In his long career, Robert Mundell made a number of influential contributions to international macroeconomics and beyond, elegantly summarised by his fellow Nobel laureate Paul Krugman (Krugman 2021). Mundell (1961), in particular, puts forward a conceptual framework to assess optimum currency areas. In this framework, the notion of labour mobility takes centre stage. Migration across the countries or regions that make up the currency union can serve as a powerful adjustment mechanism which limits the adverse employment effects of asymmetric shocks if a common currency prevents a fast exchange-rate adjustments. And, indeed, prior to the inception of the euro, many economists – though not Mundell himself – warned that a currency union including all members of the EU would face severe problems, exactly because it lacked sufficient labour mobility (e.g. Feldstein 1997).
The ongoing discussion about the functioning of the euro area, however, largely ignores this issue. Large shocks such as the economic fallout from the Covid-19 pandemic or the global crisis did not hurt all member countries of the euro area alike. This fuelled an intensive debate about devising appropriate risk-sharing mechanisms for the euro area and triggered new policy initiatives (Cecchetti and Schoenholtz 2020, Giovannini et al. 2021). When it comes to risk sharing in the euro area, the focus has traditionally been on financial market regulation, the extent of fiscal integration, and not least the role of the ECB (Bénassy-Quéré et al. 2020). Migration across the euro area is rarely discussed.
Migration as a distinct channel of risk sharing
In a new paper, we address this shortcoming and study – very much in the spirit of Mundell – migration as a channel of risk sharing in currency unions (Kohler et al. 2021). We do so as we extend the by now classic framework of Asdrubali et al. (1996) to account for migration as a distinct channel of risk sharing. The framework as originally introduced relies on a simple variance decomposition of output per capita that allows measuring the extent of consumption smoothing in a regression framework. Intuitively, without any consumption smoothing, changes in consumption per capita are perfectly ‘explained’ by changes in output per capita. This corresponds to the limiting case of no risk sharing. Conversely, with perfect consumption smoothing, changes in output and changes in consumption are disconnected; consumption is perfectly insulated from output shocks; risk sharing is complete. Allowing for intermediate cases, the approach offers a straightforward way to quantify the fraction of output fluctuations that is smoothed via different channels such as the factor-trade, the transfer, and the credit channel. The approach has been frequently applied in various contexts and a robust result of earlier work is that risk sharing is considerably lower across the member states of the euro area than across US states (for recent estimates, see Cimadomo et al. 2020).
We formally extend the original framework in order to measure the contribution of migration to risk sharing. The intuition is straightforward. Absent any inflow or outflow of labour, a demand or supply shock will affect aggregate output and output per capita by the same percentage amount. Applying the above logic, variations in output per capita are perfectly ‘explained’ by variations in aggregate output. But with labour mobility, an adverse shock is likely to generate an incentive for outward migration, and conversely for positive shocks. To the extent that migration does take place, the shocks will be absorbed with a muted reaction of output per capita relative to that of aggregate output. In our empirical analysis, we thus infer the role of migration in international risk sharing from differential movements of aggregate output and output per capita. The validity of this approach hinges on low natural population changes (births and deaths) relative to changes through net migration. We verify that this condition is met in our data.
Likewise, we stress that in our analysis migration is conceptually different from commuting because migration involves movement of people across countries, rather than people moving their jobs (without changing their residence). Still, commuting also contributes to risk sharing because it permits taking up jobs across the border in the face of deteriorating domestic employment perspectives and generates factor income – as such its contribution is captured by the factor trade channel which is distinct from the migration channel. For completeness, we also note that migration may contribute to risk sharing to the extent that it leads to remittances. These, in turn, are captured by the transfer channel of risk sharing.
Migration makes a sizeable contribution to risk sharing in the US, but none in the euro area
We summarise the results of our analysis in Figure 1. The left panel shows our estimates for risk sharing across US states, the right panel for the euro area. For US states, our sample contains annual observations and runs from 1963 to 2017, for the euro area we use quarterly observations for the period from 1998 to 2018. In our econometric specification we include both time and country fixed effects. In this way we control for aggregate shocks as well as for long-run trends in migration.
Figure 1 Risk sharing channels in the US and the euro area
Notes: The pie charts show the fraction of fluctuations of aggregate output (in percent) that is smoothed by specific risk-sharing channel; ‘unsmoothed’ corresponds to the fraction of aggregate output fluctuations that are reflected in fluctuations of consumption per capita.
The pie charts show the fraction of fluctuations of aggregate output that is smoothed by a specific risk sharing channel. In each panel, the blue piece corresponds to the fraction of aggregate output fluctuations that is reflected in fluctuations of per capita consumption. These ‘unsmoothed’ fluctuations indicate a lack of risk sharing. It is particularly severe for the euro area but less dramatic for the US: 65% versus 20%. Our main result concerns the role of migration as a risk sharing channel (the orange piece): migration makes a sizeable (and statistically significant) contribution to risk sharing in the US. It smooths 8% of output fluctuations across states. In the euro area, in stark contrast, the contribution is 2%, though not significantly different from zero.
Direct evidence on migration rates
Our empirical approach speaks to the question of whether migration contributes to risk sharing without actually using data on migration flows. But migration data do in fact lend additional support to our findings. In Figure 2, we show gross migration rates for US states and the euro area. In each case we compute the average of in-migration and out-migration to/from the remaining regions/countries, both for the US in the left panel and the euro area in the right panel. In each case we look only at migration flows within the currency union. The result is quite strong: migration rates are about 20 times higher in the US, and this ratio has been stable over the last 15 years.
Figure 2 Gross migration rates
a) US states
b) Euro area members
Notes: Gross migration rates for US states and regions and euro area member states (average of in-migration and out-migration from/to rest of US/euro area, in percent of population).
Data sources: American Community Survey (ACS) run by the US Census Bureau and the Labour Force Survey (LFS) of Eurostat.
Against this background our results regarding the importance of migration as a risk-sharing channel are perhaps not surprising. Still, by formally integrating the migration channel into the accounting framework of risk sharing, we are able to quantify its role for risk sharing in a unified framework. As our results for the US show, this role can be sizeable. We think this is an important insight for the policy debate in Europe. After all, the initial concern that the euro area fails to qualify as an optimum currency area for lack labour mobility appears just as legitimate today – in the year of Mundell’s death and 60 years after he put forward the criterion.
Asdrubali, P, B E Sorensen and O Yosha (1996), “Channels of interstate risk sharing: United States 1963-1990”, The Quarterly Journal of Economics 111(4): 1081-1110.
Cecchetti, S and K Schoenholtz (2020), “The euro area in the age of COVID-19”, VoxEU.org, 22 May.
Cimadomo, J, G Ciminelli, O Furtuna and M Giuliodori (2020), “Private and public risk sharing in the Euro Area”, European Economic Review 121, 103347.
Bénassy-Quéré, A, G Corsetti, A Fatas, G Felbermayr, M Fratzscher, C Fuest, F Giavazzi, R Marimon, P Martin, J Pisani-Ferry, L Reichlin, H Rey, M Schularick, J Sudekum, P Teles, N Veron and B Weder di Mauro (2020), “COVID-19 economic crisis: Europe needs more than one instrument”, VoxEU.org, 5 April.
Feldstein, M (1997), “The political economy of the European economic and monetary union: political sources of an economic liability”, Journal of Economic Perspectives 11(4): 23–42.
Giovannini, A, C-W Horn and F P Mongelli (2021), “An early view on euro area risk-sharing during the COVID-19 crisis”, VoxEU.org, 10 January.
Krugman, P (2021), “The Mundell difference”, VoxEU.org, 12 April.
Kohler, W, G Müller and S Wellmann (2021), “Risk Sharing in Currency Unions”, CEPR Discussion Paper 16178.
Mundell, R A (1961), “A theory of optimum currency areas”, The American Economic Review 51(4): 657–665.