Since the VAR technique is the most commonly used approach to address this key question (as noted earlier), the discussion in this subsection is predominantly based on the literature employing VAR techniques. In fact, we shall begin with an overview of the results from the VAR studies, as summarised in Table 5.1 below.
Table 5.1 Overview of results from the VAR studies
Persistence in Regional Variables Persistence in National Variables Low High Low • United States
Blanchard & Katz (1992) Migration
High
• Europe
Decressin & Fatas (1995) Labour force participation
• Italy
Decressin & Fatas (1995) • Spain
Jimeno & Bentolila (1998)
Note: The term in bold denotes the main regional labour market adjustment mechanism in response to a shock in employment.
Table 5.1 provides a good starting point for the results. It classifies countries along two dimensions – whether they exhibit high persistence in national variables and whether there is persistence of regional relative variables. This classification illustrates that there is a diverse range of experiences in terms of labour market adjustment (at the aggregate and regional levels). The United States falls into the category where there is a low degree of persistence of the aggregate and regional relative unemployment rates (for example). The EU as a whole is a case where persistence of aggregate labour market variables is high but persistence of regional labour market variables is low. Exceptions within Europe include Italy and Spain, where the degrees of persistence at the aggregate and regional levels are both high.
Not only were these countries different in their classification, where there is low persistence in regional variables (the United States and Europe), the main regional labour market adjustment channel is different in the two cases. In particular, in the United States, migration plays a substantial role as a regional labour market adjustment mechanism (Blanchard and Katz (1992)). In other words, adjustment to labour demand shocks appears to occur mainly through migration flows (i.e., laid-off workers leave depressed areas to find jobs elsewhere). Following a state-specific shock, the migration response is strong even in the first year after the shock. For example, if relative state employment falls by 10 workers, in the initial year, unemployment rises by 3 workers, participation falls by 0.5 workers, and 6.5 workers migrate out of the state. In the long run (after 7 to 10 years), employment falls by approximately 13 workers, all of whom have migrated to other states (as cited from Debelle and Vickery (1999)). Blanchard and Katz (1992) also conclude that wages
decrease and dampen the employment response, but by relatively little. This evidence suggests that in the US, wages play a limited role as a regional labour market adjustment mechanism in response to economic shocks. Meanwhile, in European labour markets, labour force participation rate changes play an important role in bringing unemployment back to trend after a region-specific shock, rather than migration (Decressin and Fatas (1995)). In other words, workers leave the labour force rather than migrate out of their region.
Of those countries that were not mentioned in Table 5.1 above, the results are summarised as follows. In Australia, inter-state migration does play an important role in reducing differences in labour market conditions across states, although permanent (or very persistent) differences between state unemployment rates remain (Debelle and Vickery (1999)). Similarly, in the Swedish case, regional adjustment has been comparatively rapid and labour mobility appears to be high by European standards (Fredriksson (1999)).
The natural question to ask is of course where New Zealand fits in. This is the gist of the second key question.
The results from Aynsley (2001) are worth highlighting here although her work focuses on the labour market adjustment at the national rather than regional or state level. An employment demand shock has a larger and more persistent impact on the New Zealand than the Australian labour market. However, international labour mobility (between the two countries) appears to be a more important adjustment mechanism for New Zealand than Australia. The author suggests that the apparent difference in trans- Tasman labour mobility may be due to the different size of the labour markets. Since Australia is larger than New Zealand, there is a greater number of local labour markets within the former, and thus there is likely to be greater opportunities for people to relocate within the country in response to a adverse shock in a local area. In fact, there is evidence that labour mobility is an important inter-state labour market adjustment mechanism in Australia (see Debelle and Vickery (1999)). In contrast, the opportunity for people in New Zealand to relocate within the country is relatively limited. Besides, Aynsley (2001) finds that labour force participation change is also an adjustment mechanism for both countries, but not the real wage.
A recent study by Chapple (2000) examines this question focusing on urban area units (which the author calls neighbourhoods). Although the paper does not use the VAR technique, it is worth noting its findings here for comparison. The author finds that an increase in labour demand (i.e. employment) has a number of effects on the neighbourhood labour market. In particular, this shock reduces the neighbourhood unemployment rate, raises labour force participation, and encourages in-migration. However, the migration response is much weaker than the impact on neighbourhood unemployment and participation.
A further New Zealand study that examines the link between migration and labour market adjustment is Morrison, Papps et al. (2000). They find a positive relationship between migration flows and wages, and emphasise the role of migration in increasing competition between firms for labour inputs, thus reducing monopsony power (especially for more mobile higher skilled workers).
The overview above summarises the relative contributions of each of the regional labour market mechanism, in response to a shock to the system. The general impression is that the significance of migration varies a lot across the countries. However, it is important to examine some quantitative estimates. There are basically
two questions - how much adjustment actually occurs in response to a shock, and given this amount of adjustment, how much adjustment actually takes place through each of the mechanisms. At this point in time, we have not been able to track down comparable estimates for the first question. On the other hand, Fredriksson (1999) provides a summary table that decomposes the response to a shock in employment growth in various countries, as reproduced below.
Table 5.2 Decomposition of the response to a one-standard-deviation shock in
employment growth (percentages of the change in log employment)
Year 1 Year 2 Europe1 (51 regions, 1975-87)
Employment rate 22 23 Participation rate 75 50 Net migration 4 27 United States2 (51 states, 1978-90)
Employment rate 34 23 Participation rate 26 23 Net migration 40 54 Spain3 (17 regions, 1976-94) Employment rate 36 39 Participation rate 23 18 Net migration 41 43 Sweden4 (24 regions, 1966-93)
Regular employment rate 8 5 Participation rate 26 9 Net migration 66 87
Source: Table 1 (p. 636) from Fredriksson (1999) 1
Decressin and Fatas (1995) 2
Blanchard and Katz (1992) 3
Jimeno and Bentolila (1995) and Jimeno and Bentolila (1998) 4
Fredriksson (1999)
Table 5.2 shows for example, that changes in participation rates have been the primary initial response to shocks in employment in Europe. In the first year after a negative shock, 75 percent of the impact is borne by workers dropping out of the labour force. Conversely, migration has been the principal response to job destruction in Sweden. By two years, the employment response consists mainly of out-migration of workers (implying that 87 percent of the affected workers would have left their region).
However, while table 5.2 tells us about the relative contribution of migration as a labour market adjustment mechanism, it does not provide much information as to how much adjustment occurs. For example, it is not clear from table 5.2 that there is high persistence of regional unemployment differentials Spain.
Up to here, we have been focusing on the mix of adjustment mechanisms. Another aspect of the dynamics is the speed of adjustment. When there is an adverse employment shock to the local labour market, how long does it take for the adjustment mechanisms to occur? In the US, net migration plays a substantial role even in the first year of an employment shock. After 5 to 7 years, the employment response consists entirely of worker migration (Blanchard and Katz (1992)). In the EU, it takes about 3
years for the effect on the labour force participation rate and 4 years for the effect on the unemployment rate to disappear (Decressin and Fatas (1995)). Meanwhile, in Australia, most of the migration takes place, on average, within four years. In particular, approximately one-third of the out-migration occurs within two years, roughly two-thirds of the net migration takes place within three years of the shock, and then the rate of out-migration flattens out. The process of adjustment is complete after seven years (Debelle and Vickery (1999)). These speed of adjustment estimates can be seen more clearly from Figure 5.2 below, which presents the impulse response results and a one standard error confidence bounds for net migration from Debelle and Vickery (1999) in response to a 1 percent negative shock to employment.
Figure 5.2 Impulse Response to a 1 percent Negative Employment Shock (in Australia)
Next, we shall examine some of the findings from studies that have used slightly different specifications. The question we are asking is whether these issues alter the results, in terms of the significance of the migration mechanism (i.e. do these specification issues matter in practice?).
Bartik (1991) and Blanchard and Katz (1992) reach different conclusions about local job growth’s long run effects. In the case of Bartik (1991), a one percent shock to employment results in the local employment reaching a new equilibrium (a one percent higher than the pre-shock level) from year one onwards. On the other hand, Blanchard and Katz (1992) find that over time, the effect on employment builds up, and reaches a peak of 2.0 percent after four years and a plateau of about 1.3 percent. Bartik (1993)
re-analyses the data set used by Blanchard and Katz (1992) to examine why this difference exists. One of the reasons put forth is that Bartik (1991) considers a once- and-for-all shock to local job growth, with subsequent growth unchanged from what it would have been (i.e. a one-time growth shock). In contrast, Blanchard and Katz (1992) allow the one-time shock to local job growth to affect subsequent growth (i.e. a shock with readjustment). As noted by Bartik (1993), both are interesting thought experiments, and one may be more appropriate than the other in some circumstances. Most VAR studies have looked at the regional labour market adjustment process in relation to the whole labour force. In contrast, Mauro and Spilimbergo (1998) and Mauro, Prasad et al. (1999) examine the same question, but focusing on the different subgroups within the population with varying levels of skills in the Spanish context. Mauro, Prasad et al. (1999) finds evidence that suggests that the high-skilled in Spain are indeed more likely to migrate than remaining unemployed or dropping out of the labour force, compared to the low-skilled. In other words, high-skilled workers migrate very promptly in response to a decline in local labour demand while low-skilled workers drop put of the labour force or stay unemployed for a long time.
There is some empirical evidence that supports the idea that active labour market programs do prevent workers from leaving the depressed region (Edin et al. (1991) and Westerlund (1995) in Fredriksson (1999)). However, using VAR techniques, Fredriksson (1999) found little evidence for this. Active labour market programs in general do not seem to have impeded regional adjustment substantially.