key: cord-0958658-7wexnggz
authors: Cörvers, Frank; Mayhew, Ken
title: Regional inequalities: causes and cures
date: 2021-04-05
journal: Oxf Rev Econ Policy
DOI: 10.1093/oxrep/graa067
sha: 030684542dbfcd7b1a0343baf8f21fbf5d95b383
doc_id: 958658
cord_uid: 7wexnggz

Significant regional inequalities of income and wealth exist in every Western European country and in North America, but their extent varies from country to country. In both Europe and the US, it is generally thought that they tended to narrow from the early 1900s until about 1980, since when they have widened. This widening has become associated with the rise of populism, while the Covid-19 crisis has thrown regional disadvantage into sharp relief. This article discusses measurement issues, traces developments over time, and explores the social and economic consequences of regional disparities. It describes the evolution of regional policy, and in particular the move to more localized approaches in Europe, analysing their strengths and weaknesses.

This issue of OxREP is concerned with economic disparities between regions in Europe and in the United States.

Significant regional inequalities of income and wealth exist in every Western European country and in North America, but their extent varies from country to country. Exactly how one country compares with another depends upon the spatial unit of analysis studied, as well as upon the measures of economic performance and inequality used. On virtually all measures, regional inequality is particularly high in the UK, higher even than in the US. Regional inequalities change across time. In both Europe and the US it is generally thought that they tended to narrow from the early 1900s until about 1980, since when they have increased. The articles in this issue are concerned with: why we should care; what exactly it is that we are measuring; how and why regional inequalities have evolved over time; and what policy-makers have done and should do to address the problem. *Maastricht University, Netherlands; e-mail: frank.corvers@maastrichtuniversity.nl **Pembroke College, Oxford, UK, and Maastricht University, Netherlands; e-mail: ken.mayhew@education.ox.ac.uk We are grateful to Chris Adam and Cameron Hepburn for their comments on an earlier draft of this paper.

A simple classical or neoclassical view of the world would predict that regional disparities were transient. In time, markets would adjust to bring different areas closer to equality. Labour would leave poorer areas for richer areas and capital would move in the opposite direction. However, in regional economics, as in in other relevant disciplines, there are contrasting strands of literature. As Van Dijk and Edzes (2016) put it: 'In economic geography circles, the debate is between space-neutral theories, where labour is seen as highly mobile, and place-based approaches that emphasize the underdevelopment traps associated with location-specific externalities and (the) potential market failures.' The latter approach is illustrated by Patricia Rice and Tony Venables (2021, this issue) , who remind us of a number of reasons why convergence forces regarding labour and capital mobility might be weak to non-existent. Because labour markets tend to be national, there is relatively little scope for wage adjustment and this lack of adjustment will dampen the willingness of investors to move into the poorer areas. The people who move out are likely to be the young and the skilled, meaning that the remaining workforce is relatively unattractive to inward investors. Rice and Venables go on to argue further that:

places that have experienced negative shocks may have adverse skill and demographic characteristics, and also weak fiscal positions, poor public services, and social and health problems associated with low employment rates. Many of these are cumulative, involving vicious feedback mechanisms with multi-generational effects.

Thus, firms are reluctant to move into such areas. In some senses, this is the inverse of the agglomeration argument. Successful areas are sustained by clusters of firms, comprising a mutually beneficial ecosystem, that are resilient after economic crises, while poorer areas can be trapped in a low-productivity, low-income equilibrium, in particular when they are hit by national macroeconomic recessions.

Disadvantaged regions fall into three categories: those which have been relatively poor in the very long term; those which failed to adjust to structural change; and those disproportionately affected by a macroeconomic shock. The first category would include the likes of the Highlands of Scotland or large parts of southern Italy. The second would include many former centres of manufacturing in Europe and North America. The third usually (but not always) contains already disadvantaged areas whose disadvantage is exacerbated by a macro shock. The articles in this issue by Joan Rosés and Nikolaus Wolf (2021) on Europe and by Trevon Logan et al. (2021) on the US encompass all three types. Thus, regional inequalities can be persistent and self-sustaining. Some regions become and remain 'left behind'. Apart from the economic consequences for many individuals across the generations, there are broader social and political implications. These areas often exhibit poorer health, higher mortality rates, lower educational attainment, and greater crime. In some countries, deprivation has become associated with various forms of political extremism. Recent studies on the geography of discontent (De Groot, 2019; Dijkstra et al., 2020) suggest that the rise of populist political parties associated with anti-establishment voting, anti-EU voting, and Brexit are concentrated in places that face population and industrial decline, have low land rents, high unemployment rates, and low levels of education of the workforce. It might also be the case that too much regional divergence acts as a drag on the growth of the national economy (de Dominicis, 2014) . For example, it is likely that labour force capabilities are under-utilized in low-productivity regions. Because of limited individual mobility, for both economic reasons (high rents and property prices in high-productivity regions, for instance) and non-economic reasons, this means that human capital is being wasted (Holmes and Mayhew, 2015) .

All these negative associations with regions that are lagging or in decline or impoverished have led to the coining of the phrase 'places that don't matter' (Rodríguez-Pose, 2018).

The variables upon which we concentrate are GDP per capita and disposable income per capita. The first is measured at the workplace. The second is a household measure and therefore relates to the place of residence.

Different authors in this issue employ different spatial units of analysis depending on the specific phenomena they are studying. The OECD has a standardized spatial classification system, the intricacies of which are well explained by McCann (2020). TL2 is the highest level of disaggregation, describing large regions. In the UK, for example, there are 12 of them. TL3 digs down into areas within these large regions-the UK has 173 of them. The third, residence-based, measure is of metropolitan urban areas containing more than half a million people and based on 'commuting flows and contiguity'. There are 17 such areas in the UK.

Eurostat and the European Commission employ a slightly different classification, labelled as NUTS 1, NUTS 2, and NUTS 3. McCann (2020) compares these with the OECD classifications. He shows that for the UK and three of the EU countries (Germany, France, and Belgium) NUTS 1 corresponds with TL2. For another 14 countries NUTS 2 is not much different from TL2. NUTS 3 and TL3 more or less correspond for all EU countries.

Unless one defines spatial areas very narrowly, then the issue of intra-area inequalities becomes potentially important; and these inequalities vary across countries. As McCann writes:

inequalities within the UK are also across such short distances with enormous local productivity variations evident within just a two-hour driving time, whereas within Spain comparable variations would only be evident across a seven-hour driving time, and in Italy and the United States across a 10-hour driving time.

This observation is reinforced in the case of the UK by the 2070 Commission:

The long-term patterns of inequalities are reflected at a neighbourhood level. This is highlighted in the research by the Geographic Data Science Lab, University of Liverpool. There is considerable intra-regional variation in the distribution of struggling neighbourhoods within more disadvantaged regions. The local patterns in neighbourhoods mirror regional disparities, illustrating the way inter-and intra-regional inequalities are reinforcing. (UK 2070 Commission) Reflecting such observations, Enrique Garcilazo et al. (2021, this issue) develop what they describe as a 'functional typology' of the OECD's TL3 for Europe and the US. They sub-divide TL3s into five categories: large and medium metropolitan regions and three regions differing according to the size of the metropolitan areas to which they have access. This enables them to conduct a fine-grained analysis of the contribution of different types of regions to national economic growth as well as of the impact of the 2008 recession on different types of spatial entities.

IV. The changing patterns of regional inequality Rosés and Wolf (2021) give a nuanced picture of the patterns of convergence in 16 countries in Europe from the beginning of the twentieth century until 2015. Initially there was little change from high levels of dispersion in the inter-war years. The significant decline in dispersion came in the years after the Second World War until about 1980. Rosés and Wolf argue that in many areas this was not driven by the classical forces of convergence, but rather by post-war reconstruction and structural change in those regions that had suffered physical destruction and massive population movements-parts of Germany, Austria, Belgium, Italy, the Netherlands, and eastern France. From 1980 there was a significant increase in regional inequality. Importantly, however, they find that many islands of prosperity have emerged within otherwise lagging regions.

Within this general picture, there was diversity of regional experience. Taking snapshots in 1900 , 1950 , 1980 Wolf describe what they term a core-periphery pattern in 1900. The regions of England, north-western Europe, and Switzerland were richer (in terms of GDP per capita) than average. The regions of France and central Europe were close to the average, while Scandinavia and southern Europe contained many poorer than average regions. Over time the spatial correlation has declined and a more complex picture has emerged. By 2015 there were metropolitan areas and islands of prosperity, such as Paris and Madrid, which were surrounded by regions with relatively low average GDP per capita. Most regions of England had experienced a relative decline. Ireland (mainly Dublin) and many parts of Scandinavia had become richer than average. Logan et al. (2021) paint a similarly detailed picture for the US. General convergence has halted in the last three or so decades. Some of the dynamics of regional inequalities are driven by major cities (see, for example, an earlier issue of OxREP on urbanization in developing countries (2017, vol. 33, no. 3)). Nevertheless, there are cities that buck the trend. The south-east of the USA contains cities which are 'among the most innovative and dynamic regions in the country'-Raleigh-Durham, Nashville, Atlanta, and Richmond. At the same time there are struggling cities in more prosperous areas-Oakland, Milwaukee, Detroit, and Baltimore.

Some disparities develop over time, others are abrupt structural breaks. With respect to the first, Rosés and Wolf describe how European regions from 1900 to date experienced a gradual and steady decline of agricultural employment shares, a rise in industrial employment shares until the 1970s followed by a decline, and a rise in the employment shares of services over the whole period. The expansion of industrial and services employment was very uneven across regions, while agricultural employment became significantly more concentrated. Today's regional disparities in income and job characteristics, as well as massive differences in agglomeration effects and human capital endowments, are the consequence of these historical developments. Logan et al. describe similar long-run developments in the United States, where in the early twentieth century some frontier regions had high levels of GDP per capita. They put particular stress on the consequences of long-run changes for the current geographical distribution of human capital. As they put it: 'Regional success is now a story of higher education, human capital, and the rising tech and service sectors. . . . Understanding regional equalities today requires understanding these dramatic differences in human capital across space in the United States.'

Not only structural trends but also structural breaks may lead to new disparities between regions that impact regional development. Examples are the abolition of slavery in the United States, the Second World War, and economic shocks such as the oil crises in the 1970s and the 2008 global financial crisis. These breaks may come with both opportunities and threats. With regard to the developments in the post-slavery era, Logan et al. state that 'the South would not develop the educational, civic, and financial institutions needed to promote innovation and diversify away from cotton.' By contrast, as is shown by Rosés and Wolf, reconstruction after the large-scale destruction during the Second World War stimulated regional economic growth on the western European continent. Any classical convergence forces at work would have had little impact but for the influence of a stable political environment and the massive Marshall Aid programme. Ironically, lack of access to Marshall Aid may have been a reason that the UK fell behind after the Second World War.

Rice and Venables (2021) explore the impact of adverse economic shocks in the 1970s. Using UK local authority district (LADs) data, they investigate the impact of the large and rapid fall in the share of the secondary sector in national output in the UK from 40 to 30 per cent in the 15 years from 1966 to 1981. They argue that if the classical forces of convergence had been at work, we would have expected to observe a negative relationship between the size of the shock in employment rates in the LADs and the subsequent growth of employment. They do not find any such relationship. Two-thirds of the Local Authority Districts with the highest deprivation rates in 2015 had experienced large negative shocks about 40 years before. They also found that 'the places that experienced negative shocks were not, on average, drawn from atypical starting points'. In other words, some previously fairly prosperous regions shared the pain. This is but one example of how fairly prosperous areas can succumb to fundamental shocks and of how difficult it can be to recover.

V. The anatomy of regional inequalities Rosés and Wolf (2021) distinguish between 'geographical factors' and 'institutional factors' that account for regional advantage or disadvantage. Disparities in geography had large and persistent effects on past regional economic developments. Although these so-called 'first nature' disparities may not be as important as they used to be, reminders of their past influence can still be very much present nowadays. Using data on 173 European regions in 16 countries between 1900 and 2015, Rosés and Wolf divide geographical factors into two types-natural and man-made. Favourable natural factors include climate, soil quality, access to coal fields, and proximity to large seaports. In similar vein, in their long-run analysis of regional inequalities of the USA from the eighteenth century to date, Logan et al. (2021) describe the geographic advantages of waterways and soil suitability for cotton as examples of regional endowments of natural resources that once led to fast growth in some areas.

Institutional or second nature factors are more the consequence of previous action by economic and governmental agents. Rosés and Wolf demonstrate how these 'second nature' disparities between regions also affect regional development. These disparities may relate to institutional differences, such as simply the country to which a region belongs, whether the region is a capital region, whether the country is part of the European Union and/or the Euro-zone. A second nature disparity of particular importance is market access-in other words, the size of nearby regional markets since they reflect purchasing power not just dependent on the size of the population but critically on its employment patterns and income. Agglomeration effects and increasing returns to scale also fall into this category. Increasing returns to scale were important for the rise of manufacturing industries but were also crucial, as Logan et al. argue, in the institutionalized slavery system of the South of the US before the Civil War. As the terms suggest, second nature disparities were frequently the consequence of first nature disparities. For example, proximity to coastlines or coal fields were often associated with the emergence of metropolitan regions. The huge regional disparities in access to cities of different population sizes and densities is the starting point of the analysis by Garcilazo et al. (2021) . They explore the contribution of different sized regions to GDP growth, categorizing five types of region:

(i) regions with a city of more than 1 million people, (ii) regions with a city of more than 250,000 people, (iii) regions near a city of more than 250,000 people, (iv) regions near a city of less than 250,000 people, (v) remote regions.

Countries differ immensely in how their populations are distributed across these regional types. Different densities are related to agglomeration economies and to regional inequalities in productivity, wages, and living standards. Disparities in population densities and sizes determine to what extent the contribution to aggregate growth is more concentrated in metropolitan regions or more distributed across regions of different sizes. They find evidence for 'agglomeration economies in regions with large cities: in the US, EU-15, and EU-25 their contribution to aggregate growth is higher than their population share'.

However, medium-sized cities play a larger role in Europe than in the US, where conversely the regions with the largest cities (of more than a million people) make a greater contribution than in Europe. Interestingly, the contribution of cities to growth is less volatile in the new member states of the EU than in the older member states. They go on to suggest two different country types:

(i) countries with 'metro-dominated growth contributions' in which regions with large cities dominate the contribution to national economic growth-Finland, France, Estonia, Greece, Lithuania, Italy, and the US; (ii) countries with 'mixed growth models', of which there are two varieties. The first comprises those countries with 'decreasing size-monotonic growth contributions' where regional growth contributions decrease with the main city sizes of the regions-Austria, the United Kingdom, Germany, the Netherlands, and Slovenia. The second have 'mixed growth regimes', 'where all regions contribute to growth in a roughly balanced way'-the Czech Republic, Denmark, Hungary, Belgium, Latvia, Portugal, Slovakia, Sweden, Poland, Spain, and Norway.

They conduct a similar exercise for regional contributions to national productivity growth and find two broad patterns:

(i) Concentrated countries where most productivity growth is contributed by the 'top productivity regions'-as in the Czech Republic, Belgium, Slovakia, Sweden, France, the UK, Greece, Lithuania, and the Netherlands. (ii) Distributed countries where 'catching up regions contributed the most to aggregate productivity growth'-as in Austria, Denmark, Germany, Estonia, Spain, Finland, Hungary, Italy, Latvia, Portugal, Slovenia, and the United States.

Such results raise two questions. The first is what impact a levelling of regional performance might have on a nation's overall economic growth. Clearly any government would hope that the productivity of a poorer region can be enhanced via policy interventions without any cost to the productivity of more successful regions, but this might in fact not be achievable. The second is where future national economic growth will come from. Noting the sort of evidence presented by Garcilazo et al., McCann (2013) , inter alia, argues that the dominance of core cities and therefore of core regions may well fall in the future quite independently of any policy initiatives. Although 'modern globalization' had made geographical proximity important for high-value knowledge activities and for service activities reliant on trust. He suggests that in the future there will be many more opportunities for non-core regions. We explore this issue further later in this article. It is not only disparities in population size that matter. So do disparities in population characteristics, not least its human capital broadly defined. Böhm et al. (2021, this issue) , studying West Germany between 1975 and 2014, provide one specific example. Their starting point is that Germany as a whole has seen rapid population and workforce ageing. Using a panel of labour market regions, they find that 'workforce mean age has considerable negative effects on the wage returns to age', which are arguably stronger in markets with more non-routine jobs. They also find that the employment rates of older workers also tend to fall with mean age. These effects vary significantly across German regions and Böhm et al. explore this further. Workforce ageing can be driven by both demand and supply influences. It may be that the demand for older workers falls in a region or that the supply of younger workers falls because of declining birth rates or outward migration. Low-income regions and those in relative decline tend to lose younger people who leave in search of better jobs but also in search of a more appealing lifestyle in the more vibrant urban centres. Böhm et al. also find a significant role for increased relative demand for younger workers, but only in these urban centres. As far as declining or left-behind regions are concerned, in most countries their working populations get older and, if Böhm et al.' s results are true beyond Germany, there are harmful consequences for these older workers.

An important dimension of human capital are leadership skills and capabilities. The more devolved responsibility for regional strategies and their implementation becomes, the more important are the qualities of local leaders. Paul Collier and David Tuckett (2021, this issue) discuss one aspect of this. They compare the political economies of Wales and the West Midlands of England. They consider the role of narratives in 'forming investment expectations', how a particular set of expectations can trap regions in 'low income equilibria' and limit 'the scope for regional leaders to reset those expectations'. Whether an area has a low income equilibrium or a high income equilibrium, it will be the consequence of a whole set of interdependencies between firms operating in the tradable sector and those in the non-tradable sector, between firms operating internationally, nationally, and locally, between the decisions made by the commercial sector and those made by the education sector, and by local government affecting things like infrastructure and local taxation. However, as Collier and Tuckett put it, 'resetting a low-income equilibrium may require a coordinated change in the narratives prevailing' in these different interest groups 'that have only limited interaction'. They pursue these ideas by interviewing representatives of the business communities in Wales and the West Midlands. The main difference they found between the two regions was that narratives were overwhelmingly negative in Wales-'narratives of identity suggest that identities are not merely fragmented, but actively oppositional. A predominant explanation for economic failure is normative: others are blamed within and outside the society, resulting in a passive mentality of victimhood'. Under what conditions, they ask, can a local leader 'reset' attitudes and actions in the local economy. The first requirement is that he must have the trust of the different parties. The second is that he has a clear, flexible, and resilient approach. Only then can what they call a Conviction Narrative be achieved.

The importance of leadership is also emphasized by Ties Vanthillo et al. (2021, this issue) in their assessment of the changing nature of regional policy in Europe. They discuss the evolving features of regional policies in four periods from the 1950s to date. The current period is characterized by place-based policies, in the construction and implementation of which they argue that political leadership (among other factors such as institutional coordination and strategic intelligence) is essential for the quality and implementation of effective measures to stimulate local development.

Populist politicians in a number of European countries have blamed membership of the EU in general and membership of the Euro area in particular for rising inequality among regions and households. There is a well-developed literature (see, for example, Beetsma and Giuliodori (2009) ) considering the impact on member countries of being members of a common currency area. Not having control over one's own exchange rate deprives a country of an important tool of macro policy. The implications for its economic fortunes are uncertain, depending as they do on other policies adopted by the country itself and by other members of the common currency area. Thus, for example, the impact of the euro on the distribution across countries of GDP per head is highly uncertain. Less studied is the impact of the euro on inequality between households and by implication between regions within the member countries. This is what Florence Bouvet (2021, this issue) investigates for the first time. She uses a synthetic counterfactual methodology, which matches individual euro countries with non-euro countries possessing similar characteristics in the period before the introduction of the euro. She then compares their trajectories after the introduction of the common currency, investigating how income inequality (gross and net of taxes and transfers) has changed within each of the Euro Area countries studied-Austria, Germany, Luxembourg, Belgium, Greece, the Netherlands, Finland, Ireland, Portugal, France, Italy, and Spain. She finds that, in the absence of the euro, gross income inequality would have been lower but net income inequality would have been higher in most countries. In other words, any 'market' effects were more than offset by transfer payments. Such in-country transfer payments were doubtless enabled directly and indirectly by monies from the Social Fund and other EU sources.

The advocates of place-based policies argue that general redistributive policies through the tax-transfer system, while necessary to alleviate hardship wherever it exists, provide no long-term solution for individuals in the left-behind and disadvantaged regions (Neumark and Simpson, 2015) . They are also sceptical of any suggestion that in implementing place-based policies there is necessarily a trade-off between equity and efficiency. This scepticism is, in part, the consequence of their doubts about the merits of arguments for agglomeration.

In the urban economics and new economic geography literatures, agglomeration effects are typically supposed to be strong in densely populated areas because of 'sharing, matching and learning' (Duranton and Puga, 2004) . This is related to the highly concentrated pools of labour and suppliers, to the excellent infrastructure with low costs of transportation and mobility, and to the easy diffusion of knowledge and innovations. The elasticities of productivity with respect to employment density are estimated to be in the range of 0.01 to 0.10 (Neumark and Simpson, 2015; De Groot, 2019) . Subsidizing people and firms to encourage them to locate in agglomerated, high-density areas can be justified on the grounds that the social returns are higher than the private returns. This policy may come to the benefit of the whole society since people will move to places with high productivity rates from where economic activity, growth, and prosperity will eventually spread or filter to the lagging and peripheral areas.

However, it is far from clear that policies exploiting agglomeration economies are beneficial to society as a whole. First, it is not certain, from historical evidence, whether the filtering effects are large enough to compensate for the adverse effects of declining employment rates and brain drain in the lagging areas. Second, policy-makers probably do not have sufficient knowledge about the magnitude and the distribution of elasticities across regions and economic activities to optimally target their investments. Third, if there is not much geographic variation in elasticities, relocating economic activities will not increase aggregate production (Neumark and Simpson, 2015) . Poorly designed regional policies could result in a zero-sum game whereby high investments in dense and prosperous regions come at the expense of regions in which people already feel 'left behind'. Fourth, diseconomies of agglomeration may emerge when real living standards in the prosperous areas are reduced by rising disamenities related to air and water pollution, traffic congestion, and more local crime.

Thus, advocates of place-based theories are usually critical of those who put their faith in agglomeration economies and they deny any necessary trade-off between equity and efficiency. They refer to the rise and fall of big cities with large agglomerations in the past, contending that high returns on public and private investments in metropolitan areas are not self-evident. They also argue in favour of tailor-made policies that seriously explore the untapped potential as well as the threats to progress in each place. Thereby they emphasize the underestimation of the economic potential of many noncore, less developed, or declining regions.

In short, there is every reason to focus on the potential of a region to achieve a situation where it has a sustainable resilient regional economy in combination with acceptable levels and distributions of wellbeing for all its inhabitants without social exclusion. (Van Dijk and Edzes, 2016, p. 178) However, as far as the lagging regions are concerned, there is the risk of failure of supply-led interventions since so many of them have attempted to boost sectors and activities that do not match local economic strengths and which become in perpetual need of assistance to survive. Furthermore, welfare-and support-based measures specifically aimed at sheltering inhabitants of poorer areas can have pitfalls. For example, place-based policies designed to stimulate local employment for the people 'left behind' incorporate the risk that residents from elsewhere profit from new economic activities, while raising rents, house prices, and land prices and increasing the share of in-commuters in local employment instead of lowering unemployment rates for people at the low end of the local labour market (Neumark and Simpson, 2015) . In Europe regional policy is said to have changed radically in recent decades. These changes are described by Vanthillo et al. (2021) . Traditional policies varied somewhat from country to country but were essentially top-down interventions. These interventions involved tax incentives and subsidies both to encourage firms to remain and grow in the poorer regions and also to attract new enterprises, not least multinationals. In many countries, government offices and parts of state-owned enterprises were moved from the centre to the periphery. As Vanthillo et al. put it, ' the policy focus of the range of instruments used in these regions was essentially centred on influencing economic activity through industrial location'. In fact, regional policy was inextricably bound up with traditional industrial policy, which can be defined as policies to stimulate growth and productivity and to rebalance the economy by altering the sectoral mix of production. Some policies were horizontal and others vertical. The former applied to all firms, whether nationally or in a particular region. The latter were applied differentially across sectors or even firms. Many horizontal policies came to be seen as often ineffectivevarious forms of investment tax incentives, for instance. But it was the vertical policies which attracted particular criticism. They often involved significant public expenditure with little return. In the British context, Crafts (2010) and others described them as 'picking losers' rather than 'picking winners', too often engaging in the forlorn task of propping up ailing industries such as shipbuilding and arguably, in the longer run, making it more difficult to cope with structural change.

From the 1980s the focus of policy started to change. Vanthillo et al. (2021) argue that there were three reasons for this. The first was broader political developments leading to devolution and decentralization in many countries. There were many complex explanations for these developments, but one was a belief that central governments had not served lagging regions well and that more could be achieved by locally led initiatives; and indeed this belief was supported by emerging research evidence that decentralized systems had been associated with less inequality in regional growth rates (McCann, 2016) . The second was a perception that policies centred on tax incentives and subsidies had failed-in part because they had led to competition for assistance between regions. Certainly, in England there is evidence that the North-west (a relatively lagging region) suffered from the fact that funds were poured much more profusely into other, and more severely, lagging regions like the North-east. Third was a shift of emphasis towards policies that were more tailor-made for individual regions. As we have already intimated, to these might be added a fourth-the realization that in the past too much public money had been wasted on trying to halt irreversible structural change. Influential in these developments was the European Commission and, in particular, the reform of the Structural Funds in the late 1980s. As Vanthillo et al. argue, the European Regional Development Fund (ERDF) had been 'complementary to national regional policies'. Now Brussels started to take a more central role. Emerging from this were 'smart specialization strategies'. Interventions were based on enhancing local competitiveness and not necessarily tied to conventional administrative regions. Critically, initiatives were placed (at least partially) in the hands of local actors. Key here was the requirement that, in order to receive funding, a region needed to articulate a 'strategy' for development. Vanthillo et al. describe how 'more than 120 regions in the European Union (EU) have recently designed a smart specialization strategy to receive funds from the ERDF in the 2014-20 programming period'. Similarly, some form of regional strategy was required to access EU Structural Funds. Key to formulating a strategy was to look forward rather than backwards and to take a realistic view of what the competitive strengths of a locality might be. At the same time, domestic spending on regional policies diminished in most countries, which came to rely ever more heavily on European funding. In recent years these initiatives have fallen within the EU Cohesion Policy whose declared aim has been 'to strengthen economic and social cohesion by reducing disparities in the level of development between regions'. The Policy accounted for no less than 32.5 per cent of the EU budget between 2014 and 2020.

If it is the case that strategy is to be devolved locally, then a necessary, but not sufficient, condition for success is the competence of those devising the strategy. This seems to be almost taken as given by national policy-makers. But when the competence of national policy-making generally cannot be taken for granted, assuming local competence across several geographical areas in a country seems dangerous. There is also the question of central funding of local initiatives. Is sufficient resource provided to allow local initiatives to flourish, or do the local entities have sufficient local revenue raising powers? (See also De Groot, 2019.) Any significant shortage of funding is likely to dictate sub-optimal strategies or sub-optimal implementation of optimal strategies. Inevitably there is an unresolved tension between the roles and powers of the centre and local administrations and, at least in some countries, it is evident that the national authorities find it difficult to let go.

In describing the eco-systems of poorly performing regions in the US, Logan et al. (2021) remind us of the dangers, as well as merits, of devolved powers and decisiontaking that are part of the federal structure of the country. They write:

Modern-day social and economic inequality is rooted in a combination of factors, including geographic endowments, agglomeration economies, regional differences in human and physical capital investments, and, importantly, persistence of past policy decisions, investments, and choices. . . . [a] broad set of sub-national policy and expenditure decisions falling within the domain of economic development, including education, social safety net transfer programmes, and labour market supports, which have helped to shape the inequality we observe today.

They argue that fiscal federalism has meant that many policy decisions have harmed sections of the population and regions. In particular, racially motivated actions against black communities in the South have had long-lingering consequences. For example, discrimination in education and restricted school funding have damaged the human capital of large swathes of the country. The example of the southern states may seem to be an extreme one to European eyes. However, the possibility of the unhealthy dominance of vested interest groups and of various forms of local corruption cannot be ignored. This is where the issue of striking an appropriate balance between national and local control becomes an important issue.

Lagging regions are more often than not in a self-reinforcing, self-sustaining equilibrium and, because of this, specific changes designed to improve performance can be ineffective since other elements of the eco-system which remain unchanged drag the local economy back to the undesirable equilibrium. Collier and Tuckett (2021) remind us of this in their arguments that Conviction Narratives are essential for buoyant local investment and the construction of a buoyant local ecosystem.

Colin Mayer et al. (2021, this issue) examine one particular aspect of the local ecosystem-banking-which is vital for financing investment by small and medium-sized enterprises (SMEs). Without vibrant local banking arrangements, they argue, devolution of economic policy would be limited in its effectiveness. They compare the British banking system with banking in Germany, Sweden, and the US. The British system, they contend, became over time highly centralized and transactional with 'weak relationships between banks and borrowers'. They contrast transactional banking with relationship banking and define decentralized banking as providing 'relationship-based banking services to its customers by operating in close proximity to them and via a business model that relies on cultivating and utilizing the strong relationship it establishes with its customers to gather and build soft information'. Banking centralization increased over a fairly long period of time in the UK but was exacerbated by the sector's response to the 2008 financial crisis. As a consequence, smaller firms in peripheral regions find it more difficult to get credit than those in London and the South-east. Mayer et al. describe the long historical evolution of the three-pillar German banking system and demonstrate that, for all its twists and turns, it serves the SME sector better than does British banking. So, they contend, does community banking in the US, though the authors recognize that its viability is under some threat. In yet another very different financial system, Sweden's Handelsbanken serves local business communities well. While acknowledging that effective regulation regimes would be needed, Mayer et al. conclude that strong local banking, based on tacit as well as codified relationships, is essential for significant improvement in the economic fortunes of lagging regions.

It is perhaps ironic that, at a time when devolution of strategy is all the rage for regions in Europe, it has been the local misuse of opportunities offered by fiscal federalism which arguably has hampered regional development in the US. This reminds us that decentralization is not a magic bullet. Nevertheless, at least in Europe, policy-making towards regions has made some progress; but there is still much to be done.

There has been greater recognition of the need to move away from emphasis on broadly defined administrative regions. Problems and their solutions are now seen to be far more spatially specific. The ability to address these problems has been massively enhanced by the emergence of robust disaggregated data. However, it is not always clear that national politicians make sensible decisions about what constitutes a locality for action. For example, some commentators argue that an obsession with city regions often overlooks the wider regional context. Indeed, if decentralization and devolution are to be the answers, then the UK 2070 Commission (2020) points to some of the difficulties in the British context:

Barriers to progress arise from: 1. Conflicting National Policies arising from an over-centralised administrative system; 2. Strained Central-Local Relationships arising from the desire for central accountability of local decision-making; 3. A Flawed Strategy for Growth that assumes the benefits of growth in London and the Wider South East will spill over to the rest of the UK; 4. Low Levels of Investment which result in under-resourced programmes of action, create a competitive project-based culture, and hold back ambition; 5. Constant Change in Policies and Delivery Agencies which does not allow sufficient time for any programme of action to have real impact; and 6. Narrow Short-Term Measures of Success that do not take account of longer-term generational and well-being impacts.

Clearly these are issues not just for the UK but also for many other countries. This is highlighted by a recent OECD (2019) report. The report makes the case for decentralization and devolution in regional policies, covering the transfer of powers and responsibilities from central to lower level authorities in three dimensions: political, administrative, and fiscal. The report shows a positive correlation at the country level between GDP per capita, public investments, and education outcomes on the one hand, and the extent of decentralization on the other. It also argues that decentralization can promote local democracy and citizen engagement, reduce corruption, stimulate efficient public service delivery, and improve regional development. Therefore, it could be a powerful instrument for reducing the 'geography of discontent'. However, echoing some of the points made by the UK 2070 Commission, the OECD also emphasizes that decentralization is not a guaranteed recipe for regional growth and development because the positive impact is very much conditional on the design and implementation of the decentralization policies themselves.

Motivated by such concerns, the OECD records the risks associated with decentralization. First, there is the risk of insufficient administrative, technical, or strategic capacities at the subnational levels. Building these capacities takes time and requires long-term commitment from central and subnational government. Then there is the risk of lack of sufficient resource-unfunded or underfunded mandates, as the OECD puts it. De Groot (2019) illustrates this for the Netherlands where municipalities are faced with growing responsibilities as a result of the country's decentralization strategy, but with hardly any ability to increase financial resources due to small local tax bases. Furthermore, governmental bodies at different levels may have overlapping responsibilities and powers, which can cause lack of clarity, conflict, and a democratic deficit. Finally, decentralization may lead to the loss of economies of scale and fragmented public policies. Policies that initially may look successful can reveal perverse consequences if the full picture is taken into account. The Dutch legislation on work and assistance in 2004 intended to provide activation and employment services that were better tailored to both the needs of the local labour market and the unemployed by decentralizing services from central government to municipalities (Van Berkel, 2006) . This implied more autonomy in the design and delivery of services to cope with local and regional circumstances and policies, but also implied a transfer of financial responsibility for the social assistance scheme. A lower enrolment in social assistance was the consequence of municipalities being incentivized to be more prudent in its allocation. This may have led to under-provision of municipal services for the unemployed, unequal treatment of individuals in similar circumstances across municipalities, and a rapidly increasing inflow of people into the centrally administered disability insurance scheme (Roelofs and Van Vuuren, 2017) .

Furthermore, if the emphasis of new policy is a region pursuing its comparative advantage, then this presumes that the region has a potential comparative advantage in something. For some localities it may be difficult to uncover exactly what this might be. It will be difficult to break out of the low income/low productivity equilibrium. Logan et al. (2021) make the case for the self-reinforcing nature of regional problems in the US. They write:

When comparing regions in the United States, a set of steady-state initial conditions, in large part shaped by the nation's pattern of economic development, and its legacy of slavery and racial exclusion, continues to shape modern-day economic and policy outcomes, helping to reinforce observable regional inequality today.

Nor can it be taken for granted that there is sufficient local political and administrative competence. Even if there is, Collier and Tuckett (2021) argue that it may be insufficient if there is no Conviction Narrative. At the same time, even if strategy is sensible and public funding appears sufficient, it may fail because of weak local institutions, as Mayer et al. (2021) contend as far as the provision of private finance is concerned. The fundamental problem is that low productivity/low income regions are experiencing systems failure. In these circumstances, attempts to improve one aspect of local performance may flounder because the other unfavourable characteristics of the locality may act like a magnet and drag it back to the original equilibrium. Indeed, it may be that a benignly intended policy has perverse effects. In the UK, for instance, the highspeed rail project is designed to cut travel times from London to the Midlands and North of the country. It is meant to stimulate these regional economies, but there is the possibility that it will simply enable more skilled workers to commute to London and further centralize economic activity.

Despite increased activism in policy, regional disparities have generally widened in Europe in the last 40 years. The balance of academic research suggests that this is mainly the consequence of the impact of globalization and changes in the sectoral composition of economies. We can be confident that this widening would have been greater but for the intervention of regional policies. This gives some reason for cautious optimism in the face of deep-seated, but not necessarily, intractable problems. In many countries Covid-19 has had a disproportionately harmful impact on poorer areas and this may well increase the focus of governments on the underlying problems of these areas. Furthermore, it is encouraging that there is some evidence for the benefits of place-based policies that build particularly on infrastructure expenditure as well as on higher education and university support (Neumark and Simpson, 2015) . Nevertheless, we still need to learn more about the long-term, redistributive, and heterogenous effects of these types of intervention. We also need to know much more about the strengths and weaknesses of devolved regional strategies more generally-about what works and what does not work.

The Macroeconomic Costs and Benefits of the EMU and other Monetary Unions: An Overview of Recent Research

Demographic Change and Regional Labour Markets

Estimating the Impact of Economic and Monetary Union on National Income Inequality: A Synthetic Counterfactual Analysis

Narratives as a Coordinating Device for Reversing Regional Disequilibrium

Overview and Policy Implications', in Learning from Some of Britain's Successful Sectors: An Historical Analysis of the Role of Government

Inequality and Growth in European Regions: Towards a Place-based Approach

Growth and Shrinkage: Challenges for Governance and Solidarity

The Geography of EU Discontent

Micro-foundations of Urban Agglomeration Economies

Regional Inequalities and Contributions to Aggregate Growth in the 2000s: An EU vs US Comparison Based on Functional Regions

Over-qualification and Skills Mismatch in the Graduate Labour Market

Long-run Analysis of Regional Inequalities in the US

Modern Urban and Regional Economics

Perceptions of Regional Inequality and the Geography of Discontent: Insights for the UK

The Structure and Relations of Banking Systems: The UK Experience and the Challenge of 'Levelling Up

Place-based Policies', ch

Making Decentralisation Work: A Handbook for Policymakers, OECD Multi-level Governance Studies

The Persistent Consequences of Adverse Shocks: How the 1970s Shaped UK Regional Inequality

The Revenge of the Places that Don't Matter (and What to Do About It)', Cambridge Journal of Regions

The Decentralization of Social Assistance and the Rise of Disability Insurance Enrolment

Regional Growth and Inequality in the Long Run: Europe

Making No Little Plans: Acting at Scale for a Fairer and Stronger Future

The Decentralisation of Social Assistance in The Netherlands

Towards Inclusive and Resilient Regional Labour Markets: Challenges for Research and Policy

The Changing Nature of Regional Policy in Europe