Assessing the scoreboard of the EU macroeconomic imbalances procedure - (machine) learning from decisions

Aug 5, 2026

This paper applies machine learning to determine which macroeconomic variables best explain how countries are classified under the EU Macroeconomic Imbalances Procedure (MIP). A random forest model uses the 14 headline indicators from the MIP scoreboard together with historical European Commission decisions. The analysis highlights unemployment, current account balance, private sector debt, and net international investment position as key drivers. The study further interprets how high versus low variable values contribute to assigning countries inside or outside each MIP category.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

The Macroeconomic Imbalances Procedure

Volume 42, Issue 4 Assessing the scoreboard of the EU macroeconomic imbalances procedure: (machine) learning from decisions Tiago Alves ISCTE-IUL João Amador Banco de Portugal and Nova School of Business and Economics Francisco Gonçalves Nova School of Business and Economics Abstract This paper uses machine learning methods to identify the macroeconomic variables that are most relevant for the classification of countries along the categories of the EU Macroeconomic Imbalances Procedure (MIP) . The random forest algorithm considers the 14 headline indicators of the MIP scoreboard and the set of past decisions taken by the European Commission when classifying countries along the MIP categories. The algorithm identifies the unemployment rate, the current account balance, the private sector debt and the net international investment position as key variables in the classification process. We explain how high vs low values for these variables contribute to classifying countries inside or outside each MIP category. The opinions expressed in this article are those of the authors and do not necessarily coincide with those of Banco de Portugal or the Eurosystem. The authors thank António R. dos Santos, two anonymous referees and the Editor for useful comments and suggestions. Code and data are available at [https://gitlab.com/alves.tiago/scoreboard. Any remaining errors and omissions are the sole responsibility of the authors](https://gitlab.com/alves.tiago/scoreboard. Any remaining errors and omissions are the sole responsibility of the authors). Citation: Tiago Alves and João Amador and Francisco Gonçalves,(2022) ''Assessing the scoreboard of the EU macroeconomic imbalances procedure: (machine) learning from decisions'', Economics Bulletin, Volume 42, Issue 4, pages 2257-2266 [Contact:](Contact: Tiago Alves-alves.tiago@posteo.net)[ Tiago Alves-alves.tiago@posteo.net](Contact: Tiago Alves-alves.tiago@posteo.net), Joã[o Amador-jamador@bportugal.pt](o Amador-jamador@bportugal.pt), Francisco Gonçalves [francisco.goncalves@posteo.co.uk](francisco.goncalves@posteo.co.uk). Submitted: June 22, 2021. Published: December 30, 2022. 1 . Introduction The euro area sovereign debt crisis of 2010-2011 demonstrated the need for stronger economic governance and enhanced policy coordination between EU Member-states to avoid the accumulation of serious imbalances, with an impact on overall macroeconomic stability. Until the sovereign debt crisis, di􀀋erent economic policy coordination procedures were implemented without articulation. Later, Member-states were asked to synchronize their timetables and existing processes were streamlined in order to align national 􀀌scal, growth and employment policies. This materialized into the designated “European Semester”. Under this setup, the supervision and coordination of macroeconomic policies was expanded to include external imbalances, as well as labour market and credit developments. The designated “Macroeconomic Imbalances Procedure” (MIP) operationalized this inherently complex process (European Commission 2016) . Overall, the main goal of the European Commission is to classify countries in terms of the seriousness of macroeconomic imbalances, using a set of common macroeconomic variables. Although initiatives to revisit the economic governance of the EU were launched recently, notably in the context of the response to the COVID 19 pandemics, the MIP remains a key surveillance tool (European Commission 2020, 2021) . Random forest algorithms are one of the most powerful machine learning methods and are suited for exercises of classi􀀌cation of observations into categories. Recent methods make it possible to identify the most important variables to the MIP classi􀀌cation, while also accounting for their positive or negative impact on the classi􀀌cation depending on the values they assume. Nevertheless, the Commission’s MIP classi􀀌cation for each country in a speci􀀌c year cannot b

FAQs

What is the main goal of the paper?

To identify which macroeconomic variables are most relevant for classifying countries under the EU Macroeconomic Imbalances Procedure (MIP).

Which model and inputs are used for classification?

A random forest algorithm uses the 14 MIP headline indicators and historical decisions made by the European Commission when classifying countries by MIP categories.

Which variables are found to be most important in the classification?

The unemployment rate, the current account balance, private sector debt, and the net international investment position are identified as key variables.

Aug 5, 2026

Upload by Evangeline