Date: 13 June, 2018


Since 2010, the European Semester has been the framework for the coordination of economic policies across the European Union. Through the semester, EU countries discuss their economic reform and budget plans and the European Commission monitors progress at specific times throughout the year. The European Semester aims to ensure sound public finances, prevent and correct excessive macroeconomic imbalances, foster structural reforms and boost jobs, growth and investment. To meet these objectives, every year the European Commission prepares recommendations for each member state on the basis of detailed country-specific analyses. The Council of the EU subsequently modifies – if necessary – and adopts the Commission proposals for these country-specific recommendations (CSRs). National governments are then supposed to implement them. The process has become the key EU mechanism for steering and guiding member states’ fiscal, macroeconomic and structural policymaking.

The European Semester incorporates three separate processes that work in parallel:

  • Fiscal surveillance based on the Stability and Growth Pact (SGP), which the EU tried to strengthen significantly with a number of new regulations that came into force in 2011 and 2013.
  • The so-called Macroeconomic Imbalances Procedure (MIP): The starting point for MIP was the realisation that large macroeconomic imbalances built-up in the euro area in the pre-crisis years and the EU lacked instruments to even monitor such imbalances1. The unwinding of such imbalances was placing a huge strain on the euro area, and the EU therefore put in place the MIP regulation in 2011 ((EU) 1176/2011) to correct imbalances2.
  • Coordination of EU countries’ economic and employment policies, as foreseen in the Treaty on the Functioning of the EU (TFEU) based on the integrated guidelines, a loose form of policy coordination3.

This Policy Contribution evaluates member-state implementation of the country-specific recommendations (CSRs) focussing on macroeconomic imbalances. We find that overall, implementation has been modest but has worsened in recent years. In particular, implementation among countries with excessive imbalances has fallen significantly. Implementation rates vary substantially for different countries and policy areas. Finally, the multi-annual assessment shows that the implementation records of some countries improve when looking at a longer period, but implementation is overall still weak.

We also assess the policy content of CSRs. Recommendations made under the MIP regulation are often irrelevant for macro-financial imbalances and rather relate to the loose policy coordination framework. Moreover, a comparison between the CSRs addressed to countries with macroeconomic imbalances and relevant recommendations given by the International Monetary Fund shows that, while in general the content, number, length, scope and level of detail of recommendations in the two surveillance frameworks are quite similar, CSRs flag problems in the financial sector much less often than the IMF. The EU recommendations therefore appear less pertinent to addressing macroeconomic imbalances than those of the IMF.

Box 1: A new dataset to analyse implementation of CSRs

To analyse trends in implementation of CSRs, we constructed a dataset of implementation scores from the European Commission’s assessments of the implementation of recommendations. There is good reason to believe these assessments are accurate. In a European Court of Auditors survey of the members of the Economic Policy Committee – a group of national officials that supports the work of ECOFIN4 – 89.5 percent of respondents described the Commission’s assessment of implementation as completely or generally accurate (European Court of Auditors, 2018).

We used the only publicly available sources of the Commission’s assessments – the Country Reports5, which are published eight months after initial recommendations are given6. In these reports, the Commission evaluates implementation at the aggregate CSR-level but also, starting with the 2013 recommendations, at the disaggregated or subpart-level. Breaking CSRs up into more specific components offers some clear advantages: evaluating implementation of recommendations at this more granular level becomes easier, more accurate and more informative. To illustrate, CSR 3 given to Spain in 2017 reads: “Ensure adequate and sustained investment in research and innovation and strengthen its governance across government levels. Ensure a thorough and timely implementation of the law on market unity for existing and forthcoming legislation”. Arguably, though bundled together in the text, this CSR consists of two distinct subparts as it recommends two totally different policy measures.

To the best of our knowledge, we are the first to systematically analyse implementation of CSRs at the subpart level outside the European Commission (EC 2016, Bricongne and Turrini 2017)7. Our dataset, therefore, covers the period from 2013 to 2017 and contains the Commission’s implementation score for each subpart, which can be one of the following: ‘no progress’, ‘limited progress’, ‘some progress’, ‘substantial progress’ and ‘fully addressed’ (see the online Annex 1). We transformed these scores into numerical variables of 0 for ‘no progress’, 25 for ‘limited progress’, 50 for ‘some progress’, 75 for ‘substantial progress’ and 100 for ‘fully addressed’, in order to compare average implementation between aggregates.

As a result, the average implementation scores we compute should be interpreted carefully: a mean implementation score of 100 (the maximum) indicates that all subparts addressed have been fully implemented, whereas a score of 0 (minimum) means that there has been no progress on any subpart. However, a score of 50 does not necessarily mean that half of recommendations have been fully implemented and half have not. Rather, it indicates that there has been ‘some progress’ with the recommendations on average.

Furthermore, we assign to each CSR subpart the policy areas that describe its content. As in the internal database of the Commission, we chose up to three policy areas for each subpart from a set of 32 areas in total (see the online Annex 2), so relative frequencies of policy areas sum up to more than 100 percent8. The dataset also contains for each subpart the corresponding legal basis on which the recommendation is made. This is derived from the recitals of the CSR text, which specify whether the CSR is relevant for the MIP.

Finally, SGP-related subparts are not publicly evaluated in the Country Reports, ie starting with the 2014 European Semester, since the fiscal data is only finalised after the publication of the reports. We therefore exclude them from the sample whenever we calculate average implementation scores over time.

1 In particular: high current-account balances, large external debts, large corporate or household debt, excessive credit growth and wage/price developments that significantly diverged from the euro-area average.

2 In cases of non-compliance, the procedure allows – as a last resort – the imposition of sanctions on the non-compliant member state, similarly to the SGP procedure.

3 Coordination is founded on the broad economic policy guidelines and employment guidelines – which together form the so-called Integrated Guidelines – adopted every year by the Council. However, since 2010, the Integrated Guidelines have been revised only once, in 2015. The Integrated Guidelines reflect the Europe 2020 strategy for smart, sustainable and inclusive growth and do not have any sanctions attached for non-compliance. They are thus the loosest form of coordination of national policies.

4 See

5 For the 2013 CSRs, the information is taken from the Commission Staff Working Documents titled ‘Assessment of the national reform programme and convergence /stability programme’.

6 Since 2015, these reports – published every February – have contained assessments of compliance with the previous year’s CSRs. These assessments thus capture progress achieved within roughly eight months following the Council of the EU’s approval of the recommendations. Commission services continue to update their assessments regularly. The European Commission also carries out an assessment each spring (May) and a multi-annual assessment of the CSRs, ie not just in the year after the recommendations are made, but in all subsequent years. However, neither these subsequent assessments nor the database maintained for that purpose are made public.

7 As a rule, CSRs are broken down into subparts in our dataset as in the Country Reports, but in some cases the breakdown into subparts in the Country Reports does not clearly map to the CSR text.

8 A correspondence between subparts and policy areas is part of the Commission database but is not made public through the Country Reports. We are confident after a number of discussions that we closely approximate the Commission’s matching.