The Economic Consequences of Migrant Occupational Downgrading in Colombia: Guest Post by Jeremy Lebow


This page in:

This is the fifth in this year’s series of posts by PhD students on the job market.

Between 2015-2019, around 1.8 million Venezuelans fled a political and economic crisis into neighboring Colombia. Despite having relatively similar education as Colombian natives, these migrants disproportionately entered occupations that employ less educated natives, such as restaurant work or domestic service. To put this in perspective, around 75% of college-educated migrants are in occupations where the modal native did not go to college, as opposed to 30% for natives. This is not unique to Colombia – around the world, it is well-documented that migrants tend to be over-educated in their occupation relative to natives.

In my job market paper, I study the effect of this migrant occupational downgrading on the labor market outcomes of native Colombians. In particular, one might worry that concentrating migrant competition among low-income natives may increase wage inequality, and this is consistent with existing evidence that migration from Venezuela decreased hourly wages more for less educated natives (for example, see Lebow 2021, Delgado-Prieto 2021, Santamaria 2020, or Caruso et al. 2019). To disentangle the role of downgrading in explaining this unequal wage effect, I estimate a model of labor demand that allows me to calculate the counterfactual effects of migration under differing scenarios of migrant downgrading. The structural approach also enables me to look under the hood at what drives the relationship between downgrading and native labor market outcomes, to understand in what settings downgrading is likely to be most consequential.

What drives the relationship between migrant occupational downgrading and native labor market outcomes?

I estimate a nested-CES model of labor demand in the spirit of Ottaviano & Peri 2012. A key component of this framework is the imperfect substitutability between migrants and natives, which determines the extent to which migration puts downward (or upward) pressure on native hourly wages, and which I allow to vary by education group. My key contribution is that, unlike previous papers, I adapt this framework to incorporate migrant occupational downgrading by assigning migrants to education groups based not on their education, but on the typical education level of their occupation. Thus, I estimate the substitutability between migrants and the natives they compete with.

I also estimate the degree of substitutability across occupation skill groups, which determines the extent to which native wages equalize across groups. For example, native workers may move into occupations less affected by migration, which would increase across-group substitutability.

One thing the model does not account for is the sorting of migrants into occupations, since I treat migrant downgrading as exogenous rather than as an equilibrium outcome of migrant occupational sorting. This is motivated by the fact that, in this setting, a variety of institutional factors have effectively closed off access to high-skill occupations for many migrants, such as gaps in legal status, educational recognition, and occupational licensing.

The model generates a set of linear equations that I estimate across 79 metropolitan areas and 6 years of analysis. Intuitively, I regress the migrant-to-native ratio in hourly wages on the migrant-to-native labor supply ratio; a low sensitivity of the wage ratio to changes in labor supply implies a high substitutability between migrants and natives. I include metro area fixed effects to capture the fixed difference in wages between migrants and natives, such that estimation is driven by changes in the ratio of outcomes over time.

Advantages of the Colombian setting for estimation

A unique benefit of the Colombian setting is the broad geographic dispersion of migrants combined with high-quality survey data on labor market outcomes of both migrants and natives from the Gran Encuesta Integrada de Hogares (GEIH), resulting in a sample of over 1.5 million natives and 30,000 migrants. To account for the endogenous sorting of migrants across metropolitan areas in Colombia, I use a shift-share instrument based on the historical location of Venezuelans in Colombia. A recent literature has evaluated the exclusion restriction in the shift-share framework, showing that it amounts to the exclusion of the initial shares – in this case, that the historical allocation of Venezuelans in Colombia is not endogenous to trends in labor market outcomes over 2015-2019. To evaluate this, I show that historical migrant shares, which were very small relative to the current migration, are uncorrelated with pre-trends in observable outcomes for natives. There was also little migration before 2015, mitigating the concern that results incorporate dynamic effects from pre-2015 migration (Jaeger et al. 2018). Moreover, because this was a sudden migration driven by factors within Venezuela, the timing of the migration is unrelated to changes happening in Colombia over this period.

What do we learn?

My first result is that migrants are more substitutable with less educated natives. Thus, when a migrant engineer, systems analyst, or accountant (among the most common high-skill occupations for Venezuelan migrants) downgrades into restaurant work, they move from a job in which they are more complementary to a job in which they are more substitutable with natives. This may be because the skills required to serve food in a restaurant are more routine and less specialized. It may also be because a large share of restaurant work in Colombia is in the informal sector, which tends to be more substitutable because of factors such as no minimum wages and high turnover rates. This may explain the larger substitutability parameters observed in this setting relative to in the U.S., where a similar elasticity is observed over many decades rather than over only a few years. I also find, consistent with existing evidence, that substitutability between workers with and without completed secondary schooling is lower in Colombia than in developed countries. As a result, wage effects on workers without completed secondary remain concentrated within this group.

I use the estimated model to calculate the total effect of the Venezuelan migration on native hourly wages. This is presented in the blue bars: the wage effect is stronger for less educated natives, at around -4% for natives without secondary schooling and close to 0 for natives with post-secondary schooling.

In the green bars, I repeat the calculation under a counterfactual in which there is no downgrading – specifically, in which migrants enter occupations that match their actual level of education. This represents a policy to improve the migrant-job educational match, such as a program to verify skills or mitigate barriers to educational recognition. The results show that undoing downgrading benefits less educated natives without substantially harming more educated natives, despite those workers facing increased competition. This is driven by the higher complementary between migrants and natives in higher-education occupations, as well as the increase in total output that results from moving migrants into more productive and relatively under-supplied jobs.

 Occupational downgrading and wage effects

This analysis assumes no capital adjustments. However, I can use the model to show that, as capital adjusts in the long-term, average wage effects disappear while the distributional consequences of downgrading persist.

Policy implications and limitations

The results suggest that policies to combat migrant downgrading are likely to have substantial benefits in terms of both wage equality and total productivity. They may be especially relevant in the developing country setting, where high informality rates and low substitutability across education groups exacerbate the consequences of downgrading. They are also especially relevant in episodes of forced displacement, which is more likely to result in migrant downgrading and which is most likely to affect developing countries (UNHCR 2020).

Finally, there are important factors not built into the model that may increase the benefits of upgrading, such as increased migrant consumption and industry-specific human capital transfers, that should be explored in future work.

Jeremy Lebow is a PhD candidate in Economics at Duke University. His personal webpage can be found here.


Join the Conversation