Immigration benefits us all – now the IMF gets in on the act

Only a short time after the Foged & Peri paper (summarised here) found that an “influx” of immigrants to Denmark benefited the both high-skilled and low-skilled workers in the local population, the IMF has examined whether or not those results apply to other advanced economies.

And, guess what? They do! Unsurprisingly.

In order to do so, the study uses a fairly nifty approach to accounting for potential reverse causation between migration and GDP per capita (since migrants might prefer moving to countries with higher GDP per capita in the first place). The study uses a “gravity model” to instrument for the share of migrants in a country, including various “push” factors (such as growth in the origin country, demographic variables etc.) and other controls, proving once again that describing something as “gobbledygook” just because you don’t understand it isn’t a particularly sensible thing to do.

The paper’s main findings are threefold. First, a 1% point increase in the proportion of population made up by migrants actually increases GDP per capita by 2%. Interestingly, this benefit arises via an increase in labour productivity, rather than an increase in the proportion of the population that is of working age.

For example, high-skilled immigrants can increase productivity through innovation and positive spillovers on native wages, while low-skilled workers can increase productivity by enabling native workers to re-train and move into more complex occupations (exactly as was found by Foged & Peri). An alternative mechanism cited by the IMF study suggests that the presence of low-skilled female immigrants increases the provision of household and child-care service, thereby increasing the labour supply of high-skilled native women. This result is robust to controlling for technology, trade openness, demographics, and country development.

Second, these benefits arise from both low-skilled and high-skilled migrants. As above, both skill-types affect GDP per capita through increasing labour productivity, rather than via increasing the proportion of the population that is of working age. However, the effect does appear to be more statistically significant for migration by low-skilled workers than it is for high-skilled migrants.

The study suggests that this difference could reflect differences in the impact of high-skilled migrants across different countries, but this seems unlikely to be sufficient to render the impact insignificant. More likely is the second reason posited by the study – namely, that high-skilled migrants initially might have to obtain jobs for which they are over-qualified, thereby meaning that their impact on the incentives of high-skilled native workers to retrain etc. is limited at first.

Third, the benefits to native workers arise across the entire income distribution. Both low-skilled and high-skilled immigration increase the GDP per capita of those in the bottom 90% of the income distribution by roughly the same amount, while high-skilled immigration increases the GDP per capita of those in the top 10% of the income distribution by roughly twice as much as does low-skilled immigration.

However, the study does not really examine the distribution within the bottom 90% particularly closely – the study just looks at the estimated impact of immigration on the Gini coefficient to conclude that the distribution within the bottom 90% would not be changed significantly. The study should, instead, have looked at, say, the impact of immigration on each decile or quintile of the income distribution separately so as to give a more complete picture of the impact of immigration across the income distribution.

The paper (and particularly the blog post linked to above) ends by getting somewhat more political. In particular the study suggests that there is a need for improvement in terms of providing support for native workers that want to re-train, find a new job etc. However, these policy suggestions are made without taking into account the fact that some countries do already have plentiful such schemes in place, to the extent that increasing the provision of such schemes in those countries might not be efficient. Of course, that’s not to say that some countries would benefit from increasing the provision of such schemes.

Pesky immigrants, coming over here, making our lives better

Further to a previous post about how immigration can benefit the global economy, it’s also important to examine how immigration can affect the people in areas to where people have migrated. One of the main arguments used by those that try to claim that immigration is harmful is that immigration hurts the wages / employment of the low-skilled because supposedly cheaper immigrants take those jobs away from domestic workers. The fact that this argument is based on the false “lump of labour” theory has been covered well elsewhere (see, for example, here), but it can also be tested empirically.

To that end, a recent paper by Foged & Peri investigates the impact of immigration on domestic workers of various different skill types. To do so, it makes use of a quirk in the system that Denmark used to re-locate successful asylum-seekers and the fact that in subsequent years, relatives of those asylum seekers came to join them. Specifically, between 1986 and 1998, Denmark distributed refugees across the country taking into account only the refugee’s family size, the nationality of the refugee (so as to try to achieve “clusters” of refugees that would help each other out), and the availability of housing in each area.

In this way, the location of refugees was almost entirely independent of local labour market conditions. (Note that it is not entirely independent because there is likely to be some relationship between an area’s labour market and the availability of housing.) Nonetheless, this distribution of refugees was followed by a period between 1995 and 2003 in which immigrants from various regions moved to Denmark (most of these were from the likes of the Former Yugoslavia, Somalia etc. and were trying to escape local conflicts).

The people in this “new wave” of immigrants tended to settle in areas where earlier refugees/immigrants with the same nationality had settled – in this way, the initial distribution of refugees (that was mostly independent of labour market conditions) also drove future immigrants’ decisions of where to settle. Hence, the increase in immigration to Denmark from 1995 on can be used to assess the impact of immigration on the labour market in Denmark since the locational decisions of those immigrants was mostly independent of the labour market conditions themselves. (This is important because otherwise the results of an analysis of the impact of immigration on the labour market would be biased if the amount of immigration to a particular area itself was affected by the labour market in that area).

Therefore, in order to assess the impact of immigration on Danish workers, the paper uses a “longitudinal cohort study” – i.e. a frequent survey of a large group of people over a prolonged period of time (in this case, the Danish Integrated Database for Labour Market Research). This survey contains information related to someone’s age, municipal location, whether or not they are employed, their occupation, the number of hours they work, their salary etc. Coupled with information on where immigrants settled as per the previous paragraph, this dataset thus enables the researchers to investigate the impact of immigration on domestic workers.

The results of the study are extremely interesting. Contrary to the fallacious argument that low-skilled workers are harmed by immigration, the results of this study indicate that immigration actually benefits low-skilled workers in multiple ways.

First, immigration motivates and enables low-skilled workers to progress into more complex occupations. Second, these more complex jobs are associated with higher wages, such that immigration actually increases the hourly wage of low-skilled workers. Third, immigration does not result in (some number) of low-skilled workers becoming unemployed. In other words, immigration doesn’t force low-skilled workers out of a job, but instead enables them to obtain better jobs.

In addition, the study also looks to see if the impact of immigration falls different across different sets of low-skilled workers. It finds that the positive impact of immigration is felt most strongly by those low-skilled workers that are either young (less than 46 years old) or have not been in their jobs very long (less than four years), but also that there are no statistically significant negative effects for older workers or those that have been in their jobs a long time.

Overall, therefore, this study is pretty conclusive that immigration is beneficial to low-skilled workers and that those that claim that immigration harms those on low wages really don’t know what they’re talking about.