3. Descriptive Evidence

We now turn to analysing the impact of immigration on labour market outcomes. We begin by examining the relationship between inflows of recent migrants to different geographic areas and labour market outcomes for different nativity groups in those areas. Before turning to regression analysis, we first examine whether there is a descriptive relationship between the concentrations of recent immigrants in different skill-groups in particular LMAs and the employment rates and average wages for each nativity group in these same skill-groups and LMAs. Economic theory suggests that the entry of immigrants should lower the wages of competing workers and increase the wages of workers whose skills become more valuable because of immigration (Borjas 1999). This increased supply of workers will also lead to reduced employment opportunities for competing workers if wages adjust slowly or institutions, such as minimum wages, make wage adjustment impossible.

Figure 1 graphs the five-yearly change in the employment rate for the New Zealand-born (left), earlier migrants (middle) and recent migrants (right) in each skill-group in each of 140 LMAs between each census versus the five-yearly change in the proportion of recent migrants in the overall population in the same skill-group and LMA between each census, with the two sets of comparisons superimposed in the same figure. The top row presents the results for age/qualification skill-groups and the bottom row presents the results for predicted occupation skill-groups. By examining changes between each census, we control for fixed differences across local skill-groups that are related to both the share of recent migrants in a particular skill-group/LMA and the employment rate for individuals in that local skill-group. The size of the plot circles are proportional to the average population of each local nativity/skill-group across the two census periods being compared and the solid line in each graph is the best linear fit of the data, with each point weighted by the average population of each local nativity/skill-group.^[17]

Figure 1: Relationships between changes in recent migrant share and changes in employment rates for each nativity group between each five-yearly census

Changes in employment rates for the New Zealand-born are weakly related to changes in the recent migrant share of the population in a particular local age/qualification group or in a particular local predicted occupation-group; a 10% increase in the recent migrant share (from the mean recent migrant share of 7% to 7.7%) in a particular local age/qualification group is associated with a 0.04 percentage point decline in the employment rate for the New Zealand-born in that local skill-group while the same change in a particular local predicted occupation-group is associated with a 0.05 percentage point increase in the employment rate for the New Zealand-born in that local skill-group. Similar to the results for the New Zealand-born, a 10% increase in the recent migrant share in a particular local age/qualification group is associated with a 0.05 percentage point decline in the employment rate for earlier migrants and a 0.32 percentage point decline for recent migrants. On the other hand, a 10% increase in the recent migrant share in a particular local predicted occupation group is associated with a 0.35 percentage point increase in the employment rate for earlier migrants and a 0.64 percentage point increase for recent migrants. Overall, a weak correlation is found between changes in the share of recent migrants in particular local skill-groups and changes in employment rates for each nativity group using both skill-group definitions.

Figure 2 graphs the five-yearly change in the mean log wage for the New Zealand-born, earlier migrants and recent migrants in each skill-group in each LMA between each census against the five-yearly change in the proportion of recent migrants in the overall population in the same skill-group and LMA between each census and presents the results in the same format as Figure 1. In local skill groups where the recent migrant share grew most, New Zealand-born workers had larger wage increases. For example, a 10% increase in the recent migrant share was associated with 0.21% higher New Zealand-born wages in the same local age/qualification groups, and 0.03% higher New Zealand-born wages in the same local predicted occupation groups. In contrast, wage growth for earlier migrants was lower in local skill-groups with greater increases in recent migrants - by 0.11% in local age/qualification groups and by 0.71% in local predicted occupation groups.

Figure 2: Relationship between changes in recent migrant share and changes in log wages for each nativity group between each five-yearly census

The results differ by skill-group when we examine the relationship between recent migrant inflows and wage changes for recent migrants themselves. Within local age/qualification groups, a 10% higher recent migrant inflow rate is associated with recent migrant wages that are 0.30% lower, whereas within local predicted occupation groups, they are 0.21% higher. As with employment rates, a weak overall correlation is found between changes in the share of recent migrants in particular local skill-groups and changes in wage rates for each nativity group using both skill-group definitions.

The associations displayed in Figure 1 and 2 will only indicate the causal relationship between inflows of recent migrants and labour market outcomes if different LMAs are equal attractive to different skill-groups, to different nativity-groups with the same skills, and their relative attractiveness does not change over time. Thus, we next extend on this graphical evidence by controlling for a range of other possible changes that may be correlated with local labour market outcomes and recent migrant inflows. In particular, we estimate reduced-form regressions models of the following form:

` Y_(jkt)=gammaR_(jkt)+alpha_(jk)+alpha_(jt)+alpha_(kt)+epsilon_(jkt) (1)

Help displaying mathmatical formuale

where j indexes age/qualification or predicted occupation skill-groups, k indexes the 140 LMAs, and t indexes time, Y_jkt is the employment rate or mean log wage for the New Zealand-born, earlier migrants or recent migrants in a particular local skill-group in a particular census year, R_jkt is the proportion of recent migrants in the overall population in a particular local skill-group, and α_jk, α_jt and α_kt are two-way interactions between skill-group fixed effects (α_j), LMA fixed effects (α_k) and time fixed effects (α_t). The coefficient of interest in this model is γ, which measures the average impact of a change in the share of recent migrants on employment rates and wages of the New Zealand-born, earlier migrants and recent migrants, controlling for unobservable fixed differences in local labour markets, skill-groups and time-periods, unobservable time-varying differences in local labour markets and skill-groups, and unobservable spatially varying differences in skill-groups.^[18]

Table 5 presents the results from estimating this regression model for employment rates (left panel) and log real wages (right panel) using both age/qualification (top panel) and predicted occupation (bottom panel) skill-groups. Each regression is variance-weighted by the population of each nativity/skill-group/LMA in a particular year and standard errors are clustered by local skill-group to adjust for possible serial correlation or error components within cells. The number of potential observations here is the number of skill groups (5 predicted occupations or 24 age/qualification skill-groups) multiplied by the number of LMAs (140) for each census; ie. 12,600 for age/qualification skill-groups and 2,100 for predicted occupations skill-groups. However, a number of skill-group/LMA/year observations are dropped because there are no individuals (for employment rates) or workers (for wages) in a particular nativity in that cell.

Table 5: Reduced-form regression of impact of recent migrant share on employment and wage in local LMAs

First, examining the results for age/qualification skill-groups, we find two statistically significant relationships. The first is between the recent migrant share and the employment rate for recent migrants, with a 10% increase in the recent migrant share (from the mean recent migrant share of 7%, to 7.7%) in a particular age/qualification-group/LMA associated with a 0.40 percentage point decrease in the employment rate for recent migrants in that age/qualification-group/LMA. The second is that a 10% increase in the recent migrant share is associated with 0.20% higher wages for New Zealand-born workers. Examining the results for predicted occupation skill-groups, we again find that a higher recent migrant share is associated with lower employment for recent migrants and higher wages for New Zealand-born workers. In addition, a higher recent migrant share in a local predicted occupation skill-group is associated with a lower employment rate for New Zealand-born workers. Converting the coefficients in this table to elasticities, a 10% increase in the recent migrant share is associated with 0.14 and 0.36 percentage point lower employment rates for the New Zealand-born and for recent migrants, respectively, and with 0.89% higher wages for the New Zealand-born.

Overall, these reduced-form models indicate that higher immigration has at most a weak negative impact on employment rates for the New Zealand-born along with a slightly stronger positive impact on their wages. In contrast, higher recent migrant inflows reduce the employment rates of recent migrants themselves, with negative but statistically insignificant impacts on wages for these individuals. However, this reduced-form approach is only appropriate if competition only occurs between recent migrants, earlier migrants and the New Zealand-born within skill-groups and does not allow us to examine whether immigration affects the wages for earlier migrants and the New Zealand-born in different skill-groups through production interactions. Thus, we next extend upon these reduced-form models by specifying the technology of the local production function. Imposing more structure on our estimates allows us to consider more complicated mechanisms through which inflows of recent migrants impact the labour market and for there to be imperfect substitutability between workers with different skills and from different nativity groups.

---

Footnotes

^[17] All summary statistics and regressions are variance weighted because the number of individuals in each LMA ranges from less than 500 in eight LMAs to over 100,000 in four LMAs and there is a large variation in the relative size of different nativity/skill-groups within LMA/years. If the variances of the estimated employment and wage rates are inversely proportional to the sample sizes for each nativity/skill-group/LMA group cell, then weighted estimates are more efficient.

^[18] LMA fixed effects control for differences in local labour market opportunities and are allowed to vary over time and skill-groups, but not both dimensions simultaneously. Skill-group fixed effects control for different returns to human capital in the labour market and are allowed to vary over time and across LMAs, but not both dimensions simultaneously, and year fixed effects control for aggregate changes in employment and wages and are allowed to vary across skill groups and LMAs, but not both dimensions simultaneously.

---