Institutional Integration and Economic Growth in Europe

Abstract

The literature on the growth effects of European integration remains inconclusive. This is due to severe methodological difficulties mostly driven by country heterogeneity. This paper addresses these concerns using the synthetic control method. It constructs counterfactuals for countries that joined the European Union (EU)from 1973 to 2004. We find that growth effects from EU membership are large and positive, with Greece as the exception. Despite substantial variation across countries and over time, we estimate that without European integration, per capita incomes would have been, on average, approximately 10% lower in the first ten years after joining the EU.

Full text

ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Journal of Monetary Economics 0 0 0 (2018) 1–17
Contents lists available at ScienceDirect
Journal of Monetary Economics
journal homepage: www.elsevier.com/locate/jmoneco
Institutional integration and economic growth in Europe
Nauro F. Campos
a , b
, Fabrizio Coricelli
c , d , ∗, Luigi Moretti
e
a
Brunel University London, United Kingdom
b
ETH-Zurich, Switzerland
c
Paris School of Economics, France
d
CEPR, United Kingdom
e
Centre d’Economie de la Sorbonne, Université Paris 1 Panthéon-Sorbonne, France
a r t i c l e i n f o
Article history:
Received 2 April 2018
Revised 8 July 2018
Accepted 1 August 2018
Available online xxx
JEL classification:
C33
F15
N14
N44
O52
Keywo rds:
Economic growth
Integration
Institutions
European union
Synthetic control method
a b s t r a c t
The literature on the growth effects of European integration remains inconclusive. This is
due to severe methodological difficulties mostly driven by country heterogeneity. This pa-
per addresses these concerns using the synthetic control method. It constructs counterfac-
tuals for countries that joined the European Union (EU) from 1973 to 20 04. We find that
growth effects from EU membership are large and positive, with Greece as the exception.
Despite substantial variation across countries and over time, we estimate that without Eu-
ropean integration, per capita incomes would have been, on average, approximately 10%
lower in the first ten years after joining the EU.
©2018 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license.
( http://creativecommons.org/licenses/by-nc-nd/4.0/ )
1. Introduction
Undoubtedly, the creation of the European Union (EU) and its subsequent enlargements represent one of the
deeper examples of voluntary institutional change involving a large number of countries during the post-war period.
The importance of such institutional integration has recently been brought to the center of political debates in re-
lation to Brexit, the first example of a country exiting the EU. The literature on the effects of European integration
on economic growth and productivity remains largely inconclusive because of well-known methodological difficulties
( Eichengreen, 2007 ; Crafts, 2016 ). Probably the most serious difficulty is the heterogeneity of country experiences before
and after their accession to the European Union (EU).
1
The aim of this paper is to present new estimates of the economic effects of European integration that address these
concerns. It does so using the synthetic control method (or “synthetic control method for causal inference in comparative
∗Corresponding author at: Paris School of Economics, 48 Boulevard Jourdan, 75014 Paris, France.
E-mail address: fabrizio.coricelli@psemail.eu (F. Coricelli).
1 We use the term European Union (or EU for short) for convenience throughout, that is, even when referring to what was then called the European
Economic Community or, later on, the European Communities.
https://doi.org/10.1016/j.jmoneco.2018.08.001
0304-3932/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license.
( http://creativecommons.org/licenses/by-nc-nd/4.0/ )
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

2 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
case studies”) pioneered by Abadie and Gardeazabal (2003) .
2 There are various important issues in assessing the benefits
from EU membership, but there is consensus that building counterfactuals is essential although, as Boltho and Eichengreen
note, “imagining the counterfactual is no easy task” (2008, p.13).
In estimating the net benefits from EU membership, we address the following main questions. What would be the level
of per capita income and productivity in a given country had it not joined the EU? Are these pay-offs from integration
temporary or permanent? Do they vary across countries?
In order to explicitly deal with country heterogeneity, these payoffs are estimated at country level for all main EU en-
largements (i.e., 1973 , 19 81, 1986, 1995 and 2004 enlargements).
3 To construct our historical counterfactuals, we take ad-
vantage of the binarity of membership in the EU (a country is or is not a full-fledged EU member). However, the timing and
the complexity of integration are two important issues to bear in mind.
Timing refers to the fact that the effects of EU entry may be felt before the formal date of accession. Economic actors
may anticipate entry. This issue is particularly relevant for the later enlargements, which were preceded by long periods of
preparation.
Complexity refers to the fact that, although EU membership may be binary, there is a continuum of degrees of economic
integration, which cannot be fully captured by a dummy variable. The extent of integration may vary across different areas
(e.g., goods, finance, services, technology, policies, etc.) and over time.
4
For instance, joining the EU in 1973 is different from
joining in 1995 (as the degree and type of integration at entry is different). Similarly, the institutional and regulatory changes
that countries need to make to become members are different: a country with a higher level of institutional development
goes through a simpler accession process than a less institutionally developed country.
Our case-study counterfactual approach addresses the difference in the degree of integration across countries and across
enlargements. Indeed, it provides measures of the effect for each single country that joined the EU and, thus, the effects
of the (binary) membership status reflect the country entering conditions and EU institutions at the time of enlargement.
Moreover, the synthetic control method allows us to assess how these effects change over time for each individual coun-
try. In contrast, standard panel or difference-in-difference approaches can only estimate “the overall average effect” of EU
membership and not the dynamic effects of EU membership on each individual country that joined the EU.
The main conclusion from our analysis is that the economic benefits from EU membership are generally positive and
large. Unsurprisingly, we find considerable heterogeneity across countries. However, our estimates indicate that only one
country, Greece, experienced smaller GDP and productivity levels after EU accession. Overall, our estimates suggest that
per capita European incomes in the absence of the institutional integration would have been about 10% lower on average
in the first ten years after joining the EU. Although this figure varies across enlargements and over time, it is within the
range of existing estimates, which go from a minimum of 5% gains in per capita income from EU accession ( Boltho and
Eichengreen, 2008 ), to a maximum of 20% gains ( Badinger, 2005 ). Differently from the rest of the literature, our estimates
are robust to a large set of sensitivity checks, including changes in the composition of the control group of non-EU countries
used to construct the counterfactuals. The latter is a methodological innovation we introduce in this paper.
The paper is organized as follows. Section 2 discusses previous attempts of estimating the growth and productivity ef-
fects from EU membership. Section 3 presents the synthetic control method and the data used in the empirical analysis.
Section 4 introduces our baseline estimates for the northern and southern enlargements, while Section 5 presents the es-
timates for the eastern enlargement. Section 6 discusses various robustness checks. Section 7 investigates the potential
reasons for the variation of the effects of EU entry across countries and over time. Section 8 concludes.
2. European integration: growth and productivity effects
Theoretically, the link between integration and growth remains a subject of debate. Using an endogenous growth frame-
work, Rivera-Batiz and Romer (1991) show that economic integration for countries at similar levels of per capita income
leads to long-run growth when it accelerates technological innovation (mostly through R&D and new ideas). Such effects
can also be achieved through trade in goods if the production of ideas does not need the stock of knowledge as an input
(this is the so-called “lab-equipment” model). In other words, the effects of economic integration on growth depend on spe-
cific channels leading to possible long-term benefits either through larger flows of goods or flows of ideas ( Ventura, 2005 ).
Further, the size of the growth dividend also depends on the similarity of per capita income levels.
These problems in deriving clear-cut effects of integration on growth are related to a lack of debate on the type of inte-
gration (e.g., deep versus shallow integration, as discussed in Brou and Ruta, 2011 , and Campos et al., 2015 ). In the economic
literature, the distinction between “shallow” and “deep” integration was introduced by Lawrence (1996) . Lawrence identified
2 A recent authoritative review of empirical methods argues that “The synthetic control approach developed by Abadie, Diamond, and Hainmueller (2010,
2015) and Abadie and Gardeazabal (2003) is arguably the most important innovation in the policy evaluation literature in the last 15 years. This method
builds on difference-in-differences estimation, but uses systematically more attractive comparisons.” ( Athey and Imbens, 2017 , p. 9).
3 There are important reasons for focusing on enlargement episodes instead of the experience of the six founders of the EU. One is that integration
was initially gradual, with trade barriers reduced over a ten-year period. By contrast, countries involved in the subsequent enlargements joined an already
largely liberalized trade area.
Moreover, there are various difficulties in building a panel dataset of candidate countries for the pre-1957 period that could
serve as potential counterfactuals for the EU founding members as this would necessarily encompass the WWII years.
4 See Dorrucci et al. (2004) for continuous indexes of economic integration in Europe.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 3
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
shallow integration with traditional trade agreements affecting tariffs and other border measures, while he identified deep
integration with trade agreements that go beyond traditional areas and affect competition and regulation policies. As an
increasing numbers of trade agreements have deep provisions ( Hoffman et al., 2017 ) here we refer to this as “institutional
integration” to distinguish it from the political integration process, which is yet another layer of depth.
In view of the theoretical and conceptual difficulties in deriving clear-cut effects of integration on growth (chiefly regard-
ing economic versus institutional integration) empirical analysis remains crucial.
2.1. A brief history of European integration and survey of the empirical literature
The massive destruction from Worl d War II was followed by swift economic recovery. By the early 1950s, most European
countries already register per capita GDPs above pre-war levels. A period known as the Golden Age of European growth
followed and between 1950 and 1973 Western and Eastern Europe grew at unprecedented rates ( Eichengreen, 2007 ). Deep
trade liberalization shore up this extraordinary economic expansion in the context of both EU-6 and EFTA.
5
European integration progressed over time in depth and extent. The deepening of trade liberalization in the 1960s was
followed by the first EU enlargement in 1973 (with the accession of the UK, Ireland and Denmark). The 198 0s saw further
increases in EU membership (Greece in 1981 and Spain and Portugal in 1986), which were followed by deepening in terms
of the Single Market. Next came another enlargement (Austria, Finland and Sweden in 1995) and then yet another deepening
with the introduction of the common currency. This was finally followed by the largest, in terms of number of countries
involved, of the enlargements (eight eastern countries plus Cyprus and Malta in 2004, Bulgaria and Romania in 2007 and
Croatia in 2013).
The deepening and broadening of European integration generated substantial growth and productivity payoffs to the
point that many scholars attach exceptionality to Europe. According to Eichengreen (2007) , Europe is the only region show-
ing evidence of unconditional beta and sigma convergences.
Focusing on the channels through which European integration affects growth, the early literature argues that the effects
of integration on growth worked mostly through the effects of trade integration (for a critical view see Slaughter, 2001 ).
Baldwin and Seghezza (1996) survey the evidence and found that the main channel through which European integration
accelerated European growth was through the boost to investment in physical capital, induced by efficiency gains brought
about by trade integration.
6
In spite of a large literature on the benefits from trade liberalization associated to the EU, from the Single Market, and
from the Euro, there is a relative dearth of econometric estimates of the benefits from EU membership.
7 Not only studies
about the benefits of EU membership are few,
8 but also the majority of these (few) studies openly warn against the lack of
robustness of their estimates.
Henrekson et al. (1997) estimate the benefits from membership to be about 0.6 to 0.8% per year but note that such es-
timates are “not completely robust” (1997, p. 15 51). Badinger (2005) estimates that “GDP per capita of the EU would be ap-
proximately one-fifth lower today if no integration had taken place since 1950” but cautions that these are “not completely
robust” (p. 50). Crespo et al. (2008) find large growth effects from EU membership, but warn that country heterogeneity
remains a severe concern.
Ben-David (1993, 1996 ) studies European integration as an engine for income per capita convergence. In his 1993 paper,
he concludes that European trade integration leads to a reduction of income dispersion. To overcome identification problems,
Ben-David (1996) contrasts the “trade-integration club” with alternative random clubs of the same size, in terms of number
of countries involved. Indeed, convergence is observed only for the trade integrated clubs.
Ben-David’s analysis does not generate a robust counterfactual, as the selection of the non-integrated clubs does not
ensure that they behave like the integrated club prior to integration. Ye t, his analysis is a main motivation for our work, as
it emphasizes the significantly different economic effects between creating an integrated area such as the EU, which involves
both economic and institutional integration, and simple trade liberalization among countries, which only involves economic
integration.
In short, most of the previous literature on the growth dividends from EU membership uses panel data econometrics
and information up to the 1990s enlargement to infer the size of these net benefits and whether they are permanent or
temporary. We echo Boltho and Eichengreen’s (2008) concern that a main difficulty in such analyses is the identification
of a benchmark, a baseline for comparison, a relevant counterfactual. The literature has not yet satisfactorily addressed this
issue.
5 The European Fre e Trade Association (EFTA) was established in 1960. The founding members were Denmark, United Kingdom, Portugal, Austr ia, Sweden,
Norway and Switzerland (only the last two are still members today).
6 This earlier literature focuses on the effects of international trade on growth and often assumes that all the increase in trade is driven purely by intra-
European integration effort s. Moreover, the extent of trade diversion of “deep” agreements such as the EU is questio nable as they contain both provisions
that
discriminate between members and non-members (such as tariffs) and provisions that favor trade with all (provisions that limit state aid to domestic
producers or that increase competition). Recent evidence finds that deep agreements increase members’ trade but do not significantly divert trade with
non-members ( Mattoo et al, 2017 ).
7 See among others Boltho and Eichengreen (2008) .
8 Crafts (2016) and Sapir (2011) survey the literature.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

4 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
3. Methodology, sample and data
The aim of this paper is to estimate what would have been the levels of per capita GDP or productivity in a given country
if it had not become a full-fledged member of the European Union. We answer this question for the countries that became
EU members in the 1973 , 1980s, 1995 and 2004 enlargements, using the “synthetic control method for causal inference in
comparative case studies” or synthetic control method (SCM), developed by Abadie and Gardeazabal (2003) and Abadie et al.
(2010 , 2015 ).
9
3.1. The synthetic control method
In general, the dynamic effect of an intervention or event occurring at a given time ( T
0
) on a country could be repre-
sented by:
τit
= Y
I
it
−Y
C
it
for all t ≥T
0 (1)
where Y
I
it
is country’s i outcome at time t , and Y
C
it
is country i ’s outcome at time t had it not been affected by the interven-
tion, that is, the counterfactual. Since we do observe Y
I
it
for t ≥T
0 but we cannot observe Y
C
it
for t ≥T
0
, we need to provide
an estimation of the counterfactual (i.e.,

Y
C
it
) to assess the dynamic effect of the intervention on country i (i.e., ˆ τit
).
In our context, countries affected by the event (EU membership) are clearly not randomly selected. Therefore, to assess
the effect of the event we need to compare the country that joined the EU against a group of control countries (non-EU
members) before and after.
To make such a comparison we use the SCM. It differs from methods such as difference-in-differences in that it “moves
away from using a single control unit or a simple average of control units, and instead uses a weighted average of the set of
control” ( Athey and Imbens, 2017 , p. 9). Indeed, the SCM focuses on the construction of the “synthetic control” or “artificial
country”, which serves as the counterfactual scenario for the country affected by the event (or “actual country”).
Consider a set of N + 1 countries for each period t ∈ [ 1 , T ] , where i = 1 is the country joining the EU at time T
0
∈
( 1 , T ) , and i = 2 , . . . , N + 1 are the non-EU control countries (or “donor pool”). Abadie et al. (2010 ) show that an unbiased
estimate of the counterfactual obtains by assigning a weight w to each control country, so that

N+1
i =2
w
∗
i
Y
it
=

Y
C
1 t
for t ≥T
0
.
An unbiased estimate of the dynamic effects of EU membership on country i can therefore be represented by:
ˆ τit
= Y
1 t
−
N+1

i =2
w
∗
i
Y
it for all t ≥T
0
. (2)
The combination of optimal weights ( w
∗
i
) assigned to the control countries in Eq. (2) is chosen to minimize the pre-
event (pre-EU accession) differences between the country in question and its synthetic (weighted) control in terms of a set
of predictors of the outcome (Z) , so that

N+1
i =2
w
∗
i
Z
i
= Z
1 and

N+1
i =2
w
∗
i
Y
it
= Y
1 t for t < T
0
.
All else the same, the longer is the pre-event period, the more accurate is the synthetic control. Indeed, the year-by-
year match during the pre-event period between the country and its synthetic control permits to deal with time-varying
omitted variables. This is a further improvement with respect to other methods, such as panel fixed-effects or difference-in-
differences, which can only account for confounders that are time-invariant or share a common trend.
10
There are two key assumptions in SCM: (1) the choice of predictors of the outcome should include variables that can
approximate the path of the country affected by the intervention but should not include variables that anticipate the effects
of the intervention; and (2) the countries used to estimate the synthetic control (the “donor pool”) should not be affected
by the event.
The first assumption implies that the effects of the event on the country are not anticipated. In our case, the absence
of anticipation effects means that the growth effects of EU membership are observed only after each candidate country
effectively becomes a full-fledged member, not before. If agents anticipate these effects, then the SCM is likely to generate
a lower-bound estimate of the true effects because part of it occurs before membership.
The second assumption requires that the performance of the countries selected in the “donor pool” should not be af-
fected by the EU accession of the country under analysis. Although this assumption obviously holds when one defines the
intervention as “full-fledged EU membership,” one should keep in mind that in a globalized world, spillovers occur and
cannot be fully excluded. Yet the direction of the bias that spillovers can introduce in our estimations is unclear. Indeed,
through their trade or financial links, some donor countries can benefit from the EU integration of the country affected by
the event, while other donor countries can lose.
3.2. Data and model specification
In our application of the SCM, we employ two alternative outcome variables ( Y
it
) GDP per capita and labor productivity
(the latter, defined as GDP per worker; both variables are from Penn World Tables).
9 See Imbens and Wooldridge (20 09) for a discussion of SCM in comparison to similar econometrics methods.
10 See Bove et al. (2017) for a comparison of SCM and panel fixed effects.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 5
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Our choice of the predictors ( Z
i
) of the outcome is based upon the specification used by Abadie et al. (2003 , 2015 )
and is in line with the empirical growth literature ( Levine and Renelt, 1992 ). The specification includes the investment
share in GDP, population growth and pre-intervention income (all from Penn World Tables), share of agriculture and share
of industry in value added, secondary and tertiary gross school enrolment (from the World Bank’s World Development
Indicators). As noted, in order to avoid the inclusion of variables that are directly affected by the event, we deliberately
exclude trade, foreign direct investment and financial integration variables. We indirectly assess the role of these latter
variables in Section 7 below, by estimating the main determinants of the effects of EU integration derived from the SCM.
The weights on the countries forming the synthetic counterfactual are determined on the basis of the match produced
by the determinants of GDP per capita or labor productivity. Therefore, countries forming the “artificial country” may differ
from the treated country in the more structural dimensions associated to geography, culture and institutions. However, those
variables typically are “slow-moving”, unlikely to change in the donor pool during the pre and post-treatment periods. In
summary, the synthetic counterfactual matches the outcome’s predictors so that the outcome variables mimic both the
levels and the trend of the actual country’s outcome before the event occurs.
11
3.3. Sample
We estimate counterfactuals for each country in four EU enlargements, namely for Denmark, Ireland and the UK in 19 73,
for Greece, Portugal and Spain in the 1980s, for Austria, Finland and Sweden in 1995 and for the eight eastern European
countries in 2004.
12 As our focus is on long run effects, our analysis stops in 2008 to avoid confounding effects from the
global financial crisis.
For the estimation of the baseline results we use a donor pool that excludes EU27 but includes OECD, EU neighbouring
countries, Mediterranean and newly industrialized countries.
13 Two points are worth stressing. First, following Abadie et al.
(2015) , the donor pool does not have to include only countries having high probability of becoming EU members in the
future (indeed, as discussed in Section 3.1 , the condition that cannot be violated is that countries in the control group are
not affected by the EU accession of the country under analysis). Second, the specific donor pool selected is important for
the point estimates but it is not critical. As shown in Section 6.2 , our results are robust to random selection of the countries
in the donor pool.
In general, the SCM addresses endogeneity and omitted variable concerns but one of its main drawbacks is that it “does
not allow assessing the significance of the results using standard (large-sample) inferential techniques, because the number
of observations in the control pool and the number of periods covered by the sample are usually quite small in comparative
case studies” ( Billmeier and Nannicini, 2013 , p. 987).
One needs to be aware of the potential dependence of our results on idiosyncratic shocks affecting countries in the donor
pool. The occurrence of such idiosyncratic shocks in the post-accession period may be incorrectly interpreted as showing
the effect of the EU membership on the new member State. Therefore, in Section 6.2 , we implement a simple yet novel
solution to test the robustness of our findings to the composition of the donor sample. Namely, for each EU country under
analysis, we construct one thousand alternative counterfactuals based on alternative donor samples that include countries
randomly selected from the full donor pool sample. We then compare our main estimates with those obtained with the
random samples, both in terms of pre-accession fit and estimated effects induced by the EU membership.
4. Results for the northern and southern enlargements
Figs. 1 and 2 present the counterfactual baseline results for the countries that joined the EU during the 1970s, 1980s,
and 1990s.
Fig. 1 reports, for the countries taking positive weights, the set of optimal weights that allows the synthetic country to
mimic as close as possible each country before the accession to the EU. For most countries, there are only a few donors with
positive weights. For instance, the set of optimal weights for “synthetic Spain” are 0.373 to Brazil, 0.358 to New Zealand,
and 0.268 to Canada (and 0 for Albania or Japan or any other country in the donor sample). This means that the “synthetic
Spain” is composed for the 37.3% by Brazil, the 35.8% by New Zealand, and the 26.8% by Canada.
Given these estimated weights obtained for each synthetic country, Fig. 2 shows the actual and counterfactual series
of GDP per capita: the continuous line represents the actual per capita GDP, while the dashed line shows its estimated
synthetic counterfactual. The vertical dashed line divides the pre-accession to the EU period (over which the weights for the
‘synthetic country’ are estimated) from the post-accession period (over which the ‘synthetic country’ is projected).
Greece is the only of the 17 countries we consider for which net benefits are negative. Our estimates show that Greek
per capita GDP would have been higher if Greece had not become a full-fledged EU member in 1981. Yet the gap shrinks
11 One should also keep in mind that growth regressions using traditional panel methods implicitly attribute weigh ts to the various countries, with no
non-negative restriction on such weights. The countries included in the panel estimations are highly heterogeneous, but it is accepted that the effect say
of investments or education on
growth is similar across countries ( Abadie et al., 2015 ).
12 We have excluded from our analysis Cyprus and Malta due to data availability and to their relative small size (and the difficulties this may generate to
find satisfactory matching countries), and Bulgaria, Croatia and Romania because the period post-EU membership is too short.
13 See the appendix for the full list of countries.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

6 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Fig. 1. Composition of the synthetic country –GDP per capita, northern and southern enlargements.
Note: In each graph, the pie chart reports the six donor countries (and the relative weights) that contribute the most to the construction of the synthetic
country in qu estion. For full details see Table A.1 in Appendix.
over time, particularly after 1995. This result does not imply Greece would be better off leaving or never joining the EU.
From 198 1 to 1995, growth rates in the EU were relatively high and Greece experienced divergence ( Vamvakidis, 2003 ). The
opening up of an uncompetitive domestic industry may have been too sudden.
14 However, entry into the economic and
monetary union represents a turnaround, with growth rates in Greece faster than in the EU for 1996–2008, mostly driven
by telecommunications, tourism and the financial sector ( Arkolakis et al., 2017 ). Interestingly, the latter is one of the few
areas in which the country implemented structural reforms ( Mitsopoulos and Pelagidis, 2012 ).
Concerning the countries that joined the EU in 1973, the results in Fig. 2 suggest that per capita GDP would be consid-
erably lower in these countries had they not joined the EU. The actual and the synthetic series are reasonably close before
1973 and diverge afterwards. The dynamics of these benefits is noteworthy. For example, the benefits from EU membership
for the UK (although substantial throughout) may have slowed down in later years while for Ireland they seem to have
instead accelerated. This would suggest that perhaps the UK benefited more from the Single Market while Ireland benefited
more from the common currency. This “sequence of deepening” also illustrates how difficult it is to separate temporary from
permanent effects.
The results for Austria, Sweden and Finland suggest that EU membership generated positive dividends in terms of per
capita GDP. Overall, the estimated payoffs from EU membership for Sweden, and to a lesser extent Austria and Finland,
are small compared to those in the 1973 enlargement. One possible explanation is that when these countries joined the
EU in 1995 they already had a relatively high level of per capita income. However, Denmark and the UK were at similar
income per capita levels relative to existing EU countries.
15 An alternative explanation is that, while the main impediment
for the 1995 countries to join was political (the Cold War), the 1973 countries designed, implemented and benefited from
14 In 1976 , the Council of Ministers extraordinarily rejected the European Commission’s view that was against opening accession negotiations with Greece
and in favor of delaying entry until Greek producers were deemed able to compete in the Common Market.
15 The “per capita income gap at entry” is the percentage difference between the per capita income average of existing members and that of candidate
countries, in USD PPP, for the official accession year. We calculate that candidate countries in 19 73 had on average 96% of the per capita income of existing
members, in the 1980 s this was 63%, in 1995 this was 10 3%, while in 2004 it was 45%.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 7
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Fig. 2. Actual vs. synthetic –GDP per capita trends, northern and southern enlargements.
Note: In each graph, the continuous line represents the trend of the real GDP per capita (PPP, 2005 International Dollars) for the actual country, while the
dashed line shows the trend for the same variable for the synthetic country. For each country the analysis ends in 2008. The composition of each synthetic
country is reported in Figure 1 and full details in Tabl e A.1 in Appendix.
the Single Market (1986–1992) and from the common currency and attendant financial integration. A possible line of inquiry
could focus on institutions. If the benefits are due to the potential positive effect of EU accession on institutional integration,
then one would expect smaller potential gains from membership in the case of Austria, Finland and Sweden in 1995, as they
had already relatively high levels of institutional development.
5. Results for the eastern enlargement
Let us now focus on the results for the eastern European countries that joined the EU in 2004. Given the shorter data
series, we must be more cautious when interpreting this set of results.
Fig. 3 reports the results for the 2004 eastern enlargement. Results are mixed, with benefits that are positive and large
for some countries, while weak or even negative for others.
In contrast with all other previous enlargements, the eastern enlargement was preceded by a long preparation process,
which entailed substantial institutional change both for entrants and for the EU itself (see Bache et al., 2011 and Campos
and Coricelli, 2002 ). Moreover, the pre-accession period involved free trade and economic integration agreements. The 2004
enlargement is the first one for which a conscious effort to guarantee satisfactory levels of institutional integration takes
place before the official accession date. Therefore, it is likely that part of the effects related to the entry in the EU manifested
before 2004 ( Bruszt and Campos, 2017 ).
Furthermore, economic agents may have anticipated the future EU membership of these countries. The main political
signal on the forthcoming enlargement was given by the European Council in December 1997, which established the proce-
dures for the eastern enlargement following the indications of the report “Agenda 20 0 0” submitted by the EU Commission.
Accordingly, in order to assess the effects for the eastern enlargement we re-estimate the synthetic counterfactuals using
199 8 as the accession year, rather than the official accession date (2004).
16
16 Other enlargements are less likely to be affected by anticipation effects: Kutan and Yigit (2007) show that structural breaks in GDP and productivity
series for the 1980 s and 1990s enlargements occur substantially close to the “official” accession dates.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

8 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Fig. 3. Actual vs. synthetic country –GDP per capita trends, eastern enlargement (2004 accession).
Note: In each graph, the continuous line represents the trend of the real GDP per capita (PPP, 2005 International Dollars) for the actual country, while the
dashed line shows the trend for the same variabl e for the synthetic country. For each country the analysis ends in 2008. The composition of the synthetic
country is reported in Figure A.1 and full details in Ta ble A.2 in Appendix.
Fig. 4 reports the set of optimal estimated weights that allows the synthetic country to mimic as close as possible each
actual country before 1998. Results in Fig. 5 show that the benefits from EU integration are positive and large across these
countries with the exception of the Slovak Republic.
Given the long preparation before accession, one can argue that in addition to anticipation effects there could be effects
due to actual implementation of pre-accession agreements leading to further integration with EU member states. Indeed,
there is evidence of such effects, which, together with the results on the anticipation effects, helps to explain the ambiguous
and not robust effects for the 2004 accession date.
17
6. Magnitude of the effects and sensitivity analysis
To further probe the robustness of our results, first we calculate the net benefits from EU membership at different time
horizons, and, then, we report estimates using randomly-generated donor samples to address concerns that the estimates
above may be driven by the specific composition of a sample of donor countries.
6.1. Post-accession effects at different time horizons
Because the time horizon over which we can reasonably attribute the dynamics of per capita GDP or productivity relative
to a synthetic counterfactual to EU accession varies, in Table 1 we report the average difference between actual and the
17 We thank an anonymous referee for this observation. In the Appendix, we provide a first and partial attempt using the SCM. The event we assess is
the joint presence of free trade and economic integration agreements between eastern candidates and EU members (the definition of the yea r of treatment
follows
Regional Trade Agreements Database from Egger and Larch, 2008 ). Figures A .8 and A .9 (and Table s A .7 and A .8) show that the effects are positive
both on GDP per capita and labor productivity. A full treatment of multiple treatment effects goes beyond the scope of this paper and it is an interesting
area for future research.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 9
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Tabl e 1
Percentage effects.
GDP per capita Labor productivity
(1) (2) (3) (4) (5) (6)
All post-accession 10 years post-accession 5 years post-accession All post-accession 10 years post-accession 5 years post-accession
Denmark 23.863 14. 298 10.292 12.673 −0.562 −0.494
United Kingdom 23.694 8.586 4.824 32.832 8.537 5.551
Ireland 48.9 9.395 5.242 28.348 8.555 6.64
Greece −19.758 −17.33 6 −11. 5 91 −12.293 −14.144 −11.456
Portugal 18.351 16. 537 11.7 33 11. 66 12. 321 10 .261
Spain 19. 806 13. 662 9.348 4.032 3.724 0.759
Austria 7.2 08 6.364 4.467 13.3 42 12.896 10.811
Finland 4.365 4.017 2.185 4.261 4.469 4.099
Sweden 3.174 2.353 0.823 2.925 2.617 1.7 13
Czech Republic 5.615 5.615 2.11 3.665 3.665 −0.563
Estonia 24.153 24.153 16.3 42 20.462 20.462 15.9 81
Hungary 12.299 12.299 8.734 17.6 97 17.6 97 12.9 37
Latvia 31.692 31.692 18. 016 19 .37 19.37 14.952
Lithuania 28.082 28.082 17.3 52 24.114 24.114 15. 31
Poland 5.93 5.93 8.67 9.387 9.387 10. 257
Slovak Republic 0.302 0.302 1. 315 −1.755 −1.755 −1.98 5
Slovenia 10.3 5 10. 35 6.327 12.778 12.778 10.84 8
Northern enlargement (1973) 32.152 10. 76 6.786 24.617 5.51 3.899
Southern enlargement (1981 and 1986) 6.133 4.288 3.164 1.133 0.633 −0.145
Southern enlargement (1986) 19. 078 15. 099 10.541 7.846 8.022 5.51
Northern enlargement (1995) 4.915 4.244 2.491 6.843 6.661 5.541
Eastern enlargement (1998 anticipation) 14 .80 3 14 .80 3 9.858 13 .215 13.215 9.717
Note: For each EU country , the % Effect is given by the percentage difference between the average real GDP per capita (or real GDP per worker, i.e. labor productivity) of the actual country in the post-accession
period (or first ten, or first five years from the accession) and the average real GDP per capita (or labor productivity) of the synthetic country in the same period. For eastern countries we consider the 19 98 as
the beginning of the post-accession period.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

10 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
43.2% Korea Rep. 22.7% Japan
21.8% Thailand 12.2% Albania
Synthetic Czech Republic
75.1% Croatia 18.1% Turkey
6.7% Colombia
Synthetic Estonia
25% Mexico 20.9% Canada
16.8% Ukraine 12.9% Colombia
8.7% Philippines 15.7% Others (19)
Synthetic Hungary
30.1% Colombia 27.1% Croatia
17.3% Morocco 15.4% Mexico
10% Turkey
Synthetic Latvia
80.4% Turkey 14.9% Russia
4.6% Ukraine
Synthetic Lithuania
54.5% Croatia 17.6% Malaysia
7.8% Colombia 5.2% Korea Rep.
3.3% Uruguay 11.5% Others (20)
Synthetic Poland
62.4% Croatia 37.6% Korea Rep.
Synthetic Slovak Republic
43.4% Korea Rep. 24.5% Chile
20.9% Canada 11.1% Colombia
0.1% Thailand
Synthetic Slovenia
Fig. 4. Composition of the synthetic country –GDP per capita, eastern enlargement (1998 anticipation).
Note: In each graph, the pie chart reports the six donor countries (and the relative weights) that contribute the most to the construction of the synthetic
country in qu estion. For full details see Table A.3 in Appendix.
synthetic country for the whole post-accession period, for the first ten, and for the first five years after accession to the EU.
For the eastern enlargement, Table 1 considers the results obtained using 19 98 as the de facto accession year.
Focusing on both per capita GDP (columns 1 to 3) and productivity (columns 4 to 6, in Table 1 ), there is little evidence
that the effects of EU accession decrease over time after each enlargement.
18 This high degree of persistence of the effects
of accession may indicate the continuous deepening of the integration process. For instance, countries involved in the 1973
and 1980s enlargements experienced a major deepening of EU integration after their accession thanks to the Single Market.
Using a medical metaphor, one may be worried that the treatment was strengthened after a given period. Of course, we
cannot claim from our evidence whether the deepening of integration through the Single Market contributed to sustain the
early effects or whether it increased the dividends from EU membership. However, it is worth noting that for these countries
we find smaller but substantial effects also at the five and ten-year windows post-accession. Thus, such strengthening of the
treatment does not seem to crucially affect the positive benefits from the membership even before the creation of the Single
Market.
19
Focusing on the more comparable “first ten years after accession,” the 1970s, 198 0s (excluding Greece), and the eastern
enlargement (considering anticipation effects) have similar net benefits. One can identify large heterogeneity of the effects
across countries, with Latvia, Lithuania and Estonia as the countries that have benefited the most and Greece as the one
that has benefited the least (to a lesser extent, the others are Sweden, Finland and the Czech and Slovak Republics).
18 Although in this section we comment results for both GDP per capita and labor productivity, for reasons of space the full set of results for labor
productivity are in the Appendix. See Figures from A.2 to A.7 and Tabl es from A.4 to A.6.
19 Note that recent literature on the effects of the Single Market suggests that such deepening had relatively modest effects on EU GDP per capita
( Mariniello et al., 2015 ).
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 11
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Fig. 5. Actual vs. synthetic country –GDP per capita trends, eastern enlargement (1998 anticipation).
Note: In each graph, the continuous line represents the trend of the real GDP per capita (PPP, 2005 International Dollars) for the actual country, while the
dashed line shows the trend for the same variable for the synthetic country. For each country the analysis ends in 2008. The composition of each synthetic
country is reported in Figure 4 and full details in Tabl e A.3 in Appendix.
6.2. Random donor samples
The other concern we must address is that our estimates could be affected by the specific composition of the donor
sample. If countries in the donor sample were affected by spillover effects, such as trade diversion induced by EU member-
ship on a non-EU trade partner country that is part of the donor sample, this would bias our results upwards. Similarly, if
a country in the donor sample experienced a positive idiosyncratic shock during the years after joining the EU, this would
bias our results downward.
In order to assess whether the estimation results are influenced by the presence of a specific country in the donor
pool, Abadie et al. (2010 , 2015) suggest excluding each time a country from the counterfactual and compare the estimates
obtained after these exclusions. Building on this idea and taking into account the uncertainty of the goodness of the choice
of the countries composing the donor pool, we propose a new systematic way to check the sensitivity of the SCM results.
20
We construct alternative donor samples and compare the obtained results with our baseline estimates. More precisely,
for each EU country under analysis, we iteratively re-estimate the synthetic counterfactual using one thousand alternative
donor samples. Each donor sample includes the same number of countries used for our main estimation, randomly drawn
from the largest set of countries for which we have available data. Therefore, each alternative donor sample has a (randomly
assigned) probability of being affected by idiosyncratic shocks, which would lead to spurious results. If (i) most of the
interval of estimates obtained with alternative donor samples is systematically different from zero, (ii) a very large share of
alternative estimates indicates effects that are of the same sign of the baseline estimate, and (iii) the baseline estimate is
not extreme with respect to these alternative estimates, then we can attach more confidence to the estimates obtained with
the preferred donor sample presented above.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

12 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Tabl e 2
GDP per capita effects after 10 years from the accession using 10 0 0 random donor samples.
(1) (2) (3) (4) (5) (6)
% Effect (our main
estimation)
Median % effect across 1,0 0 0
random samples
Average % effect across 1, 0 0 0
random samples
% of estimations with negative
effects (out of 10 0 0 random
samples)
% of estimations with positive
effects (out of 10 0 0 random
samples)
% effect using the best
pre-accession fit
Denmark 14. 298 −8.008 −4.407 78.5 21.5 −3.067
United Kingdom 8.586 0.922 0.342 41 59 2.041
Ireland 9.395 1.0 12 1.958 44.3 55.7 4.237
Greece −17.3 36 −13.655 −11.8 21 94.2 5.8 −16.283
Spain 13.662 13.695 14.4 00 0.1 99.9 13.696
Portugal 16.537 19. 318 20.518 0 100 18 .2 64
Austria 6.364 2.548 3.700 38.6 61.4 3.547
Finland 4.017 6.630 7.899 5.1 94.9 12. 497
Sweden 2.353 −0.380 2.461 53.6 46.4 4.472
Czech Republic 5.615 1.169 −1.3 58 41.9 58.1 2.515
Estonia 24.153 30.563 29.966 0.1 99.9 21.423
Hungary 12.299 15.465 15.293 0.1 99.9 16. 411
Latvia 31.692 30.86 6 31.489 0 100 26.259
Lithuania 28.082 27.022 24.979 0 10 0 28.082
Poland 5.930 8.085 7.558 7.5 92.5 2.432
Slovak Republic 0.302 6.642 7. 30 2 3.7 96.3 0.302
Slovenia 10.3 50 12 .591 12.4 06 5.3 94.7 16 .05 7
Note: For each EU country , the % Effect is given by the percentage difference between the average real GDP per capita of the actual country in the first ten years from the accession and the average real GDP per
capita of the synthetic country in the same period. For eastern countries we consider the period 1998–2008.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 13
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Fig. 6. Random donor samples (1,0 0 0 replications) –GDP per capita, northern and southern enlargements.
Note: In each graph, on the y-axis is represented the difference between the real GDP per capita of the actual country in question and its synthetic
counterfactual. The black line represents this difference for our main estimation reported in Figure 2 for northern and southern enlargements. The grey
lines represent this difference for the estimations obtained using 1, 0 0 0 alternative, and randomly chosen, donor samples. Each alternative donor sample
includes the same number of countries than the donor sample used for the main estimation.
Figs. 6 and 7 display these results for GDP per capita, while Table 2 compares our baseline estimated effects from EU
accession with those obtained with the random donor samples in the first ten years of membership.
21 In this exercise, we
consider again for the eastern countries the results obtained using the 1998 as the accession year. In column 1 of Table 2 we
report our main estimated effects. Columns 2 and 3 show the median and the mean, respectively, of the estimated effects
obtained with the one thousand alternative donor samples. Column 4 and 5 show the percentages of the estimations for
the alternative donor samples with a negative or positive (respectively) sign of the effects.
Despite a few interesting differences, results broadly confirm the baseline estimations. For five countries (Denmark, Ire-
land, United Kingdom, Austria, and Czech Republic) our baseline estimates clearly overestimate the effects, while for four
countries (Finland, Estonia, Poland and Slovakia) our baseline estimates are clearly lower than the median or average effect
obtained with the alternative donor samples. For all countries (except for Denmark and Sweden) most random donor sam-
ples estimates have the same sign as our main estimated effects. Overall, the average across countries of the mean (median)
effects obtained with the alternative donor samples indicates that these countries would have had an income per capita 10%
(9%) lower in the absence of EU membership after ten years from the accession. This value is similar to the average across
countries of our baseline estimates for the same period (which is also 10 %) .
Results are even more in line with our baseline estimates if we concentrate on the random estimates having the best
pre-treatment fit, which are more comparable with our baseline estimates (column 6).
20 In other words, we want also to test whether the specific choice we made to build our donor sample drives our results. In the Appendix, we provide
further discussion about the complementarity between the placebo tests and our approach.
21 Figures 6 and 7 display, for each country, the differences in GDP per capita between the actual and the synthetic country for our main estimation
and for the estimations obtained with the random samples that have comparable pre-accession matches (i.e., lower than 3 times the root mean squared
prediction error
of our main estimation).
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

14 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Fig. 7. Random donor samples (10 0 0 replications) –GDP per capita, eastern enlargement (1998 anticipation).
Note: In each graph, on the y-axis is represented the difference between the real GDP per capita of the actual country in question and its synthetic
counterfactual. The black line represents this difference for our main estimation reported in Figure 5 for the eastern enlargement. The grey lines represent
this difference for the estimations obtained using 1, 0 0 0 alternative, and randomly chosen, donor samples. Each alternative donor sample includes the same
number of countries than the donor sample used for the main estimation.
For most countries, the first ten years after accession seem to generate clear, robust, positive net benefits in terms of
higher per capita income levels and higher levels of labor productivity.
22
In summary, our results strongly support the crucial role of case studies, as the effects of integration are highly hetero-
geneous both across countries and over time. Nevertheless, the approach allows us to infer an average effect by averaging
the country-level gains from EU integration.
23
7. Correlates of the benefits from EU membership
Why do some countries benefit a lot while others benefit relatively little from joining the EU? Has the introduction
of the common currency (the Euro) and the extensive preparations that preceded it, affected the growth payoffs from EU
membership? To answer these questions and shed some light on the variation across countries and over time of the benefits
from EU membership that we estimated above, it is worth using a more systematic approach.
In addition to the traditional channel of trade integration, we focus on the relative roles of institutional quality, financial
development, and financial globalization. More financially developed countries are expected to be better able to exploit
(and distribute) the benefits of integration. This is a complex relationship that may depend on the level of development
22 The ten-year interval is clearly arbitrary, as we cannot a priori identify the relevant horizon for long-run effects of accession to the EU. However, the
only main difference of the effects over the entire post-entry period, with respect to the ten-year horizon, is that Ireland now displays ve ry large positive
effects, which are determined by large gains occurring in the 1990s and especially the 20 0 0s. See Ta ble A.9 in Appendix for the comparison of our baseline
estimated effects with those obtained with the random donor samples for GDP per capita in the whole post-accession period, and Table s A.10 and A.11 for
the comparisons for productivity.
23 We also tested whether the average post-accession difference between the actual and synthetic series are statistically different. Considering the lim-
itations due to the small number of observations, results in Tab le A.12 in Appendix show that, for the whole post-accession period, the differences are
statistically different from zero in most
of the countries and in every enlargement.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 15
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
Tabl e 3
Determinants of the growth dividends from EU membership.
VARIABLES (1) (2) (3) (4) (5) (6)
Lag percentage gap 0.887
∗∗∗ 0.877
∗∗∗ 0.861
∗∗∗ 0.880
∗∗∗ 0.847
∗∗∗ 0.856
∗∗∗
(0.035) (0.034) (0.042) (0.036) (0.046) (0.047)
Trade openness 0.164
∗∗∗ 0.143
∗∗∗ 0.153
∗∗∗ 0.142
∗∗∗ 0.155
∗∗∗ 0.134
∗∗∗
(0.031) (0.026) (0.028) (0.027) (0.029) (0.032)
Financial int. −0.0 01 0.012
∗∗∗ 0.012
∗∗∗ 0.012
∗∗∗ 0.012
∗∗∗ 0.012
∗∗
(0.002) (0.005) (0.004) (0.005) (0.005) (0.005)
Financial int. (sq) −0.0 0 0
∗∗∗ −0.0 0 0
∗∗∗ −0.0 0 0
∗∗∗ −0.0 0 0
∗∗∗ −0.0 0 0
∗∗∗
(0.0 0 0) (0.0 0 0) (0.0 0 0) (0.0 0 0) (0.0 0 0)
Euro 0.014
∗0.014
∗0.011 0.014
∗0.013
∗0.026
∗∗∗
(0.008) (0.007) (0.008) (0.008) (0.008) (0.010)
EPL −0.004 −0.003 −0.078
∗∗∗
(0.007) (0.007) (0.025)
EPL (sq) 0.016
∗∗∗
(0.005)
ETCR 0.013
∗∗ 0.015
∗∗ 0.022
∗
(0.005) (0.006) (0.011)
ETCR (sq) −0.0 0 0
(0.002)
Polity2 0.003 −0.007 −0.687
∗
(0.004) (0.007) (0.376)
Polity2 (sq) 0.037
∗
(0.021)
POLCON 0.007 −0.007 −0.063
(0.027) (0.034) (0.256)
POLCON (sq.) 0.045
(0.321)
Yea r of memb. 0.002
∗∗∗ 0.003
∗∗∗ 0.004
∗∗∗ 0.003
∗∗∗ 0.004
∗∗∗ 0.003
∗∗
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Country dummies Yes Yes Ye s Ye s Ye s Yes
Yea r dummies Yes Yes Yes Yes Yes Ye s
Observations 295 295 239 295 239 239
R-squared 0.986 0.987 0.991 0.987 0.991 0.992
Note: OLS estimates with robust standard errors in parentheses. Inference:
∗∗∗ p < 0.01,
∗∗ p < 0.05,
∗
p < 0.1. The dependent variable is the percentage difference between the actual and the synthetic se-
ries of real GDP per capita for each country and each yea r post accession for countries that joined in the
northern and southern enlargements and each year after 1998 for the eastern European countries. The
covariates are: Lag Percentage gap : the (1-year) lag of the dependent variable; Trade openness is open-
ness at 2005 constant prices from Penn Wor ld Tables. Financial int. : an indicator of financial integration
computed as the sum between total assets and total liabilities over GDP (source: Lane and
Milesi-Ferretti,
2007); Euro : a dummy var iable that takes value 1 if the country has joined the Euro area, the value 0
otherwise; EPL : an indicator of employment protection legislation (source: OECD; missing va lues were
interpolated using data from Allard, 2005 ); ETCR : an indicator of regulation in non-manufacturing sectors
(source: OECD; missing value s for 1973, 19 74 and 20 08); Po lity2 from the Polity IV project is a measure
of a country’s political regime; POLCON (political constraints) is a measure for “the feasibility of policy
change (the extent to which a change in the preferences of any one actor may
lead to a change in gov-
ernment policy)” (POLCON_2005 codebook); Yea r of memb. is a count variable that indicates the years the
country has been member of EU for countries in the northern and southern enlargements, and the years
post-1998 for the countries in the eastern enlargement. In each model, country and year fixed effects
are included. Note that the number of observations change because both EPL and ETCR are missing for
non-OECD countries or because we do not have information for some countries.
achieved by domestic political institutions (
Campos and Coricelli, 2012 ). By the same token, this reasoning should hold for
those countries that are better integrated internationally through, for example, foreign direct investment and cross-border
banking.
Table 3 presents a set of panel OLS estimates in which the dependent variable is the “EU membership-induced gap”,
that is the yearly percentage difference between the actual level of per capita GDP and that estimated from the synthetic
counterfactuals for the 17 countries we analyze after they joined the EU (for eastern countries we consider the anticipation
effects from 199 8) . The estimated model specifications include inertia (“lagged gap”) and allow an evaluation of various
different potential determinants: trade openness, international financial integration, adoption of the common currency (a
dummy variable for the adoption of the Euro) and economic and political institutions. Further, two key structural reforms are
captured by measures of labor market flexibility (EPL, employment protection legislation) and economic regulation (ECTR,
competition regulation in utilities industries).
24 The two reported measures of political institutions are a general index of
24 ETCR is the measure constructed by the OECD summarizing indicators of regulation in energy, transport and communications.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

16 N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
democracy (from Polity IV) and an index of political constraints on the executive (POLCON).
25 All specifications include the
number of years of EU membership and country and year fixed-effects.
The results in Table 3 suggest that three main factors are closely associated with the magnitude of the net benefits from
membership in the EU across countries and over time: trade openness, financial integration and the adoption of the Euro.
It should be clear from this exercise that we are highlighting association and not a causal relationship. With this in
mind, the coefficient for Euro membership suggests that countries that (later on) adopted the Euro have pay-offs from EU
membership (i.e., percentage differences between actual and synthetic levels of per capita GDP) that are approximately 2
percentage points larger, on average, than for those countries that have not yet adopted the Euro (recall the average payoff
is about 10%). Similar statements apply to both trade openness and financial integration.
26
A second set of results refers to employment protection legislation and utilities regulation. As it can be seen from Table 3 ,
the effects of EPL are ambiguous. Yet the results for the stringency of product market regulation (ECTR) suggest that coun-
tries that have successfully converged to the EU policy framework seem to benefit more from EU membership. It should be
noted that the source of these two reform variables is the OECD and that data are available exclusively for OECD members
during the period of analysis. The fact that various eastern European countries that joined the EU are not OECD mem-
bers explains the discrepancy between the number of observations of the first two columns and the remainder of Table 3 .
Thus, we consider the EPL and ECTR results in column 6 useful mainly for checking for possible non-linearities and to as-
sess whether the fullest specification would affect the results for what we consider the three key factors (namely, trade
openness, financial integration and the Euro). We find that controlling for EPL and ECTR does not qualitatively affect these
conclusions.
Regarding political institutions, none of the relevant coefficients are statistically significant at conventional levels (except
for democracy, Polity 2, in the full specification of column 6, but this may be capturing unduly the effects of the smaller
sample size). Perhaps, this is because after accession there is little variation among EU members regarding levels of devel-
opment of political institutions and thus we should not expect it to be a key factor in explaining cross-country variation.
Nevertheless, we believe a fruitful avenue for future research would be to extend the set of political institutions and to
investigate further their pre- and post-accession dynamics and how they may affect differently the pace and magnitude of
the estimated net benefits.
8. Conclusions
In this paper, we attempted to provide a novel and more satisfactory answer to the question of whether there are signif-
icant and substantial net benefits from deep economic integration in terms of higher per capita GDP and labor productivity
using the European experience as a case study. The main finding is that of strong evidence for positive net benefits from
EU membership, despite considerable heterogeneity across countries. More specifically, focusing on the 1973, 1980s, 1995
and 2004 enlargements, we find that per capita GDP and labor productivity increase with EU membership in Ireland, United
Kingdom, Portugal, Spain, Austria, Estonia, Hungary, Latvia, Slovenia and Lithuania. The effects tend to be smaller, albeit still
mostly positive, for Finland, Sweden, Poland, Czech Republic and Slovakia. Finally, our evidence shows that only one country
(Greece) experienced lower per capita GDP and labor productivity after EU accession than its counterfactual.
We identify three main directions for further research. First, we think research is needed to provide a fuller understand-
ing of why Greece turned out to have such an exceptionally negative economic performance since EU accession. Second,
further research should focus on the specific mechanisms and channels through which EU membership seems able to sup-
port faster GDP and productivity growth rates, as these mechanisms, and their effectiveness, may change over time and
particularly after the Great Recession. Above we document that trade openness, financial integration and the adoption of
the Euro are important factors in driving these benefits so future research should investigate the inter-relationships among
these factors as well as how they change over time. Finally, future research should focus on disentangling the various aspects
of the integration process, including the political economy dimension. Future research should focus not only on economic
and institutional integration but also on the political support for European integration, which ultimately affect reform poli-
cies in the EU and in its member states.
Acknowledgments
We would like to thank Thomas Bassetti, Laszlo Bruszt, Willem Buiter, Youssef Cassis, Efrem Castelnuovo, Nicholas Crafts,
Saul Estrin, Davide Fiaschi, Nikolaos Georgantzís, Seppo Honkapohja, Iikka Korhonen, Tommaso Nannicini, Jeffrey Nugent,
Jan Svejnar, Paola Valbonesi, and seminar participants at the University College London, INSEAD Fontainebleau, University
of Padova, University of Pisa, Central Bank of Finland, European Investment Bank, Annual Dubrovnik Economics Conference,
Meetings of the American Economic Association, European Economic Association, ISNIE, French Economic Association, Ital-
ian Economic Association, and Royal Economic Society, the editor (Urban Jermann) and an anonymous referee for valuable
comments on previous versions. Previous versions of this paper circulated with the title: “Economic growth and political
25 POLCON is described in detail in Henisz (20 0 0) and the source for the democracy variable is the Polity IV dataset.
26 Adding the linear and the squared term yields an average positive effect of financial integration.
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001

N.F. Campos et al. / Journal of Monetary Economics 0 0 0 (2018) 1–17 17
ARTICLE IN PRESS
JID: MONEC [m3Gsc; October 10, 2018;11:41 ]
integration: estimating the benefits from membership in the European Union using the synthetic counterfactuals method”.
All remaining errors are our own.
Supplementary materials
Supplementary material associated with this article can be found, in the online version, at doi: 10.1016/j.jmoneco.2018.08.
001 .
References
Abadie, A . , Diamond, A . , Hainmueller, J. , 2010. Synthetic control methods for comparative case studies: Estimating the effect of California’s tobacco control
program. J. Am. Stat. Assoc. 105 , 493–505 .
Abadie, A. , Diamond, A. , Hainmueller, J. , 2015. Comparative politics and the synthetic control method. Am. J. Polit. Sci. 59 (2), 495–510 .
Abadie, A. , Gardeazabal, J. , 2003. The economic costs of conflict: a case study of the Basque country. Am. Econ. Rev. 93, 113–132 .
Allard, G., 2005. Measuring job security over time: In search of a historical Indicator for EPL. Working
Papers Economia wp05-17, Instituto de Empresa.
Arkolakis, C. , Doxiadis, A. , Galenianos, M. , 2017. The challenge of trade adjustment in Greece. In: Meghir, C., Pissarides, C., Vayanos, D., Vettas, N. (Eds.),
Beyond Austerity: Reforming the Greek Economy. MIT Press .
Athey, S. , Imbens, G. , 2017. The state of applied econometrics: causality and policy evaluation. J. Econ. Perspect. 31 (2), 3–32 .
Bache, I. , George, S. , Bulmer, S. , 2011. Politics in the European Union. Oxford Un iversity Press .
Badinger, H. , 2005. Growth effects of economic integration: Evidence from the EU member states. Rev. Worl d
Econ. 141, 50–78 .
Baldwin, R.E., Seghezza, E., 1996. Growth and European Integration: To wards an Empirical Assessment. CEPR Discussion Paper No. 1393.
Ben-David, D. , 1993. Equalizing exchange: Trade liberalization and income convergence. Q. J. Econ. 108, 653–679 .
Ben-David, D. , 1996. Trade and convergence among countries. J. Int. Econ. 40, 279–298 .
Billmeier, A. , Nannicini, T. , 2013. Assessing economic liberalization episodes: a synthetic control approach. Rev. Econ. Stat. 95 (3), 983–1001 .
Boltho, A., Eichengreen, B., 2008. The Economic Impact of European Integration. CEPR Discussion Paper No. 6820.
Bove, V. , Elia, L. , Smith, R.P. ,
2017. On the heterogeneous consequences of civil wa r. Oxford Econ. Pap. 69 (3), 550–568 .
Brou, D. , Ruta, M. , 2011. Economic integration, political integration or both? J. Eur. Econ. Assoc. 9 (6), 1143–1167 .
Bruszt, L., Campos, N., 2017. State Capacity and Economic Integration: Evidence from the Eastern Enlargement. EUI RSCAS WP 2017/52.
Campos, N. , Coricelli, F. , 2002. Growth in transition: What we know, what we don’t, and what we should. J. Econ. Lit. XL (3), 793–836 .
Campos, N. , Coricelli, F. , 2012. Financial liberalization and its reversals: economic and political determinants. Econ. Policy 27
(71), 483–513 .
Campos, N., Coricelli, F. , Moretti, L., 2015. Norwegian Rhapsody? The Political Economy Benefits of Regional Integration. CEPR Discussion Paper No. 10653.
Crafts, N. , 2016. West European economic integration since 1950. In: Badinger, H., Nitsch, V. (Eds.), Routledge Handbook of the Economics of European
Integration. Rou tledge, London .
Crespo, J. , Silgoner, M. , Ritzberger-Grünwald, D. , 2008. Growth, convergence and EU membership. Appl. Econ. 40, 643–656 .
Dorrucci, E. , Firop, S. , Fratzscher, M. , Mongelli, F. , 2004. The link between institutional and economic integration: Insights for Latin America from the
European experience. Open Econ.
Rev. 15 , 239–260 .
Egger, P. H. , Larch, M. , 2008. Interdependent preferential trade agreement memberships: an empirical analysis. J. Int. Econ. 76 (2), 384–399 .
Eichengreen, B. , 2007. The European Economy Since 1945: Coordinated Capitalism and Beyond. Princeton University Press, Princeton, NJ .
Henisz, W. , 20 0 0. The institutional environment for economic growth. Econ. Polit. 12 (1), 1–31 .
Henrekson, M. , Torstensson, J. , Torstensson, R. , 199 7. Growth effects of European integration. Eur. Econ. Rev. 41, 1537–1557 .
Hoffman, C. , Osnago, A. , Ruta, M. , 2017. Horizontal depth: A new database on the
content of preferential trade agreements. Policy Research Working Paper
Series 7981. World Bank, Washington, DC .
Imbens, G. , Wooldridge, J. , 2009. Recent developments in the econometrics of program evaluation. J. Econ. Lit. 47 (1), 5–86 .
Kutan, A. , Yigit, T. , 2007. European integration, productivity growth and real convergence. Eur. Econ. Rev. 51, 1370–1395 .
Lawrence, R.Z. , 1996. Regionalism, Multilateralism, and Deeper Integration. Brookings Institution Press .
Levine, R. , Renelt, D. , 1992. A sensitivity analysis of cross-country growth regressions. Am. Econ. Rev. 82 (4), 942–963 .
Mariniello M., Sapir, A., Te rzi , A., 2015. The Long
Road Toward s the European Single Market. Bruegel Workin g Paper 2015/01.
Mattoo, A., Mulabdic, A., Ruta, M., 2017. Trade Creation and Trade Diversion in Deep Agreements. Wo rld Bank Policy Research Work ing Paper No. 8206.
Mitsopoulos, M. , Pelagidis, T. , 2012. Understanding the Crisis in Greece: From Boom to Bust. Palgrave MacMillan, London .
Rivera-Batiz, L. , Romer, P. , 19 91. Economic integration and endogenous growth. Q. J. Econ. 106 (2), 531–555 .
Sapir, A. , 2011. European integration at the crossroads: a review essay on the 50th anniversary of Bela Balassa’s theory of economic integration. J. Econ. Lit.
49 (4), 1200–1229 .
Slaughter, M. , 2001. Trade liberalization and per capita income convergence: a difference-in-differences analysis. J. Int. Econ. 55, 203–228 .
Vamvakidis, A. , 2003. The Convergence Experience of the Greek economy in the EU. IMF. Mimeo, Washington, D.C .
Ventura, J. , 2005. A global view of economic growth. In: Aghion, P., Durlauf, S.N. (Eds.). In: Handbook of Economic Growth, 1B. Elsevier, North Holland,
pp. 1419–1497 .
Please cite this article as: N.F. Campos et al., Institutional integration and economic growth in Europe, Journal of Monetary
Economics (2018), https://doi.org/10.1016/j.jmoneco.2018.08.001