Drowned by numbers? Stabilization properties of an EU-wide unemployment insurance system

ABSTRACT

The recent crisis has given rise to proposals for the creation of a European unemployment insurance system. We simulate an EU-wide mechanism under various scenarios, varying methods of financing (common or country-specific contribution rates) and triggers for pay-outs (all time or contingent clauses). We analyse the impact of the system using different measures of stabilization under different fiscal multipliers. A system operating during bad times (periods where the increase in unemployment is large) would reduce GDP growth variability but also growth correlation among member countries. Hence, there is a trade-off between stabilization and synchronization of national business cycles.

KEYWORDS:

• Intergovernmental transfers
• fiscal union
• monetary union
• European integration
• unemployment insurance

JEL CLASSIFICATION:

• F45
• F36
• H77
• H87
• E60

Acknowledgements

We thank the editor and referee of the journal, as well as Christophe Blot, Marcel Boyer, Gérard Cornilleau, Jérôme Creel, Gregory Claeys, Peter Claeys, Matthias Dolls, Kodjovi Ekelou, Gilles Le Garrec, Timothy Goodspeed, Eric Hayer, Hiraoki Ino, Marcelin Joanis, Richard Lewney, Norimichi Matsueda, Claude Montmarquette, Massimo Morelli, Claire Naiditch, Mathieu Plane, Vincent Touzé, Lukas Vogel and participants in conferences (3rd EURINT International Conference in Iasi, Macroeconomics Workshop “The Euro Crisis: Where Do We Stand?” in Strasbourg; ZEW Public Finance Conference in Mannheim; 10th Inha-Le Havre International Conference in Seoul) and seminars (in CIRANO – Montréal, La Sapienza University – Roma, Kansei Gakuin University, OFCE – Paris, and SEPOL – Lille) for useful comments and remarks. The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ A review is made in Fatás (1998). For an appraisal of the OCA literature considering the EMU experience, see Dellas and Tavlas (2009).

² Imperfect information at the centralized level of government about the actual realization of local shocks might make common insurance not optimal (Lockwood 1999). There might be over-insurance due to a common pool problem (Sanguinetti and Tommasi 2004). Centralized insurance is not optimal either if preferences of voters across countries are heterogenous or if regional shocks cause uncertainty on the tax rate (Alesina and Perotti, 1998).

³ An exception is Beblavý and Maselli (2014) who also consider the EU.

⁴ The particularity of the UK has been reinforced with the issue of Brexit, following the June 2016 referendum. We also ran simulations without both countries.

⁵ For the computations of net transfers, our formulas are similar to those presented in Lellouche and Sode (2014), though we compute the stabilization impact (which they do not study).

⁶ Heterogeneity concerns the nature of funding as well. For instance, in some countries, unemployment insurance is not funded by a specific, dedicated social security contribution (SSC) rate, but via overall SSC (Ireland), income tax (Luxembourg), or lump sum contribution (for employees in Denmark). See European Commission (2015).

⁷ Our discussion here encompasses the following studies: Beblavý and Maselli (2014), Brandolini, Carta, and D’Amuri (2016), Dolls et al. (2018), Dullien (2013), Dullien and Schwarzer (2009), Fichtner and Haan (2014) and Lellouche and Sode (2014).

⁸ There are actually optional state-specific trigger thresholds. Since the Great Recession, a series of new benefit extensions has sparked interest in the analysis of their impact on unemployment. We review these developments below.

⁹ In Dolls et al. (2018), the contribution rate is updated every 3-year and must balance the financial position every year or over the whole simulation period.

¹⁰ They do not precise the value of the propensity to consume.

¹¹ On microeconomic grounds, this can be justified by the larger marginal propensity to consume of unemployed people, see McKay and Reis (2016).

¹² Detailed results for each scenario (in € million and as a percentage of GDP for each year and each country) are displayed in the Appendix. For the baseline scenario, see Table A1.

¹³ The first term of the trigger is a level of unemployment above 5%. In Europe, most countries qualified for this criterion during the crisis (bar the Netherlands, Luxembourg and Austria).

¹⁴ In the literature, the stabilization impact ranges between 4% (Brandolini, Carta, and D’Amuri 2016) and 36% (Dolls et al. 2018).

¹⁵ We also computed the effects on synchronization of business cycles under higher fiscal multipliers during recessions and under low and high multipliers. The decrease in GDP growth correlation is even reinforced. Results are available upon request.

¹⁶ Fichtner and Haan (2014) make this assumption explicitly.

¹⁷ Besides, job seekers find a job according to their search effort, which is not observable. This relationship is assumed to be in the form of an exponential distribution in Hopenhayn and Nicolini (1997) or a Poisson process in Coles (2008) for example. The empirically appropriate form in the design of a European insurance fund is an unanswered question.