New research shows immigration has only a minor effect on wages
Economic arguments against immigration often take two forms – immigrants either suppress the wages of workers, or immigration creates higher unemployment. But our research shows that the impact of immigration on the labour market in Australia over the last 15 years is negligible.
The notion that immigration impacts wages or employment is largely based on a simplistic analysis of supply and demand. The idea is that immigration increases the supply of labour and, if everything else holds constant, this results in lower wages.
But the world is not this simple.
If the demand for labour is also increasing, then a greater supply of workers might be absorbed without reducing wages or increasing unemployment. Failure to attract new workers to a country could also lead to under-utilisation of capital. Mining equipment, for instance, might sit idle if there are no workers to run the machines. This would reduce economic growth.
But, at the end of the day, the question of whether immigration hurts the wage and employment prospects of in-country workers is an empirical question. So my colleagues, Nathan Deutscher and Hang Thi To, and I decided to investigate.
Let's look at two groups of Australian workers. The first are young, with five or fewer years of work experience and a high school education. The second group is older, with 21-25 years of work experience and a tertiary education.
Between 2001 and 2006, earnings for the first group grew 15.2% and grew another 13.1% between 2006 and 2011. The second group's earnings grew 21.2% from 2001 to 2006 and 32% from 2006 to 2011.
As we can see, the difference in income growth between the groups was 5% from 2001 to 2006, and 19% from 2006 to 2011. The older, more educated group fared better in both periods.
But identifying how much of an impact immigration had on this is tricky, for several reasons. Two of the most important are the selectivity of migrants and the behavioural response of incumbent businesses and workers.
Immigrants choose whether or not to come to Australia and (largely) choose where to live once they get here. Immigrants are not randomly allocated to jobs and cities, which complicates testing for causal effects. There is also evidence that where natives tend to be fully compensated for their ability and motivation, immigrants don't.
Past attempts at gauging the impact of immigration on the labour market compared geographical areas with different percentages of immigrants. The problem with this approach is that it assumes that geographic labour markets are fixed and distinct. It rules out the selectiveness of migrants and whether incumbents react to new migrants by moving to other neighbourhoods.
To get around this we used an approach pioneered by George Borjas, who found that immigration significantly impacted low-skill US workers who were at the same skill level as new immigrants. We looked at changes in immigration rates into different skill groups in Australia to identify the effects of immigration on the earnings and employment prospects of Australian workers.
There are some drawbacks to using these data sets. We are missing migrants who reside in Australia for less than one year and are not picked up by the census, such as backpackers and short-term 457 visa holders. The time period was also one of robust economic growth – we can't say what would happen with the same level of immigration in a period of much slower economic growth.
In the end we defined 40 distinct skill groups at a national level, identifying them with a combination of educational attainment and workforce experience. Unlike their place of work and residence, workers cannot readily change their skill group. People in the same skill group compete against one another and can only be imperfectly replaced by workers from other skill groups.
In our study we looked at six outcomes – annual earnings, weekly earnings, wage rates, hours worked, participation rate and unemployment. We explored 114 different possibilities in all. We estimated the model for both HILDA and the SIH data across the whole population and separately by male and female. We restricted it to young people and broadened our definition of skill groups. We also controlled for overall macroeconomic conditions.
What is left are the differences in outcomes over time, across skill groups.
Once we controlled for the fact that immigrants to Australia disproportionately flow into high-skill groups with higher wages and other positive outcomes, we found immigration had no impact on the wages of incumbent workers.
Some of our estimates showed immigration had a negative effect on incumbent workers. But positive effects outnumbered negative effects three-to-one, and the overwhelming impression is that immigration has no effect.
If we return to our two groups from before, the older, more experienced group saw their earnings grow faster because of their education. Between 2001 and 2006 all educated workers had earnings that grew faster, irrespective of experience and unrelated to how many similar (in experience and education) migrants entered the country. Between 2006 and 2011, more experienced workers saw their incomes grow faster. In neither case did changing ratios of immigration to these groups provide any increase or decrease in earnings.
Our research only looked at one, very limited, aspect of immigration. Immigrants can also bring cultural and culinary diversity, innovation and creativity. But while employment is only one aspect of the costs and benefits of immigration, it is not as impacted by immigration as we may think.