Moving closer? Changing patterns of labour mobility in Europe and the US

Mai Dao, Davide Furceri, Prakash Loungani, 1 December 2013

a

A

On 21 September 1992, four famous professors – Olivier Blanchard, Rudi Dornbusch, Stan Fischer, and Paul Krugman – took part in a panel discussion at MIT on the merits of the proposed European currency union. Echoing a view common among US-based academics at the time, all four agreed, according to a record of the event, that “a common European currency would have unfavourable economic repercussions.” Blanchard noted that “currency unification works in the US because labour can move between states. The labour mobility in Europe is negligible.”

How has labour mobility evolved over the subsequent two decades? Our recent research suggests intra-US mobility had been slowing down prior to the Great Recession while intra-European mobility had been picking up, bringing the two currency unions closer in this respect.

Implied mobility

The canonical way of describing regional evolutions is due to Blanchard and Katz (1992). They traced out the dynamic impact on state unemployment and labour force participation of a state employment shock, which was assumed to be a labour demand shock. Their results suggested a very rapid and robust response of inter-state mobility to regional shocks in the US. Extensions of the Blanchard-Katz methodology to Europe (Decressin and Fatas 1995) suggested a much weaker response of mobility, establishing the conventional wisdom that Europe lacked one of the necessary mechanisms for a successful currency union.

Figure 1 updates the Blanchard-Katz empirical exercise with 20 additional years of US data. A 1% shock to state employment raises state unemployment by about 0.2 percentage points and lowers state participation rates by about 0.25 percentage points in the first year. The implied labour mobility is the difference between the size of the employment shock and the sum of changes in unemployment and participation. Put in terms of numbers of people, out of every ten workers who lose jobs in a US state as a result of an adverse shock, two workers become unemployed, two workers drop out of the labour force, and six workers move out of the state.

Compared with the original Blanchard-Katz results, the participation response is much greater and the response of unemployment much smaller. By (divine?) coincidence, these offsetting changes leave the estimate of implied mobility identical to that of their original work.

Figure 1. Response of US states to regional shocks – Blanchard-Katz, 20 years later

Note: the horizontal axis in this figure and subsequent ones measures years.

Declining US mobility

While the update shows broad support for the overall thrust of the Blanchard-Katz findings, a closer look suggests two important qualifications. First, the response of mobility may be much less immediate. Second the response of mobility may be declining over time. We discuss each of these points in turn.

Muted mobility: As noted above, in the Blanchard-Katz methodology the shocks to state employment are assumed to be labour demand shocks. One can test the validity of this assumption by using instrumental variable estimation. Shocks to the industry mix of state employment serve as a good instrument, as discussed extensively in the urban and regional economics literature (Bartik 1991). As shown in the left panel of Figure 2, the instrumental variable estimates suggest that the response of mobility is much less immediate and much less strong in the short run, and also somewhat more muted over the long run. Instead of suggesting that six out of every ten workers who lose their jobs leave the state within the first year, the instrumental variable estimates suggest that only between one and three workers do so. The estimates of implied mobility using the instrumental variable estimation match up well with direct estimates of mobility using inter-state migration data, as shown in the right panel of Figure 2.

Declining mobility: Both implied mobility using the Blanchard-Katz methodology and direct estimates of mobility using migration data point to declining mobility over time in the US. While inter-state migration still spikes up during recessions, during normal times there has been a trend decline in the adjustment through migration. Between 1985 and just prior to the Great Recession, the propensity to migrate within the first year in response to a state labour demand shock decreased by more than half. Instead, workers were much more likely to drop out of the state labour force.

Figure 2. US migration response – more muted in the short run

Note: the left-hand panel compares implied migration results from OLS estimation with IV estimation using the Blanchard-Katz methodology. The right-hand panel compares the estimate of implied mobility using the IV estimation with direct estimates from migration data.

Increasing European mobility?

In Europe, the trend appears to have been in the opposite direction. We replicated the Blanchard-Katz analysis using data for 173 regions in 21 European countries over the period 1998–2009. On average over the whole sample, a 1% adverse shock to regional labour demand raises the unemployment rate in the first year by about 0.15 percentage points and lowers the participation rate by 0.6 percentage points. That is, out of every ten workers that lose employment, one worker becomes unemployed, six drop out of the labour force, and three workers migrate out of the region within the first year following the shock.

However, there is evidence of increasing mobility over time in Europe. When we repeat the analysis by looking at sub-periods, we find that the medium-run response of migration over the 2004–2009 sub-sample has increased when compared with the full sample results, while the role of the participation rate has decreased (Figure 3, left panel). We also find that the response of migration is stronger outside the EU15 countries (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, Sweden, and the UK.) As shown in the right-hand panel of Figure 3, the role of migration as a shock absorber is considerably larger in the case of the EU21 than when we restrict the sample to the EU15. Hence the response of migration to regional labour demand shocks in Central and Eastern European countries has been on average larger than in Western European countries.

Figure 3. Response of migration in European regions

 

Moving closer?

To summarise, prior to the Great Recession, the role of migration as a shock absorber increased in Europe and declined in the US. During the Great Recession, inter-state migration again spiked in US. Hence, while mobility was declining in importance as an adjustment mechanism, it is certainly not dead and does respond in the face of severe regional shocks. Migration also remained high among European regions, though perhaps fell short of what was needed given the magnitude of the differences in inter-regional shocks. Overall, the results suggest that Europe has been moving to strengthen one of the adjustment mechanisms needed for a successful currency union.

References

Bartik, T J (1991), “Who Benefits from State and Local Economic Development Policies”, W E Upjohn Institute for Employment Research

Blanchard, O and L F Katz (1992), “Regional Evolutions”, Brookings Papers on Economic Activity, 1: 1–77.

Dao, M, D Furceri, and P Loungani (2013), “Regional labour market adjustments in the US and Europe”, forthcoming IMF Working Paper.

Decressin, J and A Fatas (1995), “Regional Labour Market Dynamics in Europe”, European Economic Review, 39: 1627–1655.

Topics: Labour markets
Tags: currency union, euro, eurozone, labour mobility

Economist in the Research Department, IMF
Economist in the Research Department, IMF
Senior resource manager and advisor in the IMF's Research Department