Friedberg, Rachel and Jennifer Hunt, 1995, “The Impact of Immigrants on Host Country Wages, Employment, and Growth,” Journal of Economic Perspective, Vol. 9, No. 2.
Gross, Dominique, 1999, “Three Million Foreigners, Three Million Unemployed? Immigration and the French Labor Market,” IMF Working Paper 99/124 (Washington: International Monetary Fund).
New Zealand Department of Labor, 1998, “The integration of Highly Skilled Migrants into the Labor Market: Implications for New Zealand Business.”
Pope, David, and Glenn Withers, 1993, “Do Migrants Rob Jobs? Lessons of Australian History, 1861-1991,” Journal of Economic History, Vol. 53, No. 4.
Reserve Bank of New Zealand, 2000, “Monetary Policy Review: Business Cycle Developments and the Role of Monetary Policy over the 1990s.”
Winkelmann, Liliana and Winkelmann, Rainer, 1998, “Immigrants in the New Zealand Labour Market: a Cohort Analysis using 1981, 1986 and 1996 Census Data,” New Zealand Labor Market Bulletin: 1 & 2.
ANNEX I.1: Employment and Migration—Empirical Model and Results
1. Following Gross (1999), a model is constructed to examine the interactions of the unemployment rate (UR), real wage in logarithm (LRW), labor force participation rate (LFP), and net migration rate (MIG). The net migration rate is defined as the net inflows of permanent and long-term migrants divided by the labor force. The sample period is from the first quarter of 1988 to the third quarter of 2004.
2. The dynamics of the variables are explored with the following vector auto-regression (VAR):
3. Since UR, LRW, and LFP contain unit roots, they enter the VAR as the first differences. Vector Z include all exogenous variables, including the first difference in the log of import prices (seasonally adjusted) to represent supply shocks, and the first difference of the Australian unemployment rate. Also included are seasonal dummies, and a dummy variable that takes the value of one between the third quarter of 2001 and the third quarter of 2003 and zero during other periods, to capture the migration flows related to 9-11 security concern.10
4. The VAR model is specified with four lags based on likelihood ratio test. The numbers of lags of the exogenous variables are selected based on Schwartz and Akaike information criteria. The diagnostic tests indicate that the residuals are reasonably well behaved, with only normality being rejected. Alternative lag specifications yield similar results, although residuals from VARs with smaller lags are less well behaved.
5. Figure I.7 shows the accumulated impulse response of ΔUR, ΔLRW, and ΔLFP to a one-percentage point increase in the net migration rate (with two times the standard errors on both sides). A positive innovation in the net migration rate results in a fall in the unemployment rate and the real wage, and a rise in the labor force participation rate.
6. To examine the possibly different impacts of migrants to and from Australia and migrants to and from non-Australian countries, the above VAR is re-estimated with a breakdown between the net migration rate from Australia and the net migration rate from other countries. The accumulated impulse response of ΔUR, ΔLRW, and ΔLFP to a one- percentage point increase in the net migration rate from all countries, that from Australia, and that from non-Australia countries are summarized in Figure I.8. The responses of the labor market indicators to the migration rate from Australia are much stronger than the responses to the migration rate from other countries, suggesting that immigrants from non-Australian countries are less well-matched in their skill set with the local labor force, and thus are less successful in finding jobs, than those who migrate to and from Australia.
7. The estimated model is simulated with the migration flows projected by the Reserve Bank of New Zealand (Reserve Bank of New Zealand, 2004), where the annual net migration inflows are expected to slow from 18600 in the year through September 2004 to 9500 in the year through September 2005. Assuming a one-time shock of about 50 percent reduction in the net migration ratio from its current level–roughly the percentage reduction in the RBNZ projection, model simulation suggests that unemployment rate will rise 0.13 percent in one year and about 0.4 percent over two years.
ANNEX I.2: Data definitions and sources
Unemployment rate (UR). Source: OECD
Log of real wage (LRW). Wage is the labor cost index, with data before December 1992 calculated based on the prevailing weekly wage index, deflated by the Consumer Price Index. Source: OECD.
Labor force participation rate (LBF). Source: OECD.
Net migration rate (MIG_NET). Net permanent and long-term migration inflows divided by the total labor force. Source: CEIC and OECD.
Unemployment rate in Australia (AUS_UR). Source: OECD.
Change in log of import price index (DLPIMT). Source: IFS.
ANNEX I.3: Do Skilled Immigrants Induce Greater Reduction in the Unemployment Rate?
1. The data on the skill level of the migrants is based on the occupational information collected on all arrival and departure cards.11 Based on the claims of occupations, migrants are classified into skilled and unskilled workers.12 The skill indicator is defined as the difference between the number of skilled and unskilled migrants in the net inflows, divided by the labor force. As seen in Figure I.6, the relative share of the skilled immigrants have fluctuated but steadily declined between 1993 and 2001, before rising again.
2. Does the relative share of skilled migrants have any significant impact on economy-wide unemployment rate, besides those brought about by migration itself? To answer this question, lagged skill indices are added to the VAR. Wald tests suggest that the coefficients of the lagged skill indices are not statistically significant.
Prepared by Li Cui.
As a result, New Zealand has a higher share of foreign-born population than most other countries, as well as one of the largest shares of people living overseas. See OECD (2002) and Bryant and Law (2004).
A recent OECD country survey (2004) postulated that migration flows themselves might generate some cyclical “amplification” effect as migrants create more demand than supply in the short-term. The survey, however, did not find such a link in simple correlations between broad cyclical indicators and migration flows.
Those going to Australia are probably representative of the remaining population, as suggested by Glass and Choy (2001). They argued that the common labor market with Australia has allowed the migration of a broad mix of New Zealanders who might otherwise has been screened out by selective immigration policies.
A survey by the Department of Labor of New Zealand (1998) concluded that many highly qualified and experienced migrants from Asian countries and the countries in the Former Soviet Union (FSU) were either unemployed or underemployed.
People seeking to migrate to New Zealand now make an expression of interest (EOI) rather than a direct application. The government then invites the potential immigrants to apply, giving more capacity to smooth inflows over time. However, these policies do not affect outflows of migrants, which account for a large part of the fluctuations in net migration. See OECD (2004) and New Zealand Department of Labor (2005).
Other exogenous variables considered include the real growth of Australia, trade weighted real growth of all foreign countries, and terms of trade. None of these variables was found to be significant.
One caveat about the data is the large number of “not actively engaged” or “not specified” responses. More details are in Glass and Choy (2001). Nevertheless, these are the only continuous series that could shed light on the skill composition of the migrants.
Skilled workers include legislators, administrators and managers, professionals, technicians and associate professionals; unskilled workers include other professions such as clerks, service and sale workers, agriculture and fishery workers, plant and machine operators and assemblers, trader workers, and laborers and related elementary services. Similar classification was used by Glass and Choy (2001), which further split the unskilled category into semi-skilled and low-skilled.