Abstract This study analyses the effect of immigrant concentration - - PDF document

1

Residential segregation and probability of being employed of recent immigrants in Montevideo 2011 Julieta Bengochea1 El Colegio de México, México Programa de Población, Facultad de Ciencias Sociales, Universidad de la República, Uruguay Abstract This study analyses the effect of immigrant concentration in neighbourhood of residence on the probability of being employed for recent immigrants in Montevideo, Uruguay (arriving in 2005-2011) who were born in Peru, Paraguay and Chile. Social and economic features of the settlement area have an effect on the kind of integration: those immigrants with less social and economic capital tend to settle in poorer neighbourhoods than those with more social capital. With the aim of elucidating the effect of neighbourhood of residence on the employability of recent immigrants, this study answers the following question: for recent immigrants, does the probability of being employed vary with residential concentration degree of immigrant population in their neighbourhood of residence and with their country of birth? Based on the Population Census of Uruguay in 2011, multilevel logistic regression models were estimated, which allow us to analyse the overall effect of individual and structural features on the probability of being employed. Keywords: Recent immigration. Residential segregation. Labour insertion of immigrants Introduction This study analyses the effect of immigrant concentration in the neighbourhood of residence on the probability of being employed for recent immigrants in Montevideo, Uruguay (arriving in 2005-2011) who were born in Peru, Paraguay and Chile. Social and economic features of the settlement area have an effect on the type of integration: those immigrants with less social and economic capital tend to settle in poorer neighbourhoods than those with more social capital. Even when both situations may imply residential segregation, both situations imply qualitatively different integration

• processes. None of them per se has a positive or negative effect on economic integration

since a higher concentration of immigrant population can provide networks facilitating labour insertion in the country of destination during the first years of settlement. In this regard, neighbourhood of residence is understood as a geographical unit of residential

• segregation. With the aim of elucidating if the concentration of recent immigrants in

their neighbourhood of residence in Montevideo, considered a geographical segregation unit, operates as a space of working opportunities or as a space constraining working

• pportunities, this study intends to answer the following question: for recent

immigrants, does the probability of being employed vary with residential concentration degree of immigrant population in their neighbourhood of residence and with their

1 Julieta Bengochea is PhD candidate in Population Studies at Colegio de México. She is teaching

assistant and researcher for the Program of Population at Facultad de Ciencias Sociales (School of Social Sciences), Universidad de la República, Uruguay.

2

country of birth? To do so, multilevel logistic regression models were estimated, with variables of level 1 considered as sex, age, age at migration, education level, marital status, country of birth, and variables of level 2 considered as coefficient of localization

• f immigrants for the 62 neighbourhoods in Montevideo.

Background Social integration of immigrants in the host country occupies a central role in debates

• n migration. Studies on such issue were initially conducted in order to understand

social inequities which would segregate foreign from native population. Concretely, the interest rises in the United States during anti-segregation fights of the twentieth century due to the formation of "ghettos" and poor "urban neigbourhoods" (Martori & Hoberg, 2004; Gans, 1997). Even when social integration of immigrated population is a multidimensional phenomenon, the main dimension to measure integration is related to the economic ability of self-sufficiency of immigrants. Although migration is a response to multiple motivations and enables diverse kinds of flows (forced migration, labour migration, study migration, highly skilled migration or circularity, family migration, etc.), labour is the main component of flows. Thus, it makes sense to analyze labour insertion as the central dimension of integration. Moreover, employment and its characteristics determine other aspects of social life such as social security benefits or accessing to housing and to other social services (Niessen & Schibel, 2004). Immigrants

• ften present high activity rates (Alarcón & Ramírez-García, 2011); however, it does

not imply a successful insertion in the labour market if conditions are poor or jobs imply low wages, low skills and menial social status (Alarcón & Ramírez-García, 2011). Factors such as age (immigrants are younger on average than native population) and the importance of labour-related motivations for migration lead many immigrants to be less selective, more vulnerable, and to accept disadvantageous working conditions, which explains their high participation in local markets (Cerrutti, 2009). Other important factors to understand high rates of immigrant occupation are the segmentation of labour markets, the precarization of work and the existence of specific niches where immigrants work. The theory of dual labour markets postulates that labour markets are divided in a sector for native workers and another for immigrants (Piore, 1979, 1986; Massey et al., 1993). Piore, in his book Birds of Passages (1979), analyses circular migration of workers from undeveloped regions to developed ones, and elaborates the theory of dual labour markets. This theoretical perspective focuses on structures of job opportunities at destination which, through segmentation of the labour market, comprise jobs in the secondary sector of industry which were rejected by native workers employed in the primary sector. In this regard, the perspective identifies diverse factors shared among immigrants and relates them to the economic and industrial structure of destination

• countries. In the same line of reasoning, Portes develops the concept of economic

enclave which argues that immigrants are employed in given labour environments depending on their origin (Piore, 1986). For instance, Mexican immigrants in the United States are employed in the secondary sector of the labour market with lower wages than native workers, while Cuban immigrants are employed in their own commercial network developed by the first generations of immigrants (Piore, 1986). Alarcón and Ramírez-García (2011) show it among immigrant population in Los Angeles city: Mexicans and Central Americans are employed in low-skill and low-wage activities, whereas Europeans and Asians are

3

employed in qualified jobs. Moreover, they find that the native workers sector is characterized by higher stability and income, while the immigrant sector is characterized by low wages, labour precariousness and low skills (Alarcón and Ramírez-García, 2011). For its part, urban neighbourhood conformation has an important role in reproducing social inequities, social exclusion and poverty, in the sense that these spaces have specific structures of opportunities (Kaztman and Retamoso, 2007; Kaztman, 2001) and specific access to education and health services (Arim, 2008), among others. Thus, urban neighbourhoods vary notoriously in relation to dimensions such as crime rate, health level and income. There are two epistemic schools on how associations between such dimensions and residential segregation are produced. One of them argues that they are related to familiar and individual context (Verbitsky Savitz and Raudenbush, 2009), and the other that they are a structural consequence of the urbanization which generates social stratification of neighbourhoods (Sampson, 2008; Sampson, Raudenbush and Earls, 1997). Taking this into account, this study considers the neighbourhood of residence as a geographic unit of residential segregation. As well, immigrants tend to settle unevenly spread in the territory, which can imply two different situations concerning the process of integration. On the one hand, it can negatively affect immigrant integration due to social and economic exclusion of the areas they live in, and imply a downward assimilation (Logan, Zhang and Alba, 2002; Vono and Bayona, 2010). On the other hand, arriving at an area with existing social contention networks can make it easier for immigrants to get integrated socioeconomically, and implies an upward assimilation (Logan, Zhang and Alba, 2002; Vono and Bayona, 2010). Concentration of immigrants in an area of the territory can be a strategy of movement of migration networks which facilitates the adaptation and settlement in the host society, which implies a positive process of integreation to the host society and not only a procces that reproduces poverty concentration (Arriagada, 2010). Arriagada (2010) develops a comparative study on residential segregation among immigrants in Montreal and Santiago de Chile. He suggests a typology of six categories: isolated communities, non-isolated communities, assimilation enclaves, mixed enclaves, polarization enclaves and ghettos. The result of his work is a consistent example of how public policy concerning residential isolation of immigrants impacts directly on what kind of social and economic integration they experience. Vono and Bayona (2010) study the settlement process of Latin Americans in Spain and they argue that the detailed study of territory segregation differentiates the process of formation of a ghetto from that of an ethnic enclave. The first one implies a stronger structural segregation, high poverty and unemployment levels, and discrimination factors against that population. The latter, the ethnic enclave, implies less territory concentration, less poverty levels, higher activity rates and a sort of internal cohesion, and it is a space of immigrant transit (Vono and Bayona, 2010). Clearly, social and economic characteristics of the settlement area affect directly the type of integration: immigrants with less social and economic capital will surely settle in poorer neighbourhoods and those with more social and economic capital will do so in richer neighbourhoods. Nonetheless, a higher residential segregation, considered as a concentration of immigrants in a neighbourhood of residence, can provide them with networks which would enable them to be employed in the host society on arrival. In this study, the double effect of such phenomenon is investigated.

4

Recent immigration in Uruguay Based on data from the 2011 population census in Uruguay, Koolhaas and Nathan (2013) find a steady increase in the stock of returnees in the previous five years and an increase of Latin American immigrants. In Uruguay, immigrant population has significantly decreased in its stock, as well as in its relative weight. Immigrants were 3% (92,378 individuals) of the total population in 1996 and 2.4% (77,003 individuals) in 2011 (Koolhaas and Nathan, 2013). From 1996 to 2011 the number of traditional immigrants (Spanish and Italian) decreased, while the remaining, including immigrants from the southern region, increased. In 2011, most of them came from Argentina (35%), Brazil (17%), Spain (16%) and Italy (7%), whereas those from Paraguay, Peru and Chile represented 2% of the total each. In terms of year of arrival, there is an increase of foreigners from 1996 to 2011 when comparing the period of five years previous to each census, although it has to be considered that 1996 census had more missing data for this

• information. The total of recent immigrants in the 2011 census was 15,842 and for 1996

it was 13,3452 (Bengochea, 2014). Interestingly, when country of birth is considered, 50% of Peruvian, 33% of Chilean and 27% of Paraguayan immigrants arrived in Uruguay between 2005 and 2011. It means that Peruvian, Chilean and Paraguayan immigrants have increased in number in the last years, even when the weight of immigrants of traditional and border origin is the most important in Uruguay. With the aim of further understanding the three groups of immigrants, this study focuses on the probability of employment of Paraguayan, Peruvian and Chilean immigrants arriving at Uruguay between 2005 and 2011. What is the level of participation of recent immigrants in the labour market? Due to selectivity of migration towards active ages and in order to control for the effect of diverse age structure of populations among the rates compared, calculations are applied to the group of individuals between 20 and 39 years of age.3 Roughly, the highest unemployment rate (UR) can be found for recent Chilean immigrants —men and women—, as compared to native population and total recent

• immigrants. On the contrary, among recent immigrants, Paraguayan men and Peruvian

women show the lowest rates, as compared to native population and total recent

• immigrants. The highest employment rates (ER) are shown by recent Peruvian

immigrants —men and women— and the lowest by Chileans. Native population show the highest activity rate (AR), followed by recent Peruvian immigrants whose rate is lower than for total recent immigrants. Among women, Paraguayans show the highest AR, followed by Peruvians and native population. Among men, recent Peruvian immigrants show the highest AR followed by native population.

2 Total does not comprise immigrants younger than five years old. 3 This range of age was chosen because it groups enough cases so as to calculate rates without problems

• f statistical representativeness.

5

Table 1 Unemployment, employment and activity rates for population between 29 and 31 years

• ld by country of birth. 2011 census

Unemployment rate (UR) (%) Employment rate (ER) (%) Activity rate (AR) (%) Men Women Total Men Women Total Men Women Total Chile 5.9 13.8 9.7 85.1 58.5 70.3 90.4 67.8 77.8 Paraguay 1.6 10.5 7.4 87.1 69.9 75.5 88.6 78.1 81.5 Peru 2.9 7.0 5.0 91.4 70.6 79.6 94.1 75.9 83.8 Recent immigrants Total 4.7 12.7 8.5 84.7 57.1 69.5 89.0 65.4 76.0 Native population Total 4.6 10.1 7.1 88.8 68.2 78.3 93.1 75.9 84.3 Source: This study, analysis based on census data from INE 2011.

With the purpose of further analyzing the data shown, they are studied in relation to immigrant education level. Recent Peruvian immigrants show the highest ER, followed by native population, whereas recent Chilean immigrants show the lowest

• ER. Chilean immigrants present a higher education level than Paraguayan and Peruvian

immigrants (Bengochea, 2014). Taking these results into account, and considering education as the provision of abilities for the performance in the labour market, it is hypothesized that higher employment rates of recent Peruvian and Paraguayan immigrants are due to their access to lower quality jobs in the informal sector. Meanwhile, lower employment rates for recent Chilean immigrants are due to their higher education level, which implies that they aim to access qualified jobs and a delay in their employment. Even when this statement is a hypothesis, Diconca (2012) and De los Campos and Paulo (2001) give evidence that Paraguayan and Peruvian immigrants work in the fishing sector and domestic services. What neighbourhoods in Montevideo do recent immigrants live in?4 Montevideo started a process of residential segregation in the eighties showing an increase in concentration of poor homes in poor neighbourhoods along with social and educational segmentation (Kaztman and Retamoso, 2007; Kaztman, 2001). It was somehow related to the fact that state schools lost their integrative feature, due to increasing privatization of educational centers (Kaztman and Retamoso, 2007; Kaztman, 2001). Increasing residential segregation in Montevideo in the last decades of the twentieth century led to changes in its urban configuration (Arim, 2008), a process further advanced during the economic crisis that the country went through between 1999 and 2002 (Arim, 2008; Veiga, 2005). Among the effects of the crisis, income regional distribution, as well as precarization of labour market and life conditions are

• ften mentioned (Veiga, 2005). In this context, the neighbourhoods of Montevideo tend

to present increasing inequities among areas and homogeneity within them (Arim, 2008; Kaztman, 2001), which produces a kind of “region effect” or poverty feedback (Arim, 2008).

4 Hereafter, reference is made to the analysis of recent immigrants in Montevideo since the question about

neighbourhood of residence was posed only for the capital city of the country.

6

Map 1 Percentage distribution of native population by neighbourhood of residence. 2011 census

Percentage distribution of native population 0,7 - 0,9 0,9 - 1,1 1,1 - 1,5 1,5 - 2,1 2,1 - 3,2 3,2 - 5,0

²

Source: This study, analysis based on census data from INE 2011.

Map 2 Percentage distribution of recent immigrants by neighbourhood of residence in

• Montevideo. 2011 census

Percentage distribution of recent immigrants

0,3 - 0,6 0,6 1,0 1,0-1,6 1,6 - 2,5 2,5 - 5,0 5,0 - 15,1

²

Source: This study, analysis based on census data from INE 2011.

7

As shown in maps 1 and 2 the patterns of settlement throughout the neighbourhoods in Montevideo differed between native population and recent

• immigrants. Recent immigrants are concentrated in neighbourhoods of the coastal area
• f Montevideo; a higher proportion in Pocitos, then in Punta Carretas and Carrasco, and

to a lesser extent in Ciudad Vieja, Centro, Cordón and neighbouring areas. Native population is located mainly in Pocitos and then in Casavalle, La Paloma, Tomkinson, Buceo, Unión and Cordón. Table 2 Five main neighbourhoods of residence of recent Peruvian, Paraguayan and Chilean

• immigrants. 2011 census

Paraguayans (%) Chileans (%) Peruvians (%) Pocitos 20.5 Pocitos 17.0 Ciudad Vieja 24.5 Carrasco 9.9 Carrasco 10.3 Centro 12.1 Punta Carretas 6.2

• Capurro. Bella Vista

7.6 Pocitos 8.9 Cordón 5.5 Cordón 4.5 Carrasco 8.2 Malvín 5.5 Carrasco Norte 4.5 Cordón 7.2 Others 52.4 Others 56.1 Others 39.1 Total 100 Total 100 Total 100 Source: This study, analysis based on census data from INE 2011.

There are also differences in the neighbourhood of residence among the country

• f birth of immigrants (Table 2). The main difference is the settlement pattern of

Peruvian immigrants since they are mostly located in the central and port area of Montevideo (Ciudad Vieja and Centro), whereas Chilean and Paraguayan immigrants are mainly located in coastal neighbourhoods Pocitos and Carrasco. For Peruvian immigrants, one out of four lives in Ciudad Vieja, one out of eight in Centro, 9% in Pocitos and 8% in Carrasco. Recent Paraguayan immigrants are mainly located in Pocitos (21%) and Carrasco (10%), while only 3% live in Ciudad Vieja. Seventeen percent of recent Chilean immigrants live in Pocitos and 10% in Carrasco, and those living in Ciudad Vieja have no significant percentage weight. Table 3 Duncan index of dissimilarity. Total of immigrants and recent immigrants by country

• f birth

Total of immigrants Recent immigrants Peru 0.466 0.551 Chile 0.323 0.491 Paraguay 0.269 0.448 Italy 0.195 0.446 Brazil 0.201 0.433 Argentina 0.158 0.363 United States 0.306 0.285 Spain 0.231 0.214 Total of immigrants 0.207 0.373 Source: This study, analysis based on census data from INE 2011.

8

Values for Duncan index of dissimilarity (ID)5 suggest residential segregation by country of birth of immigrants (Table 3). They also suggest higher residential segregation among recent immigrants than among the total of immigrants, which may show that immigrants settle in specific enclaves on arrival. Residence time in the destination country is key to analyze the socioeconomic integration, and thus it is expected for recent immigrants to present higher ID values. Likewise, ID values show that recent and total immigrants from the southern region present higher residential

• segregation. Firstly, Peruvian immigrants show the highest ID values and the lowest gap

for ID values between total and recent immigrants. In this regard, 47% of the total of Peruvian immigrants would have to transfer between neighborhoods to reach evenness in residential segregation, and such percentage increases to 55% for recent immigrants. Secondly, for the total of Chilean immigrants the ID value is 0.323, and 0.491 for recent

• immigrants. Thirdly, for recent Paraguayan immigrants the ID value was 0.448. A total
• f 45% of recent Paraguayan immigrants and 49% of recent Chilean immigrants would

have to transfer between neighbourhoods of residence to reach evenness in territory

• distribution. Bengochea (2014) finds that recent immigrants living in Ciudad Vieja

(62%) and Casavalle (60%) are those with the highest percentage of population presenting an uncovered basic need (UBN). On the other hand, immigrants living in Carrasco (3%) present the lowest percentage of UBN of the population, and there is a gap of 59 percentage points between immigrants living in Carrasco and Ciudad Vieja. The case of Ciudad Vieja is paradigmatic since six out of ten recent immigants have at least one UBN (Bengochea, 2014). Tables 4, 5 and 6 show the five main neighbourhoods of residence for Peruvian, Chilean and Paraguayan immigrants, and poverty incidence measured as percentage of poor homes in each neighbourhood (MIDES, s/d). Table 4 Five main neighbourhoods of residence for recent Paraguayan immigrants, and poverty

• incidence. 2011 census

Percentage of residents in the neighbourhood Percentage of poor homes 1 Pocitos 20.5 1 2 Carrasco 9.9 1 3 Punta Carretas 6.2 1 4 Cordón 5.5 5 5 Malvín 5.5 2 Source: This study, analysis based on census data from INE 2011 and available information from MIDES (s/d).

5 Duncan index of dissimilarity measures whether the distribution of a minoritary group in a territory is

even or not: if not, it is territorially segregated (Martori and Hoberg, 2004). ID take values between 0 and 1, indicating minimal and mamximal residential segregation respectively. If taken as a percentage, it indicates the number of individuals that would have to change their neighbourhood of residence in order to obtain evenness of distribution (Martori and Hoberg, 2004)

| ) / ( ) / ( | ) 2 / 1 (

1

Y Y X X I

i n i i D

 

 

9

Table 5 Five main neighbourhoods of residence for recent Chilean immigrants, and poverty

• incidence. 2011 census

Percentage of residents in the neighbourhood Percentage of poor homes 1 Pocitos 17.0 1 2 Carrasco 10.3 1 3 Capurro, Bella Vista 7.6 7 4 Cordón 4.5 5 5 Carrasco Norte 4.5 9 Source: This study, analysis based on census data from INE 2011 and available information from MIDES (s/d).

Table 6 Five main neighbourhoods of residence for recent Peruvian immigrants, and poverty

• incidence. 2011 census

Percentage of residents in the neighbourhood Percentage of poor homes 1 Ciudad Vieja 24.5 11 2 Centro 12.1 2 3 Pocitos 8.9 1 4 Carrasco 8.2 1 5 Cordón 7.2 5 Source: This study, analysis based on census data from INE 2011 and available information from MIDES (s/d).

Results show that none of the five main neighbourhoods of residence accumulates the highest percentage of poor homes in Montevideo (over 34%) (MIDES, s/d). However, results differ for recent Peruvian immigrants; they live mainly in Ciudad Vieja, a neighbourhood with 11% of poor homes. Values of ID analyzed by UBN and poverty incidence in the main neighbourhoods of residence for immigrants of the southern region suggest that recent Peruvian immigrants show characteristics of unfavourable settlement. These data also show a pattern of settlement for recent immigrants which varies among them according to their country of birth. Such pattern implies that recent Peruvian immigrants settle mostly in the central region of the capital city; they live mainly in Ciudad Vieja, a neighborhood which concentrates the highest percentage of poor homes among the neighbourhoods where recent immigrants settle. They also present the highest incidence of population with at least one UBN, particularly related to housing (Bengochea, 2014). In this regard, the settlement pattern

• f recent Peruvian immigrants could be showing an integration process of a descendent

kind (Kaztman, 2001). It would be so, under the understanding that residential segregation works as a poverty feedback mechanism, which Arim (2008) identifies as a poverty trap and which is also expressed in the labour market in diverse areas of Montevideo.

10

Methods Investigation questions With the aim of elucidating if the neighbourhood of residence, considered a unit

• f geographical segregation, operates among recent immigrants either as a space of

working opportunities or as a space constraining working opportunities, this study intends to answer the following questions: Does the level of residential concentration of immigrants affect the probability of employment of recent immigrants? does the country

• f birth of recent immigrants affect their probability of employment? And the following

general question: for recent immigrants, does the probability of being employed vary with residential concentration degree of immigrant population in their neighbourhood

• f residence and with their country of birth?

Justification of the method Immigrant labour insertion in the host country is due to individual and structural

• characteristics. Among others, individual characteristics refer to sex, age, education

level, residence time at destination country and country of birth. Meanwhile, neighbourhood of residence at destination country can be found among the structural characteristics, which has certain economic, cultural and social features since it is a geographical unit and a social space. Thus, the process of integration of immigrants cannot be analyzed in relation to individual characteristics exclusively; structural aspects should also be taken into account. In this regard, multilevel analysis of the probability of being employed among immigrants in Uruguay is the most suitable method to study the effect of individual and structural characteristics on such probability. Data: variables and information source Chart 1 shows the independent variables estimated using models, and their expected effect on the dependent variable as hypothesis. Chart 1 Independent variables and expected effects

Variable Hypothesis Sex H1: The probability of employment is higher for men than for women. Age H2: The probability of employment increases with age, though at a negative rate. Age at migration H3: The probability of employment increases with residence time at destination. Conjugal union H4: Being in a mixed union increases the probability of being employed since such type of union implies a higher level of integration in the destination country Education level H5: Education level among immigrants does not have an effect on the probability of being employed since immigrants are less selective than the native population when choosing a job, and present more difficulties when trying to validate their education certificates. Country of birth H6: The probability of employment differs among countries of birth due to segmentation of labour markets at destination countries. Coefficient of localization H7: The probability of employment increases with the value of the coefficient of localization

• f the neighbourhood of residence, since immigrants have spatially segregated networks.

11

Data come from the 2011 population census in Uruguay. The universe of analysis for estimated multilevel logistic regression models includes all the people born

• utside Uruguay who declared arrival at the country between 2005 and 2011, were

between 15 and 64 years old by 2011, and were residing at the capital city (Montevideo).6 Thus, the universe is comprised of 5,716 individuals (recent immigrants) residing in 62 neighbourhoods of Montevideo. The dependent variable to be estimated is the probability of being employed, versus the probability of not being employed. Independent variables are divided into two levels. Variables in level 1 are: 1) sex 0=woman, 1=man), 2) age (continuous), 3) age at migration (continuous), 4) education level (no education, preschool, primary, secondary, tertiary), 5) country of birth (Peru, Chile, Paraguay and others), and 6) conjugal union (endogamous7, mixed8, single, separated, divorced, widow or widower). The variable in level 2 is the coefficient of localization (CL) for the total of immigrants in the 62 neighbourhoods in Montevideo, as a structural measure of spatial concentration of immigrants. Immigrant concentration in each neighbourhood is compared to the average concentration for the total population by dividing the percentage of immigrants in each neighbourhood over the percentage of immigrants in Montevideo. Values are equal or greater than zero; when CL is equal or greater than 1 it means there is a higher concentration of immigrants in the neighbourhood than proportional, whereas if CL is less than 1 it means such concentration is less than proportional. For further interpretation of the effect of CL

• ver the probability of being employed, the coefficient was transformed into a

dichotomic variable: 1 denotes values equal or greater than 1, and 0 denotes values lower than 1. In this study, values ranged between 0.248 and 3.043 (Table 7). The lower value

• f the range was obtained for a neighbourhood with a fourth of the immigrant

concentration value for the city of Montevideo. Meanwhile, the upper value triples the average value. The index IC is expressed as:

6The analysis was focused in this range of ages since it is key to labour activity. Even when ages can take

values as low as 12, and without an upper limit, only few values fell out of the selected range.

7An Endogamous union is composed by immigrants of one country. 8A Mixed union is composed by immigrants of two countries.

12

Table 7 Summary of variables used in model estimation

Variable Media Standard deviation Mínimum Maximum Probability of being employed (1=employed, 0=unemployed) 0.671 0.470 0.000 1.000 Sex (1=man, 0 =woman) 0.485 0.500 0.000 1.000 Age (continuous) 34.800 11.788 15.000 64.000 Age2 (continuous) 1349.953 902.837 225.000 4096.000 Age at migration (continuous) 32.612 11.867 9.000 64.000 No education or preschool 0.006 0.079 0.000 1.000 Primary education 0.034 0.181 0.000 1.000 Secondary education 0.294 0.456 0.000 1.000 Tertiary education 0.666 0.472 0.000 1.000 Born in Chile 0.039 0.194 0.000 1.000 Born in Paraguay 0.037 0.189 0.000 1.000 Born in Peru 0.082 0.275 0.000 1.000 Born in other countries 0.842 0.365 0.000 1.000 Endogamous union 0.325 0.468 0.000 1.000 Mixed union 0.272 0.445 0.000 1.000 Single 0.311 0.463 0.000 1.000 Separated, divorced or widow(er) 0.092 0.289 0.000 1.000 Neighbourhood of residence 33.486 18.626 1.000 62.000 Coefficient of localization 1.679 0.728 0.248 3.043 Coefficient of localization (1=concentrated, 0=not concentrated) 0.842 0.365 0.000 1.000 Coefficient of localization*Born in Peru 0.070 0.256 0.000 1.000 Coefficient of localization*Born in Paraguay 0.030 0.169 0.000 1.000 Note: Calculations are based on 5,719 observations. Source: This study, analysis based on census data from INE 2011

Analysis Table 8 shows all estimated models obtained to reach the final model defined by the following equation: logit (Probability of being employed ij) = γ00+ γ10Sex ij + γ01Age ij + γ20 Squared Age ij + γ02Age at migration ij + γ30Mixed union ij + γ03Single ij + γ40Separated, divorced or widow(er) ij + γ50Education level primary ij+ γ05 Education level secondary ij+ γ60 Education level tertiary ij + γ70 Paraguay ij + γ07 Peru ij + γ80 Chile ij + γ11I CL Coefficient of localization of neighbourhood. j + γ11 CL Coefficient of localization of neighbourhood. j * γ07 Peru ij+ γ11 CL Coefficient of localization of

• neighbourhood. j * γ70 Paraguay ij+ U1j Peru ij + U2j Paraguay ij +U0j + eij

Even when coefficients in table 8 are shown as logit of multilevel logistic regression, they are interpreted in terms of chances [((exp (βx)-1)*100] of occurrence of the event.

13

Table 8 Logit multilevel model for probability of being employed (logit coefficient) of recent immigrants between 15 and 64 years old. Montevideo, 2011

Empty Model Model 1 Model 2 Model 3 Model 4 Model 5 Sex (1=man, 0=woman) 1.355*** [0.0680] 1.398*** [0.0687] 1.400*** [0.0687] 1.409*** [0.0690] 1.409*** [0.0690] Age (continuous) 0.482*** [0.0254] 0.470*** [0.0256] 0.471*** [0.0256] 0.473*** [0.0257] 0.473*** [0.0257] Age2 (continuous)

• 0.00488***

[0.000229]

• 0.00480***

[0.000230]

• 0.00480***

[0.000230]

• 0.00481***

[0.000231]

• 0.00481***

[0.000231] Age at migration (continuous)

• 0.0852***

[0.0173]

• 0.0782***

[0.0174]

• 0.0798***

[0.0174]

• 0.0812***

[0.0175]

• 0.0813***

[0.0175] Conjugal union (Ref.: endogamous union): Mixed union (1=mixed union, 0=other) 0.398*** [0.0874] 0.432*** [0.0875] 0.464*** [0.0879] 0.473*** [0.0888] 0.472*** [0.0888] Single (1=single, 0=other) 0.607*** [0.0969] 0..574*** [0.0977] 0.573*** [0.0976] 0.580*** [0.0987] 0.576*** [0.0988] Separated, divorced or widow(er) (1=sep., div., wid, 0=other) 1.359*** [0.148] 1.316*** [0.148] 1.334*** [0.148] 1.329*** [0.149] 1.325*** [0.149] Education level (Ref.:no education and preschool): Primary (1=primary, 0=other) 0.410 [0.424] 0.341 [0.425] 0.399 [0.427] 0.385 [0.428] 0.380 [0.429] Secondary (1=secondary, 0=other) 0.308 [0.390] 0.295 [0.392] 0.325 [0.393] 0.313 [0.393] 0.312 [0.394] Tertiary (1=tertiary, 0=other) 0.530 [0.388] 0.626 [0.389] 0.611 [0.390] 0.621 [0.390] 0.621 [0.391] Country of birth (Ref.: other immigrants): Chile (1=Chilean, 0=other) 0.148 [0.167] 0.132 [0.167] 0.135 [0.167] 0.136 [0.167] Paraguay (1=Paraguayan, 0=other) 0.703*** [0.177] 0.703*** [0.177] 0.643*** [0.222] 0.631 [0.417]

14 Peru (1=Peruvian, 0=other) 1.028*** [0.140] 1.016*** [0.139] 0.789*** [0.197] 0.524 [0.340] Coefficient of localization (1=concentration higher than proportional, 0=evenness) 0.329*** [0.108] 0.297*** [0.108] 0.280** [0.111] Coefficient of localization *Peru 0.391 [0.411] Coefficient of localization *Paraguay 0.0547 [0.496] Constant β0 0.617*** [0.0537]

• 8.125***

[0.541]

• 8.188***

[0.544]

• 8.416***

[0.551]

• 8.425***

[0.553]

• 8.401***

[0.555] Variance components: Var (_cons) 0.0720** 0.0790** 0.0542** 0.0390** 0.0349** 0.0346** Var (Peruvians) 0.3572** 0.2747** Var (Paraguayans) 0.4206** 0.4076** Cov (Peruvians, Paraguayans) 0.3812 0.3299 Cov (Peruvians, _cons)

• 0.0004
• 0.0034

Cov (Paraguayans, _cons)

• 0.0224
• 0.0238

% Correlation within class (rho1) 2.1 2.3 1.6 1.2 LR test vs. linear model: 0.000 0.000 0.000 0.003 0.001 0.004 Deviance

• 6021. 1

5949.3 5940.5 5932.5 5931.6 AIC 6045.1 5979.3 5972.5 5974.5 5977.6 Note: based on 5,719 observations and 62 groups (neighbourhoods) Standard errors in square brackets / *** p<0.01, ** p<0.05, * p<0.1 Source: This study, analysis based on census data from INE 2011.

15

The empty model was estimated to analize if probabilities of being employed vary among neighbourhoods. The positive sign of the coefficient denotes that individuals are more likely to be employed than not, on average and across

• neighbourhoods. Logit of individuals to be employed was 0.617 for average typical
• effects. LR test shows higher goodness of fit for the multilevel logistic model than for a

logistic model, evidencing the effect of neighbourhoods on the logit of being employed; thus, it was decided to conduct the multilevel logistic model. In Model 1, the following level 1 variables were added: sex, age, squared age, education level and conjugal union. Results show that being a man increases the chances of being employed, if the remaining variables are kept constant. The affect of age is also significant and positive, indicating that for each additional year of age the chances of being employed also increase, if the remaining variables are kept constant. Nevertheless, the negative sign of the quadratic term of age shows that the effect is positive with a negative rate, i.e. after a given inflexion point the positive effect of age

• n the probability of being employed starts decreasing. The age at migration shows a

significant and negative effect, which means that for each additional year of age chances

• f being employed decrease, if the remaining variables are kept constant. Variables

accounting for conjugal union are positive and significant. In this regard, the fact of being part of a mixed union, separated, divorced or widow(er) increases the chances of being employed, as compared to being part of an endogamous union, if the remaining variables are kept constant. Variables related to education level are not significant in the estimated model. Lastly, it is worth noting that there is still unexplained variability in the chance of being employed once it is controlled for level 1 variables, as shown in table 5. Model 2 was estimated using level 1 variable "country of birth" of recent immigrants, with the aim of answering the question: Does the country of birth of recent immigrants affect their probability of employment? Results of the model indicate that it does, and LR test in relation to model 1 shows that the loss of parsimony equals the gain in goodness of fit. Keeping constant the effect of the rest of the variables, the fact of being Peruvian increases 179.4 times on average the chances of being employed. Being Paraguayan increases such chances 102 times on average, as compared to other immigrants and keeping constant the rest of the variables. In the model estimated, there is no statistically significant evidence to argue that being born in Chile affects in any way the probability of being employed. Model 3 adds the coefficient of localization of immigrants in the neighbourhoods as level 2 variable in order to answer the following question: Does the level of residential concentration of immigrants affect the probability of employment of recent immigrants? Given that the coefficient of localization is significant and positive, the answer is that it

• does. Thus, the fact of living in a neighbourhood with a higher concentration of

migrants than proportional increases the chances of being employed, as compared to being unemployed, and according to the estimated model, where the rest of the variables remain constant. LR test estimated, in relation to model 2, rejects H0, i.e. the loss of parsimony due to the addition of level 2 variable is gained in goodness of fit. The process conducted from the nule model to model 3 shows that the latter includes the appropriate variables to contrast the hypothesis.

16

Model 4 aims to answer if the country of origin of recent immigrants has an effect on the probability of them being employed, and depending on the neighbourhood

• f residence. This model, of random slopes, allows variation of the slopes associated to

country of birth for Peruvian and Paraguayan immigrants. The decision of using such model is based on three aspects. Firstly, model 2 and model 3 provide no statistical evidence to argue that being born in Chile affects the probability of being employed. Secondly, it implies a theoretical decision; since the descriptive analysis showed that Peruvian and Paraguayan immigrants are more residentially segregated according to ID, the aim is to determine the effect for those two groups. Thirdly, the estimated model shows an AIC9 value slightly higher than model 3 of fixed slopes, though it also shows a lower deviance value. Model 4 maintains the sign and significance of estimated coefficients of previous models. Results show that the coefficient for sex is positive and significant, which means that being a man increases 309.2 times on average the chances

• f being employed versus being unemployed, when the rest of the variables are kept
• constant. The coefficient of age is positive and significative, meaning that the chances

for recent immigrants to be employed increased 60.5 on average for each year older they are, when the rest of the variables are kept constant. However, the quadratic term for age is significant and negative, which means that after a given age such effect on the probability of being employed decreases. Coefficients associated to conjugal union are significant and positive. In particular, being in a mixed union, as compared to being part

• f an endogamous union, increases 60.5 times on average the chances of being

employed versus being unemployed, when the rest of the variables are kept constant. The condition of being single increases 78.6 times on average the chances of being employed as compared to being part of an endogamous union, and if the rest of variables are kept constant. The condition of being separated, widow(er) or divorced, as compared to being part of an endogamous union, increases 36.7 times on average the chances of being employed, and if the rest of variables are kept constant. Coefficients for being born in Peru and Paraguay are positive and significant. In the estimated model, the coefficient for being born in Chile does not show statistical evidence of having an effect on the probability of being employed. Concretely, the effect of being a recent Paraguayan immigrant, as compared to other nationalities, increases 90.2 times

• n average the chances of being employed versus being unemployed, if the rest of

variables are kept constant. Meanwhile, the effect of being a recent Peruvian immigrant, as compared to other nationalities, increases 120.1 times on average the chances of being employed versus being unemployed, if the rest of variables are kept constant. The coefficient of localization for recent immigrants in the neighbourhoods of Montevideo is significant and positive. It indicates that living in a neighbourhood with a higher concentration of immigrants than proportional, as compared to living in a neighbourhood witha a proportional concentration of immigrants, increases 34.63 times

• n average the chances of being employed versus being unemployed, if the rest of

variables are kept constant. Coefficients related to education level are not significant, so the estimated model does not show association between the level of education of recent immigrants and their probability of being employed. Given that the variance of the random term of Peruvians and Paraguayans is significant, the following hypothesis is confirmed: the country of birth has a varying effect on the probability of being employed, depending on the neighbourhood of residence, i.e. the effect of being Peruvian or Paraguayan on the probability of being employed is not the same for all neighbourhoods.

9 Akaike Information Criterion

17

Lastly, with the aim of assessing whether the effect found by model 4 is related to residential concentration of immigrants in the neighbourhood, model 5 was estimated in order to answer the general question of this study: for recent immigrants, does the probability of being employed vary with residential concentration degree of immigrant population in their neighbourhood of residence and with their countries of birth? By using a model with translevel interactions, it is studied if the effect of origin of immigrants over the probability of being employed, according to neighbourhood of residence, is related to the degree of residential concentration of immigrants. In the estimated model 5, translevel interactions are not significant, so there is no statistical evidence to confirm the initial hypothesis. The loss of parsimony due to the addition of two terms in the model is not reflected on the goodness of fit. It means that, the effect of being Peruvian or Paraguayan on the probability of being employed, depending on the neighbourhood of residence, is the same for diverse degrees of residential segregation. Discussion The thread of this study was composed by three questions: does the level of residential concentration of immigrants affect the probability of employment of recent immigrants? does the country of birth of recent immigrants affect their probability of employment? and the general question for recent immigrants, does the probability of being employed vary with residential concentration degree of immigrant population in their neighbourhood of residence and with their countries of birth? The descriptive analysis presented in this study shows a situation of residential concentration among recent Peruvian and Paraguayan immigrants, in which the effect among Peruvians is stronger. The question of whether the neighbourhood of residence

• perates as a space of working opportunities or as a space constraining working
• pportunities was answered by estimating models of multilevel logistic regressions.

Such models allowed the analysis of individual and structural characteristics on the probability of being employed, versus being unemployed, among recent immigrants. Results for the estimated model 4 enables us to enunciate the answers below. Effect of sex, age, age at migration and conjugal union (H1, H2, H3 and H4): the effects of being man, increasing one year of age, being younger at migration, being in a mixed union, single, separated, divorced or widow(er) are positive and significant, i.e. they increase the chances of being employed, as compared to being unemployed. Effect of education level (H5): the effect of the education level on the probability of being employed is not significant among recent immigrants. This could be due to recent immigrants being less selective when chosing a job in the first years of residence or to a higher difficulty in validating their education certificates. Effect of the country of birth (H6): the country of birth affect the probability of being

• employed. The fact of being a recent immigrant from Paraguay increases 90.2 times on

average the chances of being employed, compared to being unemployed, and the fact of being a recent immigrant from Peru increases 120.1 times on average such chances, compared to the other countries of birth. Variances of the random term for Peruvians and Paraguayans are significant, which implies that the effect of being from Peru or Paraguay on the probability of being employed varies among the diverse neighbourhoods.

18

Effect of degree of spatial segregation (H7): the degree of residential concentration of immigrants affects the probability of recent immigrants of being employed. The coefficient of localization was calculated as a measure of spatial concentration, and it was found that living in a neighbourhood with a higher concentration of immigrants than proportional increases 34.63 on average the chances of being employed as compared to being unemployed. Noteworthy, results obtained by model 5 do not show statistical evidence to affirmatively answer the main question. Even when there are diverse effects of the

• rigin of migrants on the probability of being employed depending on the

neighbourhood of residence, the last estimated model does not show statistical evidence

• f such effect being associated to the residential concentration of immigrants in the
• neighbourhood. That is to say, the effect of being Peruvian or Paraguayan on the

probability of being employed does not vary among the diverse degrees of residential concentration. The analysis performed shows that integration to the labour market for recent immigrants in Montevideo is associated to individual and structural traits, according to the estimated models. The effect of neighbourhood of residence and its level of coefficient of localization operate as a working opportunity since the fact of living in a neighbourhood with higher concentration of immigrants than proportional increases their chances of being employed. The effect to the country of birth on the probability of being employed is stronger for Peruvian immigrants, who are more segregated in the territory and are more affected by the number of UBN. Moreover, the effect of being Peruvian or Paraguayan

• n the probability of being employed differs among neighbourhoods, though it does

have the same effect on diverse degrees of residential concentration of immigrants. Nevertheless, due to the characteristics of the neighbourhoods of residence, and the degree of UBN of these groups, it is a case of descendent residential segregation, a feature that in stead of promoting social integration, reproduces spatial segregation and its consequences for native and immigrant people. The results exposed open new dimensions of research for the future. It is considered important to analyze not only the probability of accessing a job but also its

• quality. If the occupational category of the jobs obtained by recent immigrants are

considered, new variables such as education level and sex could become explanatory. In particular, taking into account the specific insertion of women in domestic services and men in the fishing sector —shown by Diconca (2012)—, raises the interest of studying the effect of sex, country of birth, occupation type, neighbourhood of residence and degree of residential concentration in the neighbourhood on the quality of the job accessed. References Alarcón, R. and Ramírez-García, T.(2011), «Integración económica de los inmigrantes mexicanos en la Zona Metropolitana de Los Ángeles», en Papeles de población,vol. 17, n.º69, pp. 73-103, en http://www.scielo.org.mx/scielo.php?script=sci_arttext&pid=S1405- 74252011000300004, acceso: 11/7/2017. Arim, R. (2008), «Crisis económica, segregación residencial y exclusión social: el caso de Montevideo», en: Ziccardi, A. (coord.), Procesos de urbanización de la

19

pobreza y nuevas formas de exclusión social: Los retos de las políticas sociales de las ciudades latinoamericanas del siglo XXI. Bogotá: Clacso. Arriagada Luco, C.(2010), «Segregación residencial según dos modelos de urbanización y bienestar: estudio comparado de las áreas metropolitanas del Gran Santiago, Toronto y Vancouver», en Notas de Población,n.º 91, en http://repositorio.cepal.org/handle/11362/12877, acceso: 11/7/2017. Bengochea, J.(2014), Inmigración reciente en Uruguay: 2005-2011, Serie Tesis de Maestría del Programa de Población, Montevideo: Universidad de la República,en: http://cienciassociales.edu.uy/unidadmultidisciplinaria/tesis- del-programa-de-poblacion/, acceso: 11/7/2017. Cerrutti, M.(2009), Diagnóstico de la población de inmigrantes en la Argentina, Serie de documentos de la Dirección Nacional de Población, Buenos Aires:Secretaria del Interior,Ministerio del Interior-OIM, en http://www.mininterior.gov.ar/poblacion/pdf/Diagnostico_de_las_poblacion es_de_inmigrantes_en_Argentina.pdf, acceso: 11/7/2017.. De los Campos, H. and Paulo, L. (2001), La migración andina en Uruguay, Estudio realizado por la Facultad de Ciencias Sociales con el apoyo de la OIM [inédito]. Diconca, B.(2012), Caracterización de las nuevas corrientes migratorias en Uruguay. Migrantes y retornados: acceso a derechos económicos sociales y culturales, Montevideo:MIDES-OIM, en http://eva.fhuce.edu.uy/pluginfile.php/40696/mod_resource/content/0/libro migrantes_versionweb_hb_1.pdf, acceso: 11/7/2017. MIDES [Ministerio de Desarrollo Social](s/d),Caracterización socioeconómica para unidades geográficas pequeñas. Unidad de Seguimientos de Programas,Montevideo: DINEM, MIDES. Gans, H.J. (1997), «Toward a reconciliation of “Assimilation”and“Pluralism”: The interplay of acculturation and ethnic retention»,enInternationalMigration Review, vol, 31, n.º 4, invierno, Special Issue: Immigrant adaptation and Native-Borne Responses in the Making of Americans, pp. 875-892, enhttp://www.jstor.org/stable/2547417?seq=1#page_scan_tab_contents, acceso: 11/7/2017. Kaztman, R. (2001), «Seducidos y abandonados: el aislamiento social de los pobres urbanos»,en Revista de la CEPAL,n.º 75, pp. 171-189,diciembre, enhttp://repositorio.cepal.org/bitstream/handle/11362/10782/075171189.pdf ?sequence=1, acceso: 11/7/2017. Kaztman, R. and Retamoso, A.(2007), «Efectos de la segregación urbana sobre la educación»,en Revista de la CEPAL,n.º91, pp. 133-152,abril, en http://biblioteca.hegoa.ehu.es/system/ebooks/16404/original/Efectos_de_la_ segregacion_urbana_sobre_la_Educacion.pdf, acceso: 11/7/2017. Koolhaas, M. and Nathan, M.(2013), Inmigrantes internacionales y retornados en

• Uruguay. Magnitud y características, Informe de resultados de población

2011. Montevideo:UNFPA-INE-OIM, en http://www.unfpa.org.uy/userfiles/publications/82_file1.pdf, acceso: 11/7/2017.

20

Logan, J. R.; Zhang, W. and Alba, R. D. (2002), «Immigrant Enclaves and Ethnic Communities in New York and Los Angeles», en American Sociological Review, vol. 67, n.º 2, pp. 299-322, abril, en http://www.jstor.org/stable/3088897?seq=1#page_scan_tab_contents, acceso: 11/7/2017. Martori, J. and Hoberg, K.(2004), «Indicadores cuantitativos de segregación residencial el caso de la población inmigrante en Barcelona», enRevista Electrónica de Geografía y Ciencias Sociales,vol. VIII, n.º 169, en http://dspace.uvic.cat/handle/10854/2398, acceso: 11/7/2017. Massey, D. S.; Arango, J.; Graeme,H.; Kouaouci, A.;Pellegrino, A. and Taylor, E.J. (1993), «Theories of international migration a review and appraisal», en Population and Development Review, vol. 19 n.º 3, pp. 431-466, en: http://www.jstor.org/stable/2938462?seq=1#page_scan_tab_contents, acceso: 11/7/2017. Niessen,J. and Schibel, Y.(2004), Manual sobre la integración para responsables de la formulación de políticas y profesionales, s/c: Comisión Europea (Dirección General de Justicia, Libertad y Seguridad)-MigrationPolicyGroup (MPG), en https://ec.europa.eu/migrant- integration/index.cfm?action=media.download&uuid=29BAA2FC-EC4C- 52F7-0E8FDA6276388C03, acceso: 17/7/2017 Piore, M.J. (1979), Birds of passage: Migrant labour and industrial societies, NuevaYork: Cambridge University. ————— (1986), «From Foreign Workers to Settlers? Transnational Migration and the Emergence of New Minorities», en The Annals of the American Academy of Political and Social Sciences, vol. 485, pp.23-33. Sampson, R. J. (2008), «Moving to Inequality: Neighborhood Effects and Experiments Meet Social Structure», en American Journal of Sociology, vol. 114, n. º 1,

• pp. 189-231.

Sampson, R.J.; Raudenbush, S.W. and Earls, F. (1997), «Neighborhoods and Violent Crime: A Multilvel Study of collective Efficacy», enScience, vol. 77 (5328), pp.918-924.doi: 10.1126/science.277.5328.918. Veiga, D.(2005), Desigualdad y exclusión social: estudio de caso del Gran Montevideo, Presentado a la Reunión Anual de Investigadores del Departamento de Sociología, Montevideo: FCSUniversidad de la República, en http://cienciassociales.edu.uy/wp- content/uploads/sites/3/2013/archivos/LasBrujas4-Veiga.pdf, acceso: 11/7/2017. VerbitskySavitz,N. and Raudenbush, S.W. (2009),«Exploiting spatial dependence to improve measurement of neighborhood social processes», en Sociological Methodology, vol. 1 39 (1), pp. 151-183, enhttp://onlinelibrary.wiley.com/doi/10.1111/j.1467- 9531.2009.01221.x/full, acceso: 11/7/2017. Vono, D. and Bayona, J.(2010), «El asentamiento residencial de los latinoamericanos en las principales ciudades españolas (2001-2009)», en Notas de Población,n.º91, pp. 129-159, en http://repositorio.cepal.org/handle/11362/12875, acceso: 11/7/2017.