The refugee crisis: Sugar in a teacup?

When the priestly leaders of the Parsis, fleeing Persia for India after the Arab conquest of the 8th century, came before local ruler Jadhav Rana asking for sanctuary, Rana asked for a bowl of milk. The bowl was filled to the brim.  How could his kingdom accommodate more, asked Rana, without the bowl spilling over? The priestly leaders, the legend goes, slipped sugar in the milk, masterfully suggesting that the Parsis would dissolve into the existing population, sweetening their lives in the process.

As through history, we are once again faced with a situation where millions have left their homes, ravaged by violence and conflict, seeking sanctuary in foreign lands. And once again, the leaders of the promised lands worry about spilt milk.

Ever since labor economist David Card showed that the 1980 Mariel boatlift, which brought Cuban refugees to Miami, had no effect on local wages and employment, the academic wars have raged (for instance, see summaries and rebuttals here and here). People argue that labor demand curves always slope down—a shift in the supply of workers must decrease wages. Therefore, results that show otherwise are incorrect.

To resolve the debate, I find it useful to think of the ideal experiment. Refugees would come to a country and locate in communities chosen at random, so that there is no correlation between the economic characteristics of the community and its refugee population. Data would track the labor market outcomes of each and every “native,” regardless of where they moved. We could then compare the wages of the natives (wherever they are) who lived in communities that received refugees versus those who did not, yielding the causal impact of refugee populations on labor market outcomes of the natives.

For obvious reasons, this is hard to do. Leading researchers instead focus on “natural experiments” and “area-wide” estimates on the impact of refugees. That is, they examine what happens to wages in the communities where refugees settle using a clever way of teasing out the “natural randomness” in settlement patterns. But if the locals move or if the refugee population itself is badly measured, this could lead to a bias towards finding no results. In essence, the results are for the natives who chose to remain behind and therefore presumably had better job options to begin with. In addition, mismeasurement of the refugee population drives results towards zero—the iron law of econometrics.

But a recently published study from Denmark by economists Mette Foged and Giovanni Peri solves all these problems at once:

  • They have data on each and every Danish worker between 1991 and 2008, so that they can track people wherever they go.
  • The refugees came in two waves. In the first wave, they waited in a queue and as communities opened up spaces, they were allocated in groups from the same country. Since the department managing the transfers had no data on the skills of these refugees, this was like a random allocation. Later, between 1995 and 2003, the refugee flow from Somalia, Iraq, and Afghanistan increased—but they settled in the same communities that their fellow countrymen had gone to. This created large differences in refugee populations across communities that only grew over time. The authors present a number of very convincing tests that the “like-random” assumption truly holds in their data.
  • Because they have fairly long-term data, they can assess both the immediate effects and the cumulative effects over time.

And the results?

Zip. Nada. Nothing. Kuch bhi nahin. No smoking gun showing that natives will be hurt when refugees enter.

More specifically, the authors generally find positive effects on the employment and wages of the natives who looked similar to the mostly unskilled refugees entering at this time. Quantitatively, they find that “a 1 percentage point increase in the share of low-skilled immigrants from refugee-sending countries increased wages for low-skilled native workers by 1 to 1.8 percent.” Nor do they find negative effects on employment—the fraction of the year that natives worked either remained the same or increased slightly, depending on the specification.

To interpret their findings, the authors say:

The panel regressions suggest that refugee-country immigrants, who specialized mainly in manual, low-skilled jobs, encouraged low-skilled natives to take more complex occupations, decreasing the manual content of their jobs, especially when changing establishment and this contributed to produce a positive effect on their wages and employment. In no specification do we find crowding-out of native unskilled workers or depressing effects on their wages”.

Could it be that this was just because the refugees did not work? Nope. Over the time period of the data, the foreign-born share of employment in Denmark rose sharply, from 3 percent to just over 6 percent.

Could it be that the refugees were “like the Danes” and therefore fitted in better? Unlikely. Their estimates are driven by a surge in immigrants from Somalia, Afghanistan, and Iraq between 1995 and 2003. Hard to mistake them for Danes.

But the earlier “area” estimates were wrong, right? Nope. Actually it turns out that all the legwork Foged and Peri do pretty much replicates the simpler area estimates that have been saying this all along.

To be sure, more could be said. For instance, the paper could talk about the precise government policies that were put in place to help the refugees when they came and the associated costs. In fact, it would be good to show the per refugee cost to the government and compare this to the taxes that the refugees paid back to the system as they joined the labor force. Since there are no “spillovers” to natives, if the taxes are higher than the initial public costs, refugees are a net gain to the treasury. Bringing them in would actually increase government budgets over time, without hurting native wages. Of course, the wages and lives of the refugees will be immeasurably better, but we have known that for a while now.

When it comes to taking in refugees, there shouldn’t be worry about the effects on wages or employment of natives. On the contrary, refugees sweeten the deal, as the Parsi priests pointed out thirteen centuries ago.