Lecture 1 & 2 Regional Policy Notes
Lecture Overview
This lecture examines the persistence of regional economic disparities within the European Union and the United Kingdom, the competing theoretical frameworks used to explain them, and the rationale and design of the EU's regional policy. The central analytical tension is between the neoclassical prediction of convergence and the new economic geography (NEG) prediction of divergence. These rival frameworks generate starkly different policy prescriptions: the first argues for laissez-faire integration, the second for active intervention to counteract agglomeration dynamics. Understanding this divide is essential for evaluating whether EU Structural Funds and Cohesion policy are justified, effective, and correctly designed.
- Regional inequality is substantial and persistent, both across EU countries and within them.
- Neoclassical theory predicts conditional convergence; NEG predicts path dependence and divergence.
- The EU allocates roughly 34% of its budget to regional/cohesion policy, targeting regions below 75% of EU average GDP per capita.
- Empirical evidence on effectiveness is mixed and conditional on institutional quality and human capital.
1. Regional Differences: The Stylised Facts
1.1 Europe-wide GDP per capita disparities
The NUTS2 map of European regional GDP per capita reveals a striking core-periphery pattern. Wealthy regions concentrate in a blue banana arc stretching from southern England through the Benelux countries, western Germany, Switzerland, and northern Italy, supplemented by Scandinavian and Alpine clusters. Eastern European regions, together with southern Iberia, southern Italy, Greece, and much of the Balkans, remain substantially poorer. The range is extreme, running from under €13,000 to over €81,000 per capita.
A single map communicates what tables obscure: the disparities are not merely statistical artefacts but are geographically clustered, with rich regions near rich regions and poor regions near poor regions. This spatial autocorrelation is precisely what neoclassical theory has the hardest time explaining and what NEG models were developed to rationalise.
1.2 Within-country dispersion
This dot plot shows the dispersion of regional GDP per capita within each EU member state. The key observation is that within-country inequality is often as large as, or larger than, between-country inequality. Germany spans roughly €27,800 to €65,200; France ranges from around €9,200 to €59,000; Ireland stretches from €27,100 to €85,500. Small countries such as Malta, Cyprus, and Luxembourg naturally show little variation because they have few internal regions. This matters because it implies that national averages mask severe internal divergence, and so effective regional policy must target sub-national units rather than simply transferring resources between member states.
Students often conflate country-level convergence with regional convergence. The EU as a whole can be converging in terms of national GDP per capita (with poorer new member states catching up) while simultaneously experiencing widening within-country gaps. Always specify the level of aggregation.
1.3 Employment and NEET rates
Employment rates for the 20-64 age group reinforce the core-periphery picture but with some interesting deviations. Nordic countries, Germany, the Netherlands, and the UK show high employment, while southern Italy, parts of Spain, Greece, and much of Turkey display employment rates below 57%. Employment gaps capture something GDP per capita cannot: the extensive margin of labour market participation, which reflects both labour demand and structural features such as female participation, informal economy size, and welfare incentives.
The map of young people Not in Employment, Education, or Training (NEET) offers a forward-looking diagnostic. NEET rates predict future human capital depreciation and social exclusion. Southern Italy, parts of Spain, Bulgaria, and Romania show alarmingly high NEET rates, consistent with the hypothesis that these regions are caught in a low-skill equilibrium, where poor labour market prospects suppress educational investment, which in turn reinforces low productivity.
1.4 The UK: extreme regional inequality within a G7 economy
The ONS map of UK household income reveals a pronounced London and South-East premium. The London-centric spatial structure of the UK economy is extreme by G7 standards, with incomes in parts of Inner London exceeding 1.3 times the UK average while large swathes of the North and the Celtic periphery lie below 0.8 times.
This infographic dramatises the UK's duality: Inner London is the richest region in northern Europe, yet nine of the ten poorest regions in northern Europe are also British (West Wales, Cornwall, Durham and Tees Valley, Lincolnshire, South Yorkshire, Shropshire and Staffordshire, Lancashire, Northern Ireland, and East Yorkshire and North Lincolnshire). This is a textbook example of agglomeration rents accruing to a dominant capital while the periphery is hollowed out.
The UK pattern is hard to reconcile with simple neoclassical logic. If capital were truly mobile and free to arbitrage, we would expect diminishing returns to push investment out of London and towards cheaper peripheral regions. That this does not happen at sufficient scale points to powerful agglomeration externalities (financial clustering, human capital spillovers, deep labour markets, global connectivity) that keep the capital attractive despite congestion and high rents.
2. Why Do Regional Differences Persist?
The lecture contrasts two frameworks: the Solow neoclassical model and New Economic Geography.
2.1 The Solow neoclassical growth model
The Solow model (Solow, 1956) rests on a production function with two factors (capital and labour), constant returns to scale overall, and diminishing returns to each factor individually. A Cobb-Douglas specification
A steady state is reached where
Conditional convergence: poor regions with similar savings rates, population growth, and human capital to rich regions will tend to grow faster and catch up, because their lower capital-labour ratio gives them higher marginal returns to capital.
Predictions of the Solow framework
- Capital-labour ratio and per-capita income rise with the savings rate
- In the long run, without technological progress, per-capita growth is zero at the steady state. A permanent increase in
- Augmenting the model with human capital (Mankiw, Romer and Weil, 1992) produces the augmented Solow model, in which investment in human capital drives technological adoption and conditional convergence becomes more plausible empirically.
If asked to "evaluate the Solow model's relevance for EU regional policy", emphasise the following: (i) Solow predicts convergence, so regional policy is at best a temporary accelerator; (ii) the model implies that removing institutional rigidities and enabling factor mobility is sufficient; (iii) the augmented version reframes policy as investment in human capital rather than raw transfers.
Implications for integration and intervention
Under Solow logic, integration promotes equality because capital flows to regions with high marginal returns and labour flows to regions with high wages, eliminating differentials. Regional inequalities are therefore either transient adjustment frictions or symptoms of institutional failure. Policy should focus on liberalisation and factor mobility, not subsidies.
It is wrong to say Solow predicts unconditional convergence of all regions. It predicts conditional convergence given similar structural parameters. Regions that differ in savings rates, demographics, or technology converge to different steady states.
2.2 New Economic Geography: the Krugman-Venables framework
NEG (Krugman, 1991; Krugman and Venables, 1990) overturns the neoclassical intuition by introducing increasing returns, monopolistic competition, and trade costs. Integration in this framework can worsen regional disparities rather than eliminate them.
The Krugman and Venables (1990) numerical example contrasts producing in "Belgium" (central, high-cost, good market access) with "Spain" (peripheral, low-cost, worse market access) and splitting production across both. Production costs are 10 in Belgium, 8 in Spain, and 12 if production is duplicated. Shipping costs fall from 3 (high barriers) to 0 (full integration). The total-cost minimising location depends on trade costs in a non-monotonic way.
With high trade barriers, firms duplicate production to avoid shipping costs (total 12). With moderate barriers, they concentrate in the larger market Belgium (10 + 1.5 = 11.5) despite its higher production costs, because good market access dominates. Only with very low barriers does production migrate to the cheap location Spain (8 + 0 = 8). The path is high-cost concentration first, low-cost relocation later, meaning integration can hurt the periphery before it helps.
A canonical question is "Does economic integration necessarily benefit peripheral regions?". Use the Krugman-Venables table to argue that the answer is non-monotonic in trade costs, that partial integration can harm the periphery, and that deeper integration or active regional policy is needed to realise peripheral gains.
2.3 Agglomeration and dispersion forces
NEG formalises the tension between forces that concentrate activity and forces that disperse it.
- Agglomeration forces: increasing returns to scale, positive externalities (spillovers), technological spillovers, labour market pooling, and demand and supply linkages.
- Dispersion forces: congestion, high rent and land prices, and product market competition.
The Duranton and Kerr (2015) figure on Silicon Valley shows how innovation activity clusters in a very small number of zip codes, with patent citations tracing a dense web of intra-cluster knowledge flows. This visualisation is the empirical incarnation of Marshallian externalities: firms cluster because proximity lowers the cost of accessing ideas, specialised labour, and intermediate suppliers. The knowledge sourcing zones are small and overlapping, suggesting that spillovers decay rapidly with distance.
The demand-linkage cycle captures a self-reinforcing agglomeration mechanism. If industry initially moves to the larger region, workers follow and spend their incomes locally, expanding the market. The bigger market attracts still more firms because of trade costs and economies of scale, prompting further production shifting. This is a positive feedback loop whose logic is fundamentally different from the diminishing-returns logic of Solow. The economy can have multiple stable equilibria, and small historical accidents can tip it into an unequal outcome.
The supply-linkage cycle complements the demand side. When firms agglomerate, intermediate goods become locally available and cheaper in the larger region, which attracts further upstream and downstream firms. This is the classic input-output complementarity at the heart of industrial clusters. The combined operation of demand and supply linkages produces what Krugman terms cumulative causation.
The dispersion side of the ledger shows why agglomeration does not proceed indefinitely. London rental costs rise towards the centre, and congestion imposes real productivity costs. At some point the marginal benefit of locating in the core is outweighed by congestion rents, stabilising the spatial equilibrium. The relative strength of agglomeration versus dispersion forces determines the equilibrium configuration.
In Krugman's core-periphery model, the key parameter is the level of trade costs. Very high trade costs make dispersion dominate (each region must produce for itself). Very low trade costs can tip the system into a core-periphery equilibrium in which one region hosts all manufacturing. The model thus predicts a non-monotonic relationship between integration and inequality, often called the U-shape or tomahawk bifurcation.
2.4 Path dependence: Bleakley and Lin (2012)
The map of the southeastern United States shows that population density in 2000 clusters visibly along the fall line, the geological boundary where rivers cross rapids. Historically, these were portage sites where boats had to unload, creating natural trading posts. The original economic logic, water transport, has been obsolete for more than a century, yet the settlements persist as large metropolitan areas (Washington, Richmond, Augusta, Columbus, Macon).
The Bleakley and Lin (2012) figure plots population density against distance to the fall line, with a sharp peak at zero. The causal interpretation is that a transient geographical advantage, eliminated by railways and modern transport, locked in population agglomerations that have persisted through path dependence. This is compelling evidence for NEG: small historical accidents, amplified by agglomeration externalities, produce permanent spatial outcomes.
Think of it as regional destiny being determined by a long-forgotten coin flip. Once workers, firms, and infrastructure coordinate on a location, the self-reinforcing dynamics of NEG preserve that configuration even after the original reason for choosing it disappears. History casts a long shadow on spatial outcomes.
When asked "What evidence supports path dependence in regional economics?", cite Bleakley and Lin (2012) on US portage sites as the cleanest natural experiment: an original advantage that is unambiguously obsolete, yet the populations persist.
2.5 Summing up the theoretical contrast
- Neoclassical (Solow): integration plus factor mobility implies convergence; policy should remove frictions.
- NEG: integration can trigger cumulative divergence through demand and supply linkages and spillovers; path dependence means that initial advantages persist.
- Prosperous regions enter a virtuous circle (clustering, high incomes, new industries).
- Poor regions face a vicious circle (brain drain, low investment, obsolete technology).
- Solow implies regional inequality is temporary; NEG implies it can be permanent.
- NEG supplies the intellectual case for active regional policy.
- Empirical path-dependence findings tilt the balance towards NEG for questions of long-run spatial structure.
3. Government Intervention: Rationale and Channels
3.1 Reasons for intervention
Rich regions and countries may wish to support poorer neighbours for several reasons:
- Equity: a normative concern for fairness and a minimum standard of living.
- Political solidarity: cohesion supports the legitimacy and stability of the Union.
- Economic self-interest: poorer regions growing faster expands future export markets and investment opportunities, and reduces pressure for labour migration.
3.2 Channels of intervention
- Provision of physical capital: infrastructure, transport links.
- Support for research and development.
- Public education and training to raise human capital.
- Income transfers to households and regions.
3.3 Why a common EU regional policy?
A single EU-level policy is justified by coordination failures: uncoordinated national aid risks duplication, free-riding, and a race to the bottom in subsidies. A common policy also allows the Union to influence recipient policy priorities and to generate sufficient scale to affect growth trajectories.
EU regional policy is an exercise in fiscal federalism. It combines (i) an insurance function against asymmetric shocks, (ii) a redistributive function to secure political consent for integration, and (iii) a developmental function to correct NEG-style market failures. The Optimum Currency Area (OCA) literature is relevant: in the absence of fiscal transfers, regions hit by asymmetric shocks have limited adjustment options inside a currency union, so regional policy partly substitutes for the automatic stabilisers that a federal fiscal system would provide.
4. The EU Regional Policy: Goals and Funds
4.1 Budget and structure
- Roughly 34% of the EU budget is dedicated to regional/cohesion policy.
- Funds are largely pre-allocated by country, with the poorest regions receiving the largest share.
- Funds co-finance national projects within agreed regional development plans.
4.2 The three main objectives
- Convergence (approximately 82% of funds): targets regions with GDP per person below 75% of the EU average.
- Regional competitiveness and employment (about 16%).
- Territorial cooperation at cross-border and transnational level (about 2%).
4.3 The specific funds
- European Regional Development Fund (ERDF): innovation, research, the digital agenda, SME support, and the low-carbon economy.
- European Social Fund (ESF): employment, education, and those at risk of poverty.
- Cohesion Fund: restricted to countries with Gross National Income below 90% of the EU average, focused on trans-European transport networks and environmental infrastructure.
- European Agricultural Fund for Rural Development (EAFRD).
- European Maritime and Fisheries Fund (EMFF).
ERDF and ESF together constitute the Structural Funds.
4.4 Geography of eligibility
The side-by-side maps show the geographic reach of the two main funding streams for 2014-2020. Structural Funds eligibility (left) is broad and graduated, with less developed regions in red (GDP per head below 75% of EU-27 average) concentrated in southern and eastern Europe plus West Wales, Cornwall, and parts of the UK. Cohesion Fund eligibility (right) is stricter, confined to member states with GNI per head below 90% of the EU-27 average, and therefore covers the central and eastern member states, Portugal, and Greece. Richer members such as Germany, France, and the Benelux countries are net contributors to the budget but do not draw on the Cohesion Fund.
Students often confuse Structural Funds eligibility (regional, GDP-per-capita-based) with Cohesion Fund eligibility (national, GNI-based). A rich country can contain Structural Fund regions, but only poor countries receive Cohesion Fund support.
5. The Economic Impact of EU Regional Policy
5.1 The identification problem
Evaluating regional policy is difficult because treatment is not random:
- All poor regions receive transfers.
- There is a mechanical negative correlation between income and transfers.
- A clean experiment would require randomly allocating transfers to some poor regions and withholding them from others, which is politically infeasible.
This is a classic endogenous treatment problem. The simple regression of growth on transfers is biased because the worst-off regions get the most transfers, producing a spurious negative correlation. Credible evaluation requires exploiting discontinuities, such as the 75% GDP threshold, as a quasi-random source of variation (a regression discontinuity design).
5.2 The Banerjee, Duflo, and Kremer RCT approach
The Nobel Prize-winning work of Banerjee, Duflo, and Kremer demonstrates the power of randomised controlled trials in development economics. The vaccination example on the slide compares three randomly assigned groups of villages: a control group, villages receiving mobile clinics, and villages receiving mobile clinics plus small incentives (lentils). Full immunisation rates rise from 6% in control to 18% with clinics and 39% with clinics plus incentives. Remarkably, the cost per fully immunised child falls from $56 (mobile clinics alone) to $28 (clinics with incentives), because incentives raise take-up enough to spread fixed costs over more vaccinations.
The design exploits randomisation to isolate causal effects. It also illustrates a subtle point: adding a small incentive can be cheaper on a per-outcome basis because of non-linearities in take-up. This is a powerful counter-example to the intuition that any subsidy increases unit costs.
If asked to "discuss how the causal impact of EU regional funds might be credibly estimated", draw the analogy to RCTs, explain why they cannot be used directly for EU policy, and mention quasi-experimental methods such as regression discontinuity at the 75% threshold.
5.3 Key empirical findings
- Boldrin and Canova (2001): no significant growth effects beyond short-term consumption stimuli.
- Midelfart-Knarvik and Overman (2002): positive effect in attracting high-tech industries, but at the expense of medium-skilled industries.
- Becker, Egger, and von Ehrlich (2010): positive growth and income effects only in regions with adequate human capital endowments and strong institutions (low corruption, functional administration).
The Becker, Egger, and von Ehrlich result has a strong theoretical interpretation. Transfers are not a substitute for absorptive capacity: without human capital to use the funds productively and institutions to prevent rent extraction, money is dissipated. This is consistent with the augmented Solow view that capital investment requires complementary factors, and with the NEG view that transfers alone cannot overcome agglomeration disadvantages unless they build genuine productive capacity.
- The causal impact of EU regional policy is hard to identify cleanly.
- Effects are generally modest and conditional on institutional quality.
- Transfers work best when combined with human capital and good governance.
- This gives a nuanced policy message: design and context matter more than raw quantity of funds.
6. The EU Funds and the UK
The map of UK Structural Funds eligibility for 2014-2020 shows that only West Wales and Cornwall qualified as less developed regions (below 75% of EU-27 GDP per head), while a belt of northern, Welsh, and Scottish regions qualified as transition regions (between 75% and 90%). Southern England, particularly London and the South-East, is classified as a more developed region. This is striking: the UK, a top-tier economy overall, contains areas with GDP per capita low enough to qualify alongside Greek and Romanian regions, illustrating the earlier point that the UK is the most regionally unequal G7 economy.
The ESIF 2014-2020 budget breakdown for the UK by theme shows SME competitiveness as the largest single category (over €2.5 billion), followed by sustainable and quality employment, environmental protection, climate adaptation, educational and vocational training, and social inclusion. The colour coding shows that SME support is dominated by ERDF (blue), whereas employment and education are largely ESF-funded (purple). This allocation reflects the UK's specific structural weaknesses: relatively low productivity in the SME sector and regional skills gaps.
6.1 Concrete UK projects
- Northwest and northeast England and Yorkshire established Jeremie funds (Joint European Resources for Micro to medium Enterprises) to improve access to finance for SMEs, tackling a persistent UK market failure.
- Inner London used funds for transport infrastructure, including the cable car over the River Thames.
- West Wales used funds to roll out faster broadband to homes and businesses, directly addressing digital exclusion in peripheral regions.
6.2 Open questions from Martin Wolf
- What should be the main levers for growth in English regions and at what level should decisions be taken?
- Is comprehensive reform of local government needed?
- How should regions be funded?
- How should the balance be struck between local fiscal autonomy and accountability on one hand and redistributive transfers from rich to poor regions on the other?
The Wolf questions capture the central tension in fiscal federalism: subsidiarity (local decision-making improves information and accountability) versus solidarity (central redistribution corrects spatial inequities). Pure devolution risks entrenching regional inequality, because poor regions have thinner tax bases. Pure centralisation risks paternalism and allocative inefficiency. The optimal design combines central financing with decentralised spending choices subject to conditionality.
7. Integrating Theory, Evidence, and Policy
The lecture's coherent narrative can be summarised as follows. Regional disparities in the EU and the UK are large, persistent, and geographically structured. The neoclassical Solow framework predicts convergence and implies that regional policy should be modest and transitional. The NEG framework, supported by evidence on path dependence (Bleakley and Lin, 2012) and innovation clustering (Duranton and Kerr, 2015), predicts that integration without intervention can entrench spatial inequality through cumulative causation. EU regional policy, structured around the ERDF, ESF, and Cohesion Fund, channels roughly one-third of the EU budget towards poorer regions, with convergence as the dominant objective. Empirical evaluations are mixed: effects are conditional on human capital and governance, consistent with the augmented Solow view and with NEG's emphasis on absorptive capacity.
A strong essay answer will do three things. First, set up the theoretical contrast between Solow and NEG, with explicit mention of convergence versus cumulative causation. Second, deploy empirical evidence for both sides: conditional convergence findings support Solow; Bleakley and Lin (2012), the core-periphery map, and UK inequality support NEG. Third, apply the framework to EU regional policy, noting that the Becker, Egger, and von Ehrlich (2010) conditionality result reconciles both views: transfers work when they build absorptive capacity, consistent with augmented Solow, and when they compensate for agglomeration disadvantages, consistent with NEG.
- Regional inequality is both a market outcome (NEG) and a diagnostic of structural weakness (absorptive capacity).
- Policy instruments include infrastructure, R&D, human capital, and income transfers.
- Effectiveness is conditional: money without complementary factors yields little.
- The UK case highlights that high national income can coexist with severe regional inequality, making robust regional policy a live concern for any integrated economy.
Bibliography
Banerjee, A., Duflo, E. and Kremer, M. (2019) Understanding Development and Poverty Alleviation. The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 2019, Popular Information. Stockholm: The Royal Swedish Academy of Sciences.
Becker, S.O., Egger, P.H. and von Ehrlich, M. (2010) 'Going NUTS: The effect of EU Structural Funds on regional performance', Journal of Public Economics, 94(9-10), pp. 578-590.
Becker, S.O., Egger, P.H. and von Ehrlich, M. (2013) 'Regional transfers in Europe: Do we need fewer of them or different ones?', VoxEU, 17 July. Available at: http://www.voxeu.org/article/regional-transfers-europe-do-we-need-fewer-them-or-different-ones.
Bleakley, H. and Lin, J. (2012) 'Portage and path dependence', The Quarterly Journal of Economics, 127(2), pp. 587-644.
Boldrin, M. and Canova, F. (2001) 'Inequality and convergence in Europe's regions: Reconsidering European regional policies', Economic Policy, 16(32), pp. 206-253.
Duranton, G. and Kerr, W.R. (2015) The Logic of Agglomeration. NBER Working Paper No. 21452. Cambridge, MA: National Bureau of Economic Research.
Krugman, P. (1991) 'Increasing returns and economic geography', Journal of Political Economy, 99(3), pp. 483-499.
Krugman, P. and Venables, A.J. (1990) 'Integration and the competitiveness of peripheral industry', in Bliss, C. and Braga de Macedo, J. (eds.) Unity with Diversity in the European Economy: The Community's Southern Frontier. Cambridge: Cambridge University Press, pp. 56-77.
Mankiw, N.G., Romer, D. and Weil, D.N. (1992) 'A contribution to the empirics of economic growth', The Quarterly Journal of Economics, 107(2), pp. 407-437.
Midelfart-Knarvik, K.H. and Overman, H.G. (2002) 'Delocation and European integration: Is structural spending justified?', Economic Policy, 17(35), pp. 321-359.
Solow, R.M. (1956) 'A contribution to the theory of economic growth', The Quarterly Journal of Economics, 70(1), pp. 65-94.
Solow, R.M. (1987) Robert M. Solow: Biographical. The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel 1987. Stockholm: The Royal Swedish Academy of Sciences.