Size and Quality of Ethnic Enclaves for Refugee Labor Market Entry

This article analyses the relationship between the size and the quality of ethnic enclaves on immigrants’ labor market integration. Using exogenously defined grid cells to delineate neighborhoods, we find robust empirical evidence that the employment rate of the respective immigrant group in the vicinity (as a measure of enclave quality) facilitates labor market integration of new immigrants. The influence of the overall employment rate and the share of co-nationals in the neighborhood tend to be positive, but less robust. We thus conclude that the quality is more important than the size of ethnic enclave in helping new immigrants finding jobs.


Introduction and motivation
Since the peak of the surge of refugee immigration to Europe in the fall of 2015, public debate has been dominated by immigration and the integration of the new comers in the receiving countries. The most important form of integration is widely seen as entering the labor market. The policy makers put much of their attention on a number schemes that can transcend the newcomers from labor market outsiders to insiders.
Following a big body of literature on ethnic enclaves, we the issue of labor market integration of new comers by examining the probability to find their first job with respect to their immediate environment. In the literature there are many arguments supporting the notion that immigrant enclaves play a key role in how immigrants are sorted into the labor market. Our results show that the important aspect of an enclave is its qualitative nature in terms of the intensity of employment among an immigrant's co-ethnic peers. Overall size or density, on the other hand, does not contribute one way or the other.
We contribute to the existing literature by emphasizing enclave quality rather than enclave size. Further, we use of two distinct immigrant groups, consisting of individuals that arrived to Sweden during two waves of forced immigration which allows us to say something about the effect size and relative importance. The use of exogenously determined physically standardized squares puts limits on some of the cross-enclave variation in different dimensions. Also, in this setting scale (size, population) and density (people per square kilometer) are one and the same thing, i.e. measured as per squarekilometer.
From earlier research we know that immigrants live more spatially concentrated than many other groups. We also know that this fact has a bearing on immigrants' integration into society and into the labor market in particular. Bartel (1989) and Borjas (1998) demonstrated that immigrants chiefly concentrate spatially to larger cities. Further, within these cities they concentrate and there is evidence of what is usually labelled segregation (Edin et. al., 2003, Cutler et. al. 2008).
There is consensus on the fact that ethnic enclaves are important for how/if immigrants enter the labor market successfully. Yet, there is no agreement about the size or even the direction of the effect. One line of argument is that the spatial concentration of immigrants facilitates employment and entrepreneurship among immigrants through social network effects (Edin et. al., 2003, Cutler et. al., 2008, Patacchini and Zenou, 2012Bayer et al., 2008). The opposite line of the argument is that spatial concentration and the resultant isolation from the native population brings about a distance from opportunities and risks lock-in effects (Borjas, 2000). What is largely an empirical literature is still out there to establish if and through which mechanisms ethnic enclaves can facilitate labor market integration of immigrants, to which we aim to contribute.

Quality of ethnic enclaves
In the initial discussions concerning ethnic enclaves focus was put on basic economic mechanisms working through supply and demand channels. More recent literature adds a further complementary focus on more quality-oriented characteristics of ethnic enclaves, which may have the potential to influence the labor market outcomes of immigrants. These mechanisms work in a rather different way. The discussion on enclave quality focus on the constituent parts making up the enclave. Depending on how the enclave is built-up there can be both negative and positive results for the people living in them (Cutler et al, 2008, Edin et al, 2003, Andersson et al, 2017. What is often discussed as a quality aspect is the transfer of useful information between co-national peers within an enclave. Some information may be more tacit. For example, it may entail information about the institutions (formal or informal) surrounding the search for a job. Other information may be more tangible. For example, it may be acquired information about an existing job opportunity or contact information to a potential employer accessed through ethnic peers living in the same enclave. The literature sometimes say that ethnic enclaves may facilitate access to the labor market through special information channels. However, the empirical evidence that this is actually the case is rather limited.
In our empirical analysis we use individual-level micro-data covering the whole Swedish population for the time period from 1992 to 2013. The data is geo-coded which allow us to pinpoint each individuals' residential locations down to exogenously assigned 1 by 1 kilometer grid cells covering the Stockholm metropolitan area. In identifying the connections between individuals and their peers in close proximity, we use two distinct immigration groups. The first group consist of people arriving from the Balkans between the years 1992-1993 following the Balkan wars. The second group is from the Middle East that arrived in Sweden between the years 2005-2006 following the Iraq war. Individuals show up in the database for the first time when they have obtained a work permit, and we observe their first contact with the labor market as our outcome variable against a vector of individual and neighborhood level variables.

Related literature -enclaves and labor market entry
It is a common and well-known pattern that immigrants in many countries live segregated as a group, but each immigrant group are sorted into different places (Borjas, 1995(Borjas, , 2000. This kind of geographical sorting may be voluntary if immigrants prefer places with a large share of co-national inhabitants sharing the same ethnic and cultural background and sharing the same language. But geographical sorting can also be the result of institutional mechanisms and/or path-dependence. The institutions may work on the national or sub-national level and may change the geographical distribution for some subpopulation. Sometimes this is the result without any explicit intentions being at hand. For example, land-use regulations or zoning laws may render some places and some parts of the housing market too expensive for a certain income group. In much the same manner it may happen that rent control reduces churn in some attractive and/or central locations. This hinders the availability of affordable dwellings in central locations. Centrally placed (by the authorities) newly arrived immigrants (refugees) in some places may also initiate pathdependence and the development of clusters of specific minorities that may evolve and grow over time.
These different possibilities do not exclude one another and may co-exist. At the same time, they are not universal in their manifestation. In reality, it is probably a combination of mechanisms at work when ethnic enclaves are formed, and segregation develops over time. Through history many great cities (or parts of them) have been characterized by concentrated diasporas. Using the existence of such areas we examine how the character of segregation and concentration relate to labor market outcomes of individual immigrants. In what way do living in an ethnic enclave relate to the labor market entry of immigrants?
Labor market sorting of immigrants and its connection to segregation and ethnic enclaves is a well-researched area. Edin et.al. (2003) made a seminal contribution when using a Swedish government placement policy of refugee immigrants. The policy was in place from 1985 until 1994. The authors make the conclusion that ethnic enclaves improve the labor market outcomes for less skilled immigrants. At the same time, they find that high income immigrants gain more from living in an ethnic enclave compared to those with low incomes. Cutler et. al. (2008) similarly find that for first generation immigrants' ethnic enclaves and segregation was useful. Also, in the American context, Beaman (2012) examines the dynamic consequences of social networks for the labor market outcomes of refugees. Beaman finds that an increase in the size of the network brings an effect where a one standard deviation increase of the number of network members in the year before the arrival of a refugee immigrant lowers his probability of being employed by 4.8 percent.
For the Canadian labor market, Warman (2007) study the effect from living in an ethnic enclave on the income increase of immigrants. The author finds generally negative effects on immigrant's weekly income increases.
Estimating the effect of enclaves comes with the obvious problem of potential endogeneity. It is hard to tease out how much of the estimated effect that come from ethnic peers in the neighborhood (Manski, 1993). The effect of ethnic enclaves and segregation risk being overestimated because of potential common factors that influence both the labor market outcome of the residents as well as the segregation as such. Another type of problem is what sort of bench-mark that should be used for assessing the effects on the labor market outcomes. A common way is to use natives too tease out if estimates are large or small. This approach may however be a bit problematic since the mechanisms at work may be very different between immigrants and natives' reasons for ending up in a specific area (Warman, 2007). An alternative is to look at several immigrant groups in the same framework. In the present study we follow this approach and study immigrants originating from both the Balkans and the Middle east.
At times there may be a "dark side" to close immigrant networks. An ethnic enclave can turn into an economic restraint by excluding network members from outside alternatives. Also, it may be hard from within the network to acquire the necessary skills needed for a successful integration into the labor market, e.g. language proficiency (Borjas, 2000). Concerning the dark side of ethnic enclaves and networks Borjas (2000) say that low-skilled individuals have a harder time realizing opportunities in the labor market outside the enclave. This lack of opportunity they substitute for with existing possibilities within the enclave. Skills such as relevant education is an important variable when assessing the effect on the labor market outcome of ethnic enclaves. Both Edin (2003) and Borjas (2000) find that for highly educated individuals there is no effects from living in a segregated neighborhood. In earlier work Borjas (1998) studied the connection between ethnicity and immigrants' choices of place to dwell and the choice of living in an ethnic enclave. Borjas (1998) provides a theoretical and empirical analysis of the determinants influencing the choice of whether to reside in a segregated residential area or not. The finding is that there is dispersion within and across ethnic groups concerning the probability of living in segregated neighborhoods. Such observation is important because it gives motivation to why one should separate different ethnic groups when studying the formation and effects of ethnic enclaves. Borjas (1998) also find that factors such as income, parenting skills and ethnic capital decides the ethnic mix of neighborhoods where people choose to live. Large income differences between groups have shown to increase segregation even further.

Data and empirical strategy
In this section we introduce our data and describe how we delineate neighborhoods (or enclaves) and how we capture neighborhood characteristics. We, finally, outline our empirical strategy to link neighborhood characteristics to individual outcome -that is getting a first job after arriving in Sweden. The empirical analysis is based on a full population registry of microdata on individuals maintained by Statistics Sweden. The data is a panel, covering the entire population in Sweden overthe period 1993 to 2015. We have access to a number of individual characteristics including the origin from where the individual came from. Note that all neighborhood characteristics are calculated based on this exhaustive dataset. Our aim is to disentangle the relationship between a number of individual and enclave level variables and the probability of individuals to get a job.
We restrict our analysis to immigrants from the Balkans, arriving in Sweden in 1993 and 1994, and immigrants from the Middle East, arriving in 2005 and 2006. Figure 1 shows migration patterns to Sweden of these two immigrant groups over the past 25 years and motivates the selection of the populations studied in the analysis. The Yugoslavia wars (and in particular the Bosnian War of 1992-1995) and the Second Iraq War (Third Gulf War, 2003-2011 led to a strong increase in immigration to Europe and also to Sweden.
The figure shows very different patterns over time for the two groups and thus allows us to evaluate heterogeneity of potential enclave effects across immigrant groups. In our empirical analysis, we focus on immigrants moving to Stockholm metropolitan region when coming to Sweden. The data allows us to track the individual immigrants over time.
As we are primarily interested in labor market integration, we keep individuals in our sample until the find their first job, and discard them afterwards. Finding a job is defined as receiving a strictly positive income (but we apply a more restrictive definition in the sensitivity analysis). Number of Immigrants Year Balkan Middle East individuals increases the availability of labor market related information in general, and job openings in particular. Theory on immigration and networks suggests that information may flow more easily between co-national peers, because ethnic and cultural peers are often members of the same social networks. New immigrants may thus benefit from labor market related information transmitted through this network. We account for this by introducing two variables related to the immigrant's own ethnic group. The first of these variables is the share of immigrants that belong to the own group. That is, the more peers you have around you the likelier it is that you can receive some useful information through this group. The second variable on ethnic peers captures the employment rate of this group.
Members of the ethnic network already active in the labor market are expected to have more valuable information to share. These variables can serve as proxies for information potentially available to the individual immigrant. Additionally, the intensity of labor market integration of the ethnic network may change the social norms in this group regarding working or work ethics, and may thus influence individual behavior (Bertrand et al., 2000). The more people that work in a neighborhood and the more of them that work belong to the same ethnic group, the likelier it is that there is some social pressure in favor of looking intensively for a job, and some stigma for those that have not yet been able to land a job.
Finally, we include the distance between the neighborhood and the central business district (CBD) of the municipality to control for possible knowledge spillovers that may come from the CBD. Since the CBD represents the central parts of the local community, the further away a neighborhood is located, the more peripheral it is relative to the other parts of the municipality. In essence, it means that we control for the neighborhoods place in the urban hierarchy.
The dataset allows as controlling for a variety of individual characteristics, like mobility (whether the individual has moved last year or at least once during the studied time period), gender, age (age, age squared, and whether the individual is within normal working age), family status (belong to a family and having children or being a single parent with children), education (seven categories regarding the highest level of education attainment) and the year of arrival to Sweden. We further include year-municipality fixed effects to account for business cycle effects and unobserved local labor market heterogeneity.  Notes: The figure relates the average wage income per worker (in 1,000 Swedish Crowns) in the location (1km x 1km grid cell) of the first place of residence of immigrants to the immigrants' educational attainments. The solid line denotes the median values of the average income, whereas the dashed lines denote the first and the third quartile, respectively. The educational attainment is taken from one year after the arrival. The respective categories refer to compulsory education of less than 9 years (1), compulsory education of nine years (2), upper secondary education of 2 years at most (3), upper secondary school of 3 years (4), post-secondary education of less than 3 years (5) and post-secondary education of 3 years or more (6). "NA" denotes that educational attainment is not available. Looking at the location characteristics of the 1 by 1 kilometer squares we observe some great differences between places. Average population size is about 3400 inhabitants, the smallest one only has one person living there while the biggest place has over 22 thousand people. The share of immigrants of the same region of origin is about 16 percent. Again, the variation between squares is staggering, going from virtually zero to 100 percent. The overall employment rate is about 60 percent, again accompanied by high variation between 0 and 96 percent. The mean employment rate of the immigrants' ethnic peers in the respective neighborhoods is only 44 percent and thus considerably lower, but variation across space is much larger compared to the general population. The distances from neighborhoods to municipality CBDs vary from zero to 34 kilometers with a mean of almost six kilometers.
( ) is a vector of parameters for neighborhood-specific (individual-specific) variables to be estimated.

Results: Enclave size, neighborhood quality and labor market integration
In this section we present the results from estimating a number of variants of the model introduced above. We start with a very sparse model and build up the main specification, and introduce various alterations in order to test the stability and robustness of our results later. All regressions are reported in tables 2 to 6. In

Main results
In the first three models reported in table 2 we include individual level characteristics only. One reason for this is to see how stable these coefficients are once we include location characteristics into the model. Additionally, we can get a sense of how changing the place of residence is correlated with getting a first job. Of course, moving from one neighborhood to another is a form of self-selection or sorting into certain locations. The risk is that such behavior may distort our results if high ability individuals (i.e. those with a high probability of getting a job) tend to move into enclaves with certain characteristics that we then misinterpret as enclave effects influencing the probability of its inhabitants on successfully finding jobs.
We first moving in all further estimations controls (at least partly) for the potential bias due to sorting by unobserved individual characteristics. Furthermore, we will revisit to this issue later and will provide additional results by splitting the sample into individuals who have changed their place of residence before finding their first job, and those who did not move before getting integrated into the labor market.
We now turn to including the location characteristics into the estimations. We do this also in a stepwise manner. First, we include more general neighborhood characteristics, namely the neighborhood population and the distance to the municipality CBD (model 4).
Both of these variables turn out insignificant in explaining an immigrant's probability of getting a first job. Next, we include the overall employment rate (model 5) and its coefficient turns out to be positive and highly significant. We interpret this result as a first evidence that it is not the (population) size, but the quality of the location that is important.
In the following, we include the two main variables of interest, namely the population share and the employment rate of the immigrant group. When including the share of the population belonging to the immigrant group alone the estimated parameter turns out to be significantly positive (model 6). Lastly, including both of them shows that both respective parameter estimates are positive and highly significant (model 7). In sum, we have found three neighborhood variables that seem important in explaining the probability that immigrants get a first job, namely the overall employment rate, the share of the respective immigrant group over all residents (enclave size), as well as the employment rate of the own immigrant group (enclave quality). Note that including neighborhood characteristics in models 4-7 hardly affects the parameter estimates of the individual characteristics, as reported in column 3 of table 2. Next, in table 3 we turn to other estimation techniques as alternative ways to account for a potential estimation bias due to sorting by unobserved characteristics and analyze the two immigrant groups separately. The last model estimated in table 2 is repeated in table 3 (model 7), as we take this to be our baseline results against which we compare other specifications to. Our baseline specification is estimated for immigrants from both ethnic groups, and includes all the explanatory variables measured in the year before we record the outcome variable. In the second specification (model 8), we use the same set of explanatory variables, but use the neighborhood characteristics of the immigrant's first place of residence in the year of arrival to Sweden. It can be argued that, for instance, employment in the immigrant group on the year of arrival is more exogenous. This does not mean that we can be sure that initial location is entirely exogenous. But for the subset of Balkan refugees, a random placement policy that allocated immigrants to municipalities was still in place. 1 In model 9, the location characteristics given at (t-1) are instrumented by the respective values at the year of arrival (t0). Model 10 and 11 use the same specification as model 7, but the observations are split into the two ethnic groups (Balkan and Middle East) and they are analyzed separately.
As before, we have two sets of explanatory variables. The first group describes the character of the location or neighborhood, and the second set consists of the individual level variables. The focus of the analysis is put on the location characteristics. In all four specifications, the size of the location in terms of population is not significant.
The share of the population consisting of the relevant ethnic group is positively related to the probability of finding a first job when analyzing both immigrant groups together, both when using the neighborhood characteristics of the previous year and initial location conditions. However, the coefficient using initial conditions is smaller. When we divide the observations into the two ethnic groups, we get heterogeneous results: For people originating from the Middle East the size of the own ethnic groups has a significant positive effect, while the parameter estimate for immigrants from the Balkans is negative, though not significantly different from zero.
The general employment rate in the location only come out significant when analyzing the two groups together and using previous years observations. The employment rate of the relevant immigrant group is significantly positive in all specifications. This means that the higher the share of your own group that are employed the higher is the probability that you will get your first job. For the people coming from the Middle East the own group has an effect through its size and its employment rate. For the Balkans only the second effect is present.
In no specification is the distance to the CBD significant. So, the probability of finding a first job is not influenced by the centrality of your place of residence. The individual level variables are mainly treated as controls in this analysis. Some of the results are worth mentioning though. Throughout all specification probabilities of finding a first job is lower for females. This penalty is especially strong for women from the Middle East. The next variable is working age. It is a dummy variable taking on the value 1 if the person is below retirement age. Normal retirement age in Sweden is 65 years of age. For the combined observations, the coefficient is positive and significant for both previous year and its initial value. When splitting the observation into the two ethnic groups we see that all of the effect seems to come from people from the Middle East for whom it is very significant. For people from the Balkans there is no significant effect. Turning to actual age there is a positive effect but at a decreasing rate since the linear part is positive and the quadratic part is negative. This goes for all specifications except the linear part for people originating in the Middle East. Comparing with the result for retirement age suggests that there is interaction going on between the dummy for retirement age and actual age.
Next, we turn to the family conditions in terms of living with a spouse and whether the individual has children below the age of 18. Overall living in a family and having children has a negative effect on the probability to get a first job. However, when we look at the different ethnic groups, we see that the signs are opposite and both statistically significant. For people from the Balkans the effect is positive and significant and for people from the Middle East the effect is negative and significant. Having children but being a single parent has a positive effect on the probability of finding a first job. This may be interpreted as a "push" effect. This effect is not present for people originating from the Middle East. For people coming from the Balkans the effect is significantly positive.
As before, a number of dummy variables are used as controls: We include municipality and year fixed effects, dummy variables indicating the immigrant's year of arrival to Sweden, as well as a set of dummy variables denoting the highest educational attainment.
The main results can be summarized as follows: The strength of the enclave effect depends essentially on the labor market success of the people living in the enclave.
Moreover, the most robust effect come from the employment share of the own immigrant group. This effect is positive and significant through all of the specifications so far. 118 Robust standard errors in parentheses (clustered at the individual level),*** p<0.01, ** p<0.05, * p<0.1 Notes: In model 8, location characteristics (i.e. population, share immigrant group, employment rate, employment rate immigrant group and distance to municipality CBD) are measured as initial values of year of arrival (t0). In model 9 the location characteristics given at (t-1) are instrumented by the respective values at the year of arrival (t0).
Next, we check robustness of results by adding incorporating a number of additional variables into our analysis that allows us to capture heterogeneity across different types of neighborhoods and different groups of individuals. In reporting these results, we focus exclusively on the results from the enclave level. For this reason, and for the sake of saving space we refrain from reporting the results from the individual level variables.

Effect of other immigrant groups
In this section our aim is to evaluate whether our results regarding the own immigrant group characteristics indeed stem from the immigrants' ethnic peers rather than from immigrants from other countries. In order to test this, we add two variables to our model, Note that including neighborhood information on immigrants from other countries hardly influences the estimated parameter on the employment rate of the own immigrant group. This can be interpreted as support for the notion that it is the quality as perceived by each group of immigrants that is essential for the probability of finding first job. 118 Robust standard errors in parentheses (clustered at the individual level),*** p<0.01, ** p<0.05, * p<0.1 Notes: In model 13, location characteristics (i.e. population, share immigrant group, employment rate, employment rate immigrant group and distance to municipality CBD) are measured as initial values of year of arrival (t0). In model 14 the location characteristics given at (t-1) are instrumented by the respective values at the year of arrival (t0).

Heterogeneous effects for different types of neighborhood
In this sensitivity analysis, we investigate whether our results are robust if we split the sample along neighborhood characteristics. We thus differentiate enclaves in two dimensions, namely on population density and the share of foreign population. Note that since we use one square kilometer squares as enclave's density is the same thing as size in terms of population. First, we compare the results between high-and low-density enclaves (or small and large enclaves in terms of population).
Regression results, reported in table 5, show that neighborhoods with high population density three variables on location characteristics are highly significant. These are the overall employment rate, the share of the population of the own immigrant group and the employment rate of the own immigrant group. For less densely populated regions parameter estimates are generally less significant. Only the share of people from the own immigrant group and their employment rate matter. The results confirm the finding from previous model specifications that the own immigrant group matters most. To allow for heterogeneous effects in more and less segregated areas we split the sample into enclaves with more than 40 percent of the residents born abroad, and in neighborhoods with less foreigners. We find positive parameter estimates for the overall employment rate and the share of the population of the own immigrant group in both samples, but the coefficients are only significantly different from zero in more segregated areas. The employment rate in the own immigrant group, however, is significantly positive in both samples. Somewhat surprisingly the distance to the municipality CBD is weakly significant and negative. This means that for these neighborhoods it is to some extent important to be centrally located.

Heterogeneous effects for different groups of individuals
In this last sensitivity analysis, we allow for heterogeneous effects for different individuals and split our sample along individual characteristics. We focus again on the issue of movement. In addition, we investigate individuals with different educational attainments separately and, lastly, we use a more restrictive wage threshold to assess whether an immigrant is successfully integrated in the labor market.
We first divide the population into those who changed their place of residence at least once before finding the first job ("movers") and those who did not ("stayers").
Starting with the movers (model 21) the significant variables are once again overall employment rate, share of the population belonging to the own immigrants' groups and employment rate in that group. The variable with highest significance level is the employment rate in the own group. For the group that never moved (model 22) results are generally weaker. For the first time the variable for enclave size (population) is weakly negatively significant, and the overall employment rate does not matter. The share of the immigrant group is significant, and the size of the coefficient is comparable to the movers.
The employment rate of the immigrant group is weakly significant, but the size of the coefficient is only about half the size compared to the respective parameter estimates for the movers.
Second, we divide the immigrants into three different educational groups (low, medium and high, see models 23 -25). The overall employment rate and the share of the immigrant group in the population have similar effects for all educational groups. For the employment rate in the own immigrant group we find positive coefficients for all three groups, but the estimated parameter is not significantly different from zero for the low education group. For the medium education group the effect seems strong and highly significant, but the parameter estimate for the high education group is considerably smaller and only weakly significant. These results suggest that the middle group is the one that benefits most from the co-nationals' labor market success.
We, finally, apply a more restrictive wage threshold: Immigrants are considered unemployed until they find a job with a gross annual income of at least 156,000 Swedish crowns (about 15,000 Euro). Using this wage threshold to assess successful labor market integration (model 26) the results suggest that the size of the neighborhood matters and that living in a larger neighborhood is beneficial. While the overall employment rate does not seem to be important, we find a significantly positive coefficient for both the share and the employment rate of the same immigrant group.

Conclusion
Since the great surge of refugee immigrants to Europe in the fall of 2015 the first topic of the public debate has been focused on immigrant integration. The most important form of integration is widely seen as entering the labor market. Policy makers thus focus on different schemes involving language training, education and labor market interventions.
In this paper analyze how ethnic enclave quality can explain the probability that an immigrant finds a job within the next year. It is well known that immigrants living in an ethnic enclave inhabited by co-nationals are influenced by their neighbors. Using a variety of model specifications, we show that the probability that an immigrant finds a job is related to enclave quality and -to a weaker and less robust extent -to enclave size. In this study neighborhoods are defined as exogenously given geographical areas in the form of grid cells of one square kilometer. The use of exogenously determined identically sized squares puts limits on some of the cross-enclave variation in different dimensions. Note 23 that in this setting scale (size, population) and density (people per square kilometer) are interchangeable, as both types of variables measured as per square-kilometer. In our empirical model neighborhoods have been characterized by population, immigrant population (by groups), and distance to the nearest CBD. Enclave quality is measured by the overall employment rate and by the employment rate of the relevant immigrant group. While controlling for many individual characteristics, the most robust result is that the employment rate of the immigrant group is a decisive factor in explaining the likelihood of finding a first job. The population share of co-nationals also plays a positive role, but this result is less robust. We contribute to the existing literature by emphasizing on enclave quality in addition to enclave size. Additionally, using two distinct immigrant groups allows us to infer on the size and importance of enclave effects across different immigrant groups. Our results suggest that enclave quality is more important than enclave size.