECD, 2025[2]), which highlights the importance of examining labour market outcomes at finer geographic scales.

Figure 2.1. Regional disparities in labour market outcomes are large in most OECD countries
Copy link to Figure 2.1. Regional disparities in labour market outcomes are large in most OECD countries

Employment and unemployment rates across TL3 regions, 2024 or latest available year

Regional employment disparities often map to long-standing differences in regional productivity (OECD, 2024[29]), as discussed further below in this chapter. In several countries, these divides follow persistent geographic cleavages: employment rates in the Italian Mezzogiorno continue to lag the more industrialised North; in Belgium, Wallonia trails Flanders; and the eastern regions of both Germany and Poland have yet to close the employment gap with the rest of the country. Regions hosting the capital tend to be the most productive (OECD, 2024[29]), and exhibit higher employment rates than elsewhere in most OECD countries. The picture is more nuanced for unemployment. Some remote regions, particularly in the Nordic countries and in New Zealand, exhibit low unemployment, but their tight local labour markets mainly reflect limited labour supply rather than economic dynamism; in Austria, Denmark and Norway, capital regions record high unemployment.

These within-country disparities in employment rates are, on average, as significant as those observed between best- and worst-performing OECD economies, yet many labour market policies are designed and implemented at the national level. In many OECD countries, a single minimum wage applies nationwide, even if wage setting institutions are often formally required to assess its impact across geographies. Other policy areas, including active labour market policies, are also relatively centralised in principle: subnational governments hold formal competences in only about two out of five OECD countries (OECD, 2023[33]). In practice, however, regional and local authorities often assume a more prominent role than their formal mandates suggest (OECD, 2025[34]).

These large regional disparities in employment outcomes persist despite a period of moderate convergence since the early 2010s. The Global Financial Crisis (GFC) had an asymmetric impact on regional employment, intensifying pressures for industrial restructuring in many local labour markets (Ahrend et al., 2026[25]) – see also Chapter 3. Capital cities felt the immediate shock more acutely (Dijkstra, Garcilazo and McCann, 2015[35]), but regions with lower pre‑crisis labour market dynamism ultimately experienced greater job losses (Fratesi and Rodríguez-Pose, 2016[36]). Across the European Union, the crisis led to a sudden increase in regional labour market disparities (European Commission, 2013[37]; European Commission, 2014[38]). The dynamics shifted from the early 2010s onwards, when steady economic growth and improving labour market conditions contributed to a narrowing of regional gaps in employment rates across most OECD countries.

Empirically, such convergence processes can be described using two well-established metrics (Barro and Sala‑i-Martin (1992[39]), see Box 2.2 for a more technical description):

Sigma ( $σ$ ) convergence, which measures whether the dispersion of employment rates across regions narrows over time. This is captured by the percentage change in the coefficient of variation, which measures dispersion relative to the mean, allowing for meaningful comparisons across countries. A negative percentage change indicates that regional employment rates are converging toward the mean.
Beta ( $β$ ) convergence, which measures whether low-employment regions are catching up with high-employment regions. A negative coefficient indicates that regions with lower initial employment rates experienced faster subsequent growth.

Box 2.2. Measuring convergence in labour market outcomes across regions

Copy link to Box 2.2. Measuring convergence in labour market outcomes across regions

This chapter uses two complementary metrics to assess whether employment rates across TL3 regions converge over time. All indicators are weighted by the working-age population to account for cross-regional differences in population.

Sigma ( $σ$ ) convergence

Sigma convergence measures whether the dispersion of employment rates across regions narrows over time. To measure it, this chapter uses the percentage change in the coefficient of variation (CV). The CV is calculated as:

$C V = \frac{σ}{μ}$

where $σ$ and $μ$ are the standard deviation and mean of employment rates across TL3 regions, both weighted by the regional working-age population. The standard deviation measures how far regional employment rates typically deviate from the average, with higher values indicating greater dispersion across regions. However, raw standard deviations are difficult to compare across countries or over time because the same absolute spread can have different implications depending on the baseline level. For example, a 2 p.p. standard deviation represents greater relative dispersion in a country with lower employment rates. The coefficient of variation addresses this issue by producing a unit-free measure that expresses dispersion as a share of the mean regional employment rate.

The percentage change in the CV over time $\frac{{C V}_{t} - {C V}_{t - n}}{{C V}_{t - n}}$ indicates how this relative dispersion has evolved. A negative value indicates that regional employment rates have become less dispersed, i.e. have converged towards the mean.

Beta ( $β$ ) convergence

Beta convergence measures whether regions with lower initial employment rates experience faster subsequent growth. A negative coefficient typically signals a catch-up process of lagging regions, but can also describe a case of leading regions falling behind. It is estimated using the following regression:

$\frac{1}{n} \frac{{E r a t e}_{i, t} - {E r a t e}_{i, t - n}}{{E r a t e}_{i, t - n}} = α + β {E r a t e}_{i, t - n} + ε_{i}$

where ${E r a t e}_{i, t}$ is the employment rate in TL3 region $i$ at time $t$ , ${E r a t e}_{i, t - n}$ is the employment rate $n$ years earlier, and therefore the dependent variable represents the average annual growth in the employment rate over the period. Observations are again weighted by the working-age population. A negative $β$ -coefficient indicates $β$ -convergence: regions with lower initial employment rates grew faster than those with higher initial rates. $β$ -convergence is necessary, but not sufficient, for $σ$ -convergence. If initially low-employment regions grow so rapidly that they overtake initially high-employment regions, the resulting re‑ranking can leave overall dispersion unchanged or even wider at the end of the period.

Based on these metrics, 21 of the 29 countries for which regional employment rates can be traced back to the early 2010s have experienced both catching up by lagging regions (β-convergence) and a narrowing of relative disparities across regions (σ-convergence) (Figure 2.2). The reduction in dispersion has been particularly strong in countries such as Slovenia, Czechia, Estonia and Ireland – close to or exceeding 50% of the mean – but modest in most other countries and almost negligible in Germany, the Netherlands and the United Kingdom. Only two countries, Latvia and Switzerland, show the opposite pattern: increasing dispersion, and high-employment regions pulling further ahead. Finally, Chile, Lithuania and Norway present a less intuitive case, combining β-convergence with σ‑divergence. This occurs when initially low‑employment regions overtake initially high-employment regions, and dispersion at the end of the period is wider than at the start.

Figure 2.2. Low-employment regions have been catching up, narrowing regional disparities
Copy link to Figure 2.2. Low-employment regions have been catching up, narrowing regional disparities

Sigma- (left-axis) and Beta-convergence (right-axis), TL3 regions, early 2010s to latest available year

Previous academic literature also documents large geographic disparities in other labour market outcomes. Recent empirical work suggests that wage inequality across local labour markets has roughly doubled in Canada, Germany and the United States since the 1970s, and more than tripled in the United Kingdom (Bauluz et al., 2024[40]). This process of divergence has reversed or stalled in the last decade in large European economies, but not in Canada and the United States. Later in this chapter, Section 2.2 illustrates that regional differences are also large for disposable household incomes, i.e. incomes after redistribution through taxes and government transfers.

That regional disparities in labour market outcomes remain so large even after years of convergence points to structural factors on both sides of the labour market. On the labour supply side, workers in some regions may have characteristics that make them less employable, such as lower levels of educational attainment. On the labour demand side, some regions benefit from a concentration of economic activity that attracts productive firms, increasing the availability of jobs. This distinction is more blurred in practice, as supply and demand reinforce each other: skilled workers attract productive firms, and productive firms attract skilled workers (Moretti, 2010[41]; Combes and Gobillon, 2015[42]). This self-reinforcing dynamic is partly what makes regional employment disparities so persistent.

The next two subsections present descriptive evidence on these dynamics, focussing primarily on employment rates, where data coverage across OECD countries is more complete. Consistent with recent empirical research (Card, Rothstein and Yi, 2023[43]), the results suggest that workers’ characteristics explain only part of geographic disparities in employment, and that place‑based factors play a significant role.

2.1.2. Workers are different across regions, but this alone cannot explain regional employment disparities

Workers differ substantially across regions, but this explains only a limited share of regional employment gaps. Working-age people in low-employment regions, defined again as regions in the bottom quintile of employment rates, have on average characteristics associated with weaker employment prospects. For instance, they are more likely to have left education before completing high school and to live with disabilities (Figure 2.3, Panels A and B). However, such differences only go so far in explaining regional disparities. Across most countries for which microdata at the TL3 level are available, residents of low-employment regions remain significantly less likely to be employed than those in high-employment regions even after accounting for differences in age, gender, education, household composition, and migration background (Figure 2.3, Panel C). Differences in people’s observable characteristics explain only between a third and half of the observed gap in regional employment rates across countries. These results align with previous empirical research on employment at higher geographic levels in the United Kingdom (Overman and Xu, 2024[44]) and Spain (López‐Bazo and Motellón, 2013[45]). Lithuania and Czechia are the only countries where the regional employment gap becomes statistically insignificant after accounting for observable characteristics, though relatively small sample sizes in these countries mean that this should be interpreted with caution.

One limitation of these results is that workers in regions with weaker labour markets may differ in ways not captured in the data, such as innate ability, motivation or propensity for risk-taking. However, empirical research addressing this point, notably by tracking people who move across regions and observing how their labour market outcomes evolve, reaches similar conclusions. Evidence from the United States suggests that the sorting of higher-skilled workers into stronger labour markets explains only around half of geographic earnings differences (Card, Rothstein and Yi, 2023[43]), while evidence from France points to a more limited role for sorting in explaining geographic differences in job separation rates (Bilal, 2023[4]). These results might even understate the importance of geography, as they do not account for the additional skills that workers gain over time by living in more productive regions (Roca and Puga, 2016[46]).

The role of place in shaping employment opportunities is particularly pronounced for groups with weaker labour market attachment, such as young people, those with lower educational attainment, and women. While these groups have lower employment rates everywhere, the gaps are often larger in regions with weaker labour markets. For instance, in Chile, the gender gap in employment rates is 18.2 p.p. in high-employment regions but rises to 27.9 p.p. in low-employment regions. Similarly, in the Slovak Republic, the employment gap between individuals with and without tertiary education is only 1.7 p.p. in high-employment regions, but nearly 20 p.p. in low-employment regions.

Figure 2.3. Workers’ observable characteristics explain a limited part of regional disparities in employment outcomes
Copy link to Figure 2.3. Workers’ observable characteristics explain a limited part of regional disparities in employment outcomes

Differences in workforce characteristics (Panels A and B) and the gap in the likelihood of employment (Panel C) between high- and low-employment TL3 regions, 2024 or latest available year

2.1.3. Regional economic performance is a major driver of employment disparities across regions

Regional economic performance plays a central role in shaping geographic disparities in labour market outcomes. This happens through two interrelated mechanisms. First, more productive firms tend to concentrate in higher-productivity regions (Lindenlaub, Oh and Peters, 2022[47]; Bilal, 2023[4]). While part of this regional productivity variation reflects differences in human capital as explored in the previous section, empirical studies suggest that these account for only 20% to 35% of the gap in the United States, (Hanushek, Ruhose and Woessmann, 2017[48]), and a smaller share in Europe (Beugelsdijk, Klasing and Milionis, 2017[49]), with place‑based factors such as infrastructure, capital, and local institutions explaining the remainder. Second, once high-productivity clusters take hold, they generate agglomeration economies – faster matching between employers and employees, knowledge transfers across firms, and denser networks of suppliers and clients (Combes and Gobillon, 2015[42]) – which further raise the productivity and attractiveness of these regions.

This mechanism can create a virtuous cycle that strengthens employment opportunities in leading regions: regions with greater productivity attract productive firms, increasing the availability and quality of employment. Lagging regions face the opposite dynamic. As productive firms relocate toward more developed areas, employment in left-behind places contracts, and the firms that remain tend to offer less attractive job opportunities. Previous OECD work relying on firm-level data for Italy and Spain suggests that firms in less developed regions are smaller, and 20% to 30% less productive than comparable firms elsewhere (Menon and Vermeulen, 2026[31]).

Indeed, across OECD countries, employment outcomes are closely tied to regional productivity, with regions that have persistently ranked in the bottom quintile of GDP per worker over the past 15 years showing employment rates 4.5 p.p. lower than the most productive regions (Figure 2.4, Panel A). Austria and Switzerland exhibit strong employment gaps in the opposite direction. In Austria, this reflects strong commuting flows into Vienna from surrounding regions. Because employment is measured at the place of residence and productivity at the place of work (see Box 2.1), commuters raise employment rates and aggregate income in their home regions while their output is recorded in Vienna. As a result, Vienna combines high productivity with a comparatively low employment rate, and notably an overrepresentation of people with weaker labour market attachment, such as students and recent immigrants (OECD, 2025[7]). In Switzerland, Geneva and Basel-Stadt boast high productivity due to the concentration of financial and pharmaceutical firms. Resident employment rates in both cantons are supressed by a large foreign-born population, who exhibits weaker labour market attachment particularly among women (Lacroix and Vidal-Coso, 2018[50]). Long-term unemployment provides further evidence of limited job availability in low‑productivity regions: rates are nearly twice as high in the least than in the most productive regions across countries with available data (Figure 2.4, Panel B). The size of this gap differs across countries: it is much narrower in Northern European countries and wider in Southern Europe, Hungary and the Slovak Republic.

Figure 2.4. People’s labour market outcomes are substantially worse in low-GDP-per-worker regions
Copy link to Figure 2.4. People’s labour market outcomes are substantially worse in low-GDP-per-worker regions

Gap in labour market outcomes between TL3 regions with persistently high and low GDP per worker, 2024 or latest available year

The progressive transformation of OECD countries into knowledge‑intensive economies since the 1970s has contributed to the concentration of high-productivity employment in fewer locations. Manufacturing firms were often spread out geographically, responding to the local availability of infrastructure, raw materials, and cheap labour; as transport costs fell production dispersed even further across regions (Glaeser and Kohlhase, 2004[51]). Knowledge‑intensive industries work differently. Their productivity depends on the density of skilled workers and the face‑to-face interactions that drive innovation and learning in specific locations, often cities (Moretti, 2012[3]). Over the previous decades, these dynamics have resulted in a growing concentration of high-productivity firms in a small number of metropolitan areas (Clarke, Martin and Tyler, 2016[52]; Kemeny and Storper, 2020[53]), even if the labour market recovery since the GFC has brought some regional convergence in employment outcomes in more recent years (Figure 2.2). Meanwhile, regions previously specialised in manufacturing have faced pressure to shift their industrial structure (Ahrend et al., 2026[24]) and have become increasingly vulnerable to employment shocks, which often leave long-lasting scars on local labour markets (Autor, Dorn and Hanson, 2016[54]; Celli, Cerqua and Pellegrini, 2023[55]; Vermeulen and Braakmann, 2023[56]) – see also Chapter 3.

In most OECD countries, employment in high value‑added services is disproportionately concentrated in capital-city regions, while the share of such jobs in other metropolitan regions is closer to that of non‑metropolitan regions (Figure 2.5). Unlike employment rates, sectoral employment is measured by place of work; these shares therefore capture where high-value‑added activity is located. On average across OECD countries where data are available, 26.6% of all jobs in capital-city regions are in high‑value‑added services, compared to 17.8% in other metropolitan regions and just 12.5% in non‑metropolitan regions. Switzerland and Australia, countries where the capital city is relatively small compared to other urban centres, represent exceptions. In Switzerland, jobs in high-value‑added services are concentrated in larger metropolitan regions, such as Geneva and Zurich, rather than in Bern. In Australia, several larger metropolitan regions beyond Canberra, including Melbourne and Sydney, account for a significant share of this type of employment. In Belgium, Greece, Italy and the Netherlands, capital‑city regions dominate, but high-value‑added services are more evenly spread across other metropolitan regions than is typical across the OECD.

A similar but much less pronounced pattern holds for employment rates more broadly. Across countries with available data, capital-city regions tend to have somewhat higher employment rates, while other metropolitan and non-metropolitan regions perform broadly similarly (Annex Figure 2.E.1). That not all metropolitan regions are thriving equally is also reflected in population dynamics: OECD metropolitan regions expanded their population by 15% on average between 2001 and 2021, outpacing non‑metropolitan regions; however, more than one‑in-five metropolitan regions experienced population decline over the same period (OECD, 2025[57]).

Figure 2.5. High-value‑added jobs are disproportionately concentrated in capital-city regions
Copy link to Figure 2.5. High-value‑added jobs are disproportionately concentrated in capital-city regions

Share of employment in high-value‑added services by region type, 2023 or latest available year

2.1.4. Workers’ mobility is not sufficient to narrow regional disparities in labour market outcomes

Within-country mobility could, in principle, contribute to narrowing regional disparities in employment outcomes if people with limited job prospects relocated from weaker to stronger labour markets. Supporting mobility for those who wish to move in search of better opportunities has been shown to improve job matching (OECD, 2025[7]), and should therefore play a role as part of broader strategies to reduce regional labour market imbalances – see also Chapter 3. However, the evidence in this section suggests that labour mobility is not sufficient to close existing regional employment gaps, and may sometimes even reinforce them.

Mobility from low- to high-employment regions is rather limited in magnitude. Across countries with available data, regions with persistently low employment – defined as those in the bottom quintile of employment rates within each country for most of the past 15 years – experienced annual net migration outflows of 0.25% of their population; meanwhile, high-employment regions experienced annual net inflows of approximately 0.15% from other parts of the country (see Figure 2.6). Mobility is considerably more pronounced in the Slovak Republic, Korea and Estonia, which drive up the OECD average, but very limited in many large economies such as Japan, Germany and Spain. The implication of such low mobility rates is that even if all out-migrants from low-employment regions were unemployed, these population flows would be too small to meaningfully reduce the regional disparities in labour market outcomes documented earlier in this chapter. Moreover, in countries such as Austria, Czechia and Israel, high-employment regions are experiencing population outflows.

Figure 2.6. People move out of low-employment regions and into high-employment ones, but at a slow pace
Copy link to Figure 2.6. People move out of low-employment regions and into high-employment ones, but at a slow pace

Average net inter-regional migration rates (% of the population) in high- and low-employment TL3 regions, early 2010s to latest available year

Mobility is low in part because people face multiple barriers to relocation. Financial costs can be significant, especially for people moving to regions with better employment prospects. Housing in these regions is typically more expensive, which deters low-skilled workers from moving (Ganong and Shoag, 2017[5]; OECD, 2020[58]; Causa, Abendschein and Cavalleri, 2021[59]). Access to welfare benefits tied to the region of residence, as well as higher childcare costs, add further disincentives for mobility (OECD, 2025[7]). Job search itself can also be costly, and costs rise with distance, such that most jobseekers focus their search on areas close to their place of residence (Manning and Petrongolo, 2017[60]; Gobillon and Selod, 2021[61]). These barriers can be lowered through policy interventions. Relaxing land-use regulations can ease housing supply constraints, reducing the costs of settling in high-employment regions. Making social housing entitlements easily transferable across regions can reduce immobility (Causa, Abendschein and Cavalleri, 2021[59]; OECD, 2021[62]). For example, in the Australian state of Victoria, social housing tenants can apply to transfer to another property if changes in their employment require them to move a significant distance from their current residence. Extending such arrangements to enable social housing transfers across broader geographies – particularly where responsibilities for social housing are fragmented across regional jurisdictions – would further foster the mobility of those with weaker labour market attachment. On the job search side, transport subsidies for jobseekers (Franklin, 2017[63]), and direct relocation subsidies (Caliendo, Künn and Mahlstedt, 2017[64]) have shown positive effects, though both can prove expensive at scale. Other barriers are much harder to address – not least because they reflect legitimate preferences, such as proximity to established social networks (Spring, Gillespie and Mulder, 2023[6]) or caregiving responsibilities that tie people to a specific location (Artamonova and Syse, 2021[65]). Recent OECD work explores these mobility barriers and potential policy responses in greater detail (OECD, 2020[58]; OECD, 2021[62]; OECD, 2025[7]).

A further challenge is that regions with weak labour markets tend to cluster geographically. Moving to places with substantially better employment prospects therefore often requires relocating over longer distances, which amplifies the barriers discussed above. This can be illustrated through Moran’s I, a widely used measure of spatial clustering in economic variables (Patacchini and Rice, 2007[66]; Cracolici, Cuffaro and Nijkamp, 2008[67]; Furková and Chocholatá, 2020[68]; OECD, 2023[1]) Two distinct Moran’s I metrics exist that provide complementary insights into these spatial patterns (see Box 2.3 for a technical description):

Global Moran’s I measures the overall degree of spatial clustering in a country by comparing each region’s value against the weighted average of its neighbours. Values range from −1 to 1; a positive value indicates that regions with similar values tend to be located near one another, while a negative value indicates that neighbouring regions tend to have dissimilar values.
Local Moran’s I identifies where spatial clustering occurs, pinpointing specific groups of regions whose values are persistently above (High-High) or below (Low-Low) the overall mean.

Applying these metrics to regional unemployment rates reveals substantial spatial clustering in most large OECD economies, with regions’ unemployment rates strongly correlated with those of their neighbours (Figure 2.7, Panel A).2 Overall clustering, as measured by Global Moran’s I, is strong, for example, in Italy and Türkiye, though cross‑country disparities should be interpreted with caution given differences in the size and number of regions. Japan is somewhat an exception in that the degree of clustering is weak and only marginally statistically significant.

This overall clustering often translates into stark divides within countries as measured by Local Moran’s I (Figure 2.7, Panel B). In Germany, low-unemployment clusters are concentrated in prosperous southern regions, such as Bavaria and parts of Baden-Württemberg, while high-unemployment clusters are found in eastern Germany – still bearing the structural legacy of reunification – and in the Ruhr area, scarred by decades of decline in the coal and steel industries. In Italy, the clustering mirrors the well‑documented North-South divide, with Mezzogiorno regions lagging the industrialised North. Poland exhibits a similar east-west gradient as Germany, with high-unemployment clusters in its less developed eastern regions, historically reliant on small-scale agriculture and characterised by lower levels of educational attainment. In the United States, low-unemployment clusters stretch across the Great Plains and Upper Midwest, while high unemployment is concentrated in the Deep South, where longstanding underinvestment in infrastructure and human capital continues to weigh on labour market outcomes. In Türkiye, high-unemployment clusters are concentrated in the South-East, reflecting persistent regional development gaps; in Japan, a few low-unemployment clusters emerge in the greater Chūbu region in Central Japan, home to a strong manufacturing base (maps not shown).

Box 2.3. Measuring spatial clustering: Global and Local Moran’s I

Copy link to Box 2.3. Measuring spatial clustering: Global and Local Moran’s I

A standard approach to measuring spatial clustering is the Moran’s I statistic, which assesses the degree to which regions with similar values of a given variable are located near one another (Moran, 1950[69]; Anselin, 1995[70]).

Two Moran’s I metrics provide complementary insights:

Global Moran’s I

It quantifies the overall degree of spatial clustering. It is defined as:

$I = \frac{n}{S_{0}} \times \frac{\sum_{i = 1}^{n} \sum_{j = 1}^{n} w_{i j} (x_{i} - \bar{x}) (x_{j} - \bar{x})}{\sum_{i = 1}^{n} {(x_{i} - \bar{x})}^{2}}$

where $n$ captures the number of regions, $x_{i}$ is the value of the variable of interest (here: the unemployment rate) in region $i$ , $\bar{x}$ is the average of that variable across all regions, $w_{i j}$ is the spatial weight describing the strength of the connection between regions $i$ and $j$ , and $S_{0} = \sum_{i} \sum_{j} w_{i j}$ is the sum of all spatial weights. In essence, the index measures the degree to which each region’s value co-varies with those of its neighbours, as defined by the spatial weight matrix. The spatial weights used in this chapter are based on whether they share a common border (a concept referred to in technical terms as rook contiguity). Each neighbour receives equal weight, such that each region is compared against the simple average of its neighbours’ values.

The Global Moran’s I score ranges from −1 (perfect dispersion) to +1 (perfect clustering), with values near zero indicating no systemic spatial pattern. To visualise it, this chapter plots each region’s unemployment rate against the average rate of its neighbours (Figure 2.7, Panel A). The slope of the fitted line in this scatterplot is identical to the Global Moran’s I statistic. Statistical significance is assessed using a permutation test, which repeatedly shuffles values across regions to evaluate whether the observed spatial pattern is unlikely to have occurred by chance.

Local Moran’s I

While the global statistic summarises the overall degree of clustering, Local Moran’s I identifies where specific clusters or outliers occur. For each region i, it is defined as:

$I_{i} = \frac{(x_{i} - \bar{x})}{S^{2}} \times \sum_{j = 1}^{n} w_{i j} (x_{j} - \bar{x})$

where $I_{i}$ is the Local Moran’s I for the region $i$ , $x_{i}$ is again the value of the variable of interest in region $i$ , $\bar{x}$ is the average of $x$ across all regions, $w_{i j}$ is the spatial weight, summing up to 1, which describes the relative connection between regions $i$ and $j$ , and $S^{2}$ is the variance of $x$ across all regions. Dividing by the variance standardises each region’s deviation from the mean to ensure comparability. In essence, the index assesses whether a region’s unemployment rate and those of its neighbours jointly deviate from the overall average.

Based on the sign and magnitude of $I_{i}$ and the value of region $i$ , four types of statistically significant clusters can be identified. A region can belong to a cluster of similar values: High-High (high-unemployment region surrounded by high-unemployment neighbours) or Low-Low (low-unemployment region surrounded by low-unemployment neighbours). Alternatively, it can be a spatial outlier: High-Low (high unemployment surrounded by low) or Low-High (low unemployment surrounded by high). As with the global statistic, statistical significance is assessed via permutation tests. For simplicity, this chapter focusses on statistically significant High-High and Low-Low clusters (Figure 2.7, Panel B).

Figure 2.7. Regions with high unemployment are spatially clustered
Copy link to Figure 2.7. Regions with high unemployment are spatially clustered

*Global* and *Local Moran’s I* for unemployment rates across TL3 regions in selected large OECD economies, 2022 or latest available year

These patterns of geographic clustering limit the employment opportunities available close to where jobseekers live. This not only discourages mobility by raising the costs of relocation, but renders commuting an unfeasible strategy to bridge the gap between places with surplus workers and those with open jobs.

Furthermore, mobility alone is unlikely to reduce regional labour market inequalities because people who move from low- to high-employment regions are often those with the strongest employment prospects. This partly reflects the fact that barriers to mobility weigh most heavily on people with weaker labour market attachment. A large body of empirical research shows that people who migrate across regions (Panichella and Impicciatore, 2024[71]), and particularly those moving to areas with greater employment opportunities (Combes, Duranton and Gobillon, 2008[72]; De la Roca, 2017[8]), have on average higher levels of education, earn more in their previous job, and work in higher-skilled occupations than those who stay behind. OECD estimates from administrative records for Belgium and Estonia confirm that unemployed people are not more likely to move than those who are employed (Box 2.4). People leaving low‑employment regions in these countries also tend to be younger, more frequently in the labour force and more educated. The outmigration of these workers can slowly erode the human capital base in low‑employment regions, further weakening labour markets in these regions. As a result, if not well‑managed, regional migration patterns can reinforce, rather than reduce, regional disparities in labour markets.

These results should not be interpreted as implying that workers’ mobility should not be encouraged. Rather, it highlights the need to reduce mobility barriers for those with weaker employment prospects, who also tend to have lower incomes. However, even if such barriers were eased and mobility converged to that of groups with stronger employment prospects, relocation would likely remain limited, with only a small share of jobseekers moving. Hence, place‑based policies remain an important lever for reducing regional labour market inequalities (OECD, 2025[13]).

Box 2.4. Who leaves low-employment regions behind? Evidence from administrative records in Belgium and Estonia

Copy link to Box 2.4. Who leaves low-employment regions behind? Evidence from administrative records in Belgium and Estonia

This box draws on census and labour administrative records to examine the mobility patterns of the working-age population in Belgium and Estonia. Consistent with previous empirical research (Combes, Duranton and Gobillon, 2008[72]; Fendel, 2014[73]; De la Roca, 2017[8]; Panichella and Impicciatore, 2024[71]), it illustrates that the people most likely to leave low-employment regions are not those with the weakest job prospects, but rather those with comparatively good chances of being in employment.

A straightforward way to illustrate this is by comparing the characteristics of those who leave low‑employment regions, defined as regions in the bottom quintile of employment rates, compared to those who stay behind. Survey data typically lack the geographic coverage or the sample size to capture such transitions reliably. Administrative records for Belgium and Estonia, however, make it possible to track the entire population geographically on a yearly basis, enabling such comparisons. Regression analysis can then isolate the independent effect of different observable characteristics, such as age, education, or employment status, net of the influence of the others – without of course accounting for factors unobservable in the data such as people’s motivation or specific skills.

In low-employment regions in Belgium and Estonia, older workers, those with lower educational attainment, and those who are inactive or unemployed in a given year are less likely to have moved to stronger labour markets a year later (Figure 2.8). Age is by far the strongest predictor, with workers aged 45‑54 almost 70% less likely to leave than those aged 25‑34. Education also plays a significant role: working-age people with at most upper-secondary education are around 6% less likely to leave low-employment regions in Belgium, and around 20% less likely in Estonia than their peers with a tertiary qualification. Crucially, those outside of the labour force are not the ones leaving low‑employment regions. In both countries, the inactive are less likely to move than those in employment. In Belgium, the unemployed are also 11.5% less likely to leave than the employed, while in Estonia they are just as likely to move.

Figure 2.8. People with better employment prospects are more likely to move out of low‑employment regions
Copy link to Figure 2.8. People with better employment prospects are more likely to move out of low‑employment regions

Differences in the probability of leaving low-employment TL3 regions from one year to the next, by socio‑economic characteristics, Belgium and Estonia

2.2. From labour market outcomes to the distribution of household incomes: Regional inequalities in living standards

Copy link to 2.2. From labour market outcomes to the distribution of household incomes: Regional inequalities in living standards

Where people live shapes not just their chances of finding work and the types of jobs available to them, but also the incomes they can generate and the living standards they can expect to achieve. For most households, earnings from work represent by far the largest source of income, making regional gaps in employment rates and earnings primary drivers of differences in living standards. Large and persistent differences in living standards across regions matter for several reasons: such disparities often reflect deeper structural factors – geographic remoteness, weak agglomeration, lack of high-value‑added industries, or the availability and quality of essential infrastructure and services – that shape the opportunities available to people across regions; they create obstacles for people to seize economic opportunities regardless of where they live, undermining investment in human capital, innovation and risk‑taking; they can fuel social discontent and erode economic and societal resilience; and they can complicate the design of labour market, training and social policies that are both efficient and well-targeted. Understanding the scale, structure, and drivers of disparities in household incomes and living standards is therefore a necessary first step for governments seeking to address them.

Despite this policy relevance, internationally comparable evidence on geographic disparities in the distribution of household incomes remains scarce at a more granular level. Household surveys – the standard source of cross-country data on incomes and living conditions – are in many countries representative only at the level of large (TL2) regions, given their limited sample sizes.3 As a result, both the OECD and Eurostat compile income distribution data only at the TL2/NUTS2 level (Eurostat, 2025[74]; OECD, 2026[75]). The empirical literature has consequently focussed largely on income disparities across large regions (Piacentini, 2014[76]; Castells-Quintana, Ramos and Royuela, 2015[77]; Royuela, Veneri and Ramos, 2018[78]; Veneri and Murtin, 2018[79]; Erfurth, 2023[80]; European Commission, 2024[81]; Savoia, 2024[82]; OECD, 2025[2]). In a small but growing number of countries, country-specific studies have pushed beyond this constraint, examining income distributions at finer geographic scales by drawing on register‑based income data.4 The most prominent example is the work of Chetty et al. (2014[83]) under the Opportunity Insights project in the United States. In Europe, Bonnet, d’Albis and Sotura (2021[84]) use tax record data to track inequality across French départements over the last century, while Dzhavatova et al. (2025[85]) document income distributions across municipalities in Denmark, Finland, Norway and Sweden.5

National accounts have served as an alternative source of regional income measures. A large literature has used GDP per capita as a widely available proxy for income (Iammarino, Rodriguez-Pose and Storper, 2018[11]; Gbohoui, Lam and Lledo, 2019[86]; Pina and Sicari, 2021[87]; Rosés and Wolf, 2021[88]; OECD, 2023[1]; European Commission, 2024[81]; Chan, Ellingsen and Simpson, 2025[89]). This reflects, in large part, that GDP-per-capita data offer much broader geographic and temporal coverage than household surveys, though, as a measure of economic output rather than household income, they are an imperfect proxy for living standards. For 14 OECD countries, regional accounts provide TL3‑level statistics on disposable household income per capita (OECD, 2024[29]; OECD, 2025[90]). However, these macro-level data do not account for individual household structure and provide no information on the distribution of incomes across households within regions – for a detailed discussion, see OECD (2023[1]).6

This section presents, for the first time, comparable statistics on income levels and distributions across small (TL3) regions based on microdata from 24 OECD countries. It relates these patterns to regional differences in economic activity, as measured by GDP per capita, and to the disparities in labour market outcomes documented in Section 2.1. Throughout this section, the term “household income” is used to refer to equivalised disposable household income – that is, income after taxes, social insurance contributions and government transfers, adjusted for household size. The analysis draws on newly collected income distribution statistics derived from administrative records and household surveys; Annex 2.B provides an overview of data sources and their coverage.

2.2.1. Regional income disparities vary across countries and align only weakly with regional disparities in GDP per capita

Many OECD countries display substantial regional disparities in household income levels. Figure 2.9, Panel A shows median disposable household incomes across TL3 regions – that is, the income of the person at the exact middle of each region’s income distribution – expressed relative to the national median. In most OECD countries, the median person in the highest-income region has a disposable income around 10‑30% above the national median. This highest-income region is often the country’s capital-city region – for example, Paris in France, Tokyo in Japan, Vilnius in Lithuania, and Wellington in New Zealand. Austria and Belgium are two exceptions: in both countries, the capital region has the lowest median income, reflecting that both capital regions are surrounded by affluent suburban regions.7 In some countries, top-income regions owe their position to natural resource wealth, as in Division No. 16 in Alberta (Canada), the southernmost Tierra del Fuego in Chile and the North Outback in Western Australia. At the other end of the distribution, the median person in the lowest-income region has a disposable income around 10‑20% below the national median in most countries. On average, across the 24 OECD countries covered, median incomes in the highest- and lowest-income regions differ by a factor of around 1.60 – meaning the median person in the highest-income region has a disposable income approximately 60% above that of their counterpart in the lowest-income region.

However, cross-country variation in regional income disparities is considerable, reflecting differences in country size, the number of regions, and overall income inequality. Disparities are widest in geographically large countries with many regions – such as Australia and Canada, where the top-to-bottom regional income ratio ranges between 2.1 and 3.0 – as well as in countries with many regions relative to their size, such as Greece, with a ratio of 2.1, and in populous countries such as France, Japan and Spain, where ratios range between 1.6 and 1.7 (Figure 2.9, Panel A).8 High levels of overall income inequality are also associated with wider regional disparities: Chile displays a top-to-bottom ratio of 2.6, Israel, Costa Rica and Lithuania display top-to-bottom ratios between 1.7 and 2.0.9 By contrast, regional income disparities are narrowest in smaller, more equal countries. In several Central European countries – Slovenia, Austria and Czechia – the top-to-bottom ratio stands at just 1.1‑1.2, with similarly narrow gaps in the Netherlands and the Slovak Republic (both 1.3) and across the Nordic countries: Norway (1.1), Sweden (1.2), Finland (1.2), and Denmark (1.3).

The greater geographic granularity of the data used in this chapter reveals a more differentiated picture of regional income disparities than earlier cross-country work based on large (TL2) regions. In Piacentini’s (2014[76]) earlier study, top-to-bottom regional income ratios stood below 1.5 in most countries: Mexico was the only country exceeding a ratio of 2 and cross-country differences were comparatively more compressed. The move to TL3 regions uncovers substantially wider disparities.

Figure 2.9. Regional disparities in median household incomes vary substantially across countries, but are much narrower than disparities in GDP per capita
Copy link to Figure 2.9. Regional disparities in median household incomes vary substantially across countries, but are much narrower than disparities in GDP per capita

Median disposable incomes (Panel A) and the relationship between median disposable incomes and GDP per capita in selected OECD countries (Panel B), TL3 regions, 2024 or latest available year

Regional household incomes are much less closely tied to regional GDP per capita than one might expect. Figure 2.9, Panel B illustrates this for a selection of OECD countries, plotting regional median disposable household income against GDP per capita, both expressed relative to their national-level values. For the remaining countries, most of them with a smaller number of regions, disparities in GDP per capita are shown in Annex Figure 2.E.2 the correlation coefficient $ρ$ of the relationship between median incomes and GDP per capita is reported in Annex Table 2.E.1. Two main findings emerge:

Regional disparities in household incomes are substantially narrower than those in GDP per capita – consistent with earlier evidence on narrower disparities in disposable income per capita compared to GDP per capita between metropolitan and non-metropolitan regions based on regional accounts data (OECD, 2024[29]). This pattern partly reflects the geographic concentration of high-value‑added economic activity in particular regions – one example being the so-called “headquarter effect”, which arises when the economic activity of large, multi-plant firms is attributed to the region in which their headquarters are located. Commuting across regional boundaries means that the income people earn and the economic value they generate may be registered in different regions. Finally, the redistributive effect of the tax and benefit system compresses income inequalities between households and, by extension, also between regions.
In some countries, regional household incomes and regional GDP per capita are only weakly correlated. In many countries, such as Spain and France, regional GDP per capita maps closely onto regional household incomes (correlation coefficient $ρ$ of 0.78 and 0.71, respectively). In four countries, however, there is no correlation between regional median household incomes and GDP per capita at all: in Austria, Belgium, Denmark and the Netherlands, $ρ$ is insignificant and close to 0 or even negative. These are mostly relatively small, densely populated countries with well‑developed transport infrastructure, where many people commute across regional boundaries. All four also have strong redistributive systems that further compress income differences between regions.

These results underscore that GDP per capita – while frequently used as a proxy for the income of a representative individual in a region (Pina and Sicari, 2021[87]; OECD, 2023[1]) – is likely a poor measure of regional living standards. Across the six countries shown in Figure 2.9, Panel B, the dispersion in GDP per capita is, on average, 3.5 times larger than that in median household incomes when measured by the unweighted coefficient of variation (not shown). These differences are again considerably larger than those documented by Piacentini (2014[76]) across large TL2 regions, reflecting that both incomes and GDP per capita can vary substantially between small TL3 regions nested within the same TL2 region. In a few cases, high regional GDP per capita coincides with low median household incomes, and vice versa. A striking illustration is Belgium, where the Brussels Region combines the highest GDP per capita of any region with the lowest median disposable household income, while precisely the opposite holds for Arlon, a region in Wallonia bordering Luxembourg.

Even data on regional disposable household incomes do not, however, provide a complete picture of disparities in living standards. The income statistics reported in this chapter do not account for regional differences in the cost of living. To the extent that prices – and notably housing costs – are higher in higher‑income regions, the income differences shown in Figure 2.9 overstate differences in living standards.10 Ongoing OECD work to develop subnational price indices across OECD regions could help address this issue in the future (Alasia, Amann and Horvat, 2025[91]). Regional income differences may also interact with geographic disparities in the accessibility and quality of key public services and infrastructure – another important determinant of living standards (OECD, 2021[92]; OECD, 2023[1]). Recent OECD research finds that people in regions with lower GDP per capita face longer travel times to reach essential services, including public employment services, primary schools, and early childhood education and care (Box 2.5 and Almeida et al., (2024[22])).

Box 2.5. Disparities in the accessibility of essential services compound regional disparities in incomes

Copy link to Box 2.5. Disparities in the accessibility of essential services compound regional disparities in incomes

Besides incomes, people’s ability to access essential services – such as childcare and education, healthcare, and employment and social support – close to where they live is a key determinant of their living standards, and a driver of labour force participation. In the OECD’s 2022 Risks that Matter survey, fewer than half of respondents considered that they could access good-quality and affordable public services in areas such as education, employment and family support (OECD, 2023[93]). At the 2022 OECD Employment and Labour Ministerial Meeting, Ministers underlined their commitment to improving the accessibility and responsiveness of employment services to support jobseekers and firms in the transition to greener jobs, and to providing accessible and affordable care systems to reduce gender gaps in employment, pay and career progression (OECD, 2022[94]).

Recent OECD work (Almeida et al., 2024[22]) has mapped the physical accessibility of selected essential services across European OECD countries, documenting substantially lower service accessibility in regions with lower GDP per capita. This work estimated travel times to the nearest service facility for three types of essential services – public employment services (PES), early childhood education and care, and primary education – combining data on the locations of service facilities with population density data and road network information.¹ Regional differences in service accessibility are large and closely related to GDP per capita: in the highest-GDP-per-capita regions of France and Germany, for example, almost all people can reach a PES centre within a 15‑minute drive, compared with fewer than 50% in the countries’ lowest-GDP-per-capita regions. A similar pattern holds for access to the nearest early childhood education and care facility and primary school within a 15‑minute walk.

These disparities partly reflect differences between metropolitan and non-metropolitan regions, but the relationship between service accessibility and GDP per capita holds more broadly. Metropolitan regions have, on average, higher GDP per capita and shorter travel times to the nearest PES facility – largely reflecting greater population density and better transport infrastructure. However, regression analysis confirms a statistically significant positive relationship between service accessibility and GDP per capita even after controlling for a range of regional characteristics, including access to cities, population density, population levels and growth, GDP-per-capita growth, the unemployment rate, and country fixed effects. Across 661 regions in 16 countries, a 10% higher GDP per capita is associated with a 2 p.p. greater share of people able to reach a PES centre within 15 minutes by motor vehicle.

This analysis illustrates that different dimensions of regional disparities can reinforce one another (see also OECD (2025[2])).² People in lower-income regions not only face fewer job opportunities and lower household incomes, but also poorer access to the essential services and infrastructure that shape living standards and enable them to seize economic opportunities.³

	% of people within 15 minutes to nearest PES by motor vehicle
	(1)	(2)	(3)	(4)
Unemployment rate in 2019 (%)	0.233			0.807***
	(0.307)			(0.253)
Children aged 5 to 9 in 2022 (%)
GDP p.c. in 2019 (Ln)		23.389***		20.531***
		(2.920)		(4.029)
Annual GDP p.c. growth 2005-2019 (%)		‑2.951***		‑4.028***
		(0.866)		(0.987)
Total population in 2022 (Ln)			8.744***	0.480
			(0.755)	(1.331)
Annual population growth 2015-2022 (%)			2.325	‑7.631***
			(1.467)	(2.021)
Population density in 2022 (Ln)				6.615***
				(1.157)
2. Metropolitan – Medium				‑2.797
				(3.153)
3. Non-metropolitan – Medium				‑4.510
				(3.838)
4. Non-metropolitan – Small				‑3.867
				(4.017)
5. Non-metropolitan – Remote				‑9.723**
				(4.557)
Country FE	Yes	Yes	Yes	Yes
Observations	692	922	1 612	661
Number of countries	18	25	32	16
Adjusted R-squared	0.383	0.455	0.546	0.617

Note: Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. The number of countries included in the regressions varies across models depending on data availability. Regional observations are weighted by the inverse of the number of regions within a country, such that all countries carry equal weight. GDP per capita and unemployment data are measured in 2019 to avoid potential distortions because of the COVID‑19 crisis. The reference category for access to cities is large metropolitan regions.

Source: Almeida et al. (2024[22]), “Geographic inequalities in accessibility of essential services”, https://doi.org/10.1787/12bab9fb-en.

1. By focussing on travel times, this analysis captures only the physical accessibility of essential services; it does not account for other dimensions of accessibility, such as capacity, opening hours or quality. Ongoing follow-up work examines competitive accessibility – a combined measure of physical distance and service capacity relative to local demand – for early childhood education and care services and primary education in Estonia and the Netherlands (Almeida et al., 2026[95]).

2. This analysis is not intended to estimate any causal relationship between service accessibility and regional economic activity, which would require a considerably more complex approach. In particular, a standard OLS model cannot account for potential feedback effects running from service accessibility to economic activity.

3. Across OECD countries, PES have made substantial progress in digitalising their services and operations, reducing the need for jobseekers to physically visit a PES centre. However, even in the digital age, physical accessibility may remain important for the provision of inclusive and effective employment support – particularly for more disadvantaged jobseekers, such as those with limited digital skills, people with disabilities, and others who benefit from face‑to-face support.

2.2.2. Median incomes are higher in regions where employment rates are higher

Regional median incomes are strongly and positively associated with regional employment rates in many countries, suggesting that regional disparities in labour market outcomes are an important driver of disparities in household incomes. Figure 2.10 illustrates this relationship for a selection of OECD countries with a sufficient number of TL3 regions, plotting regional median disposable household incomes against regional employment rates. For the remaining countries, the correlation coefficient between median incomes and employment rates is again reported in Annex Table 2.E.1. The relationship is strongly positive in most countries for which data on both variables are available, and somewhat stronger overall than the relationship between median incomes and GDP per capita shown in Figure 2.9, Panel B. As with GDP per capita, however, a few countries – Austria being a notable example – display no clear association between regional employment rates and median incomes.

Figure 2.10. Regional median incomes increase with employment rates in most countries
Copy link to Figure 2.10. Regional median incomes increase with employment rates in most countries

Median disposable household incomes vs. employment rates across TL3 regions, selected countries, 2023 or latest available year

The somewhat stronger association of regional median incomes with employment rates than with GDP per capita likely reflects the central role of labour earnings as the primary channel through which regional economic conditions translate into household incomes and, hence, living standards. GDP per capita captures a broader range of economic activity – including the capital intensity of local industries and the value added generated by commuters – that does not necessarily translate into higher incomes for resident households. Regional employment rates, by contrast, directly determine the share of working-age adults with access to labour income, which is the principal source of income for most households and the prime determinant of their living standards. Differences in earnings levels across regions – driven by variation in industry composition, productivity gains associated with agglomeration, skill intensity and local labour market conditions – likely reinforce this relationship further, as implied by the earlier finding that high-value‑added jobs are disproportionately concentrated in capital-city regions. Those results are consistent with earlier work on productivity in capital-city regions (OECD, 2024[29]) and the spatial concentration of high-skilled workers in cities (Autor, 2019[96]).

2.2.3. Regional incomes have converged across regions in most countries with available data, reflecting convergence in employment rates

Beyond the cross-sectional disparities in income levels documented above, a key question for policy is whether lagging regions are catching up with more prosperous ones – or falling further behind. Due to data constraints, research on within-country trends in regional disparities has mostly focussed on GDP per capita, painting at best a mixed picture. OECD analysis covering the period starting before the Global Financial Crisis finds that disparities in GDP per capita across regions increased in approximately half of the OECD countries with available data (OECD, 2020[97]; OECD, 2023[1]). The most recent EU Cohesion Report similarly presents mixed evidence, documenting significant convergence in five European OECD countries but divergence in six others (European Commission, 2024[81]). Over a longer time horizon, Ganong and Shoag (2017[5]) show that the long-run convergence of US state incomes – documented for much of the twentieth century by Barro and Sala‑i-Martin (1992[39]) – has sharply declined over the past three decades.

The analysis presented in this section points to a somewhat more encouraging picture for median disposable household incomes for the period from around 2010. Across the 19 OECD countries with available time series data, a majority have experienced convergence in regional median disposable household incomes, according to both convergence metrics introduced in Section 2.1 (see Figure 2.11, Panel A). Specifically, $σ$ ‑convergence – a decline in the relative dispersion of regional median incomes – is observed in 13 out of 19 countries, and $β$ -convergence – catch-up growth by regions with relatively lower median incomes at the start of the observation period – in 15 out of 19 countries. Czechia, Slovenia and Chile experienced the strongest reductions in regional dispersion, with their coefficients of variation declining by more than 20%, driven in all cases by rapid catch-up growth in lower-income regions. The Slovak Republic recorded the most rapid catch-up of all countries but a comparatively slower decline in the overall dispersion of regional median incomes.

Divergence in regional incomes across both measures was confined to a small number of countries. The most notable cases are Lithuania, where the relative dispersion of regional median incomes nearly doubled over the observation period, as well as Sweden and Costa Rica, where it rose by about 23% and 12%, respectively. In these countries, the widening of regional income dispersion reflects a combination of rapid income growth in the top-income capital regions Vilnius, Stockholm and Central, combined with slower‑than‑average growth in many regions that were lagging at the start of the period.11

While these results point to relatively broad convergence in regional incomes since 2010, the magnitude of this convergence is modest in most countries. This is particularly noteworthy given that the observation period coincides with a sustained expansion in employment across most OECD countries, which might have been expected to compress regional income disparities more substantially. That said, the results shown in Figure 2.11 likely understate real convergence in living standards, because the analysis corrects for national-level changes in prices, but not for differential changes in price levels across regions. If regional prices levels – and housing costs in particular – have risen more rapidly in higher‑income regions, as suggested, for example, by previous work on Germany (Ahrend and Lembcke, 2016[98]) and the United States (Hsieh and Moretti, 2019[99]; Fraisse and Pionnier, 2020[100]; Hornbeck and Moretti, 2024[101]), then convergence in living standards was stronger than suggested by Figure 2.11.

Patterns of regional income convergence closely mirror those of convergence in employment rates (Figure 2.11, Panel B), reinforcing the finding of the previous subsection that employment is central to narrowing regional divides in household incomes. This close association is reflected in a relatively high correlation coefficient $ρ$ of 0.67. Most countries are clustered in bottom-left quadrant: 10 out of the 14 countries that saw a narrowing in the dispersion of regional employment rates also saw regional income dispersion decline; in all 3 countries where regional disparities in employment rates widened, this was associated with a simultaneous increase in regional income disparities (top-right quadrant). However, the relationship is less than one‑for-one: the slope of the line of best fit is much flatter than the 45‑degree line, suggesting that convergence in employment rates translates only partially into convergence in median incomes. This may reflect disparities in the type of jobs created in lower‑ versus higher-income regions, as well as the equalising effects of redistributive tax and transfer systems. Sweden stands out as an outlier: while the dispersion in regional employment rates declined, regional income dispersion substantially widened. This reflects a similarly-sized divergence in market incomes, i.e. household incomes before taxes and transfers – with particularly rapid growth in Stockholm and slower-than-average growth in most other regions – which more than offset the equalising effect of employment convergence. This pattern is likely related to the marked increase in overall income inequality in Sweden over the same period, with an increase in the Gini coefficient by 2.5 points, or approximately 10%, attributed in part to the welfare state cutbacks and consequent declines in redistribution (Causa and Hermansen, 2017[102]; Søgaard et al., 2018[103]).

Patterns of regional income convergence also align broadly with corresponding trends in the dispersion of GDP per capita over the same period, though the relationship is considerably weaker (Figure 2.11, Panel C), with a correlation coefficient $ρ$ of just 0.29. Most countries that experienced a decline in the dispersion of regional GDP per capita also saw a reduction in the dispersion of regional median incomes, as shown by the cluster of countries in the bottom-left quadrant of Figure 2.11, Panel C. At the other end of the spectrum, Lithuania, which experienced the greatest divergence in regional median incomes, also saw divergence in regional GDP per capita (top-right quadrant). For several countries, however, the two convergence patterns do not align: Austria, Latvia, the Netherlands and Sweden all combine regional income divergence with convergence in GDP per capita – placing them in the top-left quadrant – while France and Slovenia show the opposite pattern. This weak and imperfect relationship is consistent with the earlier finding that the regional dispersion of GDP per capita and of median disposable household incomes are not always strongly correlated in the cross-section (Figure 2.9). Trends in the dispersion of regional GDP per capita should therefore not be interpreted as a reliable proxy for changes in household incomes or living standards.

An important limitation of the analysis presented in this section is that the finding of relatively broad, if modest, income convergence may not necessarily generalise beyond the countries covered here. The analysis focusses on 19 OECD countries for which time series data on regional median disposable household incomes could be collected; the correlation of convergence trends in median incomes and GDP per capita in Figure 2.11, Panel C draws on a more restricted sample of 15 countries for which also comparable time series on regional GDP per capita are available. These country samples may not be representative of trends across OECD countries more broadly. Indeed, earlier OECD analysis (OECD, 2020[97]; OECD, 2023[1]) had documented growing disparities in GDP per capita since before the Global Financial Crisis in a number of countries not covered in the present analysis for lack of comparable regional income data – such as in Poland, the United Kingdom and the United States.

Figure 2.11. Median incomes have converged across regions since 2010 in most countries, and convergence in incomes closely aligns with convergence in employment rates
Copy link to Figure 2.11. Median incomes have converged across regions since 2010 in most countries, and convergence in incomes closely aligns with convergence in employment rates

Sigma- (left-axis) and Beta-convergence (right-axis) in median disposable incomes (Panel A) and the relationship between the relative change in the coefficient of variation for median disposable income vs. employment rate (Panel B) and vs. GDP per capita (Panel C), TL3 regions, 2010 to 2024 or closest available years

2.2.4. Capital regions are marked by both higher incomes and greater inequality, but more broadly there is no clear link between regional income levels and inequality

The preceding subsection showed that regional income convergence closely relates to convergence in employment rates, but that the relationship is less than one‑for-one. This suggests that differences in the type and quality of jobs across regions, including the concentration of high-value‑added employment in capital regions documented in Section 2.1.3, matter alongside differences in employment levels. This section examines a further dimension of regional income disparities that employment rates and median incomes alone cannot capture: the distribution of incomes across households within regions. Indeed, one of the advantages of using administrative or survey microdata rather than regional accounts to study regional income disparities is precisely that these data can shed light on income inequalities within regions, alongside differences in income levels across regions. This can provide further evidence on the role of spatial sorting for both regional income levels and within-region inequalities, and helps linking regional income disparities and overall inequalities in household incomes.

Within-region income inequality, as measured by the regional Gini coefficient expressed relative to the national Gini, varies across countries to a similar degree as the dispersion in median incomes documented earlier. In many OECD countries, the Gini in the most unequal region is around 10‑30% higher than national-level Gini, while the least unequal regions typically record inequality levels around 10‑20% below the national value (Figure 2.12, Panel A). Capital regions are often the most unequal, while the least unequal regions are nearly always non-metropolitan. Cross-country differences in regional disparities in income inequality align only partially with earlier evidence produced at TL2 level (OECD, 2022[104]), and some countries with many TL3 regions, such as Canada, show considerably greater heterogeneity in the more granular data.

There is no universal relationship between a region’s income level and the degree of within-region inequality across the countries with available data. That said, a positive relationship – whereby higher‑income regions tend to have more unequal income distributions – is the most common pattern. This is the case, for example, in Chile, France, Spain and Sweden, where the most populous regions often combine high income levels with high inequality (Figure 2.12, Panel B). This pattern holds also in most countries (11 out of 14) with a smaller number of regions, for which the correlation coefficient is again reported in Annex Table 2.E.1, though in these cases the relationship is more likely to be driven by outliers. The reverse relationship holds in Austria and Belgium, where the capital regions of Vienna and Brussels record low median incomes alongside relatively high inequality; as described earlier, these low median incomes partly reflect the fact that the Vienna and Brussels Functional Urban Areas stretch over several small regions, with the higher-income suburbs falling into regions neighbouring the capital region.12 No systematic relationship is apparent in the Netherlands and in Japan, where median income and inequality levels do not differ consistently between metropolitan and non-metropolitan regions.

Figure 2.12. Within-region income inequality also varies widely and, in many countries, is greater in higher‑income regions
Copy link to Figure 2.12. Within-region income inequality also varies widely and, in many countries, is greater in higher‑income regions

Gini coefficient for disposable incomes (Panel A) and its relationship with median disposable incomes in selected OECD countries (Panel B), TL3 regions, 2024 or latest available year

Capital regions stand out, on average, for both their higher income levels and greater income inequality relative to the rest of the country. In most countries (15 out of 22), they record median incomes at least 10% above the national median, but the gap is much larger than that in countries such as Australia, France, Lithuania and Chile (Figure 2.13, Panel A).13 These results are consistent with the metropolitan/non‑metropolitan gap in disposable income per capita in regional accounts data (OECD, 2024[29]). Belgium and Austria are again the main exceptions: in both countries, the capital region records median incomes substantially below the national median and those of other region types. In most countries, the higher median incomes in capital regions coincide with substantially higher income inequality (Figure 2.13, Panel B), often reflecting large disparities between urban centres and their surrounding commuting zones (OECD, 2024[29]). Notable exceptions are Australia and the Slovak Republic – where Canberra and Bratislava combine a very high median incomes with low inequality. Beyond the capital, metropolitan regions record median incomes and inequality levels broadly in line with the national values, while non-metropolitan areas have both lower median incomes and inequality. These averages conceal substantial variation within both groups, however, as illustrated in Figure 2.12, and consistent with earlier work on income inequalities within Functional Urban Areas (Boulant, Brezzi and Veneri, 2016[105]).

Figure 2.13. Capital regions record, on average, higher median incomes and higher inequality
Copy link to Figure 2.13. Capital regions record, on average, higher median incomes and higher inequality

Regional median disposable incomes (Panel A) and Gini coefficients (Panel B) averaged across metropolitan and non-metropolitan TL3 regions, 2024 or latest available year

The higher inequality in capital regions reflects the concentration of high‑income households. In most countries with available data, households in the top 5% of the income distribution are significantly overrepresented in capital regions relative to both other metropolitan and non-metropolitan regions (Figure 2.14). Exceptions are once more Austria and Belgium, where, as highlighted earlier, the capital region is surrounded by affluent suburban regions. The distribution of low-income households, defined here as those in the bottom 20% of the distribution, follows a similar but less pronounced pattern: in most countries, low-income households are underrepresented in capital regions and overrepresented in non‑metropolitan regions. Austria and Belgium, as well as Denmark, are again exceptions.

Figure 2.14. High-income households are overrepresented and lower-income households are underrepresented in capital regions
Copy link to Figure 2.14. High-income households are overrepresented and lower-income households are underrepresented in capital regions

Share of the population who are in the top 5% (Panel A) and the bottom 20% (Panel B) of the income distribution, by type of region, 2024 or latest year available

2.2.5. Most household income inequality occurs within regions, not between them

While the regional disparities documented in this chapter are substantial in several countries, a useful way of putting them in perspective is to assess how much they contribute to overall inequality in household incomes. A standard way of doing so is to decompose aggregate income inequality into the part reflecting differences in average incomes between regions and the part due to income differences across households within regions. The Gini index – the most widely used measure of income inequality, and the one used earlier in this chapter to characterise within-region income distributions – cannot be cleanly decomposed in this way, as doing so produces a residual interaction term that cannot be attributed to either component. Alternative, commonly used inequality measures from the generalised entropy family, such as the Theil index and the Mean Logarithmic Deviation, do not suffer from this limitation and can be additively decomposed into between- and within-region components (Bourguignon, 1979[106]; Shorrocks, 1980[107]). Box 2.6 sets out the technical details of the Theil index decomposition used in this subsection.

Box 2.6. Decomposing overall household income inequality into between- and within-region components

Copy link to Box 2.6. Decomposing overall household income inequality into between- and within-region components

One standard approach to evaluating the contribution of regional disparities to overall income inequality is to decompose the Theil index – one of the most widely used measures of income inequality alongside the Gini coefficient – into its between- and within-region components.

The Theil index is defined as:

$Τ = \sum_{r} \sum_{i} (\frac{y_{r i}}{Y}) l n (\frac{y_{r i} / Y}{n_{r i} / N}),$

where $y_{r i}$ is the income of household i residing in region r, and the ratios $y_{r i} / Y$ and $n_{r i} / N$ give the income and population of household i as shares of total national income Y and total population N.

This index can be decomposed as $Τ_{T o t a l} = Τ_{B e t w e e n} + Τ_{W i t h i n}$ , where:

$Τ_{B e t w e e n} = \sum_{r} (\frac{Y_{r}}{Y}) l n (\frac{Y_{r} / Y}{N_{r} / N})$

is the between-region Theil index, with $Y_{r}$ denoting the total income of region r, and the ratios $Y_{r} / Y$ and $N_{r} / N$ giving the income and population of region r as shares of national totals; and.

$Τ_{W i t h i n} = \sum_{r} (\frac{Y_{r}}{Y}) Τ_{r}$

is the within-region component, defined as an income‑weighted average of the Theil indices computed across households within each region r, with $Τ_{r} = \sum_{i \in r} (\frac{y_{r i}}{Y_{r}}) l n (\frac{y_{r i} / Y_{r}}{n_{r i} / N_{r}})$ .

The analysis covers a selection of 20 countries for which Theil indices could be constructed at the TL3 level.

This decomposition shows that regional disparities account for only a small share of overall inequality in disposable household incomes. Across OECD countries with available data, differences in average incomes between regions explain, on average, about 5% of total household income inequality – the remaining 95% reflect differences in incomes across households living within the same region (Figure 2.15).14 The between-region share is largest in countries characterised by wide regional disparities in median incomes, such as Israel, Greece, Lithuania, Chile and Australia (see Figure 2.9); even in Israel though – the only country where the between-region share exceeds 10% – the within-region component remains dominant by a wide margin. At the other end of the spectrum, the between-region share is negligible in New Zealand, Austria, Norway, Denmark and Finland, where it amounts to just 1‑3% of total inequality.

Figure 2.15. Between-region differences account for only a fraction of overall household income inequality
Copy link to Figure 2.15. Between-region differences account for only a fraction of overall household income inequality

Share of overall household income inequality that results from *between-* vs. *within-*region inequalities, TL3 regions, 2024 or latest year available

This finding is consistent with a longer stream of work on the decomposition of income inequality into between- and within-region components (Shorrocks and Wan, 2005[108]; Novotný, 2007[109]; Carrascal-Incera et al., 2020[110]), though most earlier studies focussed on GDP per capita and/or larger geographic units.15 In his study of regional disparities in household disposable income across OECD countries, Piacentini (2014[76]) finds that between-TL2‑region differences account for between 1% and 6% of overall income inequality across eight European countries with available data. A study exploiting more granular data for Chile attributes 5.6% of overall disposable income inequality to variation between TL2‑level regions, a further 2.2% to variation between TL3‑level provinces, and 9% to variation between local communities – with the remaining 83% reflecting income differences across households within the same community (Paredes, Iturra and Lufin, 2014[111]). Together, these results suggest that the geographic contribution to income inequality increases as the unit of analysis becomes more granular, but that within‑area inequality across households remains dominant at every level of aggregation. Comparable results emerge for labour earnings, with Bauluz et al. (2024[40]) finding that between 2% and 8% of national wage inequality across five OECD countries can be explained by differences in mean wages across local labour markets. A cross-country perspective reinforces this picture: Blanchet, Chancel and Gethin (2022[112]) use a Theil decomposition to show that only around 3% of pre‑tax income inequality in the United States is attributable to variation between states – while in Europe, 17% of pre‑tax income inequality reflects variation between countries, rather than within them.

However, the small contribution of between-region income variation to overall inequality should not be read as evidence that place of residence does not matter. It reflects, first and foremost, the fact that any regional aggregation will group together households with widely varying incomes and living standards, and that these within-region differences strongly dominate differences in average incomes across regions. This is illustrated in Annex Figure 2.E.3, which plots the distribution of disposable household incomes across TL3 regions in Belgium and Estonia drawing on administrative microdata. Median disposable household incomes vary considerably across regions in both countries – from 81% of the national median in Belgium’s lowest-income region of Brussels to 121% in the highest-income region of Arlon, and from 75% in Northeast Estonia to 116% in North Estonia. Yet this variation is dwarfed by the spread of incomes within each region, as shown by the 10th and 90th percentile markers – the regional income distributions overlap strongly as a result. The evidence from Chile cited above (Paredes, Iturra and Lufin, 2014[111]) further suggests that part of this within-TL3‑region variation reflects income differences between local communities, implying that the contribution of geographic disparities to overall income inequality may be larger when more granular spatial units are considered. OECD calculations for two countries for which local-level Theil indices could be collected also suggest this is the case: for Austria and Finland, the between-territorial component increases, respectively, from 0.7% to 2.3% and from 2.3% to 3.8% when moving from the TL3 to the LAU level (OECD, forthcoming[20]), underlining the importance of fine‑grained data collection for accurately measuring people’s living conditions.

Moreover, this decomposition analysis provides only a snapshot of spatial inequalities, overlooking how regional disparities shape people’s income trajectories over time. Even in countries where regional gaps in median incomes are relatively modest, residents of low-income regions experience more limited economic opportunities. Compounded over the course of their lives, these constraints translate into weaker prospects for upward mobility and greater exposure to downward mobility. The following section turns to this dynamic dimension.

2.3. Moving up the ladder: Regional disparities in people’s income mobility

Copy link to 2.3. Moving up the ladder: Regional disparities in people’s income mobility

As documented earlier in this chapter, higher regional income levels are closely linked to greater employment opportunities (see Figure 2.10). Previous research using GDP per capita as a proxy for incomes finds that low-income regions also face multiple reinforcing disadvantages along other dimensions, including lesser accessibility of essential services (see Box 2.5), poorer infrastructure and weaker institutions (Charron, Dijkstra and Lapuente, 2014[113]; OECD, 2023[1]; OECD, 2024[29]; OECD, 2025[2]). Low rates of inter-regional mobility documented earlier in this chapter (see Figure 2.6) imply that workers are unlikely to relocate in search of better prospects. The more limited economic opportunities in lower-income regions may therefore imply less scope for income progression for people living in those regions.

This section provides evidence on these regional disparities in income mobility for a selection of countries drawing on a new OECD infrastructure for the study of income dynamics (see details in Annex 2.C). This infrastructure leverages matched tax and other administrative records that track the entire population – or large samples of it – over time, making it possible to observe people’s income and place of residence over longer time periods. The analysis builds on previous OECD work using this infrastructure that studied income mobility at the national level (Königs and Terrero-Dávila, 2025[114]), and represents the first cross-country analysis of mobility in disposable incomes over the life course at TL3 level. The analysis complements previous evidence highlighting the role of place in shaping income mobility across generations (OECD, 2025[2]; Chetty et al., 2026[115]).

People in low-income regions experience, on average, weaker upward income mobility over the life course than those in high-income regions, reflecting inequalities in opportunity across places. To illustrate this, the analysis tracks the income dynamics of low-income working-age individuals over a six‑year period. The analysis focusses on people around the 15th percentile of the national income distribution who remain in the same region throughout. This allows to compare low-income earners with the same starting income who are continuously exposed to different local conditions.16 Given the limited inter-regional mobility in OECD countries (Figure 2.6), however, results do not change substantially when movers are included.

Among these low-income earners, people living in low-income regions – defined as regions in the bottom quintile of regional median incomes – experience both lower relative and absolute income mobility. In particular, they:

Advance less, on average, in their country’s income distribution. For example, in Finland, low‑income earners living in low-income regions move up, on average, 10 percentiles over six years, while their counterparts living in the capital region of Helsinki-Uusimaa advance by 17 percentiles on average (Figure 2.16, Panel A). The gap between people living in low- and high‑income regions is large also in Estonia and Spain, relatively smaller in Norway, and minimal in Canada. Mean advancement in the income distribution is a frequently used measure of relative mobility (Loisel and Sicsic, 2024[116]; Chetty et al., 2014[83]).17
Face a greater risk of stagnating living standards. In most countries, the share of people experiencing no real income growth over six years is also higher for those living in low-income regions (Figure 2.16, Panel B). In Belgium, for example, 31% of low-income earners living in La Louvière, a low-income region in Wallonia, experience no income growth after adjusting for inflation, compared to 22% in the high-income Flemish region of Roeselare. Regional disparities are narrower in Nordic countries and Austria.

As a consequence, there is also greater persistence on lower incomes in low-income regions. In all countries with available data, people at the bottom of the income distribution who live in low-income regions tend to remain on low incomes for longer, while those in high-income regions experience shorter low‑income spells (Figure 2.16, Panel C). Overall, persistence at the bottom of the distribution over six years is greatest in Norway (42% in low-income regions and 37% in high-income regions), and lowest in Estonia (27% vs. 18%).

Weaker upward mobility in low-income regions may partly reflect the characteristics of the people living in those regions, for example if they are older or have lower levels of educational attainment. However, there is strong evidence that regional disparities in resident characteristics are not the main driver of differential income mobility patterns. Previous OECD work for Belgium (OECD, 2025[2]) using the same administrative microdata finds persistent regional differences in income mobility even after accounting for individual characteristics, a finding in line with earlier research from the United States using a similar methodology (Auten and Gee, 2009[117]). Instead, the results suggest that the more limited economic and labour market opportunities in lower-income regions (see Section 2.1) translate into fewer opportunities for income progression and, hence, permanently lower living standards over the life course.

People in economically less dynamic regions are also more exposed to risks of downward mobility, including job loss. In past recessions, regions with a lower employment share in low-productivity sectors (Di Caro, 2014[118]; Crescenzi, Luca and Milio, 2016[119]) and less dynamic labour markets (Fratesi and Rodríguez-Pose, 2016[36]) suffered greater employment loses. Research also shows that workers in low-employment regions face a higher risk of job displacement (Bilal, 2023[4]). For those who do lose their jobs, higher initial unemployment and fewer high-productivity firms in the local labour market are associated with longer unemployment spells and deeper, more persistent earnings losses (Jacobson, Lalonde and Sullivan, 1993[120]; Athey et al., 2024[121]) – see also Chapter 3.

Figure 2.16. Low-income regions offer less opportunity for upward income mobility
Copy link to Figure 2.16. Low-income regions offer less opportunity for upward income mobility

Mobility outcomes over the following five years for people initially in the 15th percentile of the national income distribution, TL3 regions, 2017 to 2022

2.4. Concluding remarks

Copy link to 2.4. Concluding remarks

Persistent regional inequalities carry large economic and societal costs. They can make economies less resilient and leave growth potential in lagging regions untapped. Among residents of lagging regions, the perception of being economically left behind can erode social cohesion and undermine trust in democratic and economic institutions.

Documenting these inequalities at a granular level requires data that go beyond what labour force and household surveys in many countries can offer. While survey data remain the backbone of the labour market analysis in this chapter, the analysis of household incomes and income mobility draws on register‑based microdata – illustrating the vast potential of such data for cross‑country analysis in this area. Administrative microdata on employment and earnings trajectories and firm characteristics have become central to OECD work on labour market dynamics, job mobility and productivity under the LinkEED 2.0 project,18 one output of which is Chapter 3 of this Employment Outlook on the impact of structural change on local labour markets. The analysis in this chapter demonstrates that such data are an equally powerful resource for cross-country analysis of income inequality and mobility – a development made possible by progress in many OECD countries in linking administrative records from various sources and making them available for research and policy analysis. Compared to most surveys, their much larger number of observations enables granular analysis across segments of the income distribution, socio‑demographic groups and geographic locations; their longitudinal dimension allows tracking individuals and households over extended periods of time. Together, these features offer considerable potential for future cross-country work on income trajectories, the role of the household and of social protection systems in cushioning income shocks, and the factors shaping living standards over the life course.

Using these different sources of data, this chapter has documented substantial regional disparities in labour market outcomes and in the distribution of household incomes across many OECD countries. It has shown that people’s labour market prospects are shaped by where they live, and that regional disparities in employment translate into disparities in household incomes and, hence, differences in living standards. While differences in average incomes across regions account for only a fraction of overall household-level income inequality at any given point in time, the chapter has also shown that income mobility is less favourable in lower-income regions. This implies that the reduced economic opportunities associated with living in a region with a weaker labour market, lower economic dynamism, and more limited access to essential infrastructure and services can accumulate over the life course, translating into persistently lower living standards.

These findings underscore the importance of place‑based policies in ensuring that people can access labour market and economic opportunities regardless of where they live. Such policies can reduce the spatial mismatch between workers and firms, align the supply of skills and labour with local demand, improve regional infrastructure and services, and help regions adapt to structural shifts with unequal territorial impact. The OECD has produced essential evidence on the geographic labour market impact of megatrends, such as automation (OECD, 2018[26]), transitions towards low-carbon economies (OECD, 2023[27]), the rise of Generative AI (OECD, 2024[28]) and the demographic transition (OECD, 2025[122]). The recent report Place‑based Policies for the Future synthesises existing evidence on the design and evaluation of place‑based interventions (OECD, 2025[13]); the OECD Recommendation on Regional Development Policy and its accompanying implementation toolkit (OECD, 2023[12]) set out guiding principles for broader regional development policies, covering capacity‑building, multi-level governance, and the role of regional and local taxes, fiscal transfers and equalisation mechanisms in reducing regional disparities. These policies are an essential complement to national-level policies that aim at developing people’s skills and employability, raising productivity and wages, and ensuring adequate social protection and effective redistribution – see also Chapter 3.

References

[24] Ahrend, R. et al. (2026), “Measuring industrial restructuring pressure in regions: An employment-based index”, OECD Local Economic and Employment Development (LEED) Papers, No. 2026/01, OECD Publishing, Paris, https://doi.org/10.1787/14baf9f3-en.

[25] Ahrend, R. et al. (2026), “Resilient, bouncing back or trapped?: Mapping responses to industrial restructuring pressure”, OECD Local Economic and Employment Development (LEED) Papers, No. 2026/02, OECD Publishing, Paris, https://doi.org/10.1787/1817c1cc-en.

[98] Ahrend, R. and A. Lembcke (2016), “Does It Pay to Live in Big(ger) Cities? : The Role of Agglomeration Benefits, Local Amenities, and Costs of Living”, OECD Regional Development Working Papers, No. 2016/9, OECD Publishing, Paris, https://doi.org/10.1787/e0490ba8-en.

[91] Alasia, A., J. Amann and P. Horvat (2025), “Towards measuring purchasing power parity across OECD regions”, Methodological background note submitted to the 49th Meeting of the Working Party on Territorial Indicators, CFE/RDPC/TI(2025)4.

[95] Almeida, V. et al. (2026), “Geographic disparities in service accessibility in Estonia and the Netherlands: How transport modes, proximity and capacity shape accessibility across cities, towns and rural areas”, OECD Social, Employment and Migration Working Papers, OECD Publishing, Paris, https://doi.org/10.1787/63b9abef-en.

[22] Almeida, V. et al. (2024), “Geographic inequalities in accessibility of essential services”, OECD Social, Employment and Migration Working Papers, No. 307, OECD Publishing, Paris, https://doi.org/10.1787/12bab9fb-en.

[70] Anselin, L. (1995), “Local Indicators of Spatial Association—LISA”, Geographical Analysis, Vol. 27/2, pp. 93-115, https://doi.org/10.1111/j.1538-4632.1995.tb00338.x.

[65] Artamonova, A. and A. Syse (2021), “Do older parents’ assistance needs deter parent-child geographic divergence in Norway?”, Health & Place, Vol. 70, p. 102599, https://doi.org/10.1016/j.healthplace.2021.102599.

[121] Athey, S. et al. (2024), “The Heterogeneous Earnings Impact of Job Loss Across Workers, Establishments, and Markets”, NBER Working Paper 34946, https://www.nber.org/system/files/working_papers/w34946/w34946.pdf.

[117] Auten, G. and G. Gee (2009), “Income Mobility in the United States: New Evidence from Income Tax Data”, National Tax Journal, Vol. 62/2, pp. 301-328, https://doi.org/10.17310/ntj.2009.2.05.

[96] Autor, D. (2019), “Work of the Past, Work of the Future”, AEA Papers and Proceedings, Vol. 109, pp. 1-32, https://doi.org/10.1257/pandp.20191110.

[54] Autor, D., D. Dorn and G. Hanson (2016), “The China Shock: Learning from Labor-Market Adjustment to Large Changes in Trade”, Annual Review of Economics, Vol. 8/1, pp. 205-240, https://doi.org/10.1146/annurev-economics-080315-015041.

[39] Barro, R. and X. Sala-i-Martin (1992), “Convergence”, Journal of Political Economy, Vol. 100/2, pp. 223-251, https://doi.org/10.1086/jpe.100.2.2138605.

[40] Bauluz, L. et al. (2024), “Spatial Wage Inequality in North America and Western Europe: Changes Between and Within Local Labour Markets 1975-2019”, Banque de France Working Paper #957, https://www.banque-france.fr/en/publications-and-statistics/publications/spatial-wage-inequality-north-america-and-western-europe-changes-between-and-within-local-labour.

[49] Beugelsdijk, S., M. Klasing and P. Milionis (2017), “Regional economic development in Europe: the role of total factor productivity”, Regional Studies, Vol. 52/4, pp. 461-476, https://doi.org/10.1080/00343404.2017.1334118.

[4] Bilal, A. (2023), “The Geography of Unemployment”, The Quarterly Journal of Economics, Vol. 138/3, pp. 1507-1576, https://doi.org/10.1093/qje/qjad010.

[112] Blanchet, T., L. Chancel and A. Gethin (2022), “Why Is Europe More Equal than the United States?”, American Economic Journal: Applied Economics, Vol. 14/4, pp. 480-518, https://doi.org/10.1257/app.20200703.

[84] Bonnet, F., H. d’Albis and A. Sotura (2021), “Income Inequality across French Departments over the Last 100 Years”, Economie et Statistique / Economics and Statistics 526-527, pp. 49-69, https://doi.org/10.24187/ecostat.2021.526d.2052.

[105] Boulant, J., M. Brezzi and P. Veneri (2016), “Income Levels And Inequality in Metropolitan Areas: A Comparative Approach in OECD Countries”, OECD Regional Development Working Papers, No. 2016/6, OECD Publishing, Paris, https://doi.org/10.1787/5jlwj02zz4mr-en.

[106] Bourguignon, F. (1979), “Decomposable Income Inequality Measures”, Econometrica, Vol. 47/4, p. 901, https://doi.org/10.2307/1914138.

[123] Breau, S. and R. Saillant (2016), “Regional income disparities in Canada: exploring the geographical dimensions of an old debate”, Regional Studies, Regional Science, Vol. 3/1, pp. 463-481, https://doi.org/10.1080/21681376.2016.1244774.

[64] Caliendo, M., S. Künn and R. Mahlstedt (2017), “The return to labor market mobility: An evaluation of relocation assistance for the unemployed”, Journal of Public Economics, Vol. 148, pp. 136-151, https://doi.org/10.1016/j.jpubeco.2017.02.008.

[43] Card, D., J. Rothstein and M. Yi (2023), Location, Location, Location, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w31587.

[110] Carrascal-Incera, A. et al. (2020), “UK interregional inequality in a historical and international comparative context”, National Institute Economic Review, Vol. 253, pp. R4-R17, https://doi.org/10.1017/nie.2020.26.

[77] Castells-Quintana, D., R. Ramos and V. Royuela (2015), “Income inequality in European Regions: Recent trends and determinants”, Review of Regional Research, Vol. 35/2, pp. 123-146, https://doi.org/10.1007/s10037-015-0098-4.

[59] Causa, O., M. Abendschein and M. Cavalleri (2021), “The laws of attraction: Economic drivers of inter-regional migration, housing costs and the role of policies”, OECD Economics Department Working Papers, No. 1679, OECD Publishing, Paris, https://doi.org/10.1787/da8e368a-en.

[102] Causa, O. and M. Hermansen (2017), “Income redistribution through taxes and transfers across OECD countries”, OECD Economics Department Working Papers, No. 1453, OECD Publishing, Paris, https://doi.org/10.1787/bc7569c6-en.

[9] Causa, O. and J. Pichelmann (2020), “Should I stay or should I go? Housing and residential mobility across OECD countries”, OECD Economics Department Working Papers, No. 1626, OECD Publishing, Paris, https://doi.org/10.1787/d91329c2-en.

[55] Celli, V., A. Cerqua and G. Pellegrini (2023), “The long-term effects of mass layoffs: do local economies (ever) recover?”, Journal of Economic Geography, Vol. 23/5, pp. 1121-1144, https://doi.org/10.1093/jeg/lbad012.

[89] Chan, J., S. Ellingsen and H. Simpson (2025), “Spatial Inequality, Regional Growth and Economic Geography”, Bank of England Staff Working Paper, Vol. 1152, https://www.bankofengland.co.uk/-/media/boe/files/working-paper/2025/spatial-inequality-regional%20growth-and-economic-geography.pdf.

[113] Charron, N., L. Dijkstra and V. Lapuente (2014), “Mapping the Regional Divide in Europe: A Measure for Assessing Quality of Government in 206 European Regions”, Social Indicators Research, Vol. 122/2, pp. 315-346, https://doi.org/10.1007/s11205-014-0702-y.

[115] Chetty, R. et al. (2026), “The Opportunity Atlas: Mapping the Childhood Roots of Social Mobility”, American Economic Review, Vol. 116/1, pp. 1-51, https://doi.org/10.1257/aer.20200108.

[83] Chetty, R. et al. (2014), “Where is the land of Opportunity? The Geography of Intergenerational Mobility in the United States *”, The Quarterly Journal of Economics, Vol. 129/4, pp. 1553-1623, https://doi.org/10.1093/qje/qju022.

[23] Cho, J. and H. Jeon (2025), “The geography of opportunity: Job accessibility and income in 10 Functional Urban Areas in the United States”, OECD Regional Development Papers, No. 157, OECD Publishing, Paris, https://doi.org/10.1787/51094842-en.

[52] Clarke, G., R. Martin and P. Tyler (2016), “Divergent cities? Unequal urban growth and development”, Cambridge Journal of Regions, Economy and Society, Vol. 9/2, pp. 259-268, https://doi.org/10.1093/cjres/rsw011.

[72] Combes, P., G. Duranton and L. Gobillon (2008), “Spatial wage disparities: Sorting matters!”, Journal of Urban Economics, Vol. 63/2, pp. 723-742, https://doi.org/10.1016/j.jue.2007.04.004.

[42] Combes, P. and L. Gobillon (2015), “The Empirics of Agglomeration Economies”, in Handbook of Regional and Urban Economics, Elsevier, https://doi.org/10.1016/b978-0-444-59517-1.00005-2.

[67] Cracolici, M., M. Cuffaro and P. Nijkamp (2008), “A spatial analysis on Italian unemployment differences”, Statistical Methods and Applications, Vol. 18/2, pp. 275-291, https://doi.org/10.1007/s10260-007-0087-z.

[119] Crescenzi, R., D. Luca and S. Milio (2016), “The geography of the economic crisis in Europe: national macroeconomic conditions, regional structural factors and short-term economic performance”, Cambridge Journal of Regions, Economy and Society, Vol. 9/1, pp. 13-32, https://doi.org/10.1093/cjres/rsv031.

[8] De la Roca, J. (2017), “Selection in initial and return migration: Evidence from moves across Spanish cities”, Journal of Urban Economics, Vol. 100, pp. 33-53, https://doi.org/10.1016/j.jue.2017.04.004.

[118] Di Caro, P. (2014), “Recessions, recoveries and regional resilience: evidence on Italy”, Cambridge Journal of Regions, Economy and Society, Vol. 8/2, pp. 273-291, https://doi.org/10.1093/cjres/rsu029.

[35] Dijkstra, L., E. Garcilazo and P. McCann (2015), “The effects of the global financial crisis on European regions and cities”, Journal of Economic Geography, Vol. 15/5, pp. 935-949, https://doi.org/10.1093/jeg/lbv032.

[85] Dzhavatova, K. et al. (2025), “Nordic spatial variations in demographic, social, and economic characteristics: A visual guide”, Nordregio Working Paper 2025:2, https://doi.org/10.6027/wp2025:2.1403-2511.

[129] Elbers, C. et al. (2007), “Reinterpreting between-group inequality”, The Journal of Economic Inequality, Vol. 6/3, pp. 231-245, https://doi.org/10.1007/s10888-007-9064-x.

[80] Erfurth, P. (2023), “Is income inequality converging at the regional level? Evidence from LIS data”, Journal of Economic Geography, Vol. 24/3, pp. 375-392, https://doi.org/10.1093/jeg/lbad029.

[21] European Commission (2024), Employment and Social Developments in Europe: Upward Social Convergence in the EU and the Role of Social Investment, Publications Office of the European Union, https://employment-social-affairs.ec.europa.eu/employment-and-social-developments-europe-2024_en.

[81] European Commission (2024), Ninth Report on Economic, Social and Territorial Cohesion, Publications Office of the European Union, https://doi.org/10.2776/585966.

[38] European Commission (2014), Investment for jobs and growth: Promoting development and good governance in EU regions and cities: Sixth report on economic, social and territorial cohesion, Publications Office, Brussels, https://data.europa.eu/doi/10.2776/15327.

[37] European Commission (2013), The urban and regional dimension of the crisis: Eighth progress report on economic, social and territorial cohesion, Publications Office, Brussels, https://data.europa.eu/doi/10.2776/74866.

[74] Eurostat (2025), Disposable income of private households by NUTS 2 region, https://doi.org/10.2908/TGS00026.

[17] Fadic, M. et al. (2019), “Classifying small (TL3) regions based on metropolitan population, low density and remoteness”, OECD Regional Development Working Papers, No. 2019/06, OECD Publishing, Paris, https://doi.org/10.1787/b902cc00-en.

[73] Fendel, T. (2014), “Work-related Migration and Unemployment”, Journal for Labour Market Research, Vol. 47/3, pp. 233-243, https://doi.org/10.1007/s12651-014-0160-7.

[127] Florida, R. and C. Mellander (2014), “The Geography of Inequality: Difference and Determinants of Wage and Income Inequality across US Metros”, Regional Studies, Vol. 50/1, pp. 79-92, https://doi.org/10.1080/00343404.2014.884275.

[100] Fraisse, A. and P. Pionnier (2020), “Location, location, location – House price developments across and within OECD countries”, OECD Statistics, https://oecdstatistics.blog/2020/07/10/location-location-location-house-price-developments-across-and-within-oecd-countries/.

[63] Franklin, S. (2017), “Location, Search Costs and Youth Unemployment: Experimental Evidence from Transport Subsidies”, The Economic Journal, Vol. 128/614, pp. 2353-2379, https://doi.org/10.1111/ecoj.12509.

[36] Fratesi, U. and A. Rodríguez-Pose (2016), “The crisis and regional employment in Europe: what role for sheltered economies?”, Cambridge Journal of Regions, Economy and Society, Vol. 9/1, pp. 33-57, https://doi.org/10.1093/cjres/rsv032.

[68] Furková, A. and M. Chocholatá (2020), “Spatial econometric approach to the EU regional employment process”, Central European Journal of Operations Research, Vol. 29/3, pp. 1037-1056, https://doi.org/10.1007/s10100-020-00714-5.

[5] Ganong, P. and D. Shoag (2017), “Why has regional income convergence in the U.S. declined?”, Journal of Urban Economics, Vol. 102, pp. 76-90, https://doi.org/10.1016/j.jue.2017.07.002.

[86] Gbohoui, W., W. Lam and V. Lledo (2019), “The Great Divide: Regional Inequality and Fiscal Policy”, IMF Working Papers, Vol. 19/88, p. 1, https://doi.org/10.5089/9781498311625.001.

[51] Glaeser, E. and J. Kohlhase (2004), “Cities, regions and the decline of transport costs”, Papers in Regional Science, Vol. 83/1, pp. 197-228, https://doi.org/10.1007/s10110-003-0183-x.

[61] Gobillon, L. and H. Selod (2021), “Spatial Mismatch, Poverty, and Vulnerable Populations”, in Handbook of Regional Science, Springer Berlin Heidelberg, Berlin, Heidelberg, https://doi.org/10.1007/978-3-662-60723-7_7.

[48] Hanushek, E., J. Ruhose and L. Woessmann (2017), “Knowledge Capital and Aggregate Income Differences: Development Accounting for US States”, American Economic Journal: Macroeconomics, Vol. 9/4, pp. 184-224, https://doi.org/10.1257/mac.20160255.

[101] Hornbeck, R. and E. Moretti (2024), “Estimating Who Benefits from Productivity Growth: Local and Distant Effects of City Productivity Growth on Wages, Rents, and Inequality”, Review of Economics and Statistics, Vol. 106/3, pp. 587-607, https://doi.org/10.1162/rest_a_01208.

[99] Hsieh, C. and E. Moretti (2019), “Housing Constraints and Spatial Misallocation”, American Economic Journal: Macroeconomics, Vol. 11/2, pp. 1-39, https://doi.org/10.1257/mac.20170388.

[11] Iammarino, S., A. Rodriguez-Pose and M. Storper (2018), “Regional inequality in Europe: evidence, theory and policy implications”, Journal of Economic Geography, Vol. 19/2, pp. 273-298, https://doi.org/10.1093/jeg/lby021.

[120] Jacobson, L., R. Lalonde and D. Sullivan (1993), “Earnings Losses of Displaced Workers”, The American Economic Review, Vol. 83/4, pp. 685-709, https://www.jstor.org/stable/2117574.

[53] Kemeny, T. and M. Storper (2020), “Superstar cities and left-behind places: disruptive innovation, labor demand, and interregional inequality”, Vol. III Working Paper 41/International Inequalities Institute, London School of Economics and Political Science, https://doi.org/10.21953/lse.8uvxv9nscxrc.

[114] Königs, S. and J. Terrero-Dávila (2025), “Who climbs the income ladder?: Cross-country evidence on income mobility from tax record data”, OECD Social, Employment and Migration Working Papers, No. 322, OECD Publishing, Paris, https://doi.org/10.1787/eadbbcbf-en.

[50] Lacroix, J. and E. Vidal-Coso (2018), “Differences in Labor Supply by Birthplace and Family Composition in Switzerland: the Role of Human Capital and Household Income”, Journal of International Migration and Integration, Vol. 20/3, pp. 659-684, https://doi.org/10.1007/s12134-018-0623-8.

[47] Lindenlaub, I., R. Oh and M. Peters (2022), Firm Sorting and Spatial Inequality, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w30637.

[116] Loisel, T. and M. Sicsic (2024), “Intragenerational Income Mobility in France Over the 2003‑2021 Period”, Economie et Statistique/Economics and Statistics, Vol. 545, pp. 65-82, https://www.insee.fr/fr/statistiques/fichier/8561915/04_ES545_Loisel-Siscic_FR.pdf.

[45] López‐Bazo, E. and E. Motellón (2013), “The regional distribution of unemployment: What do micro‐data tell us?”, Papers in Regional Science, Vol. 92/2, pp. 383-406, https://doi.org/10.1111/j.1435-5957.2012.00456.x.

[60] Manning, A. and B. Petrongolo (2017), “How Local Are Labor Markets? Evidence from a Spatial Job Search Model”, American Economic Review, Vol. 107/10, pp. 2877-2907, https://doi.org/10.1257/aer.20131026.

[124] Marchand, Y., J. Dubé and S. Breau (2020), “Exploring the Causes and Consequences of Regional Income Inequality in Canada”, Economic Geography, Vol. 96/2, pp. 83-107, https://doi.org/10.1080/00130095.2020.1715793.

[14] McKay, L., W. Jennings and G. Stoker (2021), “Political Trust in the “Places That Don’t Matter””, Frontiers in Political Science, Vol. 3, https://doi.org/10.3389/fpos.2021.642236.

[31] Menon, C. and W. Vermeulen (2026), “What drives regional productivity differentials?: Evidence from firm-level data in Italy and Spain”, OECD Local Economic and Employment Development (LEED) Papers, No. 2026/3, OECD Publishing, Paris, https://doi.org/10.1787/8fba4b12-en.

[15] Mitsch, F., N. Lee and E. Ralph Morrow (2021), “Faith no more? The divergence of political trust between urban and rural Europe”, Political Geography, Vol. 89, p. 102426, https://doi.org/10.1016/j.polgeo.2021.102426.

[126] Moller, S., A. Alderson and F. Nielsen (2009), “Changing Patterns of Income Inequality in U.S. Counties, 1970–2000”, American Journal of Sociology, Vol. 114/4, pp. 1037-1101, https://doi.org/10.1086/595943.

[69] Moran, P. (1950), “Biometrika, 37(1/2)”, Biometrika 37(1/2), pp. 17-23, https://www.jstor.org/stable/2332142.

[3] Moretti, E. (2012), The New Geography of Jobs, Houghton Mifflin Harcourt.

[41] Moretti, E. (2010), Local Labor Markets, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w15947.

[109] Novotný, J. (2007), “On the measurement of regional inequality: does spatial dimension of income inequality matter?”, The Annals of Regional Science, Vol. 41/3, pp. 563-580, https://doi.org/10.1007/s00168-007-0113-y.

[75] OECD (2026), OECD Regions, Cities and Local Areas Database - Income inequality - Regions, https://www.oecd.org/en/topics/regions-cities-and-local-statistics.html.

[7] OECD (2025), Addressing Regional Labour Market Imbalances in Austria, OECD Reviews on Local Job Creation, OECD Publishing, Paris, https://doi.org/10.1787/0cf6186b-en.

[57] OECD (2025), “OECD guidelines on the multi-level governance of demographic change”, OECD Regional Development Papers, No. 164, OECD Publishing, Paris, https://doi.org/10.1787/f66426ea-en.

[128] OECD (2025), OECD Income Distribution Database, https://www.oecd.org/en/data/datasets/income-and-wealth-distribution-database.html.

[90] OECD (2025), OECD Regions, Cities and Local Areas database, https://www.oecd.org/en/topics/regions-cities-and-local-statistics.html.

[13] OECD (2025), Place-Based Policies for the Future, OECD Regional Development Studies, OECD Publishing, Paris, https://doi.org/10.1787/e5ff6716-en.

[122] OECD (2025), Shrinking Smartly and Sustainably: Strategies for Action, OECD Rural Studies, OECD Publishing, Paris, https://doi.org/10.1787/f91693e3-en.

[34] OECD (2025), “The role of subnational governments in adult skills systems”, OECD Local Economic and Employment Development (LEED) Papers, No. 2025/02, OECD Publishing, Paris, https://doi.org/10.1787/d452e8b7-en.

[2] OECD (2025), To Have and Have Not – How to Bridge the Gap in Opportunities, OECD Publishing, Paris, https://doi.org/10.1787/dec143ad-en.

[28] OECD (2024), Job Creation and Local Economic Development 2024: The Geography of Generative AI, OECD Publishing, Paris, https://doi.org/10.1787/83325127-en.

[29] OECD (2024), OECD Regions and Cities at a Glance 2024, OECD Publishing, Paris, https://doi.org/10.1787/f42db3bf-en.

[33] OECD (2023), ““Who does what” for active labour market policies: A zoom on the role of subnational governments”, OECD Local Economic and Employment Development (LEED) Papers, No. 2023/09, OECD Publishing, Paris, https://doi.org/10.1787/d8d6868d-en.

[27] OECD (2023), Job Creation and Local Economic Development 2023: Bridging the Great Green Divide, OECD Publishing, Paris, https://doi.org/10.1787/21db61c1-en.

[93] OECD (2023), Main Findings from the 2022 OECD Risks that Matter Survey, OECD Publishing, Paris, https://doi.org/10.1787/70aea928-en.

[30] OECD (2023), “Migration and regional productivity: Evidence from individual wages in Australia”, OECD Regional Development Papers, No. 60, OECD Publishing, Paris, https://doi.org/10.1787/7bc64c78-en.

[1] OECD (2023), OECD Regional Outlook 2023: The Longstanding Geography of Inequalities, OECD Publishing, Paris, https://doi.org/10.1787/92cd40a0-en.

[12] OECD (2023), Recommendation of the Council on Regional Development Policy, OECD/LEGAL/0492, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0492.

[104] OECD (2022), OECD Regions and Cities at a Glance 2022, OECD Publishing, Paris, https://doi.org/10.1787/14108660-en.

[94] OECD (2022), Policy Statement on Moving Beyond the COVID-19 Crisis to a Better Labour Market that Works for All, OECD/LEGAL/0477, https://legalinstruments.oecd.org/en/instruments/OECD-LEGAL-0477.

[62] OECD (2021), Brick by Brick: Building Better Housing Policies, OECD Publishing, Paris, https://doi.org/10.1787/b453b043-en.

[92] OECD (2021), Delivering Quality Education and Health Care to All: Preparing Regions for Demographic Change, OECD Rural Studies, OECD Publishing, Paris, https://doi.org/10.1787/83025c02-en.

[58] OECD (2020), Housing and Inclusive Growth, OECD Publishing, Paris, https://doi.org/10.1787/6ef36f4b-en.

[97] OECD (2020), OECD Regions and Cities at a Glance 2020, OECD Publishing, Paris, https://doi.org/10.1787/959d5ba0-en.

[26] OECD (2018), Job Creation and Local Economic Development 2018: Preparing for the Future of Work, OECD Publishing, Paris, https://doi.org/10.1787/9789264305342-en.

[10] OECD (2012), Promoting Growth in All Regions, OECD Regional Development Studies, OECD Publishing, Paris, https://doi.org/10.1787/9789264174634-en.

[18] OECD (2005), OECD Employment Outlook 2005, OECD Publishing, Paris, https://doi.org/10.1787/empl_outlook-2005-en.

[19] OECD (forthcoming), “Geographic disparities in labour market outcomes in EU and OECD countries”, OECD Social, Employment and Migration Working Papers, OECD Publishing, Paris.

[20] OECD (forthcoming), “The geography of income inequalities in OECD countries”, OECD Social, Employment and Migration Working Papers, OECD Publishing, Paris.

[44] Overman, H. and X. Xu (2024), “Spatial disparities across labour markets”, Oxford Open Economics, Vol. 3/Supplement_1, pp. i585-i610, https://doi.org/10.1093/ooec/odae005.

[71] Panichella, N. and R. Impicciatore (2024), “The selectivity of internal movers: An analysis of the relationship between education, social origin, and geographical mobility in Europe”, Population Studies, Vol. 78/3, pp. 393-411, https://doi.org/10.1080/00324728.2024.2388555.

[111] Paredes, D., V. Iturra and M. Lufin (2014), “A Spatial Decomposition of Income Inequality in Chile”, Regional Studies, Vol. 50/5, pp. 771-789, https://doi.org/10.1080/00343404.2014.933798.

[125] Partridge, M. (2005), “Does Income Distribution Affect U.S. State Economic Growth?”, Journal of Regional Science, Vol. 45/2, pp. 363-394, https://doi.org/10.1111/j.0022-4146.2005.00375.x.

[66] Patacchini, E. and P. Rice (2007), “Geography and Economic Performance: Exploratory Spatial Data Analysis for Great Britain”, Regional Studies, Vol. 41/4, pp. 489-508, https://doi.org/10.1080/00343400600928384.

[76] Piacentini, M. (2014), “Measuring Income Inequality and Poverty at the Regional Level in OECD Countries”, OECD Statistics Working Papers, No. 2014/3, OECD Publishing, Paris, https://doi.org/10.1787/5jxzf5khtg9t-en.

[87] Pina, Á. and P. Sicari (2021), “Enhancing regional convergence in the European Union”, OECD Economics Department Working Papers, No. 1696, OECD Publishing, Paris, https://doi.org/10.1787/253dd6ee-en.

[46] Roca, J. and D. Puga (2016), “Learning by Working in Big Cities”, The Review of Economic Studies, Vol. 84/1, pp. 106-142, https://doi.org/10.1093/restud/rdw031.

[88] Rosés, J. and N. Wolf (2021), “Regional growth and inequality in the long-run: Europe, 1900–2015”, Oxford Review of Economic Policy, Vol. 37/1, pp. 17-48, https://doi.org/10.1093/oxrep/graa062.

[78] Royuela, V., P. Veneri and R. Ramos (2018), “The short-run relationship between inequality and growth: evidence from OECD regions during the Great Recession”, Regional Studies, Vol. 53/4, pp. 574-586, https://doi.org/10.1080/00343404.2018.1476752.

[82] Savoia, F. (2024), “Income inequality convergence among EU regions”, Socio-Economic Planning Sciences, Vol. 92, p. 101803, https://doi.org/10.1016/j.seps.2024.101803.

[107] Shorrocks, A. (1980), “The Class of Additively Decomposable Inequality Measures”, Econometrica, Vol. 48/3, p. 613, https://doi.org/10.2307/1913126.

[108] Shorrocks, A. and G. Wan (2005), “Spatial decomposition of inequality”, Journal of Economic Geography, Vol. 5/1, pp. 59-81, https://doi.org/10.1093/jnlecg/lbh054.

[103] Søgaard, J. et al. (2018), Increasing Income Inequality in the Nordics, Nordic Council of Ministers, Copenhagen, https://doi.org/10.6027/tn2018-519.

[6] Spring, A., B. Gillespie and C. Mulder (2023), “Internal migration following adverse life events: Assessing the likelihood of return migration and migration toward family”, Population, Space and Place, Vol. 30/3, https://doi.org/10.1002/psp.2711.

[16] Stroppe, A. (2023), “Left behind in a public services wasteland? On the accessibility of public services and political trust”, Political Geography, Vol. 105, p. 102905, https://doi.org/10.1016/j.polgeo.2023.102905.

[32] Usai, S. (2011), “The Geography of Inventive Activity in OECD Regions”, Regional Studies, Vol. 45/6, pp. 711-731, https://doi.org/10.1080/00343401003792492.

[79] Veneri, P. and F. Murtin (2018), “Where are the highest living standards? Measuring well-being and inclusiveness in OECD regions”, Regional Studies, Vol. 53/5, pp. 657-666, https://doi.org/10.1080/00343404.2018.1463091.

[56] Vermeulen, W. and N. Braakmann (2023), “How do mass lay-offs affect regional economies?”, OECD Local Economic and Employment Development (LEED) Papers, No. 2023/01, OECD Publishing, Paris, https://doi.org/10.1787/99d48aeb-en.

Annex 2.A. Sources of data on regional labour market outcomes

Copy link to Annex 2.A. Sources of data on regional labour market outcomes

The analysis in Section 2.1 relies on aggregate regional employment and unemployment rates. For 18 of the 32 countries covered, these indicators are drawn from the OECD Database on Regions, Cities and Local Areas (https://www.oecd.org/en/topics/regions-cities-and-local-statistics.html) and refer to the population aged 15 to 64. For the remaining countries, data were collected from statistical authorities’ websites, which has permitted extending data coverage or including more recent years. Exceptions to this are Chile (own calculations based on CASEN) and Poland (data provided by national authorities). The newly collected data are mostly drawn from LFS, though in some countries they derive from census or administrative records to achieve TL3‑level granularity. Cross-country comparability of results based on these sources may therefore be somewhat lower than for standard OECD indicators, given different definitions of employment status. In France, for example, census data – in which employment status is self-declared – yield higher unemployment rates and lower employment rates than the LFS; in Denmark register-based unemployment rates, which count as unemployed only those enrolled with the public employment service, fall below LFS estimates. Cross-country comparisons should accordingly be interpreted with caution wherever the underlying data source differs. For most countries, the data refer to 15‑to‑64‑year‑olds or a similar age group (e.g. 16‑64 for Denmark); where only a substantially different age group was available, this is clearly flagged in the chapter. Annex Table 2.A.1 below provides full overview.

Country	Data source	Years	Age group
AUS	Downloaded from Australian Bureau of Statistics Underlying data: national labour force survey + estimation model	2012, 2024	15‑64
AUT	Employment: OECD Database on Regions, Cities and Local Areas Unemployment: AMS (Arbeitsmarktservice)	2010, 2024	Employment: 15‑64 Unemployment: Labour force potential
BEL	Downloaded from the Institute Wallon De L’Évaluation, De La Prospective Et De La Statistique (IWEPS) Underlying data: national labour force survey + calibration with administrative records	2010, 2022	15‑64
CAN	Downloaded from Statistics Canada Underlying data: population census	2021	15‑64
CHE	Downloaded from Federal Statistical Office Underlying data: population census	2010, 2024	15‑64
CHL	Own calculations based on CASEN microdata	2011, 2022	15‑64
CZE	OECD Database on Regions, Cities and Local Areas	2010, 2023	15‑64
DEU	OECD Database on Regions, Cities and Local Areas	2010, 2023	15‑64
DNK	Downloaded from Statistics Denmark Underlying data: administrative records (RAS/AMR_UN)	2010, 2024	16‑64
ESP	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
EST	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
FIN	Downloaded from Statistics Finland Underlying data: national labour force survey	2010, 2024	Employment: 15‑64 Unemployment: 15‑74
FRA	Downloaded from INSEE Underlying data: population census	2010, 2022	15‑64
GBR	Downloaded from Office for National Statistics Underlying data: national labour force survey	2010, 2024	16‑64
HUN	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
IRL	OECD Database on Regions, Cities and Local Areas	Employment: 2010, 2024 Unemployment: 2010, 2023	15‑64
ISR	OECD Database on Regions, Cities and Local Areas	2010-2023	15‑64
ITA	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
JPN	Downloaded from e‑Stat Japan Underlying data: national labour force survey + estimation model	2010-2024	15+
KOR	OECD Database on Regions, Cities and Local Areas	2010, 2023	15‑64
LTU	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
LVA	Downloaded from Central Statistical Bureau Underlying data: national labour force survey	2010-2023	15‑64
NLD	Central Statistical Bureau Underlying data: national labour force Survey	2013-2023	15‑75
NOR	OECD Database on Regions, Cities and Local Areas	2010, 2023	15‑64
NZL	OECD Database on Regions, Cities and Local Areas	2010, 2023	15‑64
POL	Tabulations provided by Statistics Poland Underlying data: administrative records	Employment: 2023 Unemployment: 2024	Employment: 15‑64 Unemployment (men): 18‑65 Unemployment (women): 18‑60
SVK	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
SVN	OECD Database on Regions, Cities and Local Areas	2010, 2022	15‑64
SWE	OECD Database on Regions, Cities and Local Areas	2010, 2024	15‑64
TUR	Downloaded from TurkStat Underlying data: national labour force survey	2023	15+
USA	OECD Database on Regions, Cities and Local Areas	2010, 2019	15‑64

* In Austria, the labour force potential is calculated as the sum of the number of unemployed and employed people registered under social insurance institutions.

Annex 2.B. Sources of data on regional income distributions

Copy link to Annex 2.B. Sources of data on regional income distributions

Section 2.2 presents newly collected income distribution statistics across TL3 regions for 22 OECD countries. For 13 countries (Austria, Belgium, Canada, Denmark, Estonia, Finland, France, Greece, Ireland, the Netherlands, Norway, Spain and Sweden), income distribution data are based on administrative registers – mainly tax records, often matched with other sources of administrative data; for the remaining 11 countries (Australia, Chile, Costa Rica, Czechia, Israel, Japan, Latvia, Lithuania, New Zealand, the Slovak Republic and Slovenia), the income data come from surveys that include TL3‑level information (see Annex Table 2.B.1 for an overview).

Country	Data provider	Data source	Years
AUS	Australian Bureau of Statistics	Survey of Income and Housing	2020
AUT	Statistics Austria	Administrative	2012-2022
BEL	OECD calculations using administrative records (see Annex Table 2.C.1)	Administrative	2016-2023
CAN	Statistics Canada	Administrative	2010, 2015, 2020
CHL	Instituto Nacional de Estadística	Casen (Encuesta de Caracterización Socioeconómica Nacional)	2011, 2022
CRI	Instituto Nacional de Estadística y Censos	Encuesta Nacional de Hogares	2010, 2023
CZE	Czech Statistical Office and LIS	EU-SILC and LIS (based on EU-SILC)	2010-2023
DNK	Statistics Denmark and LIS	Administrative	2010-2023
ESP	OECD calculations using administrative records (see Annex Table 2.C.1)	Administrative	2016-2022
EST	OECD calculations using administrative records (see Annex Table 2.C.1)	Administrative	2017-2024
FIN	Statistics Finland	Administrative	2010-2024
FRA	Institut National De La Statistique Et Des Études Économiques (Insee)	Administrative	2012-2021
GRC	OECD calculations using administrative records (see Annex Table 2.C.1)	Administrative	2017-2023
IRL	Central Statistics Office	Administrative	2022
ISR	OECD calculations using LIS	LIS (based on Household Expenditure Survey)	2010-2022
JPN	Statistics Bureau of the Ministry of Internal Affairs and Communications	National Survey of Family Income, Consumption and Wealth	2019
LTU	Statistics Lithuania and OECD calculations using LIS	EU-SILC and LIS (based on SILC)	2010-2021, 2023
LVA	Central Statistical Bureau of Latvia	EU-SILC	2010-2023
NLD	Statistics Netherlands	Administrative	2011-2023
NOR	Statistics Norway	Administrative	2023
NZL	Stats NZ, provided to the OECD by the New Zealand Treasury	Household Economic Survey	2019-2024
SVK	OECD calculations using LIS	LIS (based on SILC)	2010, 2013-2018
SVN	Statistical Office of the Republic of Slovenia	EU-SILC	2010-2023
SWE	Statistics Sweden	Administrative	2011-2024

Administrative data can have several advantages over survey data for studying regional income distributions, notably their much larger number of observations, which permit mapping household incomes with great granularity. However, they also come with certain limitations, reflecting that these data originate as a by-product of administrative processes, such as tax collection, rather than to have been produced with the explicit intent to facilitate empirical analysis. Administrative data are therefore less standardised than “traditional” household survey data, notably in their coverage of income sources and definition of the household as the relevant unit of observation.

Great care has been taken to ensure a high degree of comparability between the survey- and register‑based income data used in this chapter, and to closely document any remaining differences in data coverage and definitions. Indeed, for all countries covered in the analysis, the data sources used for this chapter provide a broad coverage of the main sources of household income, permitting the reliable calculation of disposable household incomes (see Annex Table 2.B.2 for an overview). In particular:

Labour income – comprising both employment and self-employment income, as well as public pensions – is typically fully captured by the data, as are the corresponding taxes and social security contributions paid.
Capital income is also well covered, though coverage is only partial in a few countries (Austria, Belgium, Finland, Ireland). This income source likely represents a more sizable share of disposable income for households at the top of the income distribution, but the partial coverage would likely not significantly affect median incomes.
Government transfers, private/occupational pensions, and inter-household transfers are fully covered in most countries. Any existing gaps likely account for only a small fraction of disposable household income.

All income distribution statistics presented in Section 2.2 have been calculated over individuals attributing to each person the equivalised disposable household income of the household they live in. In most countries, household incomes were adjusted for household size by dividing disposable household incomes by the square root of the household size; in some, the modified OECD scale or national equivalence scales were used.

In the analysis presented in Section 2.2, regional disparities in income levels and inequalities are in all cases expressed relative to the national values, rather than in absolute currency values or as Gini coefficients. Those relative values are likely less sensitive to smaller differences in methodology or coverage of certain income sources across countries.

	AUS	AUT	BEL	CAN	CHL
Data source	Survey	Administrative	Administrative	Administrative	Survey
Equivalence scale	Sq. root	Mod. OECD	Sq. root	Sq. root	Sq. root
Employment income	Yes	Yes	Yes	Yes	Yes
Self-employment income	Yes	Yes	Yes	Yes	Yes
Capital / property income	Yes	Partial	Partial	Yes	Yes
Public pensions	Yes	Yes	Yes	Yes	Yes
Private/occupational pensions	Yes	Partial	Partial	Yes	Yes
Net inter-household transfers (e.g. alimony)	Yes	No	Yes	Yes	Yes
After taxes on labour and pension income (deducted)	Yes	Yes	Yes	Yes	Yes
After taxes on capital income (deducted)	Yes	Yes	Yes	Yes	Yes
After social security contributions (deducted)	Yes	Yes	Yes	No	Yes
Unemployment benefits	Yes	Yes	Yes	Yes	Yes
Family / parental and child benefits	Yes	Yes	Yes	Yes	Yes
Health and disability benefits	Yes	Yes	Yes	Yes	Yes
Education-related benefits	Yes	No	No	Yes	Yes
Housing benefits	Yes	No	No	Yes	Yes
Other old age and survivor benefits	Yes	No	Yes	Yes	Yes
Other social assistance benefits	Yes	Yes	Yes	Yes	Yes
	CRI	CZE	DNK	ESP	EST
Data source	Survey	Survey	Administrative	Administrative	Administrative
Equivalence scale	Sq. root	Sq. root	Mod. OECD	Sq. root	Sq. root
Employment income	Yes	Yes	Yes	Yes	Yes
Self-employment income	Yes	Yes	Yes	Yes	Yes
Capital / property income	Yes	Yes	Yes	Yes	Yes
Public pensions	Yes	Yes	Yes	Yes	Yes
Private/occupational pensions	Yes	Partial	Yes	Yes	Yes
Net inter-household transfers (e.g. alimony)	Partial	Yes	Yes	Yes	Yes
After taxes on labour and pension income (deducted)	Yes	Yes	Yes	Yes	Yes
After taxes on capital income (deducted)	No	Yes	Yes	Yes	Yes
After social security contributions (deducted)	Yes	Yes	Yes	Yes	Yes
Unemployment benefits	Yes	Yes	Yes	Yes	Yes
Family / parental and child benefits	No	Yes	Yes	Yes	Yes
Health and disability benefits	No	Yes	Yes	Yes	No
Education-related benefits	Yes	Yes	Yes	Yes	Yes
Housing benefits	Yes	Yes	Yes	Yes	Yes
Other old age and survivor benefits	Yes	Yes	Yes	Yes	Yes
Other social assistance benefits	No	Yes	Yes	Yes	Yes
	FIN	FRA	GRC	IRL	ISR
Data source	Administrative	Administrative	Administrative	Administrative	Survey
Equivalence scale	Mod. OECD	Mod. OECD	Sq. root	Sq. root	Sq. root
Employment income	Yes	Yes	Yes	Yes	Yes
Self-employment income	Yes	Yes	Yes	Yes	Yes
Capital / property income	Partial	Yes	Yes	Partial	Yes
Public pensions	Yes	Yes	Yes	Yes	Yes
Private/occupational pensions	Yes	Partial	Yes	Yes	Yes
Net inter-household transfers (e.g. alimony)	Yes	Yes	No	No	Yes
After taxes on labour and pension income (deducted)	Yes	Yes	Yes	Yes	Yes
After taxes on capital income (deducted)	Yes	Yes	Yes	Yes	Yes
After social security contributions (deducted)	Yes	Yes	Yes	Yes	Yes
Unemployment benefits	Yes	Yes	Yes	Yes	Yes
Family / parental and child benefits	Yes	Yes	Yes	Yes	Yes
Health and disability benefits	Yes	Yes	Yes	Yes	Yes
Education-related benefits	Yes	Yes	No	Yes	Yes
Housing benefits	Yes	Yes	Yes	Yes	Yes
Other old age and survivor benefits	Yes	Yes	Yes	Yes	Yes
Other social assistance benefits	Yes	Yes	Yes	Yes	Yes
	JPN	LTU	LVA	NLD	NOR
Data source	Survey	Survey	Survey	Administrative	Administrative
Equivalence scale	Sq. root	Sq. root / Mod. OECD	Mod. OECD	National scale	Sq. root
Employment income	Yes	Yes	Yes	Yes	Yes
Self-employment income	Yes	Yes	Yes	Yes	Yes
Capital / property income	Yes	Yes	Yes	Yes	Yes
Public pensions	Yes	Yes	Yes	Yes	Yes
Private/occupational pensions	Yes	Yes	Yes	Yes	Yes
Net inter-household transfers (e.g. alimony)	Yes	Partial	Yes	Yes	Yes
After taxes on labour and pension income (deducted)	Yes	Yes	Yes	Yes	Yes
After taxes on capital income (deducted)	Yes	Yes	Yes	Yes	Yes
After social security contributions (deducted)	Yes	Yes	Yes	Yes	Yes
Unemployment benefits	Yes	Yes	Yes	Yes	Yes
Family / parental and child benefits	Yes	Yes	Yes	Yes	Yes
Health and disability benefits	Yes	Yes	Yes	Yes	Yes
Education-related benefits	Yes	Yes	Yes	Yes	Yes
Housing benefits	Yes	Yes	Yes	Yes	Yes
Other old age and survivor benefits	Yes	Yes	Yes	Yes	Yes
Other social assistance benefits	Yes	Yes	Yes	Yes	Yes
	NZL	SVK	SVN	SWE
Data source	Survey	Survey	Survey	Administrative
Equivalence scale	Sq. root	Sq. root	Mod. OECD	National scale
Employment income	Yes	Yes	Yes	Yes
Self-employment income	Yes	Yes	Yes	Yes
Capital / property income	Yes	Yes	Yes	Yes
Public pensions	Yes	Yes	Yes	Yes
Private/occupational pensions	Yes	Yes	Yes	Yes
Net inter-household transfers (e.g. alimony)	Partial	Yes	Yes	Yes
After taxes on labour and pension income (deducted)	Yes	Yes	Yes	Yes
After taxes on capital income (deducted)	Yes	Yes	Yes	Yes
After social security contributions (deducted)	N/A	Yes	Yes	Yes
Unemployment benefits	Yes	Yes	Yes	Yes
Family / parental and child benefits	Yes	Yes	Yes	Yes
Health and disability benefits	Yes	Yes	Yes	Yes
Education-related benefits	Yes	Yes	Yes	Yes
Housing benefits	Partial	Yes	Yes	Yes
Other old age and survivor benefits	Yes	Yes	Yes	Yes
Other social assistance benefits	Yes	Yes	Yes	Yes

Annex 2.C. Sources of data on regional income mobility

Copy link to Annex 2.C. Sources of data on regional income mobility

The analysis in Section 2.3 presents evidence on people’s mobility in disposable household incomes across regions for 9 OECD countries, drawing on a new OECD microdata infrastructure for the study of income dynamics. The infrastructure relies on administrative data, mainly tax records linked with other administrative income data and population registries, which permit tracking people both geographically and over time. In 5 countries (AUT, BEL, ESP, EST, GRC), the OECD has access to the administrative microdata and the results presented in the chapter come from OECD calculations. For the 4 other countries (CAN, FIN, NOR, SWE), the results come from tabulations provided by the national authorities especially prepared for this chapter. Annex Table 2.C.1 below provides a detailed overview.

Great care has been taken to ensure a high degree of comparability across administrative sources from different countries, even as the presented analysis of disparities in income mobility within countries is likely less sensitive to smaller cross-country differences in methodology or income coverage. The definition of disposable income mirrors the one used in Section 2.2, and the underlying administrative microdata are the same in both sections. For a detailed description of the income and tax components covered in the data for each country, see Annex Table 2.B.2.

Country	Data source
AUT	Own calculations using administrative records from Statistics Austria, in particular: LUE – Integrierte Lohn- und Einkommensteuerstatistik (LUE – Integrated Wage and Income Tax Statistics) AEST REGZ – Abgestimmte Erwerbsstatistik und Registerzählung (AEST REGZ – Co‑ordinated Employment Statistics and Register Census).
BEL	OECD calculations using administrative records from StatBel, in particular: Administrative Disposable Incomes (ADI) Database Population census
CAN	Tabulations provided by Statistics Canada based on tax and population records
ESP	OECD calculations using administrative records from Institute for Fiscal Studies, in particular: Panel de Hogares (Household Panel) Muestra Continua de Vidas Laborales (Continuous Work History Sample)
EST	OECD calculations using administrative records from Statistics Estonia, in particular: EUROMOD-SILC administrative database
FIN	Tabulations provided by Esa Karonen (University of Turku) using tax and population records in co‑ordination with the Finnish Ministry of Economic Affairs and Employment and the Ministry of Social Affairs and Health
GRC	OECD calculations replicated by the Ministry of Labour and Social Security using administrative records, in particular: Tax records (AADE)
NOR	Tabulations provided by Statistics Norway based on tax and population records
SWE	Tabulations provided by Statistics Sweden based on tax and population records

Annex 2.D. Regions excluded from the analysis

Copy link to Annex 2.D. Regions excluded from the analysis

Figure	Regions excluded from the analysis
Figure 2.1. Regional disparities in labour market outcomes are large in most OECD countries (Panel A. Employment)	Regions with missing data: AUS: AU510 (Western Australia – Outback (North), AU511 (Western Australia – Outback (South) CHE: CH054 (Appenzell Innerrhoden) CHL: CL012 (Tamarugal), CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL114 (General Carrera), CL122 (Antártica Chilena), CL152 (Parinacota) DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis) ESP: ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES703 (El Hierro), ES704 (Fuerteventura), ES705 (Gran Canaria), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES709 (Tenerife) FRA: FRY50 (Mayotte) GBR: UKM61 (Caithness & Sutherland and Ross & Cromarty), UKM62 (Inverness & Nairn and Moray, Badenoch & Strathspey), UKM63 (Lochaber, Skye & Lochalsh, Arran & Cumbrae and Argyll & Bute), UKM64 (Eilean Siar (Western Isles)), UKM65 (Orkney Islands), UKM66 (Shetland Islands), UKE13 (North and North East Lincolnshire), UKF25 (North Northamptonshire), UKN06 (Belfast), UKN07 (Armagh City, Banbridge and Craigavon), UKN08 (Newry, Mourne and Down), UKN09 (Ards and North Down), UKN0A (Derry City and Strabane), UKN0B (Mid Ulster), UKN0C (Causeway Coast and Glens), UKN0D (Antrim and Newtownabbey), UKN0E (Lisburn and Castlereagh), UKN0F (Mid and East Antrim), UKN0G (Fermanagh and Omagh) ITA: ITG2D (Sassari), ITG2E (Nuoro), ITG2F (Cagliari), ITG2G (Oristano), ITG2H (Sud Sardegna) KOR: KR054 (Sejong) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) NZL: NZ016 (Hawke’s Bay Region), NZ022 (West Coast Region) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.1. Regional disparities in labour market outcomes are large in most OECD countries (Panel B. Unemployment)	Regions with missing data: AUS: AU510 (Western Australia – Outback (North), AU511 (Western Australia – Outback (South) CHE: CH054 (Appenzell Innerrhoden) CHL: CL012 (Tamarugal), CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL114 (General Carrera), CL122 (Antártica Chilena), CL152 (Parinacota) DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis) ESP: ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES703 (El Hierro), ES704 (Fuerteventura), ES705 (Gran Canaria), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES709 (Tenerife) FIN: FI1D5 (Central Ostrobothnia), FI1D8 (Kainuu), FI200 (Åland) FRA: FRY50 (Mayotte) ITA: ITG2D (Sassari), ITG2E (Nuoro), ITG2F (Cagliari), ITG2G (Oristano), ITG2H (Sud Sardegna) KOR: KR054 (Sejong) BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.2. Low-employment regions have been catching up, narrowing regional disparities	Regions with missing data: AUS: AU510 (Western Australia – Outback (North), AU511 (Western Australia – Outback (South) CHE: CH054 (Appenzell Innerrhoden) CHL: CL012 (Tamarugal), CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL114 (General Carrera), CL122 (Antártica Chilena), CL152 (Parinacota) DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis) ESP: ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES703 (El Hierro), ES704 (Fuerteventura), ES705 (Gran Canaria), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES709 (Tenerife) FRA: FRY50 (Mayotte) GBR: UKM61 (Caithness & Sutherland and Ross & Cromarty), UKM62 (Inverness & Nairn and Moray, Badenoch & Strathspey), UKM63 (Lochaber, Skye & Lochalsh, Arran & Cumbrae and Argyll & Bute), UKM64 (Eilean Siar (Western Isles)), UKM65 (Orkney Islands), UKM66 (Shetland Islands), UKE13 (North and North East Lincolnshire), UKF25 (North Northamptonshire), UKN06 (Belfast), UKN07 (Armagh City, Banbridge and Craigavon), UKN08 (Newry, Mourne and Down), UKN09 (Ards and North Down), UKN0A (Derry City and Strabane), UKN0B (Mid Ulster), UKN0C (Causeway Coast and Glens), UKN0D (Antrim and Newtownabbey), UKN0E (Lisburn and Castlereagh), UKN0F (Mid and East Antrim), UKN0G (Fermanagh and Omagh) ITA: ITG2D (Sassari), ITG2E (Nuoro), ITG2F (Cagliari), ITG2G (Oristano), ITG2H (Sud Sardegna) KOR: KR054 (Sejong) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.3. Workers’ observable characteristics explain a limited part of regional disparities in employment outcomes	Regions with missing data: CHL: CL012 (Tamarugal), CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL114 (General Carrera), CL122 (Antártica Chilena), CL152 (Parinacota) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft)
Figure 2.4. People’s labour market outcomes are substantially worse in low-GDP-per-worker regions	Regions with missing data: CHE: CH054 (Appenzell Innerrhoden) DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis) ESP: ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES703 (El Hierro), ES704 (Fuerteventura), ES705 (Gran Canaria), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES709 (Tenerife) · FRA: FRY50 (Mayotte) GBR: UKN06 (Belfast), UKN07 (Armagh City, Banbridge and Craigavon), UKN08 (Newry, Mourne and Down), UKN09 (Ards and North Down), UKN0A (Derry City and Strabane), UKN0B (Mid Ulster), UKN0C (Causeway Coast and Glens), UKN0D (Antrim and Newtownabbey), UKN0E (Lisburn and Castlereagh), UKN0F (Mid and East Antrim), UKN0G (Fermanagh and Omagh) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: NZL: NZ016 (Hawke’s Bay Region), NZ022 (West Coast Region)
Figure 2.6. People move out of low-employment regions and into high-employment ones, but at a slow pace	Regions with missing data: DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis) ESP: ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES703 (El Hierro), ES704 (Fuerteventura), ES705 (Gran Canaria), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES709 (Tenerife) ITA: ITG2D (Sassari), ITG2E (Nuoro), ITG2F (Cagliari), ITG2G (Oristano), ITG2H (Sud Sardegna) KOR: KR054 (Sejong)
Figure 2.7. Regions with high unemployment are spatially clustered (Panel A)	Regions without neighbours: USA: US008 (Anchorage, AK), US074 (Honolulu, HI) JPN: JPJ47 (Okinawa), JPA01 (Hokkaido) Regions with missing data: DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis)
Figure 2.7. Regions with high unemployment are spatially clustered (Panel B)	Regions without neighbours: USA: US008 (Anchorage, AK), US074 (Honolulu, HI) Regions with missing data: DEU: DEG04 (Suhl, Kreisfreie Stadt), DEG0B (Schmalkalden-Meiningen), DEG0F (Ilm-Kreis), DEG0H (Sonneberg), DEG0I (Saalfeld-Rudolstadt), DEG0N (Eisenach, Kreisfreie Stadt), DEG0P (Wartburgkreis)
Figure 2.9. Regional disparities in median household incomes vary substantially across countries, but are much narrower than disparities in GDP per capita (Panel A. Regional median equivalised disposable incomes across TL3 regions)	Regions with missing data: CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre). FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.9. Regional disparities in median household incomes vary substantially across countries, but are much narrower than disparities in GDP per capita (Panel B. Median disposable incomes vs. GDP per capita across TL3 regions, selected OECD countries)	Regions with missing data: ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre). FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife)
Figure 2.10. Regional median incomes increase with employment rates in most countries	Regions with missing data: CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre), ES703 (El Hierro), ES704 (Fuerteventura), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES705 (Gran Canaria), ES709 (Tenerife) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft)
Figure 2.11 (Panel A. Change in coefficient of variation and β-convergence)	Regions with missing data: BEL: BE335 (Arr. Verviers – Communes francophones), BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL122 (Antártica Chilena) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NZL: NZ015 (Gisborne), NZ016 (Hawke’s Bay), NZ021 (Tasman‑Nelson‑Marlborough), NZ022 (West Coast) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.11 (Panel B. Median income vs. Employment rate)	Regions with missing data: CHL: CL012 (Tamarugal), CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL114 (General Carrera), CL122 (Antártica Chilena), CL152 (Parinacota) ESP: ES531 (Eivissa y Formentera), ES532 (Mallorca), ES533 (Menorca), ES703 (El Hierro), ES704 (Fuerteventura), ES705 (Gran Canaria), ES706 (La Gomera), ES707 (La Palma), ES708 (Lanzarote), ES709 (Tenerife), ES211 (Araba/Álava), ES212 (Gipuzkoa), E213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte)
Figure 2.11 (Panel C. Median income vs. GDP per capita)	Regions with missing data: CHL: CL052 (Isla de Pascua), CL104 (Palena), CL122 (Antártica Chilena) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.12. Within-region income inequality also varies widely and, in many countries, is greater in higher‑income regions (Panel A. Gini coefficient)	Regions with missing data: AUS: AU315 (Outback, Qld) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP : ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.12. Within-region income inequality also varies widely and, in many countries, is greater in higher‑income regions (Panel B. Gini coefficient vs. median income)	Regions with missing data: AUS: AU315 (Outback, Qld), AU204 (Hume, Vic), AU217 (Warrnambool and S.W., Vic) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP : ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.13. Capital regions record, on average, higher median incomes and higher inequality (Panel A. Median income)	Regions with missing data: AUS: AU315 (Outback, Qld) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.13. Capital regions record, on average, higher median incomes and higher inequality (Panel B. Gini coefficient)	Regions with missing data: AUS: AU315 (Outback, Qld) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.14. High-income households are overrepresented and lower-income households are underrepresented in capital regions.	ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) Regions aggregated due to data availability constraints: ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife)
Figure 2.15. Between-region differences account for only a fraction of overall household income inequality	Regions with missing data: AUS: AU315 (Outback, Qld), AU204 (Hume, Vic), AU217 (Warrnambool and S.W., Vic), AU308 (Central, Qld), AU201 (Ballarat, Vic), AU113 (Riverina, NSW) CHL: CL052 (Isla de Pascua), CL104 (Palena), CL113 (Capitán Prat), CL122 (Antártica Chilena), CL152 (Parinacota) ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) FRA: FRY10 (Guadeloupe), FRY20 (Martinique), FRY30 (French Guiana), FRY40 (La Réunion), FRY50 (Mayotte) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife) NZL: NZ016 (Hawke’s Bay Region) aggregated with NZ015 (Gisborne Region); NZ022 (West Coast Region) aggregated with NZ021 (Tasman – Nelson – Marlborough)
Figure 2.16. Low-income regions offer less opportunity for upward income mobility	Regions with missing data: ESP: ES211 (Araba/Álava), ES212 (Gipuzkoa), ES213 (Biscay), ES220 (Navarre) NOR: NO0B1 (Jan Mayen), NO0B2 (Svalbard) Regions aggregated due to data availability constraints: BEL: BE335 (Arr. Verviers – communes francophones) aggregated with BE336 (Bezirk Verviers – Deutschsprachige Gemeinschaft) ESP: ES704 (Fuerteventura) and ES708 (Lanzarote) aggregated with ES705 (Gran Canaria); ES706 (La Gomera), ES703 (El Hierro) and ES707 (La Palma) aggregated with ES709 (Tenerife)

Annex 2.E. Additional results

Copy link to Annex 2.E. Additional results

Annex Figure 2.E.1. Capital-city regions often exhibit higher-employment rates
Copy link to Annex Figure 2.E.1. Capital-city regions often exhibit higher-employment rates

Employment rates by access to cities typology, 2024 or latest available year

Annex Figure 2.E.2. Regional disparities in GDP per capita are large in many OECD countries
Copy link to Annex Figure 2.E.2. Regional disparities in GDP per capita are large in many OECD countries

GDP per capita, TL3 regions for selected OECD countries, 2024 or latest available year

Annex Figure 2.E.3. Regional income distributions have a very large degree of overlap
Copy link to Annex Figure 2.E.3. Regional income distributions have a very large degree of overlap

Median, percentile 10 and percentile 90 of the disposable household income distribution for Belgium in 2023 (Panel A) and Estonia in 2024 (Panel B), TL3 regions

Country	Number of TL3 regions	Correlation coefficient ρ of median income vs. GDP per capita	Correlation coefficient ρ of median income vs. employment rate	Correlation coefficient ρ of median income vs. Gini coefficient
AUS	49	n/a	0.40	0.36
AUT	35	0.05	0.02	‑0.67
BEL	42	0.06	0.60	‑0.57
CAN	293	n/a	n/a	0.15
CHL	51	n/a	0.84	0.86
CRI	6	n/a	0.79	0.22
CZE	14	0.74	0.26	0.58
DNK	11	0.18	0.91	0.46
ESP	50	0.78	0.68	0.32
EST	5	0.82	0.96	0.95
FIN	19	0.80	0.68	0.91
FRA	96	0.71	0.62	0.72
GRC	52	0.60	n/a	0.58
IRL	8	0.64	0.45	0.89
ISR	6	n/a	0.99	‑0.73
JPN	47	0.57	0.64	‑0.10
LTU	10	0.96	0.85	0.40
LVA	6	0.60	0.91	0.73
NLD	40	‑0.07	0.69	0.07
NOR	11	n/a	0.59	0.69
NZL	10	0.82	0.47	0.65
SVK	8	0.85	0.86	‑0.75
SVN	12	0.51	0.65	n/a
SWE	21	0.76	0.20	0.92

Note: Entries marked n/a indicate cases where TL3‑level data on employment rates, GDP per capita or the Gini coefficient were not available. The number of regions refers to the number of TL3 regions covered by the microdata. For information on observation years and data sources, see Figure 2.9, Figure 2.10 and Figure 2.12.

Source: Income distribution indicators collected from national authorities (AUS, AUT, CAN, CRI, DNK, FIN, FRA, IRL, JPN, LVA, NLD, NOR, NZL, SVN, SWE) and OECD calculations based on microdata from administrative records (BEL, ESP, EST, GRC), the LIS database (CZE, ISR, LTU, SVK) and CASEN (CHL). GDP per capita from the OECD Database on Regions, Cities and Local Areas, https://www.oecd.org/en/topics/regions-cities-and-local-statistics.html (accessed in February 2026).

Notes

Copy link to Notes

← 1. These data are available in the OECD Database on Regions, Cities and Local Areas (https://www.oecd.org/en/topics/regions-cities-and-local-statistics.html).

← 2. Similar clustering patterns are observed also for employment rates.

← 3. This applies notably to the EU-SILC and to most surveys covered by the Luxembourg Income Study (LIS), the two most important sources of cross-country microdata on household incomes.

← 4. In a few non-European countries, studies of geographic income inequalities have also been conducted using large survey data, including the Chilean CASEN (Paredes, Iturra and Lufin, 2014[111]), the Canadian Census (Breau and Saillant, 2016[123]; Marchand, Dubé and Breau, 2020[124]), the US Census (Partridge, 2005[125]; Moller, Alderson and Nielsen, 2009[126]) and the US American Communities Survey (Florida and Mellander, 2014[127]).

← 5. In an earlier OECD study, Boulant, Brezzi and Veneri (2016[105]) use micro‑aggregated administrative data to assess income inequalities, but only within large Functional Urban Areas (those with at least 500 000 inhabitants).

← 6. A related literature examines geographic inequalities in employment income or wages; see, for example, Bauluz et al. (2024[40]) on long-run trends in spatial wage disparities between and within local labour market areas in Canada, France, West Germany, the United Kingdom and the United States, and Cho and Jeon (2025[23]) on disparities in jobs and earnings across functional urban areas in the United States.

← 7. In Austria, Vienna’s two surrounding suburbs – Wiener Umland/Nordteil and Wiener Umland/Südteil – are the country’s highest-income regions. In Belgium, the two regions surrounding Brussels – Leuven and Halle‑Vilvoorde – have the 2nd and 4th highest median incomes in the country, respectively. Earlier OECD work has aggregated these regions into a single unit for empirical analysis, see Box 2.1, but such an approach that is not possible with regional median incomes and Gini coefficients.

← 8. Of the countries covered in this section, Canada has by far the largest number of regions (293, all covered), followed by France (101, 96 covered), Spain (59, 50 covered), Chile (56, 51 covered), Greece (52, all covered), and Australia (50, 49 covered).

← 9. Costa Rica, Chile, Lithuania and Israel are among the OECD countries with the highest income inequality, with Gini coefficients of 0.46, 0.45, 0.35 and 0.34, respectively, compared to an OECD average of 0.31 (OECD, 2025[128]).

← 10. The same point applies though to standard national-level income distributions, which equally do not account for the fact that households at different income levels may face systematically different costs of living depending on where they live.

← 11. Austria presents a less intuitive case combining $β$ -convergence with $σ$ ‑divergence. As in Section 2.1.1 for employment rates, this occurs when regions with low median incomes at the outset rapidly overtake regions with initially high median incomes, but dispersion at the end of the period is wider than at the start.

← 12. In both countries, dropping the capital region attenuates the relationship observed in Figure 2.12, Panel B, but the correlation coefficient still takes values very close to zero.

← 13. These gaps are considerably smaller again than those observed for GDP per capita: for countries with regional data available for both indicators, the average capital to non-metropolitan gap in median incomes amounts to 17% of the national median, less than one‑third of the 59% gap observed for GDP per capita. This figure is based on a slightly smaller number of countries than those shown in Figure 2.13, as regional GDP-per-capita data at TL3 level are not available for Australia, Canada, Chile and Norway.

← 14. Elbers et al. (2007[129]) have proposed an alternative measure for this type of inequality decomposition that accounts for the number of different groups as well as their relative sizes. However, as the authors indicate, if the largest region accounts for less than 30% of the population, their alternative measure yields nearly identical results as the conventional decomposition approach followed here. Out of the 780 regions across the 20 countries with available data for the Theil decomposition, only 7 account for more than 30% of the country’s population and only 2 account for more than 33%.

← 15. Related OECD analysis examines the contribution of regions to disparities in GDP per capita across OECD and EU countries, finding an overall decline in inequality over the last two decades, but slowing convergence between countries – as measured by the between-country Theil index – and divergence within countries, as measured by the within-country between-TL3‑region Theil index (Pina and Sicari, 2021[87]; OECD, 2023[1]; OECD, 2024[29]).

← 16. Using a narrow percentile band of the national distribution avoids comparing individuals with different starting incomes, as people in broader groups, such as the bottom quintile, already show substantial differences in incomes across regions. Using the national, as opposed to the regional, income distribution as the benchmark for mobility means that, in low-income regions, people at a given national percentile sit higher in the regional income distribution than their counterparts in high-income regions. Where regional income inequalities are large, this can imply comparing individuals who differ in important ways and may have different mobility prospects. However, using regional distributions instead would also not provide an apples-to‑apples comparison: it would compare individuals with very different absolute starting incomes, which directly affects mobility estimates. Moreover, differences in the shape of regional income distributions could translate equal income gains into very different relative mobility patterns across regions. More importantly, this approach would obscure part of the regional variation that the analysis aims to capture, namely that in low-income regions higher incomes may be structurally out of reach.

The sample further excludes individuals with zero or negative incomes in any year of the mobility window. This serves to limit the influence of cross-country differences in how administrative records treat zero and missing incomes, to exclude individuals who arrive mid-year in the country, and to remove people whose large negative capital income in one period offsets positive capital income in others, distorting mobility transitions in “real” living standards.

← 17. On average, people at the lower end of the distribution tend to move up in the distribution over time, while those at the top tend to move down – a phenomenon referred to as “mean reversion”.

← 18. See https://www.oecd.org/en/about/projects/linkeed-200.html.

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