VoxEU Column EU institutions Europe's nations and regions

European integration: Benefits of deep versus shallow integration

The imminence of the British referendum lays the European integration project at a crossroads. One tabled policy proposal is to offer different membership options – shallow integration (economic only) and deep integration (economic and political). This column presents new evidence comparing these two options. Focusing on Norway, a country that is economically but not politically associated with the EU, deep integration is estimated to bring a 6% productivity gain in the first five years, compared with shallow integration. These findings bring new economic arguments to debates about EU integration and membership.

A referendum on whether or not the UK remains a full-fledged member of the EU will take place before the end of 2017. It puts the European project at a crossroads. On the one hand, the UK is a major economic and political power so one expects current members will do their best to avoid a ‘Brexit’. On the other, no country has ever left the Union hence one may expect the EU to react with harsh post-exit conditions in order to discourage potential followers.

A clear indication that Europe is at a crossroad is that one tabled policy option is the establishment of different types of membership.1 This idea has recently been echoed by former Italian (Letta 2015) and Belgian prime ministers (Verhofstadt 2015), the French Economy Minister and the (German) President of the European Parliament.2 The difference between associate and full membership corresponds to what economists call shallow and deep integration – the former means purely economic, while the latter involves both economic and political integration (see Baldwin and Jaimovich 2012, Brou and Ruta 2011, Martin et al. 2012, and Sapir 2011). Differentiated memberships would entail a ‘two-speed Europe’ (Bagehot 2011).

This column takes the suggestion of differentiated EU memberships seriously and presents new econometric evidence comparing net benefits from these two options.  

Norway and the EU

The EU-Norway relationship shows it is possible for a country to be economically associated to and, at the same time, politically disassociated from the EU. One important yet difficult question is of course whether this ‘economic only’ type of membership is superior to the ‘economic plus political’ type enjoyed by the 28 full EU members. Disentangling the net benefits of such integration is embroiled with difficulties and has not, to the best of our knowledge, been attempted previously.

The identification strategy we put forward takes advantage of an unique natural experiment – Norway fulfilled all EU entry requirements, completed accession negotiations, accepted founding membership in the European Economic Area (EEA, which gave it unrestricted access to the Single Market), but in a national referendum in 1994 decided to reject full-fledged EU membership.

 According to the European Commission, Norway was as ready to join the EU in 1994 as the other candidates at the time – Sweden, Finland and Austria. The discovery of natural gas reserves preceded the first and second EU referendums (in 1972 and 1994), so Norway is the only country to have voted twice to reject EU membership (House of Commons 2013). Because it rejected full-fledged membership in 1994, Norway enjoys the benefits of economic integration with the EU but not the benefits of political integration.

In the deep end

In recent research, we use differences-in-differences and synthetic control methods (SCM) to estimate the consequences in productivity terms of a hypothetical EU entry of Norwegian regions, before and after the 1995 EU enlargement (Campos, Coricelli and Moretti 2015).  We use regional annual data from Norway, Austria, Finland and Sweden spanning 1985 to 2000.

Table 1 summarises our differences-in-differences results. We estimate that the average change before and after 1995 in the level of productivity in Norwegian regions was between 1.6 and 2 euros (or about 8%) lower than the average change before and after 1995 in regions of the three countries that joined the EU in 1995. Oslo has a distinct behaviour and its exclusion from the sample means the magnitude of the effect is higher.

Table 1. Differences-in-differences estimations on regional level data (1985-2000)

Note: Robust standard errors in parentheses. Inference: ***p<0.01, ** p<0.05, * p<0.1. Norway is a dummy variable that takes the value 1 for Norwegian regions, the value 0 for Austrian, Finnish, and Swedish regions. Post95 is a dummy variable that takes the value 1 for years from 1995 to 2000, the value 0 for years from 1985 to 1994. Controls RY is a set of control variables with region-year dimension (such as the shares of industries employment on total employment, population density, investment share, education).

Figure 1 shows our SCM estimations in a differences-in-differences fashion with the pre- (1985-1994) and post-treatment (1995-2000) differences in productivity (output per hour worked) between each of the seven Norwegian NUTS2 regions and their (synthetic) counterfactuals. It shows that during the pre-treatment period the actual and synthetic regions show similar levels of productivity, while we estimate large differences for the post-treatment period. Again, the Oslo region is exceptional – the post-treatment difference between actual and synthetic is positive.

Figure 1. Difference (%) between actual and synthetic counterfactual region

The average of our SMC results is not far from the effect from differences-in-differences (averaged across regions). One reason they differ is that the SCM fully accounts for the presence of time-variant unobservable characteristics (Abadie et al 2010).

Figure 2 has our full SCM estimations, with the red line showing the year-by-year difference between the observed levels of productivity and our estimated counterfactuals for each of Norway NUTS2 regions. If the values post-1995 are negative (positive), productivity would be higher (lower) if that individual region instead were a full-fledged member state of the EU. The figure shows that the difference between productivity levels in actual and synthetic regions is not constant but increases over time suggesting these politically-driven productivity benefits are not temporary.

Figure 2. Productivity difference (euros) between actual and synthetic counterfactual region (1985-2000)

Notes: The red line represents our baseline estimation that is obtained using in the donor sample all of the 22 NUTS2 regions of Austria, Finland, and Sweden. The grey lines represent 100 alternative estimations that are obtained using in the donor sample 11 regions randomly chosen from the pool of 22 NUTS2 regions of Austria, Finland, and Sweden.

Each region in the donor pool that takes a positive weight influences, by construction, the post-treatment dynamics of the synthetic region and thus the estimated effects. One of the main concerns in estimating synthetic counterfactuals is that the results could be driven by individual regions in the donor pool experiencing idiosyncratic shocks unrelated to the treatment. In order to deal with this possibility, we iteratively re-estimate the synthetic counterfactuals using a large number (one hundred) of alternative donor pools. Each alternative donor pool randomly selects half of the regions used for the baseline estimations (11 regions out of 22).  The grey lines in Figure 2 show that the direction of the effects obtained in the baseline estimation (red line) is robust to different donor samples. If anything, the baseline estimations underestimate the effects vis-à-vis the random donor pools. We conclude that the estimated productivity effects from our baseline estimations are not unduly driven by the specific composition of the donor pool.

Overall, the main finding is that there are substantial economic benefits from full EU membership (deep integration) when compared to associated membership (shallow integration). If Norway had joined the EU in 1995, productivity levels would have been 6% higher in an average Norwegian region between 1995 and 2000.

Of the seven official (NUTS2) Norwegian regions, only Oslo shows lower productivity with hypothetical full EU membership. This exception reinforces our interpretation of these effects as politically-driven benefits from integration. Even without full membership, the ‘economic only’ integration boosts the productivity of the capital city region thanks to its high level of human capital, growing financial sector, and efficient public sector. In contrast, other regions remained shielded from competition through generous subsidies and protection of traditional Norwegian activities such as fisheries and agriculture, which full membership would not have permitted.

At the crossroad

We provide new evidence suggesting that the net benefits from deep integration significantly outstrip those from shallow integration. Since Norway already enjoys economic integration benefits via EEA, our results suggest substantial additional benefits from political integration. This challenges the conventional wisdom that benefits are mostly related to economic integration, while costs are mostly due to political dimensions of integration. It also further shores up the notion that deeper integration could be instrumental towards a more efficient monetary union (De Grauwe and Ji 2014).  

Considering this evidence in light of the British question, these results would suggest that a Brexit may generate heavy losses for the UK – any economic losses3 are likely to be augmented by the complementary loss of politically-driven benefits. 

Our results also have implications for the broader European project. They show that deeper integration pays off. European integration has always been explicitly a political and an economic process. The choice of a customs union model instead of a free trade area, enshrined in the Treaty of Rome, underscores the agreed direction of travel as one towards deep integration. Considering different types of membership lessens the ideals encapsulated in the Treaty of Rome. Our results suggest that attempts to dilute these ideals should be resisted not morally or politically but on economic grounds. ‘Ever closer union’ is often considered (and often criticised for being) a soft and exclusively political argument. Our results suggest that by being at the heart of deep integration, ‘ever closer union’ actually embodies a powerful economic argument. Should the EU offer the possibility of associated membership to its members? And how would this potential ‘hollowing’ affect the payoffs from deep integration? Further research on these questions is urgently needed. 


Abadie, A, A Diamond and J Hainmueller (2010) “Synthetic control methods for comparative case studies: Estimating the effect of California’s tobacco control program”, Journal of American Statistical Association 105(490): 493–505.

Bagehot (2011) “Britain changes its mind about a two-speed Europe”, The Economist, 21 July.

Baldwin, R and D Jaimovich (2012) “Are free trade agreements contagious?”, Journal of International Economics, 88(1): 1–16.

Brou, D and M Ruta (2011) “Economic integration, political integration or both?”, Journal of the European Economic Association, 9(6): 1143–1167.

Campos, N, F Coricelli and L Moretti (2014) “Economic growth and political integration: Estimating the benefits from membership in the European Union”, CEPR Discussion Paper 9968.

Campos, N, F Coricelli and L Moretti (2015) “Norwegian rhapsody? The political economy benefits of regional integration”, CEPR Discussion Paper 10653.

De Grauwe, P and Y Ji (2014) "The future of the Eurozone", Manchester School, 82(1): 15-34.

European Council (2014), “Conclusions”, European Council, June 2014.

House of Commons (2013) “Norway’s relationship with the EU”, London: UK House of Commons Library Standard Note SN06522.

Letta, E (2015) “We can rescue Europe by letting Britain move into the slow lane”, The Guardian, 15 May.

Lichfield and Patterson (2015) “EU referendum: David Cameron boosted by support from European politicians for a 'two-speed Europe'”, The Independent, 1 June.

Martin, P, T Mayer and M Thoenig (2012) “The geography of conflicts and regional trade agreements”, American Economic Journal: Macroeconomics, 4(4): 1-35.

Sapir, A (2011) “European integration at the crossroads: A review essay on the 50th anniversary of Bela Balassa’s theory of economic integration”, Journal of Economic Literature, 49(4): 1200–1229.

Verhofstadt, G (2015) “An EU for ‘full’ members only”, Politico, 6 May.


1 Conclusion 27 from the June 2014 European Council reads: “The UK raised some concerns related to the future development of the EU. These concerns will need to be addressed. In this context, the European Council noted that the concept of ever closer union allows for different paths of integration for different countries, allowing those that want to deepen integration to move ahead, while respecting the wish of those who do not want to deepen any further” (2014, p. 12).

2 The latter two are quoted in Lichfeld and Patterson (2015).

3 For example, due to diminished international trade, foreign investment, financial globalisation and migration.

6,404 Reads