• Aucun résultat trouvé

Literature review

equilibrium analysis of immigration

1. Literature review

Understanding the economic impact of immigration has long been an important goal for many economists.

Over the years, several methodologies have been developed, each focusing on a particular aspect of the economy. Two main strands in this literature can be identified, namely a strand focusing on the public finance impact of immigration and one focusing on the labour market (and especially the wage) impact of immigration.

An  elaborate overview of the impact of immigration on public finance can be found in Part I of the present report. Therefore, this review focuses primarily on the wage impact of immigration.

From a labour market perspective, immigration can be interpreted as a positive labour supply shock. The canonical labour market model then dictates that in the short run – when capital stock has not yet adapted to the new stock of labour supply – this should lead to a fall in the average wage of workers. When capital is adapted to the new stock of labour supply, the average wage level should be restored to its previous equilibrium level. This is conditional on the assumption that immigrants and natives are perfectly substitutable. If the labour

1 Those we consider here as natives include second‑generation immigrants because of data availability.

2 Data that are used to calibrate the model are from other sources than the CBSS data since we need macroeconomic variables. This allows us to take the most recent data at the time of the analysis, which was 2017.

provided by immigrants and natives is imperfectly substitutable, the average wage effect of immigration on natives should be (slightly) positive, while immigrants wages decrease. The mechanism of imperfect substitution will be discussed in more detail in the next section. Finally, the average wage impact of immigration conceals the differing wage impact of immigration on smaller groups of individuals, identified by age, skill and origin, for instance. The inequality impact of immigration should therefore also be considered. Studies empirically assessing the wage impact of immigration can be harboured in two broad categories, namely the spatial correlation approach and the skill‑cell approach. Both have their advantages and disadvantages, which will be presented in the following sections.

1.1 Spatial correlation

The concept of the spatial correlation relates the development of wages and immigrant stock across regions to see if there is a significant relation between the two. Assuming the regions of interest would have developed similarly if there had been no immigration, the differing wage development may be attributed to immigration.

But it is very likely that immigrants choose their destination based on the economic opportunities it provides. The pure spatial correlation approach is thus likely to obtain a spurious positive impact of immigration on wages, i.e.

it is likely to point to a positive association between immigration and wages though the direction of causality remains unclear.

To account for this problem Altonji and Card (1991) developed an instrumental variable‑strategy – the so‑called shift‑share methodology – where the location of previously established immigrants is used as an instrument to approximate the destination of recent immigrants. The argument goes that immigrants are more strongly attracted by networks of earlier immigrants, rather than by economic opportunities. Many studies using this methodology appear to find negligible effects of immigration on native wages (e.g. Winter‑Ebmer and Zweimüller (1996) for Austria ; Pischke and Velling (1997) for Germany ; Zorlu and Hartog (2005) for the Netherlands, Norway and the United Kingdom). However, some studies present slightly positive wage effects of immigration. Mitaritonna et al.

(2017) show that immigration tends to increase local productivity. They relate it to specialisation of immigrants and natives in complementary tasks (Peri and Sparber, 2009 ; D’amuri and Peri 2014) on the one hand, and to a potential innovation increase on the other. Dustmann et al. (2012) also find a slight positive effect of immigration on wages. Moving beyond wage effects at the mean, immigration appears to exert a downward pressure on wages in the 20th bottom percentile of the wage distribution. In return, immigration appears to slightly increase the wage in the upper part of the wage distribution.

However, Jaeger et al. (2018) argue that the shift‑share methodology does not sufficiently manage to distinguish short‑ and long‑run effects of immigration. The negligible or slightly positive wage effects of immigration should therefore primarily be interpreted as long‑run impacts, while little can be said about the short‑run wage impact of immigration.

To avoid conflation of long‑ and short‑run effects of immigration on wages, one can make use of periods of exceptional and unexpected immigration in certain regions. By comparing the wage change in this region with a region that was previously similar, the impact of immigration on wages can be observed. The most famous example of this type of natural experiment is the Mariel Boatlift. It refers to the influx of over 100 000 Cuban refugees from the port of Mariel in Miami (Card, 1990 ; Borjas, 2017 ; Peri and Yasenov, 2019). Other examples are the repatriation from Algeria to France in 1962 after the end of the Algerian independence war (Hunt, 1992 ; Edo, 2017), the lifting of emigration restrictions in the Soviet Union that led to huge immigrant flows of Russian Jews into Israel in the early 1990s (Friedberg, 2001 ; Cohen‑Goldner and Paserman, 2011) and the massive inflow of Syrian refugees into Turkey in response to the Syrian war (Tumen, 2016).

Even so, neither of these methodologies accounts for the potential displacement of natives or earlier immigrants as a consequence of immigration (e.g. Borjas, 2006 ; Mocetti and Porello, 2010 ; Basile et al., 2020). This reaction may greatly reduce or even mitigate the actual wage impact of immigration on the incumbent population,

although Peri and Sparber (2011) argue that “the cross‑region analyses of immigration’s effect on wages are still informative”.

1.2 Skill cells and structure

To address the two main problems of the spatial correlation approach, the skill‑cell approach was developed by Borjas (2003). Instead of sub‑dividing a country into regions, the approach creates cells of individuals by education, experience and origin, the main indicators of productivity. It assumes that immigrants and natives in each cell compete for similar jobs, so heterogeneity in the inflow of immigrants across skill cells can be assessed.

In its initial iteration, the approach filtered out the cross‑cell impact of immigration on wages, leaving only the average in‑cell impact. Since immigrants and natives are assumed to compete for similar jobs within these cells, immigration of individuals with similar characteristics is found to reduce the wage of natives. A 10 % increase in labour supply is estimated to reduce native wages by 3 % to 4 % in the short run (Borjas, 2003).

Although in‑cell wage effects of immigration are interesting in their own right, it is crucial to understand the relation of wages to immigration in other groups as well to get a complete image. As it is not feasible to estimate the relation of each cell to all other cells at the same time, a structure needs to be imposed on the production function. In this way, the number of elasticities to be estimated is reduced to a manageable quantity. Two key findings emerge from these studies.

First, immigration appears to have a slight positive effect on the average native wage in the long run (when capital has adapted to the labour supply), if imperfect substitution between natives and immigrants is observed (Gerfin and Kaiser,  2010  for Switzerland ; D’amuri et  al.,  2010 for Germany ; Manacorda et  al.,  2012 for the United Kingdom ; Ottaviano and Peri, 2012 for the United States ; Brücker et al., 2014 for Denmark). In contrast, the average wage of previously established immigrants decreased because of immigration. When immigrants and natives of similar skill are found to be perfectly substitutable, there appears to be no long‑run effect of immigration on the average wage (Borjas, 2014 for the United States ; Edo and Toubal, 2015 for France).

Second, the skill composition of the immigrant wave has a significant impact on the relative distribution of native wages. By increasing the labour supply in some skill cells, immigration will decrease the wage of natives with similar skill, while raising the wages of natives with complementary skills. This implies that immigration may reduce inequality among natives, conditional on the fact that the share of high skilled among recent immigrants is larger than in the population. On the other hand, if the share of low skilled dominates, inequality is found to increase (e.g. Docquier et al., 2014).

The skill cell approach has some downsides of its own, though. Imposing a structure on the production function requires assumptions on behaviour of firms and individuals. The assumption that immigrants and natives of the same skill (education – experience) compete for the same job is one such example. Per cell, it appears as if immigrants earn a lower wage than natives, which is often related to a lower productivity in the same type of jobs. However, several studies have argued that immigrants experience skill downgrading when entering the host economy (e.g. Dustmann et al., 2012 ; Peri and Sparber, 2009). They therefore do not compete with natives of similar skill, which may have significant implications for the obtained outcomes of the model.

1.3 Interacting economic impact channels

Studies estimating the impact of immigration on wages or public finance offer a valuable contribution to understanding the overall economic impact of immigration. However, these studies only present a partial image of the economic impact of immigration. Other channels, such as employment, labour market frictions, market size or trade have not been taken into consideration.

To account for (some of) the additional impact channels, several authors have constructed models describing a wider image of the economy. Aubry et al. (2016), for instance, include employment, market size and trade effects into a model of 34 OECD countries. They find that immigration has improved the welfare for 69 % of the non‑migrant OECD population, and for 83 % of non‑migrant citizens of the 22 richest OECD countries. Although the wage and fiscal effects are significant in some countries, it appears that the market size effect – the number of varieties available to individuals – is the strongest contributor to the welfare increase. In the case of Belgium, they find that average welfare gains are combined with close to zero welfare losses for the low‑skilled.

Using a similar model structure, barring the trade effect of immigration, Burzynski et  al. (2018) compare the welfare effect of two pre‑crisis immigration waves (1991–2000  and  2001–2010) and of the post‑crisis wave (2011–2015) for native citizens in 20 OECD countries. They confirm that immigration has a positive impact on the real income of natives but note that this effect is strongly heterogenous across countries and skill groups.

The post‑crisis wave of immigrants appears to bring smaller welfare gains compared to previous waves. This is driven by the change in origin mix of immigrants, leading to lower levels of human capital. Although the welfare increase of immigration in Belgium is also weaker post‑crisis than pre‑crisis, this is not driven by lower levels of human capital. The 2011‑2015 immigrant wave has reduced human capital less severely than immigration from 2001‑2010 (respectively –1.03 % and –1.66 %). Still, the overall welfare effect is lower post‑crisis (+2.01 % compared to 2.29 %) because the immigrant wave is smaller, leading to a smaller positive market size effect.

Finally, Battisti et  al. (2018) include search frictions and wage bargaining in their model economy. It  appears that immigration attenuates the effect of search frictions, by increasing the size of the labour market. Especially in countries with high native unemployment and large immigrant wage gaps, the job creation effect is found to be strong. Although the gains from immigration tend to outweigh the welfare costs of redistribution for most countries, this is not the case in Belgium. A one percentage point increase in immigration is found to slightly reduce native welfare by 0.02 %. In contrast, it appears to increase welfare of previously established immigrants by 0.12 %, because they are significantly more likely to be unemployed than natives.

Overall, it appears that additional impact channels may play an important role in understanding the wage and welfare effects of immigration on the native population.