You may notice some images loading slow across the Open Collections website. Thank you for your patience as we rebuild the cache to make images load faster.

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Essays on immigrant assimilation Torres, Javier 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2013_fall_torres_javier.pdf [ 2.03MB ]
Metadata
JSON: 24-1.0074045.json
JSON-LD: 24-1.0074045-ld.json
RDF/XML (Pretty): 24-1.0074045-rdf.xml
RDF/JSON: 24-1.0074045-rdf.json
Turtle: 24-1.0074045-turtle.txt
N-Triples: 24-1.0074045-rdf-ntriples.txt
Original Record: 24-1.0074045-source.json
Full Text
24-1.0074045-fulltext.txt
Citation
24-1.0074045.ris

Full Text

Essays on Immigrant AssimilationbyJavier TorresB.A., Pacifico University, 2000M.A., University of British Columbia, 2005A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate Studies(Economics)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2013c? Javier Torres 2013AbstractThis dissertation examines immigrants (to Canada) assimilation prob-lems from a perspective of imperfect human capital transferability. Chapter2 discusses how much of the immigrant wage gap can be explained by the un-dervaluation of foreign human capital (education and work experience). Theidentification of the human capital source (using information available in the2006 Canadian census) can explain up to 70% of the native-immigrant wagegap. The foreign-born dummy coefficient goes from around -11% to closeto -3%. Education acquired in Asia tends to be valued less than educationfrom South America, Africa and East Europe; which in turn is less valuedthan education from Oceania, the U.S. and the rest of continental Europe.Studying in the UK consistently appears more beneficial than studying inCanada. When incorporating country of origin fixed effects, the differentspecifications visibly reduce the heterogeneity of country coefficients. Thereduction is sizeable for Pakistan, India, China and the Philippines; thoughtheir coefficients remain negative. A smaller reduction for Europe, South-East Asia, Hong Kong and the US drives their coefficients close to zero. TheUK country of origin dummy has the only persistently positive coefficient.Chapter 3 describes the occupational assimilation process of 2000-2001 im-migrants in their first four years. The results show that those with highlevels of education experience a more significant decline in their first occu-pation. Education, though, has a positive and significant effect on occupa-tional improvement; which reduces the size and significance of the negativeeffect of education on the second occupational gap. It, however, does notchange its sign. The same pattern is observed when analyzing occupationalgaps through time. Chapter 4 focuses on immigrants? English proficiencyimprovement. Overall, immigrants show relatively small improvements inlanguage proficiency in the first four years in Canada. Still, those arriv-ing under the family immigrant category with an intermediate or advancedlevel are less likely to improve and more likely to decrease their English pro-ficiency. Human capital variables (age and education) are also consistentlyrelevant for English proficiency improvement.iiPrefaceChapter 2 comes from an on going joint work with Professors NicoleFortin and Thomas Lemieux. I was responsible for managing the data setand designing the empirical strategy to test the main hypotheses. Likewise,I conducted all the statistical estimations and wrote the manuscript. Pro-fessors Fortin and Lemieux provided the original research idea as well asinvaluable research advice and guidance throughout the investigation. Fi-nal edition suggestions were given by Professor Lemieux. This research waspresented the University of Montreal, the University of Chicago and theUniversity of Ottawa.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Foreign education and immigrants earnings gap . . . . . . . 21.2 Occupational mismatches of recent immigrants . . . . . . . . 41.3 Immigrants English Proficiency Improvement . . . . . . . . . 72 Foreign Education and Immigrants Earnings Gap. . . . . . 102.1 Data and Descriptive Statistics . . . . . . . . . . . . . . . . . 112.1.1 The 2006 Canadian Census . . . . . . . . . . . . . . . 112.1.2 Descriptive Statistics . . . . . . . . . . . . . . . . . . 152.2 Empirical strategy . . . . . . . . . . . . . . . . . . . . . . . . 182.3 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.3.1 Replicating Friedberg?s results . . . . . . . . . . . . . 202.3.2 Base Specifications . . . . . . . . . . . . . . . . . . . 222.3.3 Separating the foreign born wage gap by country oforigin . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.3.4 Identifying the effect of field of study . . . . . . . . . 282.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30ivTable of Contents3 Occupational Mismatches of Recent Immigrants . . . . . . 793.1 Data Set and Occupational Quality . . . . . . . . . . . . . . 803.1.1 Working Sample . . . . . . . . . . . . . . . . . . . . . 823.1.2 Occupational quality . . . . . . . . . . . . . . . . . . 843.1.3 Occupational mobility . . . . . . . . . . . . . . . . . . 863.1.4 Hypotheses summary . . . . . . . . . . . . . . . . . . 893.2 Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . 903.2.1 City self-selection and the identification method . . . 913.2.2 Probability of occupational change . . . . . . . . . . . 933.2.3 Occupational Improvement . . . . . . . . . . . . . . . 943.2.4 Occupational gap through time . . . . . . . . . . . . 943.3 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 953.3.1 First Occupational Gap . . . . . . . . . . . . . . . . . 953.3.2 The Second Occupational Gap . . . . . . . . . . . . . 973.3.3 Occupational Improvement . . . . . . . . . . . . . . . 983.3.4 Probability of switching occupations . . . . . . . . . . 993.3.5 Occupational Gap through time . . . . . . . . . . . . 1003.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1014 Immigrant?s English Proficiency . . . . . . . . . . . . . . . . 1254.1 Language proficiency meassures in the LSIC . . . . . . . . . 1254.2 Requirements for Becoming a Landed Immigrant . . . . . . . 1294.3 Descriptive statistics . . . . . . . . . . . . . . . . . . . . . . . 1304.4 Econometric Considerations . . . . . . . . . . . . . . . . . . 1334.5 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1364.5.1 Levels of English Proficiency . . . . . . . . . . . . . . 1364.5.2 Language Proficiency Improvement . . . . . . . . . . 1374.5.3 Decline in Language Proficiency . . . . . . . . . . . . 1384.5.4 Ordered Probit Conditioning on English Level . . . . 1394.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1395 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160AppendicesA Appendix to Chapter 2 . . . . . . . . . . . . . . . . . . . . . . 165A.1 Sample Restrictions . . . . . . . . . . . . . . . . . . . . . . . 165vTable of ContentsA.2 Minimum Hourly Wage by Province . . . . . . . . . . . . . . 166A.3 Years of Education by Highest Degree Attained . . . . . . . 167A.4 Region of Origin and Location of Study . . . . . . . . . . . . 168B Appendix to Chapter 3 . . . . . . . . . . . . . . . . . . . . . . 169B.1 Sample Attrition . . . . . . . . . . . . . . . . . . . . . . . . . 169B.2 Ranking of Occupations a Sample . . . . . . . . . . . . . . . 171B.3 Individual Characteristics and the Occupational Gap . . . . 172B.4 Motivational Model . . . . . . . . . . . . . . . . . . . . . . . 173B.5 Dictionary of Variables . . . . . . . . . . . . . . . . . . . . . 175B.6 Wage Improvement . . . . . . . . . . . . . . . . . . . . . . . 177C Appendix to Chapter 4 . . . . . . . . . . . . . . . . . . . . . . 178C.1 Dictionary of Variables . . . . . . . . . . . . . . . . . . . . . 178viList of Tables2.1 Immigrant?s Top Countries of Origin by Year of Arrival . . . 322.2 Immigrant?s Top Locations of Study by Year of Arrival . . . . 332.3 Distribution of Immigrants and Natives by Field of Study . . 342.4 Distribution of Immigrants and Natives by CMA of Residence 352.5 Age Distribution of Immigrants and Natives . . . . . . . . . . 362.6 Distribution of Immigrants and Natives by Years of Education 372.7 Immigrants with a Canadian Degree by Education Category . 382.8 Immigrants with a Canadian Degree by Age at Immigration . 382.9 Summary Statistics Immigrants vs Natives . . . . . . . . . . . 392.10 Replication of Friedberg?s table 4 . . . . . . . . . . . . . . . . 402.11 Replication of Friedberg?s table 5 . . . . . . . . . . . . . . . . 412.12 Replication of Friedberg?s table 4 - Sample Extension Immi-grants Arriving at 15 and older . . . . . . . . . . . . . . . . . 422.13 Replication of Friedberg?s table 5 - Sample Extension Immi-grants Arriving at 15 and older . . . . . . . . . . . . . . . . . 432.14 Adaptation of Friedberg?s table 4 with Location of Study In-formation - Immigrants Arriving at 15 and older . . . . . . . 442.15 Adaptation of Friedberg?s table 5 with Location of Study In-formation - Immigrants Arriving at 15 and older . . . . . . . 452.16 Base Specification . . . . . . . . . . . . . . . . . . . . . . . . 462.17 Separating Male vs Female Immigrant Wage Gap . . . . . . . 472.18 Separating Years of Educ. by Degree Achieved . . . . . . . . 482.19 Educational Group Dummies . . . . . . . . . . . . . . . . . . 492.20 Interacting Education with Location of Study . . . . . . . . . 502.21 Base Specification - Sample Extension: Immigrants Arrivingat 15 and older . . . . . . . . . . . . . . . . . . . . . . . . . . 552.22 Educational Group Dummies - Immigrants Arriving at 15 andolder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562.23 Immigrant Wage Gap by Country of Origin - Base Specification 572.24 Immigrant Wage Gap by Country of Origin - Separating Yearsof Educ. by Degree Achieved . . . . . . . . . . . . . . . . . . 58viiList of Tables2.25 Immigrant Wage Gap by Country of Origin - EducationalGroup Dummies . . . . . . . . . . . . . . . . . . . . . . . . . 592.26 Immigrant Wage Gap by Country of Origin - EducationalGroup Dummies- Immig 15 plus . . . . . . . . . . . . . . . . 602.27 Exploring the Effect of Field of Study . . . . . . . . . . . . . 612.28 Interacting Field of Study with Foreign-born Dummy . . . . . 622.29 Interacting Field of Study with Location of Study Groups . . 633.1 Distribution of Immigrants by Category . . . . . . . . . . . . 1033.2 Employment by Immigration Category . . . . . . . . . . . . . 1043.3 Education by Immigration Category . . . . . . . . . . . . . . 1053.4 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . 1063.5 Distribution of the Population by Metropolitan Areas . . . . 1073.6 First Occupational Gap . . . . . . . . . . . . . . . . . . . . . 1083.7 First Occupational Gap with Self-selection Correction . . . . 1093.8 Second Occupational Gap . . . . . . . . . . . . . . . . . . . . 1103.9 Second Occupational Gap with Self-selection Correction . . . 1113.10 Occupational Improvement - First to Second Occupation . . . 1123.11 Probability of Having More than One Occupation . . . . . . . 1133.12 Occupational Gap Through Time . . . . . . . . . . . . . . . . 1144.1 Age Distribution by Immigration Category . . . . . . . . . . 1414.2 Average Savings 6 Months After Arrival by Immigration Group1424.3 Government Social Assistance by Immigration Group . . . . . 1424.4 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . 1434.5 Share of Interviews in English . . . . . . . . . . . . . . . . . . 1444.6 Evolution of English Proficiency . . . . . . . . . . . . . . . . 1454.7 Changes in English Proficiency . . . . . . . . . . . . . . . . . 1464.8 Communication Abilities Changes . . . . . . . . . . . . . . . 1474.9 Ordered Probit - Level of Speaking at Wave 3 (Coefficients) . 1484.10 Ordered Probit - Level of Reading at Wave 3 (Coefficients) . 1494.11 Ordered Probit - Level of Writing at Wave 3 (Coefficients) . . 1504.12 Improvement for People Beginning at a Basic Level (wave 1to wave 3) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1514.13 Improvement for People Beginning at an Intermediate Level(wave 1 to wave 3) . . . . . . . . . . . . . . . . . . . . . . . . 1524.14 Improvement for People Beginning at Intermediate or BasicLevel (wave 1 to wave 3) . . . . . . . . . . . . . . . . . . . . . 1534.15 Language Proficiency Decrease for People Beginning at anIntermediate Level (wave 1 to wave 3) . . . . . . . . . . . . . 154viiiList of Tables4.16 Language Proficiency Decrease for People Beginning at anAdvanced Level (wave 1 to wave 3) . . . . . . . . . . . . . . . 1554.17 Lang. Prof. Decrease for People Beginning at an Intermedi-ate or Advanced Level . . . . . . . . . . . . . . . . . . . . . . 1564.18 Ordered Probit (coefficients) - Language Proficiency at Wave3 Conditioning on Initial Level . . . . . . . . . . . . . . . . . 157B.1 Probit model marginal effect . . . . . . . . . . . . . . . . . . 169ixList of Figures2.1 Location of Study Fixed Effects Table 2.16 Column 2 . . . . . 642.2 Location of Study Fixed Effects - Table 2.16 Columns 2,4 and 6 652.3 Country of Origin Fixed Effects Table 2.23 Columns 1,2 and 3 662.4 Country of Origin Fixed Effects Table 2.23 adding groups oflocation of study Fixed Effects . . . . . . . . . . . . . . . . . 672.5 Country of Origin Fixed Effects Table 2.23 adding Locationand Field Study Fixed Effects . . . . . . . . . . . . . . . . . . 682.6 Country of Origin Fixed Effects Table 2.24 . . . . . . . . . . 692.7 Country of Origin Fixed Effects Table 2.24 . . . . . . . . . . 702.8 Country of Origin Fixed Effects Table 2.25 . . . . . . . . . . 712.9 Country of Origin Fixed Effects Table 2.25 . . . . . . . . . . 722.10 Country of Origin Fixed Effects Table 2.26 . . . . . . . . . . 732.11 Field of Study Fixed Effects Table 2.27 . . . . . . . . . . . . 742.12 Interacting Field of Study with Foreign-born Dummy - Table2.28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 752.13 Location of Study table 2.29 . . . . . . . . . . . . . . . . . . . 762.14 Interaction: Location + Field of Study Table 2.29 - Part 1 . . 772.15 Interaction: Location + Field of Study Table 2.29- Part 2 . . 783.1 Kernel Density - Ln weekly wages by Occupation for nativeCanadians . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1153.2 Kernel Density - Ln Weekly Wages Across Waves by Numberof Occupations . . . . . . . . . . . . . . . . . . . . . . . . . . 1163.3 Kernel Density - Occupational Distribution of the Intendedand Wanted Occupations . . . . . . . . . . . . . . . . . . . . 1173.4 Occupations held before and after arriving in Canada . . . . 1183.5 Occupational Gap by Education and Number of Occupations 1193.6 Occupational Gap by Language Proficiency and Number ofOccupations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1203.7 Occupational Gap by Immigration Category and Number ofOccupations . . . . . . . . . . . . . . . . . . . . . . . . . . . 121xList of Figures3.8 Occupational Gap by Network Job and Number of Occupa-tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1223.9 Occupational Gap by number of Occupations . . . . . . . . . 1233.10 Kernel Density - Ln Weekly Wages Across Occupations . . . 124xiAcknowledgementsI have many to thank. Throughout the PhD program I have benefitedfrom the support, encouragement and generosity of many people. I wish tothank my committee members Nicole Fortin, Thomas Lemieux and DavidGreen. Their insightful comments and feedback greatly improved both thecoverage and exposition of my work.I owe my deepest gratitude to Professor Fortin for her guidance, extraor-dinary support and patient over my years of study at UBC. Without herinvolvement accomplishing this project would not have been possible.Particular thanks go to David Green and Thomas Lemieux for manystimulating conversations and helpful advice at various stages of my work.Special thanks go to Mauricio Drelichman who was readily available andgenerously offered his time to answer my questions and concerns inherentto the doctoral process.I also want to thank my fellow students at UBC. My time in the PhDprogram vastly benefitted from interacting with them. Among all, I am indebt to Marcos Agurto, Nishant Chadha, Matias Cortes, Thomas Fujiwaraand Changhua Yu for their help, especially in trial times.I gratefully thank my parents and my wife Nadia, who have supportedme throughout my years of education, both morally and financially. Finally,financial support from the Canadian Labour Market and Skills ResearcherNetwork (CLSRN) is gratefully acknowledged.xiiDedicationA mi NadixiiiChapter 1IntroductionImmigrants have not fared well in the Canadian labour market lately.A number of studies have documented a steady decline in their earningsrelative to the Canadian born over the last three decades. They have foundthat immigrants who arrived in Canada in the 1990s earned around 30 per-cent less than Canadian-born workers (see for instance Green and Worswick(2004) and Aydemir and Skuterud (2005)). By contrast, earlier cohortsof immigrants who arrived in the 1970s were earning about the same asCanadian-born workers. Starting with Chiswick (1978), studies have sug-gested that the lack of transferability of human capital is a key reason whyimmigrants tend to earn less than natives. While most immigrants in the1960s were from countries culturally close to Canada (Europe and the UnitedStates), about two thirds of immigrants who arrived in the 1980s and 1990swere from Asia, Africa, and Central and South America. Given the large andincreasing fraction of foreign-born population in Canada (currently around20%), the appropriate assimilation of recent immigrants in the Canadianlabor market is a priority.1This dissertation studies immigrants assimilation problems in Canada.Its three chapters are introduced in the following sections. The first chapterdiscusses how much of the immigrant wage gap can be explained by theundervaluation of foreign human capital. The next two chapters describethe occupational and language assimilation process of 2000-2001 immigrantsin their first four years. In particular, the second chapter explores the declinein the occupational status of immigrants in the short and medium run, whilethe last chapter examines immigrants? English proficiency improvement andhow it relates to their immigration category.1The proportion of foreign-born fluctuated between 15% and 16% from 1961 to 1991.The fraction rose to 18.4% in 2001 and 19.8% in 200611.1. Foreign education and immigrants earnings gap1.1 Foreign education and immigrants earningsgapThe country of origin of recent immigrants may account for the growingearnings gap between immigrants and Canadian-born workers if the humancapital acquired abroad is not fully transferable into the Canadian labourmarket. The influential study by Friedberg (2000) uses the 1983 Israeli Cen-sus to look at the contribution of the differences in the returns to foreign andnative schooling and labour market experience to the immigrant-native wagegap.2 Friedberg also highlights the level of heterogeneity in the returns toforeign schooling by source country. Returns to education abroad are higherfor immigrants from Europe and the Western Hemisphere (in comparisonto immigrants from Asia and Africa). The paper also suggests that acquir-ing further education in the host country may increase the overall return toeducation.The Canadian literature (Ferrer and Riddell (2008) and Ferrer et al.(2006) among others) also suggests that years of schooling and experienceaccumulated before arrival are much less valued than the host country ones.3Ferrer and Riddell study the period from the 1980s to the early 2000s. Theircentral focus is on the effect of credentials (degrees and diplomas) on theearnings of immigrants, holding constant the number of years of education.Using public-use Census files from 1981 to 2001, they estimate a flexiblefunction form for wages where education and experience are interacted withan immigrant indicator. They find substantially lower returns to foreigneducation and experience. Ferrer et al. (2006) use a different approach.They incorporate measures of literacy skills in addition to educational levelsand separate degrees acquired in the home and host country. They findthat, among the university-educated, literacy skills explain about two thirdsof earnings difference between immigrants and the Canadian born. Thissuggests that the quality of foreign education, as measured by literacy skills,is not as high as Canadian education.The main goal of this first chapter is to estimate how much of the earn-ings gap between immigrants and the Canadian born is due to the fact2The 1972 Israeli Census is used to argue that the results found using the 1983 Censuscome from an assimilation process and not a change in cohort quality over time.3Another possibility is that recent immigrants may not have good enough languageskills (in English or French) to get some high-paying jobs. Aydemir and Skuterud (2005)indeed find that language skills accounts for a share of the earnings gap between immi-grants and the Canadian born. Still, this can also be viewed as a human capital explana-tion.21.1. Foreign education and immigrants earnings gapthat foreign education is not as valued as much as Canadian education inthe labour market. We use the 2006 Canadian Census to estimate detailedearnings penalties to foreign born linked to country or region of study; con-trolling for gender, academic degree, work experience, metropolitan area ofresidence, mother tongue (English or French only) and field of study. Wego further and identify for which of the principal countries/regions of ori-gin does the inclusion of the location of study substantially decrease theearnings gap. Our sample are full-time workers between the age of 20 and64 with an education level higher than high school, and with positive wagein 2005. For immigrants, we additionally restrict the sample to those whoarrived in Canada between the age of 15 and 29.4One challenge when trying to estimate the role of foreign education in theearnings gap between immigrants and Canadian-born workers is that explicitinformation on the location of study is typically not available in data sets likethe Census. Researchers have tried to infer where education was obtained bycomparing age at immigration to the age at which an immigrant should havenormally completed her highest degree.5 Indeed, Friedberg estimates yearsof schooling in the home country under the assumptions that children startschooling at age 7 and, more importantly, that they attend school withoutinterruption. Bratsberg and Ragan Jr (2002) follow a similar strategy usingthe 1990 U.S. Census to estimate differences in the return to educationfor immigrants with and without U.S. schooling.6 Their main findings aresimilar to those of Friedberg for Israel.7The procedure, however, may mis-attribute the country where educationwas acquired. Immigrants may have worked in Canada for a number of yearsbefore completing their final degree.8 Likewise, foreign-born individuals may4We do this trying to focus on immigrants who would face problems adapting to thenew environment, but who would still be able to invest in education upon arrival. Still,for some specifications we broaden the age range.5For instance, one may assume that an immigrant with a BA degree who came toCanada at age 30 has completed that degree abroad prior to immigrating to Canada.6They determine an age of graduation based on the reported years of education andthe assumption of starting schooling at age at 6. Immigrants arriving at an age lower thanthe presumed age at graduation (years of completed schooling plus six) are classified ashaving U.S. schooling.7 They find that immigrants with U.S. schooling earn higher wages than immigrantswithout U.S. schooling. Their results also indicate that returns to foreign schooling aresignificantly higher for immigrants who completed some of their studies in the UnitedStates. Though this last finding is based on a small (351 immigrants) and not represen-tative survey of U.S. immigrants (the National Longitudinal Survey of Youth)8For instance, a 40 year old immigrant with a MBA who came to Canada at age 30may very well have completed that degree in a Canadian university after the age of 30.31.2. Occupational mismatches of recent immigrantsfinish their studies in Canada before being officially considered immigrants.9Fortunately, the long form of the 2006 Canadian Census includes an ex-plicit question about where one?s highest degree was obtained. The locationof study is either recorded as a country, for those who studied abroad, or asa province, for those who studied in Canada. While the location of studyis also available in some smaller data sets, we can perform a much moredetailed analysis in this chapter thanks to the large sample available in themaster files of the 2006 Canadian Census (20 percent of the population).The Census also includes information on field of study, which enables usto see whether degrees obtained in a different country are more portable insome fields of study (e.g. math and computer science) than in others (e.g.education and humanities).1.2 Occupational mismatches of recentimmigrantsThe second chapter focuses on the quality of the occupations of immi-grants instead of their earnings. The seminal work of Chiswick (1977, 1978),based on cross-section data (1970 U.S. census), found that the difference inthe skill distribution of immigrants and natives reduces throughout the yearsafter migration. Using the long form the 2001 Canadian census, this chap-ter constructs an index of occupational quality based on the average of thelogarithm of weekly wages. With it, the chapter documents the change inthe quality of occupations in the short and medium run.10Borjas (1985) calls attention to the possible biases in cross-section anal-ysis. In particular, he argues that the cross-sectional results could be causedby differences in the relative skill levels of immigrant cohorts. Borjas (1985,1995) emphasizes the change in the postwar immigration policy and its ef-fects on the national composition of the immigrant flow. If the changes re-sulted in a less-skilled immigrant flow, then the cross-section results wouldreflect differences in ability or skills across cohorts.11 As Borjas (1994)9 In the Canadian Census age at immigration is the age an individual had whenhe/she first became a permanent resident of Canada. Foreign students who went to uni-versity in Canada and became permanent residents after finishing school would, therefore,be misleadingly classified as having a foreign degree using a Friedberg-type imputationprocedure.10Specifically, it defines an occupational gap as the quality of the occupation in Canadaminus the quality of the occupation held in the home country.11The cohort differences may also occur if return migration is not exogenous from labormarket integration (Borjas, 1994)41.2. Occupational mismatches of recent immigrantsnotes, the ideal solution for the identification problem that surfaces in thecross-section analysis would be a longitudinal data set tracking a particularworker over time.12This chapter uses, precisely, the Longitudinal Survey of Immigrants toCanada (LSIC) to document the occupational matches of immigrants intheir first four years.13 It highlights the relation between occupational de-cline (or rise) and education levels and language proficiency of immigrants.In particular, it tests if immigrants with high education levels experience alarger occupational decline in their first occupation. It also investigates ifthe probability of changing occupations is influenced by education or lan-guage proficiency; and if among those who change occupations educationis positively correlated with the magnitude of the occupational quality im-provement.Chiswick, Lee and Miller (2003, 2005) are among the first longitudinaloccupational studies of immigrants. They use the Longitudinal Survey ofImmigrants to Australia (LSIA 1) to classify immigrant?s occupations by theeight broadest level of the Australian Standard Classification of Occupations(ASCO).14 They find that the occupational path of immigrants follows a U-shaped pattern.15 Immigrants experience a decline in their occupational atarrival, but improve the quality of their occupation with time. The negativeslope seems to be steeper for immigrants from countries that do not speakEnglish and for refugees and family immigrants (relative to economic immi-grants). Subsequent improvements were greater for those with high levelsof education.For Canada, different studies have taken advantage of the LSIC.16 Goeland Lang (2007) use it to analyze the effect of ethnic network size and ethnicnetwork strength on the earnings of recent immigrants as well as the prob-ability of being employed by the ethnic network (or outside of it).17 Theyfind that ethnic network size and strength are significant determinants of12Concretely, Borjas asks for a longitudinal data set where both immigrants and nativesare tracked to correctly identify wage convergence.13Longitudinal surveys of immigrants began appearing by the middle of the 1990s.Chiswick et al. (2003, 2005) extensively analyze the Longitudinal Survey of Immigrants toAustralia (LSIA) while Akresh (2008) uses the U.S. New Immigrants Survey (NIS), andGoel and Lang (2007) and Grenier and Xue (2009) employ the LSIC.14They also use a measure of occupational socio-economic status developed by Jones(1989) known as ANU3.15Their research follows hypotheses made first by Chiswick (1977, 1978).16Statistics Canada has also written a series of working papers on the economic assim-ilation of recent immigrants using the LSIC. The most relate to our topic are the onesfrom Chui and Tran (2005, 2006) and Grondin (2005)17 They define ethnic network size by the log of the share of working age immigrants in51.2. Occupational mismatches of recent immigrantsgetting a job through the ethnic network.18 Grenier and Xue (2009) exam-ine the immigrants duration of access to a job on the intended occupation.They find that immigrants who intent to work in non-professional jobs havea shorter waiting period before reaching their desired occupations. Theirmost important result is, however, that the first year in the labor market iscritical for finding employment in the desired field.In our case, the evaluation of an occupational mismatch requires specialconsiderations. The discrete choice characteristic of occupational decisionsleads to two possible paths of analysis. The first is through a multinomialregression of job selection, while the second requires the transformation ofthe discrete occupation decision into an index of occupational quality.The first method provides a clear interpretation of the explanatory vari-ables effects on the probability of choosing one occupation versus a basecase, but it requires for the number of options (to choose from) to be lim-ited. For instance, Green (1999) uses a multinomial analysis on samplesfrom the Canadian censuses of 1980s and immigrant landing records from1973 to 1991 to compare the occupational distribution (and mobility) ofmale immigrants with those of natives.19 The second method enables one tomeasure the difference between the original occupation and that in the hostcountry, but there is no standard procedure to classify occupations, makingthe construction of the index relatively arbitrary. For instance, Chiswickand Miller (2010) employ the U.S. 2000 census to construct a measure ofeducational mismatch by calculating the modal level of education for eachoccupation.20 Also, Imai, Stacey and Warman (2011) merge the informationof O*NET with the LSIC data to generate a continuous variable related todifferent skills being evaluated.Given that we want to measure the magnitude of the occupational changein the short and medium run, we choose to construct an occupational qual-ity index. The analysis supplements the existing literature by focusing onthe same city and from the same country of origin. Also, ethnic network strength is definedby the presence of at least one close friend or relative in the city. Finally, employmentthrough the ethnic social network is described as finding a job through a relative or friend.18They also find that for immigrants with a high probability of having a network jobthose who have a non-network job have higher wages; the result, though, is robust onlyfor the lower wage percentiles.19He finds immigrants to be relatively over-represented in high-skilled occupations butthis over-representation declines across successive cohorts. He also finds a strong rela-tionship between language fluency (English or French) or being accepted under the pointsystem and having a professional occupation.20They calculate the modal level of education of native-born workers for 22 of the 23major group occupations included in the Census.61.3. Immigrants English Proficiency Improvementwork-oriented male immigrants with high labor market participation rates.Furthermore, it distinguishes between the first and the second occupationand the effect of key variables (i.e., education) in each case. Unlike previousstudies, it addresses concerns about city self-selection by analyzing immi-grants who chose a city for non-economic reasons and incorporating controlfactors based on the probability of an immigrant selecting a particularly city.Overall, the approach followed here is in line with the studies of Chiswick etal. (2003, 2005) regarding the theoretical assimilation of immigrants; and itcomplements Goel and Lang (2009) and Grenier and Xue (2009) concerningthe study of recent immigrants to Canada.1.3 Immigrants English Proficiency ImprovementIn addition to the mentioned effects language proficiency has on theearnings and occupational status of immigrants, it defines the extent oftheir interactions with the new environment (e.g., use of medical services,over the phone communications, day-to-day interactions). Yet over the lastdecade about 37% of accepted permanent residents spoke neither of theofficial languages (see Citizenship and Canada (2009)). The proportion ofnon-speakers has declined over time (48% in 2000 to 29% in 2009), but thecumulative effect might have a significant impact on language assimilationfor present and future newcomers.Most of the literature on language assimilation focuses on the effects ofhost-country language proficiency on earnings. These studies commonly finda positive and statistically significant relation between language proficiencyand earnings.21 Naturally, those able to communicate proficiently in thehost-country language will have a better chance to use their skills efficiently.Furthermore, the limited research on the determinants of host-countrylanguage proficiency mainly uses a cross section perspective, focusing onthe characteristics of language proficient migrants. Few papers examinethe particular determinants of language improvement. The more robustresults of these cross-section analyses indicate that language proficiency ispositively related to education and time spent in the host country.22 For21A literature review list should include but not be limited to Carliner (1981), Grenier(1984), Rivera-Batiz (1990), Chiswick (1991), Dustmann (1994) and Chiswick and Miller(1995 and 2002)22This literature has generated some interesting ideas. Chiswick and Miller (1995, 1996and 2001) propose a human capital approach to identify the main factors that influencelanguage proficiency, classifying the variables in broad categories: exposure, efficiency,economic incentives and wealth. They also put forward compelling hypotheses, e.g. linking71.3. Immigrants English Proficiency Improvementexample, Dustmann (1994) uses the German Socio-Economic survey of 1984to analyze immigrants from Italy, Spain, Jugoslavia, Turkey and Greecewho legally arrive in Germany after 1956. He finds that older migrantshave a lower probability of achieving high language proficiency, that yearsof residence in the host-country improve speaking and writing fluency andthat language proficiency is an important determinant of migrant earnings.Among the studies that directly investigate changes in host-country lan-guage proficiency, Chiswick (1991) and Hou and Beiser (2006) are of par-ticular interest. Chiswick (1991) uses data from a survey of over 800 appre-hended illegal immigrants in the U.S. (from October 1986 to October 1987) that retrospectively self-asses their present and past language skills. Heshows that both speaking and reading skills improve with duration of resi-dence and more so for people with high levels of education. Hou and Beiser(2006) utilize the three surveys of the Refugee Resettlement Project (RRP)of Canada, a decade-long longitudinal study, and concluded that the majorlanguage improvements occurred in the first years.23 Initial education andage at arrival are statistically important in the first years. For languageassimilation in the long run they put emphasis on Canadian education andemployment in Canada. However, the visibly endogeneity of these last vari-ables coupled with the limited number of observations (1,349 initially) andlarge attrition (only 647 observations by the third survey) hinders the va-lidity of the study.24This last chapter also takes advantage of the panel data features of theLSIC. It examines the characteristics of immigrants who change their En-glish proficiency within their first four years in Canada. I categorize new-immigrant?s origin language to the degree of difficulty with regard to learning the host-country language (language distance). Chiswick and Miller (2001) apply this humancapital model to male immigrants using the 1991 Canadian census. The authors arguethat the use of English (or French) increases with years of residence and relates positivelyto education. They also find that proficiency is greater the younger the age at migrationand the linguistically closer the mother tongue is to English (or French).23The Refugee Resettlement Project (RRP) interviewed 1,349 Southeast AsianRefugees in and nearby Vancouver in 1981. Two follow up interviews were done. One in1983 and the other in 1991; being able to locate 1,169 and 647 members of the originalsample, respectively.24 Dustmann and Van Soest (2001, 2002) studies should also be mentioned. They usea panel data of immigrants based on seven waves of the German Socio-Economic Panelfrom 1984 to 1993. They employ the panel to circumvent potential problems relating tounobserved heterogeneity and measurement error in language proficiency. Though themain contribution of both papers hinges on a better estimation of the effect of languageproficiency on earnings, a section in both is dedicated to evaluate the determinants oflanguage proficiency.81.3. Immigrants English Proficiency Improvementcomers? initial English proficiency in three groups (basic, intermediate andadvanced) and employ probit and ordered probit estimations to evaluate theprobability they would raise or lower their language proficiency between thefirst and third (last) interview of the survey. I relate this change in languageproficiency to human capital variables, such as education and age; and putemphasis on the effect of immigration categories arguing that they could re-flect either inherent non-observable characteristics or specific environmentsthat may hinder English proficiency improvement.9Chapter 2Foreign Education and TheEarnings Gap BetweenImmigrants andCanadian-born WorkersThis chapter investigates how much of the earnings gap between im-migrants and the Canadian born is due to the fact that foreign education(and work experience) is not valued as much as Canadian education in thelabour market. Unlike Friedberg (2000), our research uses a direct mea-sure of where one?s highest degree was obtained based on a new question ofthe 2006 Canadian Census. This new information goes a long way towardsidentifying how much of an immigrant?s education was obtained in the homecountry and how much was obtained in Canada.25An important limitation of the imputation approach ?a la? Friedbergfor Canada is that immigrants may go back to school after having immi-grated.26 Consider an immigrant with a high school diploma who comesto Canada at age 25 and then completes a two-year community collegeprogram. Since one would normally complete such a program at age 20,the imputation procedure would suggest that the immigrant got all of herschooling abroad despite the fact the two-year community college programwas actually acquired in Canada. With the new information available wecan clearly distinguish both sources of human capital (and correspondinglyrecalculate years of work experience abroad and in Canada).It should be noted that relatively recent studies (such as Clark and Jaeger(2006) and Hartog and Zorlu (2009)) have also directly identify the origin ofimmigrants education for other countries or for other analysis. For example,25This information, in turn, helps distinguish work experience acquired aborad fromwork experience acquired in Canada.26As mentioned in the introduction, Friedberg estimates years of schooling in the homecountry under the assumptions that children start schooling at age 7 and that they attendschool without interruption.102.1. Data and Descriptive StatisticsClark and Jaeger (2006) distinguish between immigrants who earned a GEDand immigrants who did not. They find immigrants with a GED earn morethan immigrants without one (but with similar foreign schooling). Still,their study relates more to sheepskin effect (or signaling) than to the un-dervaluation of foreign education. Hartog and Zorlu (2009) follow refugeesto The Netherlands for their first five years (1995 to 2000) and use the ad-ministrative immigration records to measure their education level. Theirmain finding is that returns to higher education are not significant. Theirstudy, however, limits itself to refugees and does not identify their locationof study. Moreover, the analysis focuses on the first five years after arrivalthey can not identify the effect of education acquired in the host country.For Canada, as mentioned in the introduction, the literature clearlypoints out that years of schooling and experience accumulated before ar-rival are much less valued than the host country ones. For instance, Greenand Worswick (2004) find that immigrants arriving by the beginning of the1990s had no significant wage differential for years of foreign experience. Onthe other hand, Ferrer et al. (2006) incorporate measures of literacy skills inaddition to educational levels and separate degrees acquired in the home andhost country. They find that, among the university-educated, literacy skillsexplain about two thirds of earnings difference between immigrants and theCanadian born. The implication is that the quality of foreign education, asmeasured by literacy skills, might not be as high as Canadian education.This chapter proceeds as follows. We present the census data and de-scriptive statistics in section 2.1. The foreign wage gap is modeled in anearnings equation framework in section 2.2. Empirical findings are reviewedin section 2.3. Conclusions are presented in section 2.4.2.1 Data and Descriptive Statistics2.1.1 The 2006 Canadian CensusThe 2006 Census was conducted by Statistics Canada on residents ofprivate dwellings as of May 16, 2006 (the reference day) between the monthsof February and August of that year.27 It enumerated Canadian citizens,landed immigrants, and non-permanent residents.28 One in five households27Statistics Canada points out that this was the period of most intense activities ondata collection.28According to the Census information, it also counted ?Canadian citizens and landedimmigrants who were temporarily outside the country on Census Day; including federal andprovincial government employees working outside Canada, Canadian embassy staff posted112.1. Data and Descriptive Statisticsreceived the long form questionnaire which, in addition to the regular eightquestions on household members, age, gender, marital status and mothertongue, contained 53 questions on topics such as education, immigration,income and employment.We focus on people between the age of 20 and 64 with an education levelhigher than high school, and who were full-time workers with positive wageincome in 2005.29 Regarding immigrants, we further focus on those whowere old enough to face problems adapting to their new environment, butwho were young enough to invest in education around the time of migra-tion. Therefore, the majority of our analysis uses immigrants who arrivedin Canada between the age of 15 and 29. Nevertheless, trying to make ourdata set more comparable with the ones used by the literature we replicatesome of our specifications using a sample of immigrants 15 or older at ar-rival. Non-permanent residents are removed from the sample since they arenot comparable to landed immigrants, and are not asked when they cameto Canada.30Our dependent variable is the logarithm of the average weekly wages.The average is constructed by dividing the total wage income in 2005 bythe declared number of weeks worked in the year. Since the census does notrecord weekly hours of work in 2005, we limit our analysis to full-time work-ers to have a better measure of the hourly price of labor. We also restrictthe sample to people with an average weekly wage earnings higher than 15times the province minimum hourly wage rate to minimize the problem oflow-wage outliers .31 Minor restrictions are imposed to exclude observationswith inconsistencies in key explanatory variables such as unspecified coun-try of origin (?Other?), location of study (?Outside Canada? or ?DistanceLearning?) or year of immigration (for the foreign born).32 A list of therestrictions is detailed in Appendix A.1. We have about 1.2 million obser-vations in our final sample, with an immigrant share of around 10 percent.We then create explanatory variables related to years of experience andyears of education. As in a standard Mincer earnings regression, total (po-to other countries, members of the Canadian Forces stationed abroad and all Canadiancrew members of merchant vessels?.29The question on the location where the highest degree obtained is only asked topeople with more than a high school degree.30Non-permanent residents are defined as persons living in Canada who have a Workor Study Permit, or who are claiming refugee status.31Appendix A.2 shows the minimum hourly wage by Province valid in 2005.32The Research Data Center data release policy prevents us from currently disclosingthe number of observations dropped with each restriction. However, the total number ofobservations eliminated are smaller than 5% of the final sample.122.1. Data and Descriptive Statisticstential) labour market experience is calculated as the difference between ageand years of education assuming that children start school at age seven.Since the 2006 census no longer asks for the number of years of schooling,this variable is imputed using the highest degree or diploma a person ob-tained (see Appendix A.3). We further separate work experience in Canadaand abroad based on age at immigration. Under the assumption that landedimmigrants who finished their studies abroad start working upon arrival, wecalculate their work experience in Canada as the difference between their ageat the time of the Census and their age at arrival. Foreign-born individualswho finish their studies in Canada are divided into three groups accordingto their age at arrival and highest degree attained. For those arriving at 18years old or younger, Canadian work experience is assumed equal to totalwork experience. For those arriving between the age of 19 and 22, Canadianwork experience is calculated as age (in 2006) minus age at arrival minus theimputed years of education in Canada assuming that those with a bache-lor?s degree or higher didn?t start their programs until arriving in Canada.33Lastly, for those who arrived after age 22 work experience is calculated asage minus age at arrival minus imputed years of education in Canada. Inthat case, however, we assume that those with a bachelor?s degree or higherfinished their bachelor?s degree before arriving in Canada. Individuals bornin Canada are simply assumed to have obtained all of their work experiencein Canada. For immigrants, work experience abroad is computed as thedifference between total work experience and work experience in Canada.The new question on location of study was introduced as part of a ma-jor revamping of the education section in the 2006 census. Another changeintroduced in 2006 was to simplify the set of questions about educationalattainment by simply asking what was the highest degree or diploma oneachieved.34 This suggests including a set of dummy variables for each edu-cation level in the wage regressions. We further simplify the empirical modelby grouping the educational degrees into four categories: trades certificates,college or university diploma below a bachelor?s degree, bachelor?s degree,and post-graduate degrees, and include them as dummies in the model (us-ing trade as the base category). The advantage of this approach relative33This implicitly assumes that there are no transfer mid-program.34In the 1981-2001 census more detailed questions were asked about years of educa-tion. This provides much more detailed information about the educational achievementof individuals without a high school diploma (e.g. 8 vs. 11 years of education) relative tothe 2006 Census that only records whether or not one completed a high school (or higher)diploma. Fortunately, this is of limited consequence for our analysis that only focuses onworkers with more than a high school diploma.132.1. Data and Descriptive Statisticsto one based on years of education is that non-linearities in returns to ed-ucation are directly captured by this set of dummy variables. To capturedifferences in the return to education obtained in Canada and abroad, theeducation dummies are interacted with a dummy for foreign education (forthe highest degree attained), which yields a total of seven binary variables(three for education attainment in general and four interaction terms). Forinstance, if a person got a bachelor?s degree outside Canada we would ob-serve the separate effects of having a bachelor?s degree in general and havinga bachelor?s degree from abroad.Since many other studies use years of education instead, we also presentresults where the information on the highest diploma is used to impute acorresponding number of years of education (e.g. 16 years of education fora bachelor?s degree). Following Friedberg we can then divide years of edu-cation into years of education in Canada and years of education obtainedabroad. We also distinguish between years of education above high schooland years of education above a bachelor?s degree. Thus, we have four vari-ables: years of education above high school in Canada, years of educationabove high school abroad, years of education above a bachelor?s degree inCanada and years of education above a bachelor?s degree abroad. This en-ables us to identify the returns to foreign-education for different categoriesof higher education. In principle, a foreign-born who finishes a master?sdegree in Canada has only one (or two) year(s) of education outside hercountry. In our specification there is a limit of four years of education abovehigh school (which is approximately the time needed to finish a bachelor?sdegree) and five years of education above bachelor?s degree (considering adoctoral degree as the highest education possible). People with bachelor?seducation and higher are inputted with the maximum years of educationabove high school. Hence, in the case of the foreign-born with a master?sdegree in Canada we would input four years of education above high schoolabroad and two years of education above bachelor?s in Canada.Country of origin and the country where the highest diploma was ob-tained are grouped in 22 and 19 categories, respectively. We identify the topten countries where immigrants get their education from (including Canada)and group the rest in broad geographic areas (such as ?South America?,?East Europe?, ?Africa?, etc.). The first ten countries represent more than80 percent of all immigrants. Similarly, we find the top ten countries oforigin and combine the rest in relatively homogenous geographic areas. Wealso include two more country of origin dummies for Pakistan and Romaniasince these two countries are among the top ten in terms of where the high-142.1. Data and Descriptive Statisticsest diploma was obtained.35 With Canada also included in the country oforigin list, in the end we have thirteen countries (the top ten plus Canada,Pakistan and Romania) and nine regions (three more categories for countryof origin than for location of study. Appendix A.4 shows the details).The 2006 census also included information about field of study. The in-formation is coded using the Classification of Instructional Programs (CIPCanada 2000) and is highly detailed. We aggregate this information intothe eleven major fields used in the public use files of the census. We thenuse this information to see to what extent foreign diplomas are differentlyportable into the Canadian labour market depending on the field of study.For example, while diplomas in education may be valued differently depend-ing on the country of origin, diplomas in fields based heavily on mathematicsmight be more portable since they are arguably less influenced by culturaland linguistic factors. Education, experience, location of study, country oforigin and field of study, along with some interactions among these variables,are the main variables in our analysis of the causes of the immigrant-nativewage gap.2.1.2 Descriptive StatisticsTable 2.1 shows the distribution of immigrants by country/area of originand year of arrival. The top ten countries account for 51% of all immigrantsin our sample. With the exception of the United Kingdom, all of the topfive countries of origin are located in Asia. Since our sample is limited toindividuals age 20 to 64 and immigrants who arrived at age 15 and above(up to 29), the earliest year of arrival is 1956. Table 2.1 shows that thedistribution of source countries has changed dramatically throughout theyears. Most of the immigrants who arrived between 1956 and 1970 werefrom the UK and the rest of Europe (21.9% from UK, 2.4% from France,1.2% from Poland, 7.7% from East Europe and 25.9% from the rest of thecontinent), whereas the majority of immigrants who arrived after 1990 arefrom Asia. Together India, Philippines, China, Hong Kong, Vietnam, andthe rest of the region made up 52% and 51% of all the immigrants whoarrived between 1991 to 2000 and after the year 2000, respectively.Table 2.2 shows that most immigrants received their highest diploma inCanada (close to 56%). Thus, most immigrants are able to invest in human35Note that these two countries are also among the top 15 countries of origin. Theclassification of country of origin must be at least as detailed as the classification forlocation of study to make sure location of study is not just a proxy for differences incountry of origin that haven?t been controlled for.152.1. Data and Descriptive Statisticscapital after first coming to Canada. Immigrants who arrived earlier aremore likely to have a Canadian diploma. In particular, more than 62% ofall immigrants who arrived before 1990 obtained their highest diploma inCanada. That number is around 55% for immigrants who arrived between1990 and 2000, and it drops to 26% for immigrants who arrived after 2000.This lower figure suggests that it takes some time for immigrants to acquiresome additional education in Canada, and that the 26% figure will likelyincrease with the passage of time.Most immigrants who did not acquire their highest degree in Canada gotit in their home country instead. Accordingly, the distribution of the locationof study for countries other than Canada closely mirrors the distribution ofcountry of origin of immigrants. Table 2.2 shows that the total share ofimmigrants who got their highest degree in countries other than Canadaranges from 0.7% to 5.8%. The U.K. and the rest of Europe are the mostimportant locations of study for immigrants who arrived between 1956 and1970 (around 23% of those immigrant got their highest there), but theirrelevance shrinks substantially after 1980. India, Philippines and Chinaincrease their positions after 1991. Nonetheless, even for immigrants whoarrived after 2000 the combined share of these countries never reaches 30%.Table 2.3 presents the distribution of natives and immigrants by field ofstudy. Foreign-born are relatively less likely to have diplomas in the fieldsof education, social sciences, agriculture and services. With the exceptionof agriculture one could argue that these fields require a higher level ofcommunication abilities, which may discourage people with these specialtiesto immigrate to Canada. Conversely, immigrants are over-represented infields that require more quantitative skills, such as mathematics, computerand information sciences, architecture, engineering, and related technologies.Consistent with other Census-based studies, table 2.4 shows a strongclustering of immigrants in large census metropolitan areas (CMAs), and inparticular Toronto, Vancouver and Montreal). 66% percent of immigrantslive in Toronto, Vancouver and Montreal, compared to only 30% of natives.The twenty largest CMAs listed in the Table account for over 90% of im-migrants. While the remaining CMAs and non-CMAs account for close to40% of natives, less than 9% of immigrants live there.The age distribution of immigrants and natives reported in table 2.5show some important demographic differences between the two groups. Im-migrants in our sample (those who came to Canada between the age of 15and 29) are generally older than natives. For instance, there is a higher frac-tion of immigrants than natives age 50 to 64 (30% and 23% respectively),but fewer immigrants of the age of 20 (12% versus 20%). Table 2.6 shows162.1. Data and Descriptive Statisticsthat immigrants are also more educated than natives. In the table we groupthe eleven levels of education from the Census into seven categories andimpute the number of years one would normally take to complete the cor-responding diplomas. Relative to natives, immigrants are less likely to havesome vocational education or a community college/CEGEP diploma (45%compared to 60% for natives), but more likely to have a diploma above abachelor?s degree (17% for immigrants compared to 11% for natives).As we mentioned earlier more than half of all immigrants obtained theirhighest degree in Canada. Table 2.8 show that this fraction depends verymuch on age at arrival. Immigrants who arrived to Canada at an olderage are more likely to have obtained their highest education in their homecountries, while immigrants who arrive young are logically more inclined tofurther invest in their education in the new country. Of the people whoarrived at age 19, only 8.4% have obtained their highest degree abroad,compared to 71% for immigrants who arrived at age 29. Moreover, given thatwe only analyze people with more than a high school degree, all immigrantswho arrived before age 19 obtained their highest diploma in Canada.Table 2.8 also clearly illustrates how the direct question on locationof study helps us identify where immigrants acquired their human capital.Only a few immigrants have diplomas like PhDs or MDs that one cannotcomplete, in principle, before the age of 25. Hence, very few immigrants age25 and over would be imputed some Canadian education using a Friedberg-type imputation approach. By contrast, Table 2.8 shows that around 40percent of immigrants who came at age 25 or 26 have a Canadian diploma.This fraction remains substantial (about 30 percent) even for immigrantswho came to Canada at age 29. More generally, the table shows a relativelysmooth decline in the share of immigrants with a Canadian degree as afunction of age at immigration. By contrast, imputation procedures thathave been used in the past would predict a much sharper drop in the shareof immigrants with a Canadian degree as a function of age at immigration.36The means and standard deviations of the main analysis variables arereported in Table 3.4. The mean of log weekly wages for natives is onlyslightly larger than the mean for immigrant (0.04 difference in logs, i.e. a 436Remember that in standard imputation procedure a la Friedberg people are assumedto start their education at age seven and continue without interruption. Under this as-sumption, people should complete a bachelor?s degree at age 23. Immigrants arrivingbefore age 23 would all be imputed a Canadian diploma, while those arriving after age 23would all get imputed a foreign. This leads to a complete drop in the fraction of immi-grants with a Canadian diploma around age 23 (from 100 to 0 percent), while the actualdata shows a much smoother decline.172.2. Empirical strategypercent difference). That said, the means reported in the table also confirmthe earlier evidence that immigrants are both older (average age of 42.37versus 40.37) and more educated (15.14 against 14.68) than natives. Sinceearnings increase in both age and education, controlling for these factors inthe earning regression should make the 4 percent earnings gap even larger,something that is confirmed in the regressions reported below. Immigrantsalso work as many weeks per year as natives (around 47) and have a largertotal work experience (21.23 versus 19.7), though understandably a smallerCanadian work experience (17.26 versus 19.7).2.2 Empirical strategyWe estimate the logarithm of weekly wages on a foreign-born dummy,a set of demographics and human capital variables (gender, education andwork experience) and several fixed effects (country of origin, location ofstudy, city of residence, etc.). The focus is on the magnitude of the foreign-born dummy as it would indicate how much of the immigrant/native gapwe are not able to explain. An initial specification restricts the regressioncoefficients (except the constant) to be the same for immigrants and theCanadian born. Consider a standard (log) earnings equation for immigrantswi = ?I +Xi?I + ?iand for Canadian-born workerswi = ?C +Xi?C + ?iwhere wi is the logarithm of weekly wage, Xi is a vector of covariates (in-cluding work experience and education), and ?i is an error term that satisfiesthe usual assumption (E(?i|Xi) = 0).The equation would condition the effect of the covariates on earnings tobe the same for immigrants and natives.wi = ?C + ?IIi +Xi? + ?i (2.1)The mean earnings gap (wC ? wI) would come from the difference inthe average value of covariates ((XC ?XI)?) and the specific constant forimmigrants (??I), where Ii is a dichotomous variable indicating whetherperson i is an immigrant. The sign, size and significance of ?I relates to theunexplained part of the earnings gap and our different specifications wouldshow how much of the wage gap we are able to disentangle.182.2. Empirical strategyThis first specification is, however, restrictive. We make efforts to dis-tinguish education and experience acquired in Canada from education andexperience obtained abroad. The estimations control for language skills (twodummies for English and French as mother tongues), metropolitan area ofresidence (being Toronto the omitted CMA), province of residence (makingOntario the base case), country/area of origin (using Eastern Europe as theomitted region) and, in some cases, field of study (using visual arts as theomitted category). Still, the innovation of the paper centers in the incor-poration of dummy variables indicating the location where people receivedtheir highest degree or diploma. The central equation can be generalized to:wilasf = ?C + ?IIi +Xi?C(1?Di) +Xi?IDi + ?l + ?a + ?s + ?f + ?ilasf(2.2)Where Di is an indicator of where the human capital was acquired, oneif it was acquired out of Canada and zero otherwise. The additional in-dexes specify the effect of knowledge of official languages (?l), residence ina particular province and city (?a), the country/area of origin (?s) and thefield of study (?f ). The use of country/area of origin fixed effects (?s) in anestimation changes the interpretation of the immigrant coefficient Ii. It nolonger represents the average unexplained earnings gap between natives andall immigrants, but only the unexplained gap between the country of originomitted category, Eastern Europe in our case, and natives. Nevertheless,Eastern Europe was chosen as the base case because its country coefficientwas similar to the original immigrant dummy. We hope its changes wouldreflect the average changes of all the countries of origin. In strict though,each country/area of origin fixed effect becomes the particular unexplainedwage gap between immigrants from that country and Canadians. The im-migrant coefficient Ii, becomes the unexplained wage gap for all countriesof origin.To further investigate the pattern of the wage gap, we estimation wageequations separating the sample by gender.37 Also, in some of the specifica-tions we decompose the immigrant indicator Ii into three dummies accordingto the immigrants? age at arrival: 15 to 19, 20 to 24 and 25 and 29 yearsold. Lastly, some of the specifications include interactions between the coun-try/region where it education was acquired and the education level or thefield of study and country/origin. These interactions enable the comparison37We present this separation for only one of our specifications but the key results aresimilar across specifications.192.3. Findingsof the effects of obtaining a degree in particular country versus obtainingit in Canada; or the comparison between studying a field in a particularcountry versus studying in Canada.2.3 Findings2.3.1 Replicating Friedberg?s resultsIn order to highlight the improvement of our education and work expe-rience imputations from the literature, we replicate some of Friedberg?s firstestimations. We want to show that we could get similar results if we restrictourselves to the standard computations. For education we create one vari-able for the total years of education in Canada and one for the total yearsof education abroad. Canadians are assumed to have no years of educationabroad and immigrants are given years of education in Canada only if theirage at arrival minus six is equal (or less) than their total years of education.In that case, years of education abroad would be their age at arrival minussix and years of education in Canada would make up the difference to thetotal. Canadians are imputed zero work experience abroad. Immigrants areonly assigned foreign work experience if, under this rule, they would havefinished their studies before coming to Canada. That is, if they have noeducation in Canada. For them the foreign work experience would be thedifference between their age at immigration and the assumed age at whichthey finish their studies (total years of education plus six).The results are indeed similar to the ones Friedberg found in her research(see tables 2.10 and 2.11). With only years of education, work experienceand years since arrival as explanatory variables, the immigrant dummy issignificant with a coefficient of -0.22, indicating the average wage gap thatimmigrants face. The identification of the source of the human capital,shows that Canadian years of education and work experience in Canadaare valued more than their foreign counterparts. Just as with Friedberg,the disentanglement of human capital turns the sign of the foreign-borncoefficient from negative to positive (to 0.24) and, in our case, statisticallysignificant. The interactions of the foreign-born dummy with measures ofCanadian human capital suggest that the return to education in Canada issimilar for natives and immigrants. Still, work experience gained in Canadawould have a greater positive effect for immigrants than for natives.Separating the analysis by regions gives us some parallel to Friedbergsfindings as well (see table 2.11). Considering all immigrants, Canadian hu-man capital (education and work experience) turns out to be more valued202.3. Findingsthan foreign human capital. However, if we compare Canadians to only im-migrants from the West the difference in the returns to education is smalland their work experience is actually more valued than the Canadian.38 Thisis also one of Friedberg?s findings. If, instead, we focus on immigrants fromIndia, China and the Phillipines we find that the return to their foreign workexperience is much lower than the return to work experience in Canada.In order to make our results more comparable to the literature, we ex-tend our sample to include all immigrants arriving at an age of 15 or olderand recalculate the tables.39 We do this in tables 2.12 and 2.13. The newresults are very similar to our initial ones. The immigrant dummy in theinitial specification of table 2.12 has a negative and significant coefficient.It seems that the inclusion of older arriving immigrants in our sample al-most doubles the size of the (negative) immigrant dummy coefficient (-0.22in table 2.10 and -0.39 in table 2.12). This is consistent with older arrivingimmigrants having more difficulty integrating to the Canadian labour mar-ket. The results in the other two columns (of table 2.12) are also similarto our previous findings. Human capital acquired in Canada is valued morethan human capital acquired abroad. The returns to education and workexperience by source is very similar in both tables. Also, just as before,the immigrant dummy changes signs in the second and third specifications(goes from an initial -0.39 to 0.345 and 0.347). A simple explanation for thischange in sign refers to an interpretation of the immigrant dummy as the(predicted) wage gap for immigrants with no education or work experience.The identification of lower slopes of education and work immigrants forcesan out of sample positive wage gap. lastly, in the third column we also findthat work experience in Canada would have a slightly more positive effect(a coefficient of 0.0041) for immigrants than for natives.Likewise, the results of table 2.13 coincide with the ones previouslyfound. Grouping all immigrants gives lower returns to foreign human cap-ital, particularly to work experience abroad (0.002 per year worked versus0.02 per year worked in Canada). This broad result comes from mixing im-migrants from the different areas of origin. Again, immigrants from westerncountries show similar returns to education and work experience abroad to38In the context of this estimation USA, Austria, Belgium, France, Ger-many,Liechtenstein, Luxembourg, Monaco, Netherlands, Switzerland, Ireland, Denmark,Finland, Iceland, Norway, Sweden, UK, Greece, Italy, Portugal, Spain, Australia and NewZeland are considered the west.39Indeed, Friedberg (2000) compares male immigrants and natives between the ages of25 to 65 without a restriction on immigrants? age at arrival. She does this to facilitate thecontrast between her results with the ones of the previous US literature.212.3. Findingsthe ones in Canada. In contrast, immigrants from India, China and thePhilippines present a reduced return to education abroad and even a nega-tive (but very close to zero) return to work experience abroad.Finally, we use the new information on the location of study of thehighest educational degree to compute human capital acquired in Canadaand abroad. In tables 2.14 and 2.15 we find that the immigrant dummycoefficient, the return to education and work experience (abroad and inCanada) and the results when we separate the sample (natives, all immi-grants, western countries, India-China-Philippines, etc.) coincide with ourprevious findings (and in broader sense, with the results of Friedberg). Theadvantage of using the location of study variable is that allow us to iden-tify more accurately where was the human capital acquired and incorporatethis information directly into our estimation. Indeed, the next section triesspecifications with and without a location of study fixed effect.402.3.2 Base SpecificationsOur first tables estimate the native/immigrant wage gap under severalspecifications (see tables 2.16,2.18,2.19 and 2.20). In all of them the coef-ficient of the foreign-born dummy experiences a reduction between 30% to50%. The largest reductions are associated with the incorporation of loca-tion of study fixed effects or the separation of the human capital acquiredin Canada from the one obtained abroad. For both measures the new in-formation available in the Census is paramount. It should be noted thatthe statistical significance of the results will not be mentioned regularly.Given the large number of observations the majority of our results are sta-tistically significant at 1%. The focus of this chapter is on interpreting ourexplanatory variables and the size of the immigrant dummy coefficient.An initial sensible finding is that immigrants arriving at a younger ageappear to have less problems integrating into the Canadian labour market.Columns five and six of table 2.16 show that foreign-born arriving at agesbetween 15 to 19 have a wage gap of about 2.6%. In comparison, the wagegap for people arriving at ages between 20 to 24 is 8%, and for peoplearriving at ages 25 to 29 is 19%. The inclusion of location of study fixed40Extending the sample of immigrants to include those who arrived at an age older than29 might have helped the similitude of results between tables 2.12 and 2.14. The differencesbetween Friedbergs imputation method and the computation using the direct location ofstudy variable are relevant for young arriving immigrants that have the opportunity ofstudy upon arrival (and could be misattribute under the Friedberg methodology). Thedirect identification of location of study is not so critical for immigrants arriving at anolder age.222.3. Findingseffects have a higher impact on the wage gap of immigrant who arrive at anolder age, reducing their wage gap between 34% to 39%. The effect on thewage gap of people arriving at 19 or younger is negligible.The incorporation of location of study fixed effects lets us graph theaverage negative effects of obtaining (the last) educational degree from aparticular country in comparison to Canada (the base category). Figures2.1 and 2.2 show that Pakistan is the most penalize location of study. India,China and the rest of Asia (South East Asia, Western and Central Asiaand the rest) follow suit as less advantageous places to study. Studying inthe U.S., Oceania and the rest of Europe (basically, continental WesternEurope) appears to be considerate as good as studying in Canada, whilestudying in the U.K. has, in fact, a positive premium payoff. This patternis consistently depicted by all the even columns of table 2.16.41The comparison of the wage gap by gender in table 2.17 gives a particularinsights on the assimilation process of immigrants. The foreign-born dummycoefficient in columns one and five illustrates that male immigrants face asomewhat larger wage gap than females (-12% and -9%, respectively). Also,location of study explains a larger share of women wage gap. The inclusionof location of study fixed effects in column six reduces the coefficient in asizeable manner for females (from -9% to -2%). Its inclusion (in column two)for the male sample produces a smaller reduction (from -12% to -8%).42Tables 2.18 and 2.19 show the largest decrease in the immigrant/nativewage gap (from around -12% to around -3%).43 They present differentways to incorporate education into the earnings equation (years above ahigh school and bachelor vs degree achieved), but for both their highestreductions in the immigrant dummy coefficient come from the separation offoreign and Canadian human capital (see columns 3,5 and 7 in both tables).We see this as evidence of the better classification we are able to achievewith the new data. Location of study fixed effects are included in all evencolumns, however their incorporation starts losing impact by column six. By41Even though the number of years after arrival is incorporated in the work experiencevariable, at this moment we can not completely rule out the possibility of a discontinuityin the immigration assimilation process for immigrants arriving before the 1970s as muchof UK immigrants did.42With the incorporation of country of origin fixed effects (on columns 3 and 4, and 7and 8) the foreign-born dummy coefficient not longer represent the average wage gap ofall immigrant but recovers the wage gap of the omitted category (Eastern Europe in thiscase). However, as our base category relates to the average changes of the countries oforigin we can still claim that incorporating the location of study in the regression helpsexplain the immigrant-native wage gap.43Note that the first two columns of both tables come from table 2.16.232.3. Findingsthen, the human capital separation is already included in the specifications.The difference between foreign and Canadian human capital is clearlyshown in table 2.19. There education abroad is specified as an additionaleffect on the achieved degree dummy variables. The major difference inthe returns to education happens at the bachelor and graduate level. Thosewith a foreign bachelor degree or higher have between 6% to 11% less incomethan people who finish a similar degree in Canada. The difference is minorfor people with university certificate below bachelor?s level and positive forforeign education in the case of trades certificates. The most straightforwardexplanation is that occupations that do not require higher levels of educationtend to involve lower levels of communication skills and a higher componentof manual skills, which are quite comparable across countries.Curiously, in most specifications, people with an above bachelor?s edu-cation have a smaller native/immigrant wage gap than those with only abachelor?s degree. Again, it could be argued that occupations for very spe-cialized workers are homogenous across countries; hence the penalty fromforeign education would be smaller.44Interactions between the highest degree achieved and the last location ofstudy are presented in table 2.20. Country of origin fixed effects are addedas extra controls in the last columns (in addition to CMA and province fixedeffects) but the main finding of the table remains. Recognition of educationalachievements is highly related to the last location of study. Even people witha trade certificate have a negative premium if their last location of study wasthe Philippines, China or Pakistan. Bachelor?s and above bachelor?s degreetend to have the largest negative coefficients but education coefficients varysubstantially by location of study. The contrast is greater between the oldsource areas of origin (UK, US and non-East Europe) and the new sourcesof immigrants (India, China, Philippines, Africa and Pakistan; with theexception of Romania). While a bachelor degree obtained in Philippines,India or China would create a negative wage gap of about 33% to 15% forthe newcomer, a bachelor degree obtained in the UK would have a positivewage gap between 8% to 13%.As before, we decide to redo some estimations after expanding the sampleto include immigrants arriving at an age older than 29 (so all immigrantsarriving at 15 and older are considered). In this case, we decide to replicatetables 2.16 and 2.19, which gives us relatively similar but interesting findings.44The possibility that immigrants with a graduate education abroad could be morelikely to arrive with an arranged employment can not be disregarded. Information onarranged employment before arriving is not available in the census.242.3. FindingsA consistent result in table 2.21 (the replication of table 2.16) is that theinclusion of location of study fixed effects markedly reduces the immigrantdummy coefficient. The comparison of columns one and two, and threeand four shows that the size of the immigrant coefficient decreases in morethan 50%. The initial magnitude of the unexplained wage gap (in columnsone and three) is, however, twice as large as previously found. Columnsfive and six shed some light on the factors behind the wage gap increase.Immigrant/native wage gap for immigrants arriving at an age older than 29is considerably larger. The immigrant coefficients for those arriving between15 and 19, 20 and 24, and 25 and 29 remain quite similar to the previouslyfound (compare columns five and six of tables 2.16 and 2.21); in a rangebetween -0.03 and -0.18. However, the coefficient jumps to a range of -0.31to -0.42 for those arriving in their thirties; it jumps again to a range of -0.53to -0.61 for those arriving in their forties and it is -0.64 for those arrivingat 50 or older (see column six). These are pretty large numbers for theimmigrant/native wage gap. Moreover, the incorporation of the locationof study fixed effects loses effect on the immigrants dummy as the age atarrival increases. While the reduction in the coefficient is about 42% forimmigrants arriving between the ages of 25 and 29, the reduction for thosearriving between 40 and 44 is about 25% and only 22% for those arriving at50 or older.The results of table 2.22 confirm that, just as in table 2.19, the iden-tification of foreign human capital explains a significant part of initial na-tive/immigrant wage gap. Specifically, the separation of education and workexperience by source (Canadian vs. foreign) decreases the immigrant dummycoefficient from -0.23 to -0.02 (column one to seven). Moreover, adding loca-tion of study fixed effects always reduces the immigrant coefficient (70% to27% depending on the specification). From column seven to eight, where thesource of both education and work experience is identified, the immigrantcoefficient goes down to -0.016 (from -0.02) when the location of study fixedeffects are added. The order of the coefficient of location of study is almostthe same as previous tables.In tables 2.16 to 2.19 the immigrant dummy represents the average wagegap between Canadians and foreign-born. Heterogeneity in the effect ofcountry of origin on the wage gap could still be present and relevant. Giventhat about half of immigrants obtained their last degree in Canada, theclear identification of country of origin and location of study is required.We explore this issue on tables 2.23 to 2.25.252.3. Findings2.3.3 Separating the foreign born wage gap by country oforiginWe repeat most of the specifications shown in tables 2.16 to 2.19 whileadding country of origin, location of study and field of study fixed effects.Given the pattern found in figures 2.1 and 2.2 we decide to group locationof study into broader areas, namely: Canada; the West (including Oceania);East Europe (including Romania and Poland); China and West and CentralAsia (including Hong Kong); India, Pakistan and the rest of Asia; the restof America; South-East Asia (including the Philippines) and Africa. A moremanageable number of locations lets us interact them with other variables(such as education in table 2.20)We find a considerable disparity in the country/area of origin fixed effectsrelated to geographical areas. Figures 2.3 to 2.9 show that countries fromAsia (not including South-East Asia or Hong Kong) have a larger negativepremium. The lowest coefficient consistently relates to Pakistan. The nextgroup includes South and Central America, Africa and East Europe withcoefficients ranging from -0.2 to -0.1. The US, France, Oceania, South-EastAsia, Hong Kong and the rest of Europe have coefficients around -0.10.United Kingdom is the only country/area that invariably shows a positivecoefficient (around 0.05).The majority of immigrants from the UK, unlike most foreign-born inthe sample, came to Canada before 1980. The UK coefficients for country oforigin and location of study fixed effects are the only ones robustly positive,suggesting that long assimilation processes (for immigrants with more than26 years in Canada) may not be fully incorporated in the work experiencevariable. At least not for UK immigrants.The disparity is reduced by incorporating information on the location ofstudy and foreign work experience. This further confirms that the negativecountry/area premium is driven by the lower valuation that the Canadianlabour market assigns to education and work experience abroad. In par-ticular, the incorporation of the location of study dummies in table 2.23(columns two, five and eight) reduces the size of the coefficients of EastEuropean countries (including Romania and Poland) and South-East Asia(see figure 2.4) to close to zero. The same pattern is observed in the moredetailed specifications of tables 2.24 and 2.25 (Figures 2.6 and 2.8 graphicthe change in the coefficients).Following our specifications tables 2.24 and 2.25 separate Canadian andforeign work experience as well as Canadian from foreign education (thoughin different manners). Throughout their columns the size of the country262.3. Findingsdummies decreases. The separation of work experience is particularly im-portant to explain the negative premium of most Asian countries. The com-parison of columns 4 and 7 of tables 2.24 and 2.25 respectively (see figures2.6 and 2.8), shows a sizeable reduction in the coefficients of Pakistan, India,China, Philippines, West and Central Asia and the rest of Asia (Africa andRomania as well). Though the coefficients stay lower than -0.10, their rangedrops from -0.11 to -0.36 (in column 1 of table 2.24) to -0.06 to -0.23 (incolumn 7). The addition of location of study fixed effects to these specifi-cations lowers their negative values even more (see columns 8 on figure 2.6and figure 2.8).Curiously, the inclusion of the field of study seems to have only a minoreffect on the country dummies. Columns 6 and 9 of tables 2.23 and 2.24 and2.25 do not present an extra visible reduction of the country dummies thanwhat is achieved by the inclusion of location of study fixed effects and thedetailed specifications that separate human capital origin (see figures 2.5,2.7 and 2.9 for each table).Overall, the education and work experience coefficients are similar tothe ones found in the initial estimations, adding country of origin and fieldof study fixed effects does not change them noticeably. Higher levels ofeducation still pay a premium; those with a bachelor?s degree have a higherwage than those without one and less than people with a graduate educationlevel.Table 2.25 shows again that, up to bachelor?s degree, the higher theeducation level the larger the difference between Canadian and foreign edu-cation. Graduate foreign education is valued less than Canadian education,but the gap is smaller (or equal) than the one for immigrants with a bache-lor?s degree. Also, for education just above high school there does not seemto be a problem of human capital transferability. The coefficient of the for-eign education premium (of trade certificates) is negative but not significant.Foreign work experience is heavily discounted though. At best, the first or-der difference (without considering the square term) between Canadian andforeign years of experience is five to one; at worst, twelve to one. Withcoefficients of foreign experience close to zero. It should be noted, however,that the average foreign years of experience in this sample is less than fouryears. Initial jobs in a given career may not be valued as much as seniorpositions.In table 2.26 we replicate some of the columns of table 2.25 adding tothe sample immigrants who arrived at an age older than 29. Regarding ed-ucation, we find very similar returns and patterns for Canadian and Foreigneducation. Though the returns for foreign education become more negative272.3. Findingsin columns three and four, most of the difference (with respect to table 2.25)goes away by identifying the source of work experience and incorporatinglocation of study fixed effects. The key results remain; Canadian educa-tion is valued more than foreign education, there is no significant negativepremium for acquiring a trade certificate abroad (instead of in Canada),and the negative premium for obtaining a graduate education abroad is atleast no larger than the premium for acquiring a bachelor?s degree. The re-turns to work experience do not present important changes. Years workingin Canada are heavily more valued than years working abroad, which haveclose to zero returns. Likewise, the country fixed effect coefficients recov-ered show similar order and about the same magnitude than the ones obtainpreviously.2.3.4 Identifying the effect of field of studyTables 2.27 to 2.29 explore the heterogenous premiums of field of study.Though the inclusion of field of study fixed effects does not seem to have asubstantial explanatory power on the country of origin wage disparity, wecompare the wage premiums for the fields of study and check if they changeaccording to where education was acquired. The base specification is thesame for all tables; three educational dummies and separated Canadian andforeign work experience variables (in addition to gender and mother tonguedummies). The differences across tables and columns lie in the fixed effectsand interactions added.Using ?Visual Arts? as the base category we find in table 2.27 (and thecorresponding figure 2.11) that two out of the four highest rewarded fieldsof study require the use of significant mathematical skills: ?Architecture,Engineering, and Related Technologies? and ?Mathematics, Computer andInformation Sciences?. The abilities they relate to are highly tradable andthe jobs associated with them are not likely to demand advanced communi-cation skills. The other two fields relate to administration, accounting andcomputer science (?Business, Management and Public Administration?),and to health (medicine, nursing, optometry, dentistry and veterinary) andrecreation (?Health, Parks, Recreation and Fitness?).45 Though they may45 The Classification of Instructional Programs separates each of these two primarygroupings into three sub-categories: ?Business, Management, Marketing and Related Sup-port Services; Accounting and Computer Science and Public Administration and SocialService Professions? for the first grouping and ?Health Professions and Related ClinicalSciences, Dental, Medical and Veterinary Residency Programs, and Parks, Recreation,Leisure and Fitness Studies? for the second one.282.3. Findingsrequire more communication skills, they also face a higher labor demand.Humanities is the only field with a negative wage premium and is markedlylower than the penultimate (Social Sciences and Law).46 In addition to hav-ing a lower requirement of mathematical abilities humanities? professions(such as languages and literature, classical studies, history, theology,etc.)need a high level communication skills.The interactions between fields of study and the foreign-born dummy arefor the most part statistically not significant. With large standard errors,it can not be argued that immigrants in general face a different premiumon their fields of study. Though exceptions can be made for ?Mathematics,Computer and Information Sciences?, ?Education? and ?Agriculture, Natu-ral Resources and Conservation?. Figure 2.12 shows that foreign-born whospecialized in mathematics have a higher premium on their wages than theirCanadian counterparts. The opposite occurs for foreign-born who special-ized in education or agriculture.We can further separate the effect of the field of study by where theeducation was acquired. Under the presumption that the wage premiumfor each field of study depends on where was the education obtained, table2.29 interacts the location of study fixed effects with the field of study coef-ficients. The general location of study fixed effects are shown in figure 2.13.Consistent with the previous results the west, which includes the U.S., U.K.,Oceania and west Europe, has the highest and closer to zero coefficient ofthe areas of study. Unlike any other area studying in the west does notcreate a structural lower wage than studying in Canada.Figures 2.14 and 2.15 present the coefficients of the interactions. Wefind dispersion in the returns to the field of study within and across thelocations. The coefficients for immigrants coming from the ?West? havethe smallest standard errors and are relatively close to the zero line. Also,as in figure 2.11, fields with high mathematical skills requirements (suchas ?Architecture, Engineering, and Related Technologies?, ?Mathematics,Computer and Information Sciences? and ?Physical and Life Sciences andTechnologies?) tend to be valuated more. The exception is ?Social andBehavioral Sciences and Law? who curiously has the the highest payoff.The rest of the regions don?t share the same order or precision of thewest estimates. Different fields of study have a different payoff dependingon the location of study. Though most of the interactions are negative theirlarge standard errors make them statistically not different from zero. Still,46 The grouping of Social Sciences includes legal professions such as law, legal researchand legal support services292.4. Conclusionsit can be seen that ?Mathematics, Computer and Information Sciences? isamong the top rewarded fields of study in all the regions.2.4 ConclusionsWe use new information available in the 2006 Canadian census regardingthe location where the highest degree of education was attained to betterestimate Canadian and foreign human capital (education and work experi-ence) for immigrants and natives. The identification of the human capitalsource explains up to 70% of the native-immigrant wage gap. Our estima-tions are able to reduce the native-immigrant wage gap from around 11%to close to 3%. The separation of both education and work experience intonative and foreign sources is a deciding factor in the reduction of the wagegap.Our base specifications show that foreign-born immigrants between theages of 15 and 19 at arrival have a lower wage gap (averaging 2.6%) asthey likely acquire a significant part of their human capital in Canada. Forthem, the inclusion of location study fixed effects does not change the wagegap. Immigrants who arrived at 25 to 29 years old have a higher wage gap(around 18.6%) and location of study fixed effects explain about a third ofthe gap (it decreases to 12.2%).The incorporation of location of study fixed effects provides evidence thateducation obtained in Asia tends to be less valued than education obtainedin South America, Africa and East Europe. In turn, education obtained inthese regions is less valued than education from Oceania, the U.S. and therest of continental Europe. Lastly, a UK education appears to have greatervalue than even its Canadian counterpart.We explore how the inclusion of location of study fixed effects affectseach country or region of origin by breaking up the immigrant dummy.The different specifications visibly and consistently reduce the country fixedeffects coefficients. The reduction is sizeable for Pakistan, India, China andPhilippines. A smaller reduction for Europe, South-East Asia, Hong Kongand the US drives their coefficients close to zero. The UK country of origindummy has the only persistently positive coefficient.The inclusion of the field of study does not seem to markedly improvethe explanation of the wage gap. Still, two out of the four highest rewardedfields of study require the use of significant mathematical skills (?Archi-tecture, Engineering, and Related Technologies? and ?Mathematics, Com-puter and Information Sciences?), while the other two relate to business and302.4. Conclusionscomputer science (?Business, Management and Public Administration?) orhealth (?Health, Parks, Recreation and Fitness?).312.4.ConclusionsTable 2.1: Immigrant?s Top Countries of Origin by Year of ArrivalTotal 1956 to 1970 1971 to 1980 1981 to 1990 1991 to 2000 After 2000Top Ten countriesIndia 9.3 3.9 7.3 6.7 11.3 16.6UK 7.8 21.9 13.0 6.6 2.9 2.2Philippines 7.6 2.5 7.5 7.0 9.5 8.1China 5.9 2.7 3.2 3.1 8.2 12.5Hong Kong 5.1 1.9 6.4 6.5 5.8 0.6US 3.9 6.7 6.7 3.6 1.9 2.2Poland 3.2 1.2 1.2 7.6 3.0 0.9Jamaica 3.0 3.9 4.5 3.8 2.0 0.7Vietnam 2.8 . 3.9 5.0 1.9 0.7France 2.3 2.4 1.7 1.6 2.3 4.3Two other CountriesPakistan 1.8 . 1.1 0.8 2.7 3.8Romania 1.7 . 0.4 1.0 2.7 3.7Rest of the WorldAfrica 7.8 3.9 7.0 7.7 8.3 10.6Rest of America 7.3 7.1 8.4 9.3 6.0 4.6Rest Europe 7.1 25.9 9.2 6.2 2.7 2.2East Europe 5.9 7.7 3.0 3.2 9.0 7.5W. and C. Asia 5.2 1.5 2.7 6.5 6.7 6.3South America 5.0 2.7 6.6 5.8 3.7 5.4Rest Asia 4.4 1.1 2.5 4.0 7.0 4.6South-East Asia 1.8 . 2.3 3.0 1.3 1.1Oceania 1.3 1.6 1.4 1.2 1.2 1.4Total 100 9.2 23.8 23.1 30.8 13.2Numb. of Observ. 651750Note: A Missing value ?.? indicates that the cell has less than 100 observations.32Table 2.2: Immigrant?s Top Locations of Study by Year of ArrivalTotal 1956 to 1970 1971 to 1980 1981 to 1990 1991 to 2000 After 2000Top Ten countriesCanada 55.91 64.85 62.03 64.03 55.09 26.25India 5.8 1.17 3.5 3.44 7.12 14.26UK 5.24 10.83 8.95 4.57 2.29 2.69Philippines 5.05 1.72 4.94 3.94 6.01 7.25US 3.07 2.65 4.26 2.56 2.43 3.6China 2.43 . . 0.75 3.72 8.12France 1.65 1.43 1.01 0.9 1.75 4.07Poland 1.63 . 0.63 4.25 1.29 0.58Romania 1.06 . . 0.42 1.62 3.26Pakistan 1.02 . 0.45 0.4 1.33 2.87Rest of the WorldEast Europe 3.23 3.55 1.65 1.92 4.55 5.08Rest Europe 2.9 8.66 3.55 2.5 1.41 1.94Africa 2.25 . 1.64 1.79 2.29 5.18Rest Asia 2.16 0.94 1.92 2.13 2.63 2.4Rest of America 1.99 1.35 1.6 2.29 1.8 3.05W. and C. Asia 1.76 . 0.72 1.58 2.26 3.87South America 1.47 . 1.41 1.21 1.15 3.53Oceania 0.7 . 0.62 0.44 0.66 1.29South-east Asia 0.7 . 0.8 0.9 0.6 0.7Total 100.0 9.2 23.8 23.1 30.8 13.2Numb. of Observ. 651750Note: A Missing value ?.? indicates that the cell has less than 100 observations.332.4. ConclusionsTable 2.3: Distribution of Immigrants and Natives by Field of StudyNative Immigrant TotalEducation 7 4 7Visual Arts 3 3 3Humanities 5 5 5Social Sciences and Law 10 8 10Business and Pub. Adm 23 24 23Physical and Life Sciences 3 4 3Math, Comp., Information Sciences 4 8 4Architecture, Engineering, others 25 28 25Agriculture, Nat. Resources 3 1 2Health, Parks, Recreation 12 11 12Serv: Pers., Transp., Security 6 4 6Total 100 100 100342.4. ConclusionsTable 2.4: Distribution of Immigrants and Natives by CMAof ResidenceImmigrant Native TotalTop 20 CMAsToronto 40.9 11.0 14.1Vancouver 13.0 5.1 5.9Montreal 12.4 12.9 12.9Calgary 4.7 4.0 4.0Ottawa 4.1 4.5 4.4Edmonton 3.7 3.7 3.7Hamilton 2.2 2.1 2.1Winnipeg 2.1 2.3 2.2Kitchener 1.5 1.4 1.4London 1.1 1.5 1.5Windsor 1.0 0.9 0.9Oshawa 0.9 1.1 1.1Victoria 0.9 1.1 1.1St. Catharines?Niagara 0.8 1.1 1.1Quebec 0.5 3.6 3.3Abbotsford 0.5 0.4 0.4Halifax 0.4 1.7 1.6Guelph 0.4 0.4 0.4Barrie 0.3 0.6 0.6Saskatoon 0.3 0.9 0.8RestOther CMAs 5.3 21.8 20.1Non-CMAs 3.2 18.0 16.4Total 10.3 89.7 100Note: Numbers are rounded up to one decimal point.352.4. ConclusionsTable 2.5: Age Distribution of Immigrants and NativesNatives Immigrants Total20 to 29 years old 20.2 12.3 19.430 to 39 years old 26.8 32.8 27.440 to 49 years old 29.8 25.1 29.350 to 59 years old 20.2 23.4 20.560 to 64 years old 3.1 6.7 3.4100 100 100362.4.ConclusionsTable 2.6: Distribution of Immigrants and Natives by Years of EducationGroup Assigned Yrs. of Educ. Natives Immigrants TotalCommunity College/CEGEP (3 - 12 months) 13 26.9 19.0 26.1Community College/CEGEP ( - more than 2 years) 14 32.6 25.8 31.9University certificate below bachelor level 15 6.6 12.0 7.1Bachelor?s degree 16 23.5 26.4 23.8University certificate above bachelor level 17 3.4 4.3 3.5Master?s degree or Degree in medicine 18 6.2 10.8 6.7Earned doctorate degree 21 0.9 1.8 0.9Total 100 100 100It includes trades certificate, registered apprenticeship certificate and CEGEP (between 3 to 12 months).It includes degrees in medicine, dentistry, veterinary and optometry372.4. ConclusionsTable 2.7: Share of Immigrants with a Canadian Degree by EducationCategoryDistribution Foreign degree Canadian DegreeTradea 19.0 16 22Below Bachelorb 37.7 36 39Bachelor 26.4 31 23Above Bachelorc 16.9 18 16Total 100 100 100a It refers to people with 13 years of education.b It refers to people with 14 or 15 years of education.b It refers to people with more than 16 years of education.Table 2.8: Share of Immigrants with a Canadian Degree by Age at Immi-grationAge at Immigration Foreign degree Canadian degree Total19 8.4 91.720 20.6 79.421 30.3 69.722 42.7 57.323 47.5 52.524 53.3 46.725 57.6 42.426 61.9 38.127 64.7 35.428 67.0 33.029 71.3 28.7Total 44.1 55.9 100All immigrants who arrived at an age younger than 19 obtained their highest degreeof education in Canada.382.4. ConclusionsTable 2.9: Summary StatisticsNative ImmigrantLog. weekly wages 6.8 6.76(0.65) (0.67)Weeks worked 47.01 46.59(10.42) (10.72)Age 40.37 42.37(10.79) (10.92)Age of Imm. . 23.42(4.04)Year of Imm. . 1986.92(11.32)Years of Educ. 14.68 15.14(1.59) (1.74)Years of Educ. (-12) 2.68 3.14(1.59) (1.74)Canadian degree 98% 56%Yrs. Educ. AboveHS - CAN 2.45 1.27(1.23) (1.51)Yrs. Educ. AboveBACH - CAN 0.18 0.2(0.63) (0.74)Yrs. Educ. AboveHS - FOR 0.03 1.52(0.29) (1.73)Yrs. Educ. AboveBACH - FOR 0.02 0.15(0.26) (0.56)Work Exp. 19.7 21.23(10.89) (11.23)Work Exp. CAN 19.7 17.26(10.89) (11.18)Work Exp. FOR . 3.96(3.04)Weighted Observations 5671380 651750392.4.ConclusionsTable 2.10: Replication of Friedberg?s table 4(1) (2) (3)Immigrant -.2234 0.2457 0.1644(0.004)??? (0.0127)??? (0.0209)???Yrs. of Educ. 0.1198(0.0004)???Friedberg: Yrs. Educ. - FOR 0.0973 0.0983(0.0009)??? (0.0012)???Friedberg: Yrs. Educ. - CAN 0.1218 0.122(0.0004)??? (0.0004)???Work Exp. 0.0157(0.00006)???Friedberg: Work Exp. - FOR 0.0091 0.0102(0.0007)??? (0.0008)???Friedberg: Work Exp. - CAN 0.0163 0.0159(0.00006)??? (0.00006)???Years since Imm. 0.0051(0.0002)???Friedberg: Imm*(Yrs. Educ. - CAN) -.0014(0.0018)Friedberg: Imm*(Work Exp. - CAN) 0.0036(0.0002)???Weighted Obs. 6323125 6323125 6323125R2 0.1387 0.1389 0.1393Note: Standard errors are in parenthesis. *, ** and *** denote significance at 10%,5% and 1% levels respectively. The variable ?Friedberg: Yrs. Educ. - CAN? indicatesthat years of education in Canada were constructed following the assumptions applied byFriedberg: Children start their education at age seven and continue without interruptions.The separation between years of education in Canada and abroad comes from the assumededucation level at the age of immigration. Canadians are presumed to have no educationabroad. Work experience is calculated under the same assumptions. Canadians are alsoassumed to have no work experience abroad.402.4.ConclusionsTable 2.11: Replication of Friedberg?s table 5Natives Immigrants West India-China-Phillip. RestFriedberg: Yrs. Educ. - FOR 0.0983 0.096 0.0933 0.1049(0.0012)??? (0.0024)??? (0.0026)??? (0.0017)???Friedberg: Yrs. Educ. - CAN 0.122 0.1206 0.1022 0.1306 0.1257(0.0004)??? (0.0017)??? (0.0036)??? (0.0041)??? (0.0023)???Friedberg: Work Exp. - FOR 0.0102 0.0147 0.0098 0.0078(0.0008)??? (0.0019)??? (0.0017)??? (0.001)???Friedberg: Work Exp. - CAN 0.0159 0.0195 0.013 0.0205 0.0198(0.00006)??? (0.0002)??? (0.0004)??? (0.0004)??? (0.0003)???Weighted Obs. 5671380 651750 148670 361645R2 0.1392 0.1361 0.0879 0.1381 0.136Note: Standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The thirdcolumn includes: USA, Austria, Belgium, France, Germany,Liechtenstein, Luxembourg, Monaco, Netherlands, Switzerland,Ireland, Denmark, Finland, Iceland, Norway, Sweden, UK, Greece, Italy, Portugal, Spain, Australia and New Zeland. Thefourth column includes India, China and Phillipines. The variable ?Friedberg: Yrs. Educ. - CAN? indicates that years ofeducation in Canada were constructed following the assumptions applied by Friedberg: Children start their education at ageseven and continue without interruptions. The separation between years of education in Canada and abroad comes from theassumed education level at the age of immigration. Canadians are presumed to have no education abroad. Work experienceis calculated under the same assumptions. Canadians are also assumed to have no work experience abroad.412.4.ConclusionsTable 2.12: Replication of Friedberg?s table 4 - Sample Extension Immigrants Arriving at 15 and older(1) (2) (3)Immigrant -.3936 0.3445 0.3469(0.0028)??? (0.0105)??? (0.0153)???Yrs. of Educ. 0.1153(0.0004)???Friedberg: Yrs. Educ. - FOR 0.0911 0.0886(0.0007)??? (0.0008)???Friedberg: Yrs. Educ. - CAN 0.1203 0.122(0.0004)??? (0.0004)???Work Exp. 0.0145(0.00006)???Friedberg: Work Exp. - FOR -.0008 0.00009(0.0002)??? (0.0002)Friedberg: Work Exp. - CAN 0.0165 0.0159(0.00006)??? (0.00006)???Years since Imm. 0.011(0.0001)???Friedberg: Imm*(Yrs. Educ. - CAN) -.0189(0.0014)???Friedberg: Imm*(Work Exp. - CAN) 0.0041(0.0002)???R2 0.133 0.1373 0.1381Note: Standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and1% levels respectively. The variable ?Friedberg: Yrs. Educ. - CAN? indicates that yearsof education in Canada were constructed following the assumptions applied by Friedberg:Children start their education at age seven and continue without interruptions. The separa-tion between years of education in Canada and abroad comes from the assumed educationlevel at the age of immigration. Canadians are presumed to have no education abroad.Work experience is calculated under the same assumptions. Canadians are also assumed tohave no work experience abroad.422.4.ConclusionsTable 2.13: Replication of Friedberg?s table 5 - Sample Extension Immigrants Arriving at 15 and olderNatives Immigrants West India-China-Phillip. RestFriedberg: Yrs. Educ. - FOR 0.0886 0.1008 0.0861 0.0893(0.0008)??? (0.0018)??? (0.0016)??? (0.0011)???Friedberg: Yrs. Educ. - CAN 0.122 0.1031 0.1001 0.1101 0.1053(0.0004)??? (0.0014)??? (0.003)??? (0.0032)??? (0.0018)???Friedberg: Work Exp. - FOR 0.00009 0.0091 -.0021 -.0005(0.0002) (0.0007)??? (0.0004)??? (0.0003)Friedberg: Work Exp. - CAN 0.0159 0.0199 0.0121 0.0214 0.02(0.00006)??? (0.0002)??? (0.0004)??? (0.0003)??? (0.0002)???R2 0.1392 0.1202 0.0915 0.1226 0.1148Note: Standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The thirdcolumn includes: USA, Austria, Belgium, France, Germany,Liechtenstein, Luxembourg, Monaco, Netherlands, Switzerland,Ireland, Denmark, Finland, Iceland, Norway, Sweden, UK, Greece, Italy, Portugal, Spain, Australia and New Zeland. Thefourth column includes India, China and Phillipines. The variable ?Friedberg: Yrs. Educ. - CAN? indicates that years ofeducation in Canada were constructed following the assumptions applied by Friedberg: Children start their education at ageseven and continue without interruptions. The separation between years of education in Canada and abroad comes from theassumed education level at the age of immigration. Canadians are presumed to have no education abroad. Work experienceis calculated under the same assumptions. Canadians are also assumed to have no work experience abroad.432.4.ConclusionsTable 2.14: Adaptation of Friedberg?s table 4 with Location ofStudy Information - Immigrants Arriving at 15 and older(1) (2) (3)Immigrant -.3936 0.3811 0.2826(0.0028)??? (0.0088)??? (0.0143)???Yrs. of Educ. 0.1153(0.0004)???Yrs. Educ. - FOR 0.0889 0.091(0.0006)??? (0.0008)???Yrs. Educ. - CAN 0.1222 0.1224(0.0004)??? (0.0004)???Work Exp. 0.0145(0.00006)???Work Exp. - FOR 0.0005 0.002(0.0002)?? (0.0002)???Work Exp. - CAN 0.0164 0.0159(0.00006)??? (0.00006)???Years since Imm. 0.011(0.0001)???Imm*(Yrs. Educ. in CAN) -0.0016(0.0012)Imm*(Work Exp. in CAN) .0041(0.0002)???R2 0.133 0.1376 0.1382Note: Standard errors are in parenthesis. *, ** and *** denote significance at10%, 5% and 1% levels respectively. The variable ?Yrs. Educ. - CAN? wasconstructed using the new information on the location of study of the highestdegree attained available in the 2006 Canadian census. The separation betweenyears of education in Canada and abroad comes from the assumed educationlevel at the age of immigration. Work experience in Canada and abroad iscalculated using age, age at immigration and years of education (abroad and inCanada). Canadians are also assumed to have no work experience abroad.442.4.ConclusionsTable 2.15: Adaptation of Friedberg?s table 5 with Location of Study Information - ImmigrantsArriving at 15 and olderNatives Immigrants West India-China-Phillip. RestYrs. Educ. - FOR 0.1080 0.0872 0.1023 0.0851 0.0880(0.0018)??? (0.0008)??? (0.0018)??? (0.0016)??? (0.0011)???Yrs. Educ. - CAN 0.1227 0.1175 0.0973 0.1346 0.1186(0.0004)??? (0.0012)??? (0.0024)??? (0.0025)??? (0.0015)???Work Exp. - FOR 0.0019 0.0086 -.0001 0.0015(0.0002)??? (0.0007)??? (0.0004)??? (0.0003)???Work Exp. - CAN 0.0159 0.0198 0.0121 0.0209 0.0198(0.00006)??? (0.0002)??? (0.0004)??? (0.0003)??? (0.0002)???R2 0.1393 0.1207 0.0918 0.1255 0.1153Note: Standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels re-spectively. The third column includes: USA, Austria, Belgium, France, Germany,Liechtenstein, Luxembourg,Monaco, Netherlands, Switzerland, Ireland, Denmark, Finland, Iceland, Norway, Sweden, UK, Greece, Italy,Portugal, Spain, Australia and New Zeland. The fourth column includes India, China and Phillipines. Thevariable ?Friedberg: Yrs. Educ. - CAN? indicates that years of education in Canada were constructed followingthe assumptions applied by Friedberg: Children start their education at age seven and continue without inter-ruptions. The separation between years of education in Canada and abroad comes from the assumed educationlevel at the age of immigration. Canadians are presumed to have no education abroad. Work experience iscalculated under the same assumptions. Canadians are also assumed to have no work experience abroad.452.4.ConclusionsTable 2.16: Base Specification(1) (2) (3) (4) (5) (6)Const. 5.8405 5.8408 5.7572 5.7567 5.7528 5.7540(0.0022)??? (0.0022)??? (0.0022)??? (0.0022)??? (0.0022)??? (0.0022)???Immigrant -.1096 -.0580 -.1180 -.0657(0.002)??? (0.0025)??? (0.002)??? (0.0025)???Below Bachelor 0.1652 0.1653(0.0014)??? (0.0014)???Bachelor 0.4456 0.4494(0.0017)??? (0.0017)???Above Bachelor 0.5906 0.5908(0.0022)??? (0.0022)???Yrs. of Educ. (-12) 0.1233 0.1242 0.1243 0.1246(0.0004)??? (0.0004)??? (0.0004)??? (0.0004)???Work Exp. 0.0518 0.052 0.051 0.0512 0.0512 0.0513(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0879 -.0889 -.0873 -.0882 -.0877 -.0884(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Gender 0.2922 0.2918 0.286 0.2859 0.2865 0.2862(0.0012)??? (0.0012)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)???Age of Imm. 15-19 -.0266 -.0261(0.0042)??? (0.0042)???Age of Imm. 20-24 -.0800 -.0491(0.0033)??? (0.0036)???Age of Imm. 25-29 -.1863 -.1227(0.0029)??? (0.0038)???Loc. of study F.E. No Yes No Yes No YesObs. 6323125 6323125 6323125 6323125 6323125 6323125R2 0.2137 0.2167 0.2115 0.2143 0.2126 0.2147Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levelsrespectively. The omitted category in the location of study fixed effects is Canada.462.4.ConclusionsTable 2.17: Separating Male vs Female Immigrant Wage GapMen Women(1) (2) (3) (4) (5) (6) (7) (8)Const. 6.1068 6.1069 6.1909 6.1943 5.8152 5.8148 5.9583 5.9607(0.003)??? (0.003)??? (0.005)??? (0.005)??? (0.003)??? (0.003)??? (0.0047)??? (0.0047)???Immigrant -.1184 -.0839 -.0757 -.0248 -.0933 -.0202 -.1533 -.0588(0.0029)??? (0.0036)??? (0.0119)??? (0.0159) (0.0028)??? (0.0033)??? (0.0109)??? (0.0143)???Below Bachelor 0.1212 0.121 0.1012 0.1012 0.2508 0.2519 0.2334 0.2345(0.0019)??? (0.0019)??? (0.0019)??? (0.0019)??? (0.0021)??? (0.0021)??? (0.0021)??? (0.0021)???Bachelor 0.3677 0.3711 0.3362 0.3382 0.5589 0.5639 0.5229 0.5268(0.0024)??? (0.0024)??? (0.0024)??? (0.0024)??? (0.0024)??? (0.0024)??? (0.0024)??? (0.0024)???Above Bachelor 0.5041 0.5016 0.4742 0.4742 0.7167 0.7209 0.6741 0.6777(0.0032)??? (0.0033)??? (0.0032)??? (0.0032)??? (0.003)??? (0.003)??? (0.0029)??? (0.003)???Work Exp. 0.0578 0.0581 0.0586 0.0586 0.0469 0.047 0.0479 0.0479(0.0003)??? (0.0003)??? (0.0003)??? (0.0003)??? (0.0003)??? (0.0003)??? (0.0003)??? (0.0003)???Work Exp. Square (/100) -.0981 -.0995 -.1001 -.1003 -.0805 -.0811 -.0825 -.0825(0.0007)??? (0.0007)??? (0.0007)??? (0.0007)??? (0.0007)??? (0.0007)??? (0.0007)??? (0.0007)???Eng. or Fren. Mother Tongue No No Yes Yes No No Yes YesCMA/Province F.E. No No Yes Yes No No Yes YesCountry/area of Origin F.E. No No Yes Yes No No Yes YesLoc. of study F.E. No Yes No Yes No Yes No YesObs. 3413975 3413975 3413975 3413975 2909155 2909155 2909155 2909155R2 0.1628 0.1658 0.2013 0.2025 0.2077 0.2118 0.2406 0.243Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omitted categoryin the location of study fixed effects is ?Canada?. The mother tongue fixed effects incorporates a dummy for English mother tongue and a dummyfor French Mother tongue. The omitted categories in the CMA and Province fixed effects are Toronto and Ontario, respectively. The omittedcategory for country/area of origin fixed effect is ?Eastern Europe?.472.4.ConclusionsTable 2.18: Separating Years of Education Above High School from Years of Education Above Bachelor?s Degree(1) (2) (3) (4) (5) (6) (7) (8)Immigrant -.1180 -.0657 -.1186 -.0655 -.0677 -.0565 -.0351 -.0303(0.002)??? (0.0025)??? (0.002)??? (0.0025)??? (0.0025)??? (0.0025)??? (0.0039)??? (0.0039)???Yrs. of Educ. (- 12) 0.1233 0.1242(0.0004)??? (0.0004)???Yrs. of Educ Above HS 0.1492 0.151(0.0005)??? (0.0005)???Yrs. of Educ Above BACH 0.068 0.0661(0.001)??? (0.001)???Yrs. of Educ Above HS - CAN 0.1516 0.1517 0.1515 0.1515(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Yrs. of Educ Above BACH - CAN 0.0726 0.0708 0.0719 0.0703(0.001)??? (0.0011)??? (0.001)??? (0.0011)???Yrs. of Educ Above HS - FOR 0.1178 0.1296 0.1258 0.1361(0.0011)??? (0.0016)??? (0.0012)??? (0.0016)???Yrs. of Educ Above BACH - FOR 0.0596 0.0441 0.0581 0.0469(0.0021)??? (0.0024)??? (0.0021)??? (0.0024)???Work Exp. 0.051 0.0512 0.0516 0.0518 0.0515 0.0518(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0873 -.0882 -.0878 -.0888 -.0876 -.0888(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - CAN 0.0522 0.052(0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0908 -.0908(0.0005)??? (0.0005)???Work Exp. - FOR 0.0066 0.0099(0.002)??? (0.002)???Work Exp. Square (/100) - FOR -.0085 -.0394(0.0211) (0.0211)?Gender 0.286 0.2859 0.2926 0.2924 0.2935 0.2927 0.2934 0.2928(0.0011)??? (0.0011)??? (0.0012)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)???Loc. of study F.E. No Yes No Yes No Yes No YesObs. 6323125 6323125 6323125 6323125 6323125 6323125 6323125 6323125R2 0.2115 0.2143 0.2152 0.2182 0.2163 0.2185 0.2178 0.2193Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omitted category in thelocation of study fixed effects is ?Canada?.482.4.ConclusionsTable 2.19: Modeling Return to Education by Including Educational Group Dummies(1) (2) (3) (4) (5) (6) (7) (8)Immigrant -.1096 -.0580 -.0833 -.0603 -.0458 -.0285 -.0373 -.0283(0.002)??? (0.0025)??? (0.0024)??? (0.0025)??? (0.0038)??? (0.0039)??? (0.0039)??? (0.0039)???Below Bachelor 0.1652 0.1653 0.1678 0.1667 0.1647 0.1649 0.1665 0.1657(0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)???Bachelor 0.4456 0.4494 0.4573 0.4557 0.447 0.4496 0.4554 0.4542(0.0017)??? (0.0017)??? (0.0017)??? (0.0017)??? (0.0017)??? (0.0017)??? (0.0017)??? (0.0017)???Above Bachelor 0.5906 0.5908 0.6013 0.5991 0.5921 0.5916 0.6004 0.5986(0.0022)??? (0.0022)??? (0.0023)??? (0.0023)??? (0.0022)??? (0.0022)??? (0.0023)??? (0.0023)???Trade - FOR 0.0499 0.0115 0.0348 0.0165(0.007)??? (0.0122) (0.007)??? (0.0122)???Below Bachelor - FOR -.0353 -.05528 -.0252 -.0297(0.005)??? (0.0116)??? (0.0051)??? (0.0116)???Bachelor - FOR -.1496 -.1410 -.1128 -.0987(0.0054)??? (0.0118)??? (0.0055)??? (0.0118)???Above Bachelor - FOR -.0783 -.1326 -.0664 -.1000(0.0057)??? (0.0118)??? (0.0057)??? (0.0118)???Work Exp. 0.0518 0.052 0.0518 0.0521(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0879 -.0889 -.0881 -.0891(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - CAN 0.0526 0.0523 0.0524 0.0522(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0914 -.0910 -.0910 .0910(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR -.0092 0.0068 0.0007 0.0077(0.0019)??? (0.002)??? (0.002) (0.002)???Work Exp. Square (/100) - FOR 0.167 0.0086 0.0557 -.0139(0.0199)??? (0.0206) (0.0206)??? (0.0207)Gender 0.2922 0.2918 0.2927 0.2919 0.2923 0.2919 0.2927 0.292(0.0012)??? (0.0012)??? (0.0012)??? (0.0012)??? (0.0012)??? (0.0012)??? (0.0012)??? (0.0012)???Loc. of study F.E. No Yes No Yes No Yes No YesObs. 6323125 6323125 6323125 6323125 6323125 6323125 6323125 6323125R2 0.2137 0.2167 0.2148 0.217 0.2156 0.2176 0.2162 0.2178Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omitted categoryin the location of study fixed effects is ?Canada?.49Table 2.20: Interacting Education with Location of Study(1) (2) (3) (4)Immigrant -.1016 -.0757 -.0402 0.0055(0.0028)??? (0.0041)??? (0.0108)??? (0.0112)Below Bachelor 0.1465 0.1457 0.1466 0.1457(0.0014)??? (0.0014)??? (0.0014)??? (0.0014)???Bachelor 0.4189 0.4176 0.4186 0.4171(0.0017)??? (0.0017)??? (0.0017)??? (0.0017)???Above Bachelor 0.5618 0.5612 0.5613 0.5602(0.0023)??? (0.0023)??? (0.0023)??? (0.0023)???Work Exp. 0.0525 0.0527(0.0002)??? (0.0002)???Work Exp.Square(/100) -.0894 -.0900(0.0005)??? (0.0005)???Work Exp. - CAN 0.0525 0.0525(0.0002)??? (0.0002)???Work Exp.Square(/100) - CAN -.0906 -.0909(0.0005)??? (0.0005)???Work Exp. - FOR 0.0117 0.0108(0.002)??? (0.002)???Work Exp.Square(/100) - FOR -.0570 -.0427(0.0206)??? (0.0206)??Gender 0.2868 0.287 0.2869 0.2869(0.0011)??? (0.0011)??? (0.0011)??? (0.0011)???Eng. or Fren. Mother Tongue Yes Yes Yes YesCMA/Province F.E. Yes Yes Yes YesCountry/area of origin F.E. No No Yes YesObs. 6323125 6323125 6323125 6323125Note: Robust standard errors are in parenthesis.*, ** and *** denote significance at 10%,5% and 1% levels respectively. The omitted categories in the CMA and Province fixedeffects are Toronto and Ontario, respectively. The omitted category for country/area oforigin fixed effect is ?Eastern Europe?.50Cont. - Interacting Education with Location of Study(1) (2) (3) (4)Trade - INDIA -.0512 -.0274 -.0418 -.0222(0.0339) (0.0345) (0.0349) (0.0354)Below Bachelor-INDIA -.1850 -.1574 -.1792 -.1543(0.0144)??? (0.0145)??? (0.0167)??? (0.0167)???Bachelor - INDIA -.3286 -.2915 -.3249 -.2895(0.0123)??? (0.0124)??? (0.015)??? (0.015)???Above Bachelor - INDIA -.3996 -.3775 -.3976 -.3762(0.0169)??? (0.017)??? (0.019)??? (0.0191)???Trade - UK 0.1535 0.0899 0.074 0.0324(0.0165)??? (0.0165)??? (0.0173)??? (0.0173)?Below Bachelor - UK 0.1191 0.0772 0.044 0.0235(0.0113)??? (0.0112)??? (0.0126)??? (0.0125)?Bachelor - UK 0.1292 0.1267 0.0738 0.0868(0.0201)??? (0.0202)??? (0.0205)??? (0.0206)???Above Bachelor - UK 0.073 0.0704 0.0553 0.0576(0.0149)??? (0.0149)??? (0.015)??? (0.015)???Trade - FR -.0321 -.0437 -.0460 -.0754(0.029) (0.0297) (0.0312) (0.0319)??Below Bachelor - FR -.0220 0.0197 -.0272 -.0052(0.0257) (0.0262) (0.0282) (0.0287)Bachelor - FR -.0667 -.0194 -.0709 -.0420(0.0374)? (0.0377) (0.0399)? (0.0403)Above Bachelor - FR 0.0158 0.0373 0.0138 0.0224(0.0202) (0.0201)? (0.022) (0.0219)Trade - PHIL -.0722 -.0579 -.0798 -.0659(0.0274)??? (0.0278)?? (0.0292)??? (0.0297)??Below Bachelor - PHIL -.1147 -.0932 -.1237 -.1025(0.014)??? (0.0138)??? (0.0168)??? (0.0168)???Bachelor - PHIL -.2012 -.1781 -.2102 -.1867(0.012)??? (0.0121)??? (0.0155)??? (0.0156)???Above Bachelor - PHIL -.3464 -.3298 -.3545 -.3370(0.0294)??? (0.0294)??? (0.0309)??? (0.0309)???Trade - US -.0054 -.0067 -.0074 -.0072(0.0199) (0.02) (0.02) (0.0201)Below Bachelor - US -.0240 -.0259 -.0273 -.0271(0.0131)? (0.0131)?? (0.0132)?? (0.0132)??Bachelor - US -.0540 -.0539 -.0560 -.0547(0.0103)??? (0.0103)??? (0.0104)??? (0.0104)???Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Robuststandard errors are in parenthesis.51Cont. - Interacting Education with Location of Study(1) (2) (3) (4)Above Bachelor - US -.0220 -.0230 -.0214 -.0223(0.0082)??? (0.0082)??? (0.0082)??? (0.0082)???Trade - CHINA -.3229 -.2635 -.2721 -.2490(0.0664)??? (0.0638)??? (0.068)??? (0.0653)???Below Bachelor - CHINA -.2580 -.1545 -.2085 -.1444(0.0232)??? (0.0229)??? (0.0251)??? (0.0248)???Bachelor - CHINA -.2619 -.1578 -.2125 -.1477(0.0173)??? (0.0174)??? (0.0197)??? (0.0198)???Above Bachelor - CHINA -.1883 -.1274 -.1400 -.1167(0.039)??? (0.0387)??? (0.0406)??? (0.0402)???Trade - POL 0.0348 0.032 -.0446 -.0557(0.0333) (0.0332) (0.0351) (0.035)Below Bachelor - POL -.0682 -.0684 -.1512 -.1581(0.0184)??? (0.0183)??? (0.0221)??? (0.022)???Bachelor - POL -.2331 -.2339 -.3155 -.3219(0.0557)??? (0.0554)??? (0.0571)??? (0.0568)???Above Bachelor - POL -.2360 -.2440 -.3184 -.3312(0.0292)??? (0.0291)??? (0.0317)??? (0.0316)???Trade - PAK -.1933 -.1884 -.1229 -.1391(0.0713)??? (0.0668)??? (0.0702)? (0.0663)??Below Bachelor - PAK -.2986 -.2600 -.2143 -.2004(0.0368)??? (0.0363)??? (0.0395)??? (0.0391)???Bachelor - PAK -.3640 -.3106 -.2767 -.2478(0.0322)??? (0.0319)??? (0.0362)??? (0.036)???Above Bachelor - PAK -.5022 -.4575 -.4147 -.3935(0.0359)??? (0.036)??? (0.0402)??? (0.0403)???Trade - ROMANIA 0.0229 0.0916 -.0570 -.0286(0.0533) (0.0527)? (0.0561) (0.0554)Below Bachelor - ROMANIA -.0292 0.0605 -.1089 -.0595(0.0371) (0.0364)? (0.0417)??? (0.0409)Bachelor - ROMANIA -.1401 -.0355 -.2225 -.1587(0.0245)??? (0.0241) (0.032)??? (0.0315)???Above Bachelor - ROMANIA -.0514 -.0031 -.1327 -.1242(0.0268)? (0.0268) (0.0338)??? (0.0335)???Trade - SOUTH-AMER -.0803 -.0877 -.0461 -.0479(0.0326)?? (0.0329)??? (0.034) (0.0343)Below Bachelor - SOUTH-AMER -.1117 -.1012 -.0759 -.0578(0.0215)??? (0.0215)??? (0.0230)??? (0.0230)??Bachelor - SOUTH-AMER -.1831 -.1152 -.1474 -.0698(0.0263)??? (0.0271)??? (0.0277)??? (0.0285)??Above Bachelor - SOUTH-AMER -.2443 -.2037 -.2097 -.1591(0.0384)??? (0.0386)??? (0.0393)??? (0.0395)???Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Robuststandard errors are in parenthesis.52Cont. - Interacting Education with Location of Study(1) (2) (3) (4)Trade - REST-AMER -.1200 -.1200 -.0713 -.0592(0.0185)??? (0.0186)??? (0.0193)??? (0.0193)???Below Bachelor - REST-AMER -.1344 -.1191 -.0843 -.0567(0.0202)??? (0.0202)??? (0.0207)??? (0.0208)???Bachelor - REST-AMER -.2830 -.2170 -.2354 -.1561(0.0297)??? (0.0298)??? (0.03)??? (0.0302)???Above Bachelor - REST-AMER -.2648 -.2338 -.2222 -.1794(0.0414)??? (0.042)??? (0.0414)??? (0.0421)???Trade - EAST-EUR 0.0982 0.0751 0.0323 -.0034(0.02)??? (0.02)??? (0.0217) (0.0217)Below Bachelor - EAST-EUR -.0525 -.0397 -.1170 -.1192(0.0207)?? (0.0204)? (0.0227)??? (0.0225)???Bachelor - EAST-EUR -.1599 -.0994 -.2223 -.1755(0.0219)??? (0.0219)??? (0.0239)??? (0.0239)???Above Bachelor - EAST-EUR -.1668 -.1327 -.2265 -.2060(0.0201)??? (0.0199)??? (0.0223)??? (0.0221)???Trade - REST-EUR 0.1166 0.0597 0.0271 -.0037(0.0174)??? (0.0173)??? (0.0184) (0.0183)Below Bachelor - REST-EUR 0.0618 0.0321 -.0212 -.0259(0.018)??? (0.0179)? (0.0189) (0.0189)Bachelor - REST-EUR -.0669 -.0680 -.1341 -.1144(0.0422) (0.0425) (0.0424)??? (0.0426)???Above Bachelor - REST-EUR -.0322 -.0253 -.0659 -.0495(0.0265) (0.0263) (0.0265)?? (0.0264)?Trade - AFRICA -.0401 -.0069 -.0089 0.001(0.0283) (0.0284) (0.0293) (0.0293)Below Bachelor - AFRICA -.1273 -.0777 -.0966 -.0696(0.0183)??? (0.0185)??? (0.0198)??? (0.0199)???Bachelor - AFRICA -.1626 -.1136 -.1312 -.1036(0.0234)??? (0.0233)??? (0.0246)??? (0.0244)???Above Bachelor - AFRICA -.0353 -.0057 -.0062 0.0036(0.0421) (0.0419) (0.0427) (0.0424)Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively Robuststandard errors are in parenthesis.53Cont. - Interacting Education with Location of Study(1) (2) (3) (4)Trade - WESTASIA -.1480 -.1187 -.1107 -.0956(0.0372)??? (0.0366)??? (0.0377)??? (0.0372)??Below Bachelor - WESTASIA -.1923 -.1561 -.1565 -.1345(0.0254)??? (0.0258)??? (0.0264)??? (0.0268)???Bachelor - WESTASIA -.2432 -.1807 -.2040 -.1561(0.0259)??? (0.026)??? (0.0273)??? (0.0273)???Above Bachelor - WESTASIA -.1917 -.1571 -.1543 -.1324(0.0396)??? (0.0394)??? (0.0406)??? (0.0404)???Trade - SOUTHEASTASIA -.1106 -.1246 -.0882 -.0957(0.0433)?? (0.0426)??? (0.043)?? (0.0422)??Below Bachelor - SOUTHEASTASIA -.1962 -.1888 -.1874 -.1734(0.0334)??? (0.0336)??? (0.0336)??? (0.0337)???Bachelor - SOUTHEASTASIA -.4380 -.4229 -.4236 -.4008(0.0462)??? (0.0465)??? (0.0464)??? (0.0466)???Above Bachelor - SOUTHEASTASIA -.3767 -.3814 -.3665 -.3729(0.088)??? (0.0885)??? (0.0897)??? (0.0899)???Trade - REST-ASIA -.1181 -.1150 -.0527 -.0684(0.0321)??? (0.0313)??? (0.0327) (0.0319)??Below Bachelor - REST-ASIA -.1515 -.1404 -.0781 -.0859(0.0196)??? (0.0196)??? (0.0205)??? (0.0205)???Bachelor - REST-ASIA -.3187 -.2989 -.2227 -.2246(0.0234)??? (0.0233)??? (0.0243)??? (0.0243)???Above Bachelor - REST-ASIA -.4439 -.4239 -.3569 -.3562(0.0498)??? (0.0495)??? (0.0496)??? (0.0496)???Trade - OCEANIA 0.09 0.0935 0.077 0.0774(0.0399)?? (0.0415)?? (0.0417)? (0.0431)?Below Bachelor - OCEANIA 0.0611 0.0685 0.0481 0.0529(0.0318)? (0.0317)?? (0.0339) (0.0337)Bachelor - OCEANIA -.0008 0.0247 -.0055 0.0178(0.0306) (0.0308) (0.0312) (0.0313)Above Bachelor - OCEANIA -.0895 -.0894 -.0902 -.0911(0.0253)??? (0.0254)??? (0.0252)??? (0.0253)???Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively Robuststandard errors are in parenthesis.542.4.ConclusionsTable 2.21: Base Specification - Sample Extension: Immigrants Arriving at 15 and older(1) (2) (3) (4) (5) (6)Const. 5.8839 5.8694 5.8133 5.7987 5.7764 5.7768(0.0022)??? (0.0022)??? (0.0022)??? (0.0022)??? (0.0022)??? (0.0022)???Immigrant -.2336 -.0973 -.2445 -.1107(0.0016)??? (0.0022)??? (0.0016)??? (0.0022)???Below Bachelor 0.1617 0.1641(0.0014)??? (0.0014)???Bachelor 0.4178 0.4351(0.0017)??? (0.0017)???Above Bachelor 0.5596 0.5678(0.0021)??? (0.0021)???Yrs. of Educ. (-12) 0.1143 0.1173 0.1193 0.1197(0.0004)??? (0.0004)??? (0.0004)??? (0.0004)???Work Exp. 0.0484 0.0499 0.0476 0.049 0.0492 0.0498(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0799 -.0842 -.0793 -.0832 -.0813 -.0836(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Gender 0.2827 0.2838 0.2756 0.2768 0.2801 0.2795(0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)???Age of Imm. 15-19 -.0258 -.0249(0.0042)??? (0.0042)???Age of Imm. 20-24 -.0831 -.0469(0.0033)??? (0.0035)???Age of Imm. 25-29 -.1825 -.1060(0.0029)??? (0.0035)???Age of Imm. 30-34 -.3091 -.2078(0.0033)??? (0.0039)???Age of Imm. 35-39 -.4179 -.2957(0.0042)??? (0.0048)???Age of Imm. 40-44 -.5285 -.3960(0.0058)??? (0.0063)???Age of Imm. 45-49 -.6128 -.4725(0.009)??? (0.0093)???Age of Imm. 50 plus -.6441 -.5053(0.0149)??? (0.0147)???Loc. of study F.E. No Yes No Yes No YesR2 0.194 0.2066 0.1927 0.2043 0.204 0.2098Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levelsrespectively. The omitted category in the location of study fixed effects is Canada.552.4.ConclusionsTable 2.22: Modeling Return to Education by Including Educational Group Dummies - Immigrants 15 and older(1) (2) (3) (4) (5) (6) (7) (8)Immigrant -.2336 -.0973 -.1405 -.1006 -.0419 -.0119 -.0212 -.0155(0.0016)??? (0.0022)??? (0.0022)??? (0.0022)??? (0.003)??? (0.0031)??? (0.0031)??? (0.0031)???Below Bachelor 0.1617 0.1641 0.1672 0.1655 0.1622 0.1637 0.1652 0.1643(0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)???Bachelor 0.4178 0.4351 0.4552 0.4529 0.4287 0.4381 0.4527 0.4512(0.0017)??? (0.0017)??? (0.0017)??? (0.0017)??? (0.0016)??? (0.0017)??? (0.0017)??? (0.0017)???Above Bachelor 0.5596 0.5678 0.5998 0.5951 0.5725 0.5758 0.5993 0.5966(0.0021)??? (0.0021)??? (0.0023)??? (0.0023)??? (0.0021)??? (0.0021)??? (0.0023)??? (0.0023)???Trade - FOR 0.0247 -.0937 0.0471 -.0180(0.0056)??? (0.0083)??? (0.0056)??? (0.0082)??Below Bachelor - FOR -.0861 -.1739 -.0240 -.0689(0.0039)??? (0.0072)??? (0.0039)??? (0.0071)???Bachelor - FOR -.2779 -.3182 -.1642 -.1809(0.004)??? (0.0073)??? (0.004)??? (0.0071)???Above Bachelor - FOR -.2117 -.3239 -.1244 -.1873(0.0044)??? (0.0074)??? (0.0043)??? (0.0072)???Work Exp. 0.0484 0.0499 0.0493 0.0503(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. 2 (/100) -.0799 -.0842 -.0822 -.0853(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - CAN 0.052 0.051 0.0512 0.0507(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. 2 (/100) - CAN -.0903 -.0887 -.0887 -.0881(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR -.0016 0.0066 0.0033 0.0072(0.0006)??? (0.0006)??? (0.0006)??? (0.0006)???Work Exp. 2 (/100) - FOR -.0026 -.0206 -.0167 -.0243(0.0023) (0.0023)??? (0.0023)??? (0.0022)???Gender 0.2827 0.2838 0.2862 0.2849 0.2869 0.2868 0.2887 0.2876(0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)???Loc. of study F.E. No Yes No Yes No Yes No YesR2 0.194 0.2066 0.2003 0.2083 0.2077 0.2132 0.21 0.2142Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omittedcategory in the location of study fixed effects is ?Canada?.562.4.ConclusionsTable 2.23: Immigrant Wage Gap by Country of Origin - Base Specification(1) (2) (3) (4) (5) (6) (7) (8) (9)Below Bachelor 0.1452 0.1456 0.1752(0.0014)??? (0.0014)??? (0.0015)???Bachelor 0.4108 0.4131 0.4798(0.0017)??? (0.0017)??? (0.0019)???Above Bachelor 0.5535 0.5542 0.617(0.0022)??? (0.0022)??? (0.0024)???Yrs. of Educ. (- 12) 0.116 0.1167 0.129 0.1166 0.1169 0.1294(0.0004)??? (0.0004)??? (0.0004)??? (0.0004)??? (0.0004)??? (0.0004)???Work Exp. 0.0526 0.0527 0.0521 0.0519 0.0519 0.0512 0.0521 0.052 0.0513(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0898 -.0899 -.0891 -.0892 -.0891 -.0882 -.0894 -.0893 -.0884(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Gender 0.287 0.2869 0.2442 0.2823 0.2823 0.2422 0.2827 0.2826 0.2422(0.0011)??? (0.0011)??? (0.0014)??? (0.0011)??? (0.0011)??? (0.0014)??? (0.0011)??? (0.0011)??? (0.0014)???Age of Imm. 15-19 0.0549 0.0166 0.0178(0.0053)??? (0.0054)??? (0.0054)???Age of Imm. 20-24Age of Imm. 25-29 -.0814 -.0573 -.0662(0.0043)??? (0.0044)??? (0.0044)???Eng. or Fren. Mother Tongue Yes Yes Yes Yes Yes Yes Yes Yes YesCMA/Province F.E. Yes Yes Yes Yes Yes Yes Yes Yes YesCountry of Orig. F.E. Yes Yes Yes Yes Yes Yes Yes Yes YesGroup Loc. of Study F.E. Yes Yes Yes Yes Yes YesField F.E. (Visual Arts) Yes Yes YesObs.R2 0.2455 0.2468 0.2611 0.2445 0.2458 0.2592 0.2452 0.246 0.2595Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omitted category in thegroups of location of study is ?Canada?. The mother tongue fixed effects incorporates a dummy for English mother tongue and a dummy for French Mothertongue. The omitted categories in the CMA and Province fixed effects are Toronto and Ontario, respectively. The omitted category for country/area oforigin fixed effect is ?Eastern Europe?. The base case for the field of study fixed effects is ?Visual Arts?.572.4.ConclusionsTable 2.24: Immigrant Wage Gap by Country of Origin - Separating Years of Education Above High School from Years of EducationAbove Bachelor?s Degree(1) (2) (3) (4) (5) (6) (7) (8) (9)Yrs. of Educ Above HS 0.139 0.1399 0.1637(0.0005)??? (0.0005)??? (0.0006)???Yrs. of Educ Above BACH 0.0677 0.0672 0.0666(0.001)??? (0.001)??? (0.0009)???Yrs. of Educ Above HS - CAN 0.1408 0.1404 0.1647 0.1406 0.1402 0.1646(0.0005)??? (0.0005)??? (0.0006)??? (0.0005)??? (0.0005)??? (0.0006)???Yrs. of Educ Above BACH - CAN 0.0724 0.0704 0.0702 0.0719 0.0699 0.0698(0.001)??? (0.001)??? (0.001)??? (0.001)??? (0.001)??? (0.001)???Yrs. of Educ Above HS - FOR 0.1133 0.1271 0.1441 0.1191 0.1326 0.1499(0.0012)??? (0.0016)??? (0.0016)??? (0.0012)??? (0.0016)??? (0.0016)???Yrs. of Educ Above BACH - FOR 0.0558 0.0514 0.0497 0.054 0.0532 0.0515(0.0021)??? (0.0024)??? (0.0023)??? (0.0021)??? (0.0024)??? (0.0024)???Work Exp. 0.0524 0.0525 0.0519 0.0524 0.0525 0.0519(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0896 -.0897 -.0889 -.0895 -.0897 -.0889(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - CAN 0.0524 0.0524 0.0519(0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0908 -.0909 -.0903(0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR 0.006 0.0116 0.0093(0.002)??? (0.002)??? (0.002)???Work Exp. Square (/100) - FOR -.0032 -.0463 -.0252(0.0208) (0.0209)?? (0.0207)Gender 0.2879 0.2879 0.2411 0.2886 0.2881 0.2409 0.2886 0.2881 0.241(0.0011)??? (0.0011)??? (0.0014)??? (0.0011)??? (0.0011)??? (0.0014)??? (0.0011)??? (0.0011)??? (0.0014)???Eng. or Fren. Mother Tongue Yes Yes Yes Yes Yes Yes Yes Yes YesCMA/Province F.E. Yes Yes Yes Yes Yes Yes Yes Yes YesCountry/area of Orig. F.E. Yes Yes Yes Yes Yes Yes Yes Yes YesGroup Loc. of Study F.E. Yes Yes Yes Yes Yes YesField of study F.E. Yes Yes YesNote: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omitted category in thegroups of location of study is ?Canada?. The mother tongue fixed effects incorporates a dummy for English mother tongue and a dummy for French Mothertongue. The omitted categories in the CMA and Province fixed effects are Toronto and Ontario, respectively. The omitted category for country/area of originfixed effect is ?Eastern Europe?. The base case for the field of study fixed effects is ?Visual Arts?.582.4.ConclusionsTable 2.25: Immigrant Wage Gap by Country of Origin - Educational Group Dummies(1) (2) (3) (4) (5) (6) (7) (8) (9)Below Bachelor 0.1452 0.1456 0.1752 0.1471 0.1465 0.1766 0.146 0.1456 0.1755(0.0014)??? (0.0014)??? (0.0015)??? (0.0014)??? (0.0014)??? (0.0015)??? (0.0014)??? (0.0014)??? (0.0015)???Bachelor 0.4108 0.4131 0.4798 0.419 0.4185 0.4881 0.4174 0.417 0.4867(0.0017)??? (0.0017)??? (0.0019)??? (0.0017)??? (0.0017)??? (0.0019)??? (0.0017)??? (0.0017)??? (0.0019)???Above Bachelor 0.5535 0.5542 0.617 0.5622 0.5613 0.6264 0.5612 0.5602 0.6255(0.0022)??? (0.0022)??? (0.0024)??? (0.0023)??? (0.0023)??? (0.0025)??? (0.0023)??? (0.0023)??? (0.0025)???Trade - FOR -.0017 -.0123 0.0081 -.0125 -.0126 0.0092(0.0069) (0.0159) (0.0158) (0.0069)? (0.0159) (0.0158)Below Bachelor - FOR -.0564 -.0581 -.0458 -.0470 -.0392 -.0251(0.005)??? (0.0151)??? (0.015)??? (0.0051)??? (0.0151)??? (0.015)?Bachelor - FOR -.1315 -.1314 -.1474 -.1050 -.0960 -.1102(0.0054)??? (0.0153)??? (0.0152)??? (0.0055)??? (0.0153)??? (0.0152)???Above Bachelor - FOR -.0817 -.1230 -.1265 -.0751 -.0989 -.1012(0.0056)??? (0.0159)??? (0.0157)??? (0.0056)??? (0.0158)??? (0.0157)???Work Exp. 0.0526 0.0527 0.0521 0.0526 0.0527 0.0522(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0898 -.0899 -.0891 -.0898 -.0900 -.0893(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - CAN 0.0526 0.0526 0.0521(0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0911 -.0911 -.0904(0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR 0.0043 0.0108 0.0074(0.002)?? (0.002)??? (0.002)???Work Exp. Square (/100) - FOR 0.0249 -.0388 -.0058(0.0203) (0.0205)? (0.0204)Gender 0.287 0.2869 0.2442 0.2876 0.287 0.2438 0.2875 0.287 0.244(0.0011)??? (0.0011)??? (0.0014)??? (0.0011)??? (0.0011)??? (0.0014)??? (0.0011)??? (0.0011)??? (0.0014)???Eng. or Fren. Mother Tongue Yes Yes Yes Yes Yes Yes Yes Yes YesCMA/Province F.E. (Ontario - Toronto) Yes Yes Yes Yes Yes Yes Yes Yes YesCountry/area of Orig. F.E. Yes Yes Yes Yes Yes Yes Yes Yes YesGroup Loc. of study F.E. Yes Yes Yes Yes Yes YesField of study F.E. Yes YesObs.R2 0.2455 0.2468 0.2611 0.2462 0.247 0.2616 0.2467 0.2473 0.2619Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively. The omitted category in the groups oflocation of study is ?Canada?. The mother tongue fixed effects incorporates a dummy for English mother tongue and a dummy for French Mother tongue. Theomitted categories in the CMA and Province fixed effects are Toronto and Ontario, respectively. The omitted category for country/area of origin fixed effect is?Eastern Europe?. The base case for the field of study fixed effects is ?Visual Arts?.592.4.ConclusionsTable 2.26: Immigrant Wage Gap by Country of Origin - Educational Group Dummies - Immigrants15 and older(1) (2) (3) (4) (5) (6)Below Bachelor 0.1426 0.1448 0.1463 0.1456 0.1451 0.1447(0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)??? (0.0014)???Bachelor 0.3918 0.4008 0.4175 0.4168 0.4159 0.4153(0.0016)??? (0.0016)??? (0.0017)??? (0.0017)??? (0.0017)??? (0.0017)???Above Bachelor 0.5341 0.5379 0.5633 0.5607 0.5628 0.561(0.0021)??? (0.0021)??? (0.0023)??? (0.0023)??? (0.0023)??? (0.0023)???Trade - FOR -.0248 -.0212 -.0011 0.0222(0.0055)??? (0.0116)? (0.0055) (0.0115)?Below Bachelor - FOR -.0939 -.0794 -.0421 -.0092(0.0038)??? (0.0107)??? (0.0039)??? (0.0107)Bachelor - FOR -.2285 -.2062 -.1451 -.1058(0.0041)??? (0.0108)??? (0.0041)??? (0.0107)???Above Bachelor - FOR -.1844 -.2056 -.1175 -.1082(0.0044)??? (0.0112)??? (0.0044)??? (0.0111)???Work Exp. 0.0502 0.0507 0.0507 0.0511(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) -.0849 -.0859 -.0861 -.0868(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - CAN 0.0509 0.0508(0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0878 -.0877(0.0005)??? (0.0005)???Work Exp. - FOR 0.0059 0.0083(0.0006)??? (0.0006)???Work Exp. Square (/100) - FOR -.0223 -.0275(0.0022)??? (0.0022)???Gender 0.2795 0.2805 0.2823 0.2815 0.2846 0.2839(0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)??? (0.0011)???Eng. or Fren. Mother Tongue Yes Yes Yes Yes Yes YesCMA/Province F.E. Yes Yes Yes Yes Yes YesCountry of Orig. F.E. Yes Yes Yes Yes Yes YesGroup Loc. of Study F.E. Yes Yes YesR2 0.2313 0.2366 0.2351 0.2378 0.241 0.2425Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respec-tively.The omitted category in the groups of location of study is ?Canada?. The mother tongue fixed effects incorporatesa dummy for English mother tongue and a dummy for French Mother tongue. The omitted categories in the CMA andProvince fixed effects are Toronto and Ontario, respectively. The omitted category for country/area of origin fixed effectis ?Eastern Europe?. The base case for the field of study fixed effects is ?Visual Arts?.602.4. ConclusionsTable 2.27: Exploring the Effect of Field of Study(1) (2) (3) (4)Immigrant -.0846 -.0967(0.0041)??? (0.0041)???Below Bachelor 0.1449 0.1743 0.145 0.1744(0.0014)??? (0.0015)??? (0.0014)??? (0.0015)???Bachelor 0.4111 0.478 0.4116 0.4791(0.0017)??? (0.0019)??? (0.0017)??? (0.0019)???Above Bachelor 0.5542 0.6182 0.5538 0.6179(0.0022)??? (0.0024)??? (0.0022)??? (0.0024)???Work Exp. - CAN 0.0528 0.0524 0.0527 0.0522(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0912 -.0906 -.0914 -.0907(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR -.0056 -.0085 -.0059 -.0086(0.0019)??? (0.0019)??? (0.0019)??? (0.0019)???Work Exp. Square (/100) - FOR 0.1309 0.1605 0.1316 0.1612(0.0198)??? (0.0196)??? (0.0198)??? (0.0197)???Gender 0.2873 0.245 0.2871 0.2447(0.0011)??? (0.0014)??? (0.0011)??? (0.0014)???Mother Tongue - ENG 0.0676 0.0747 0.0521 0.0549(0.0023)??? (0.0023)??? (0.0025)??? (0.0025)???Mother Tongue - FREN 0.0866 0.0884 0.073 0.071(0.0029)??? (0.0029)??? (0.003)??? (0.003)???Prov + CMA Fixed Effects Yes Yes Yes YesCountry/area Fixed Effects No No Yes YesField of Study Fixed Effects No Yes No YesWeighted Obs.R2 0.2448 0.2592 0.2462 0.2608Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5%and 1% levels respectively. The omitted categories in the CMA and Province fixed effects areToronto and Ontario, respectively. The omitted category for country/area of origin fixed effectis ?Eastern Europe?. The base case for the field of study fixed effects is ?Visual Arts?.612.4. ConclusionsTable 2.28: Interacting Field of Study with Foreign-born Dummy(1) (2) (3) (4)Immigrant -.0886 -.0893(0.0041)??? (0.0041)???Below Bachelor 0.1748 0.1747 0.1747 0.1749(0.0015)??? (0.0015)??? (0.0015)??? (0.0015)???Bachelor 0.4802 0.4793 0.4804 0.4799(0.0019)??? (0.0019)??? (0.0019)??? (0.0019)???Above Bachelor 0.6161 0.6154 0.6176 0.6171(0.0024)??? (0.0024)??? (0.0024)??? (0.0024)???Work Exp. - CAN 0.0521 0.0521 0.0522 0.0522(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0902 -.0902 -.0905 -.0906(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR 0.0093 0.0087 0.0068 0.0064(0.002)??? (0.002)??? (0.002)??? (0.002)???Work Exp. Square (/100) - FOR -.0141 -.0075 0.0151 0.0199(0.0203) (0.0203) (0.0203) (0.0203)Gender 0.2441 0.2441 0.2443 0.2442(0.0014)??? (0.0014)??? (0.0014)??? (0.0014)???Mother Tongue - ENG 0.0581 0.0578 0.0526 0.0526(0.0023)??? (0.0023)??? (0.0025)??? (0.0025)???Mother Tongue - FREN 0.0755 0.075 0.0691 0.0689(0.0029)??? (0.0029)??? (0.003)??? (0.003)???Prov + CMA Fixed Effects Yes Yes Yes YesCountry/area Fixed Effects No No Yes YesGROUP Location of Study Yes Yes Yes YesField of study Fixed Effects Yes Yes Yes YesInteraction: Field of Study * Foreign-born No Yes No YesWeighted Obs.R2 0.2609 0.2610 0.2616 0.2618Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1%levels respectively. The omitted categories in the CMA and Province fixed effects are Toronto andOntario, respectively. The omitted category in the groups of location of study is ?Canada?. The omittedcategory for country/area of origin fixed effect is ?Eastern Europe?. The base case for the field of studyfixed effects is ?Visual Arts?.622.4. ConclusionsTable 2.29: Interacting Field of Study with Location of Study Groups(1) (2) (3) (4) (5)ImmigrantBelow Bachelor 0.145 0.1744 0.1452 0.1747 0.1749(0.0014)??? (0.0015)??? (0.0014)??? (0.0015)??? (0.0015)???Bachelor 0.4116 0.4791 0.4133 0.4804 0.4801(0.0017)??? (0.0019)??? (0.0017)??? (0.0019)??? (0.0019)???Above Bachelor 0.5538 0.6179 0.5544 0.6176 0.6171(0.0022)??? (0.0024)??? (0.0022)??? (0.0024)??? (0.0024)???Work Exp. - CAN 0.0527 0.0522 0.0526 0.0522 0.0522(0.0002)??? (0.0002)??? (0.0002)??? (0.0002)??? (0.0002)???Work Exp. Square (/100) - CAN -.0914 -.0907 -.0911 -.0905 -.0906(0.0005)??? (0.0005)??? (0.0005)??? (0.0005)??? (0.0005)???Work Exp. - FOR -.0059 -.0086 0.0105 0.0068 0.0066(0.0019)??? (0.0019)??? (0.002)??? (0.002)??? (0.002)???Work Exp. Square (/100) - FOR 0.1316 0.1612 -.0245 0.0151 0.0186(0.0198)??? (0.0197)??? (0.0204) (0.0203) (0.0203)Gender 0.2871 0.2447 0.2869 0.2443 0.2441(0.0011)??? (0.0014)??? (0.0011)??? (0.0014)??? (0.0014)???Mother Tongue - ENG 0.0521 0.0549 0.0496 0.0526 0.0526(0.0025)??? (0.0025)??? (0.0025)??? (0.0025)??? (0.0025)???Mother Tongue - FREN 0.073 0.071 0.071 0.0691 0.0689(0.003)??? (0.003)??? (0.003)??? (0.003)??? (0.003)???Prov + CMA Fixed Effects Yes Yes Yes Yes YesCountry Fixed Effects Yes Yes Yes Yes YesGROUP Location of Study No No Yes Yes YesField of Study Fixed Effects No Yes No Yes YesInteraction: Group Location & Field No No No No YesWeighted Obs.R2 0.2462 0.2608 0.2471 0.2616 0.2621Note: Robust standard errors are in parenthesis. *, ** and *** denote significance at 10%, 5% and 1% levels respectively.The omitted categories in the CMA and Province fixed effects are Toronto and Ontario, respectively. The omitted categoryin the groups of location of study is ?Canada?. The omitted category for country/area of origin fixed effect is ?EasternEurope?. The base case for the field of study fixed effects is ?Visual Arts?. The interactions are between the groups oflocations of study (excluding Canada) and the eleven fields of study (excluding ?Visual Arts?).632.4.ConclusionsFigure 2.1: Location of Study Fixed Effects Table 2.16 Column 2642.4.ConclusionsFigure 2.2: Location of Study Fixed Effects - Table 2.16 Columns 2,4 and 6652.4.ConclusionsFigure 2.3: Country of Origin Fixed Effects Table 2.23 Columns 1,2 and 3662.4.ConclusionsFigure 2.4: Country of Origin Fixed Effects Table 2.23 adding groups of location of study Fixed Effects672.4.ConclusionsFigure 2.5: Country of Origin Fixed Effects Table 2.23 adding Location and Field Study Fixed Effects682.4.ConclusionsFigure 2.6: Country of Origin Fixed Effects Table 2.24692.4.ConclusionsFigure 2.7: Country of Origin Fixed Effects Table 2.24702.4.ConclusionsFigure 2.8: Country of Origin Fixed Effects Table 2.25712.4.ConclusionsFigure 2.9: Country of Origin Fixed Effects Table 2.25722.4.ConclusionsFigure 2.10: Country of Origin Fixed Effects Table 2.26732.4.ConclusionsFigure 2.11: Field of Study Fixed Effects Table 2.27742.4.ConclusionsFigure 2.12: Interacting Field of Study with Foreign-born Dummy - Table 2.28752.4.ConclusionsFigure 2.13: Location of Study table 2.29762.4.ConclusionsFigure 2.14: Interaction: Location + Field of Study Table 2.29 - Part 1772.4.ConclusionsFigure 2.15: Interaction: Location + Field of Study Table 2.29- Part 278Chapter 3Short and Medium RunOccupational Mismatches ofRecent ImmigrantsThis chapter shifts the focus of the analysis from the earnings to thequality of occupations immigrants attain. It evaluates the imperfect inte-gration of recent immigrants to the Canadian labour market by tracking theoccupational matches of those with different educational levels and languageproficiency. The research complements the existing literature by distinguish-ing between the first and second occupations and by accurately separatingthe short from the medium run effects of key variables. Overall, the ap-proach followed is in line with the studies of Chiswick et al. (2003, 2005)regarding the theoretical assimilation of immigrants, and Goel and Lang(2009) and Grenier and Xue (2009) concerning employment patterns of re-cent immigrants to Canada.The issue is of particular importance as a significant fraction of foreign-born is admitted to Canada on the basis of their individual and professionalskills. Specifically, Citizenship and Immigration Canada (CIC) grants thecategory of skilled-worker to those who possess individual characteristicsand occupational skills that are deemed to be required by the economyand that will help them become established. In this sense, a permanentand significant occupational mismatch represents an inefficient allocationof human resources and reduces the economic benefits to be derived fromimmigration.The rest of the chapter is organized as follows. Section 3.1 describes thedata, presents the basic patterns of immigrants? occupational matches andpresents the basic hypotheses. Sections 3.2 and 3.3 describes the empiricalmodel and presents the main results, respectively. Section 3.4 concludes.793.1. Data Set and Occupational Quality3.1 Data Set and Occupational QualityThe Longitudinal Survey of Immigrants to Canada (LSIC) is the maindata set used on this chapter. The LSIC was conducted by Statistics Canadaand Citizenship and Immigration Canada (CIC). The main topics includedemographics (sex, age, country of origin, etc.), language proficiency, educa-tion, foreign credentials, employment (job hold before migrating, labor forcestatus, etc.), health, social networks, income and settlement perception.The complete survey consists of three waves: six months, two years andfour years following immigrants? arrival in Canada. It was conducted onimmigrants who arrived between October 1, 2000 and September 30, 2001;were 15 years or older at the time of landing; and landed from abroad(individuals who applied and landed from within Canada are excluded fromthe survey).Regarding employment information, the LSIC includes a record of allthe jobs immigrants held in the first four years in Canada and the last oneheld in the home country. Specifically, in each wave all immigrants wereasked about all the jobs they held since their last interview (or arrival, if itis the first interview). Only one job is identified as the principal occupationper wave. Likewise, they are also asked to provide information regardingtheir ?intended? occupation before coming to Canada and the occupation?wanted? six months after arriving.The survey uses a ?funnel-shaped? approach, that is, only immigrantswho responded to the first wave are tracked for wave 2; and only thosewho responded to the second wave for wave 3. The number of observationsper wave are 12040, 9322 and 7716 respectively. This study uses the 7716immigrants who were in all the waves.The overall attrition of 4284 observations (2,700 from wave 1 to wave2 and 1,600 from wave 2 to wave 3) might be worrisome if those leavingthe sample are significantly different from those who remain. In that case,the significance of a particular variable could be due to the analysis of abiased sample (as the individuals for which the variable of interest was notsignificant decided to leave the survey).I address the attrition fears by evaluating those in the second wave whowere interviewed in wave 3 versus those who were not interviewed on wave 3.Specifically, I run a probit model with a dichotomic dependent variable thatis one if the immigrant was interviewed in wave 3 and zero otherwise.47 The47Access to the immigrants who were interviewed in wave 1 but not in wave 2 was notgranted by the Research Data Center803.1. Data Set and Occupational Qualityestimation (see Appendix B.1) shows that attrition (from wave 2 to wave 3)is not related to English proficiency, employment, gender, education, numberof months since arriving and region of birth. The only variable that appearssignificant (at 5 percent) is age with a low marginal effect.48It is important to remember that not all immigrants come to Canada forthe same reasons.49 Some came to work as employees while others come look-ing for investment opportunities, and still others come as family dependants.Broadly we can identify five classes of immigrants: ?Family class?,?Businessimmigrants?, ?Skilled workers and professionals?, ?Provincial Nominees?and ?Refugees?.50Under the current immigration policy, family immigrants are sponsoredby permanent residents or citizens, and in general they are the sponsor?sspouse, parent, grandparent or children.51 The sponsor commits to pro-vide financial support for his/her relative if necessary. On the other hand,Business immigrants are experienced business people that are accepted onthe grounds that they will help develop the Canadian economy. They arerequired to have a minimum net worth of $ 800,000 (Cdn.) and make an in-vestment of $ 400,000 (Cdn.) to CIC for 62 months.52 Provincial Nomineesare nominated by a province requiring their skills, education or work expe-rience.53 In the same way, skilled workers are selected at the federal levelby a point system based on their education, work experience, knowledgeof English and/or French, age, adaptability and on whether they have ar-ranged employment. They are supposed to have overall skills that will helpthem become economically established and that are needed in the country.5448The drop in observations may be explained in part by immigrants? refusal to par-ticipate in a long survey every two years. According to Statistics Canada the interviewslasted approximately between 65 to 90 minutes.49 All the immigrants in the LSIC were admitted under the Immigration Act of 1976.50 For detailed information on the current different types of immigrants and the re-quirements for each see http://www.cic.gc.ca/english/immigrate/index.asp51 They can also be brothers or sisters, nephews or nieces, granddaughters or grandsonsas long as they are orphaned, under 18 years of age and not married.52There are two other classes of business immigrants: entrepreneurs and self-employedpersons. Entrepreneurs must have a a minimum a net worth of C$ 300,000 and managea Canadian business creating at least one full-time job. Self-employed must demonstratethat they posses self-employment experience, farm management experience or have par-ticipated in world-class cultural activities or athletics.53As the Provincial Nominee program was fairly new at the time of the first wave ofthe LSIC, the number of immigrants interviewed in this category is substantially small.54Under the 1976 Immigration Act a nine-factor points system was used to select skilledimmigrants. The system assigned values to education attainments, vocational preparation,occupation demand, experience, arranged employment, demography, age, language, and813.1. Data Set and Occupational QualityLastly, refugees are defined as those in or outside Canada who fear returningto their home country. The majority are victims of civil war, armed conflictor violations of human rights (Citizenship and Immigration Canada, 2006).Some are eligible for the governmental Resettlement Assistance Programwhich provides income support for their first year.55The remainder of the information used is derived from the 20% sam-ple of the 2001 Canadian Census. The Census sample is used to calculatethe average quality of each of the 520 occupations of the (Canadian) Na-tional Occupation Classification list (NOC-S 2001); as well as to measurethe occupational quality of an ethnicity in a given labor market. Trying toincorporate the notion of a geographical labor area and following Goel andLang (2009), I selected Census Metropolitan Area or Census Agglomeration(CMA/CA) as the unit of reference.563.1.1 Working SampleUnlike most of the literature I focus on work oriented immigrants, soself-selection participation in the labour market does not become a consid-erable problem. Specifically, I use male immigrants between 24 and 55 yearsold who belong to one of four detailed immigration categories: ?Skilled Im-migrant - Principal Applicant (PA)?, ?Skilled Immigrant - Spouse and De-pendent (S and D)?, ?Family Class - Spouses and Fiances? or ?Family Class- Other?.57 Given the age, gender and immigration category restrictions,the sample consists of around 2,300 immigrants, with high labor marketparticipation rate and low rates of unemployment.In contrast to the rest of immigrants (refugees, Business immigrants,Family Class - Father and Grandfather, etc.) those comprising the samplenot only needed to work but most likely came to work. At least for theSkilled Immigrants (PA), a large number of them were assessed for theirpersonal suitability. The original point system was amended several times, including majorrevisions in 1985 and 1992, before being replaced in 2002 by the Immigration and RefugeeProtection Act.55 Aydemir (2009) notes that the provinces may provide refugees with financial supportin the second year if they look for work or take classes.56According to Statistics Canada, a Census Metropolitan Area or Census Agglomera-tion consists of one or more adjacent municipalities centered on a large urban area, knownas the urban core. The census population of the urban core must be at least 10,000 toform a census agglomeration, and at least 100,000 to form a census metropolitan area. Inthe 2001 Census there were 27 CMAs and 113 CAs.57 The third (and last) category of Family class immigrants, that is not considered, is?Family Class - Parents and Grandparents?.823.1. Data Set and Occupational Qualitylabor skills. This makes them a particular good sample to measure occupa-tional mismatch in the short and medium-run.Tables 3.1, 3.2, and 3.3 present descriptive statistics of the selected im-migrants using the survey weights of the LSIC. 58 Overall, they show a labormarket participation rate of over 90% even after only 6 months in the coun-try; and about 90% of them are employed by wave 2. In addition, they havehigher levels of education than the majority of immigrants. The principaleducational differences lie at the tails of the distribution, the proportion ofthe selected immigrants with a high school diploma or less is 6% while thepercentage of those with more than a Bachelor?s degree is about 30%. Forother Immigrants categories, the proportion of individuals with high schoolor less is between 30% and 84%; and at most only 8% have a graduateeducation.An important fact to consider is that not all immigrants find an occupa-tion at the same time. Although labor market oriented, 6% of the selectedimmigrants remain without a job by the third wave and about 28% dur-ing the first 6 months. Because the conditions that led an immigrant todecide not to have a job two years after arriving might differ from thosefaced by the rest of the selected sample and because late arrivals into thelabor market will significantly reduce the employment history, I decided toreduce the sample size. The estimations will be conducted using immigrantsthat found their first job within 18 months of arriving. This will allow meto focus on those who needed or desired to work quickly and will provideme with lengthy employment records. Further restrictions excluded obser-vations with missing values or errors in key explanatory variables, like theoccupational code. In addition, following Goel and Lang (2009) I limit thesample to CMA/CAs and countries of origin with at least ten individuals inthe LSIC. The final working sample consists of 1,579 immigrants.59Table 3.4 presents a summary statistics of the working sample. Thetable shows that during the period of analysis 34% of immigrants had onlyone job, 50% two jobs and a relatively small share (16%) three jobs. Itshould be noted that even if an immigrant has only one job, the LSIC couldrecord up to 3 answers (on the 3 interviews) on the occupational code. Thischange could denote an occupational improvement while working for the58All descriptive statistic are weighted according to Research Data Center data releasedpolicy.59Goel and Lang also restrict their analysis to immigrants for whom which the 2001Census lists 500 or more countrymen in the CMA/CA of residence. The aim is to ensurean accurate measure of ethnic networks. However, ethnic networks is not among the keyvariables of interest in this study.833.1. Data Set and Occupational Qualitysame employer, but it could also reflect measurement error in coding.60However, the majority of single-job immigrants give a very similar an-swer for the occupational code in all three waves. Figure 3.2 graphs thedistribution of occupations in each wave (according to the average loga-rithm of weekly wages) and confirms that immigrants who change jobs areresponsible for the bulk of the occupational mobility.61 Thus, for the studyof occupational change (from the first to the second occupation) I choose thefirst answer given as the one representing the occupational code for all thetime employed. In contrast, the analysis of occupational match through timeestimations is conducted using the occupational codes given in each wave,as if all immigrants had three occupations (one occupation per wave).623.1.2 Occupational qualityI use the 2001 Census to construct an occupational quality index. Specif-ically, I calculate the average of the logarithm of weekly wages for full-timemale Canadian paid workers for each of the NOC-S occupations. In thisway I am able to evaluate an improvement (or decline) in the occupationalstatus of a recent immigrant. Note that foreign-born workers are excludedin the calculations in order to avoid incorporating possible occupationalmismatches.In should be noted that other measures of evaluating occupational qual-ity where explored. For example, the literature had used the occupationalranking provided by O*NET to construct indexes of cognitive and non-cognitive skills (Imai et al. (2011) use the O*NET on their analysis of theLSIC). In Canada, the Human Resources and Skills Development Canadaconstructed a Career Handbook that also ranked occupations according totheir different features (among them, numerical, manual and verbal apti-tudes). The results found were very similar regardless of the occupationalquality measure used. However, the use of the logarithm of weekly wagesallows for a simple interpretation of occupational mobility. For example, ifthe difference between two occupations is -0.10 that could be read as a re-duction in a worker?s weekly wage of about 10% (though the approximationloses accuracy as the gap between occupations widens).60There exists literature that discusses the presence of measurement error in occupa-tional codes and the bias this generates in the study of occupational choices. For example,analyzing the U.S. National Longitudinal Survey of Youth, Sullivan (2009) finds that 9%of occupational choices are misclassified61Figure 3.2 was calculated only for those who have an occupation in wave 162The exception would immigrants who found a job after their first interview. Theywould only have two main occupations.843.1. Data Set and Occupational QualityIn the end, I compile a list of 520 occupations from the NOC-S 2001and I assign a value to each representing occupational quality. AppendixB.2 provides a sample that illustrates how the occupations were ordered.Listed are the names of 22 occupations ordered according to their logarithmof weekly wages (there are 22 or 23 occupations between each one). At thetop of the list we find ?Judges? and ?Geologists? with (rounded) values of7.89 and 7.13, respectively; while at the end of the table we have ?Babysitters? and ?Furniture Assemblers?with values of 5.98 and 5.36. Using theoccupational information from the employment roster of the LSIC, I assigna quality level to each occupation foreign-born workers have in their firstfour years in Canada and the last one held in the home country.63 Withthis information I was able to define the occupational gap as the quality ofan occupation in Canada minus the quality of the home-country occupation(evaluated using Canadian wage information).Occupational Gap = Occupational Quality Canada?Occupational Quality Home CountryIf the immigrant improved his occupational status the occupational gapwould be positive; if however, there was a decline then the occupationalgap would be negative. Moreover, because I can distinguish between oc-cupations, I can separately evaluate the first and second occupational gapand check whether improvement occurs over time. In a regression set upa positive coefficient would indicate that the variable helps improve theoccupational status of a recent immigrant (relative to his home countryoccupation), while a negative coefficient would signal the opposite.The definition of occupational quality seeks to capture an objective (av-erage) job productivity measure, so the distance from the home-county oc-cupation could be understood as a (mean) productivity difference. Otherpossible points of comparison could be the self-reported ?intended? occu-pation, which refers to the desire occupation before arriving in Canada,or the ?wanted? occupation reported in the first interview of the LSIC (6months after arriving). However these two measures could incorporate anendogenous change in expectations due to the forthcoming change in the la-bor environment (or verified change in the case of ?wanted? occupation).64In contrast, the code for the home country occupation is derived from theadministrative database of landed immigrants. Moreover, in my working63As the LSIC uses the 1991 ?Standard Occupational Code? (SOC) to classify occupa-tions, a concordance between NOC-S 2001 and SOC 1991 is employed.64Figure 3.3 shows the distribution of the intended and wanted occupation versus theactual occupation held in the home country.853.1. Data Set and Occupational Qualitysample these other variables have a large number of missing values (seetable 3.4).Concerning ethnic networks, I estimate the average occupational qualityof every ethnicity in every CMA/CAs (except for those in which they don?treside). That is, I attached an occupational quality value to each full-timeforeign-born paid-worker (male or female) and calculate an average by eth-nicity and CMA/CA. The 2001 Census is able to distinguish ethnicity indetail but the LSIC is not, so I use country of origin as a proxy for eth-nicity. Next, I use the demographic variables and the geographic referencecoding of the LSIC to match each recent immigrant with his correspondingCMA/CA and country of origin. In this way I am able to link the qualityof an immigrant?s occupation to the occupational quality of his ethnicity inthe CMA/CA he lives in. Trying to achieve a better proxy for ethnicity,I group Latin-America and Spain because of the predominance of Spanishand I separate Hong Kong from mainland China.3.1.3 Occupational mobilityFigure 3.4 shows the occupational quality of the first three occupa-tions held in Canada as well as the quality of the occupation held in thehome country and the desired occupations before (?intended?) and after(?wanted?) arriving in Canada. The behavior shown is consistent with theassimilation literature. Immigrants first occupations in Canada are of sub-stantial lower quality than the those held back home. The difference betweenthe average occupational quality in the home country and the occupationalquality of the first occupation is about 0.34 log points (which is the distancebetween the occupational quality of ?Financial and investment analysts? andthat of ?Land surveyors?and can be understood as a reduction of around30% on the weekly wage)The average skill level of their second and third occupations are between0.08 to 0.12 points higher than their first. Hence, figure 3.4 presents apicture of occupational integration, although the occupational gap closedonly by a third over the course of 4 years (and only for those who changesoccupations). The same behavior is observed when analyzing occupationalstatus at given intervals (6 months, 2 years and 4 years after arrival).Figures 3.5, 3.6 , 3.7, and 3.8 explore the initial hypothesis, presented inthe literature, regarding the transferability of individual skills. I evaluate theoccupational mismatch of immigrants with different levels of education andlanguage proficiency and in different immigration categories; I also analyzethe occupational mismatch of immigrants who find a job directly through863.1. Data Set and Occupational Qualityfamily or friends. In figure 3.5 the sample is divided in two parts, onecomprised of immigrants with an education level equivalent to or higher thanan undergraduate bachelor?s degree and another made up of those with lessthan a bachelor?s degree. For both groups I calculate the occupational gapof their first occupations. I also produce a graph for all immigrants (left sideof the graph) and for those who have more than one occupation (right sideof the graph). I find that immigrants with a high level of education tend toassimilate less effectively than those with a low level. The former experiencea larger decline in their occupational status (0.38 points drop versus 0.20),which significantly affects their assimilation in the short and medium run.Even though the second occupations of educated immigrants show a largerincrease in quality than those with a low level (reducing the gap in 0.09with respect to the first occupation), their level of occupational mismatchis still higher. Just as skill recognition problems an entry effect might, inpart, explain the downward occupational mobility of immigrants. People withigh levels of education would have a high occupational status in their homecountry. Obtaining those few high status occupations would be difficult ina new labour market. In principle, the same entry problems might be facedby educated Canadians entering the labour force. Unfortunately, we do nothave a adequate native-born control group to identify the effect.Analysis of language proficiency shows a pattern consistent with the lit-erature on assimilation. Language proficient immigrants face a significantlysmaller decline in the quality of their first occupation than non-proficientimmigrants (a decline around 0.29 versus 0.42). That is, an immigrants ca-pable of communicating well in the language of the city (in which he resides)would be better able to transfer his skills than one with communication dif-ficulties. I define an immigrant as language proficient if he reports writingEnglish very well in a English speaking city or French very well in a Frenchspeaking city (an exception is made in Montreal, where English proficiencyis also consider). As before non-proficient immigrants are found to experi-ence a larger improvement in their second and third occupation, reducingthe gap with their proficient counterparts. However, even in the second andthird occupations language proficient immigrants have a lower occupationalgap (between 0.09 and 0.04).Regarding immigration type, we observe a pattern similar to what Gre-nier and Xue (2009) found in their study. Family immigrants tend to havea lower occupational mismatch than skilled workers (both PA and S andD). Unlike most of the other variables the gap between family immigrantsand skilled workers remains relative stable. The most likely cause reviewedin the literature for this occupational pattern is that the skills family im-873.1. Data Set and Occupational Qualitymigrants require in their jobs are easier to transfer than the ones skilledworkers possess.Figure 3.8 shows the occupational gap of people who found their firstoccupation through family or friends and those who obtained it throughthe market (directly contacting the employer, using an employment agency,posting or answering an add, etc.). In the case of immigrants who findtheir first job through their social networks, the occupational gap is larger.Although a simple average, this finding conflicts with earlier studies that finda positive effect of ethnic networks on employment and on helping achievethe intended occupation. It is important to note that most of the differencesbetween the groups mentioned are statistically significant. Appendix B.2presents regressions of the first and second occupational gap on each of thevariables analyzed here. Almost all the variables appear relevant to explainthe occupational gaps.As mentioned above most immigrants have more than one occupationin their first 4 years. I disaggregate immigrants based on the number ofoccupations they had in Canada (figure 3.9). The most salient feature liesin the difference between the initial occupational gap of immigrants withone occupation and those with 2 or 3 occupations. Immigrants with oneoccupation experience less occupational downward mobility than those with2 or 3 occupations, which explains the need to change occupations. Figure3.10 presents the occupational distribution of immigrants by their numberof occupations. We see, again, that the overall distribution movement (froma concentration in high skill occupations to a concentration in low skilloccupations) is smaller for immigrants with only one occupation and thatthe occupational distribution prior to coming to Canada is very similar forall groups.Considering the information on figures 3.2, 3.9 and 3.10 as well as thetable of descriptive statistics, a rough calculation on the improvement ofoccupational quality can be made. Specifically, we could separate immi-grants by the number of occupations they have (one, two or three) and tryto infer the quality of their occupations at beginning and end of the panel.This would allow us not only to calculate the general occupational qualityimprovement from the labour market entry to the end of the survey, butalso broadly identify what share of the improvement is due to change inoccupation.65From figure 3.2 we already know that immigrants with only one oc-65To calculate how much of the wage improvement is due to occupational mobility wewould need express the average wage in the first and last occupation as weighted averages883.1. Data Set and Occupational Qualitycupation do not importantly improve their occupational quality, but ourcalculation would help quantify the importance of changing occupations.Combining the information of the average quality of the occupations heldin the home country, the share of immigrants by occupations (both in thedescriptive statistics) and the occupational gaps (in figure 3.9) we could veryroughly infer the initial and final quality of occupations.66 Assuming thatthe people who change occupations would have not improve their occupa-tional status have they stayed, we can claim that the change in occupationsis responsible for around 90% of the quality of occupational improvement inthe first four years.A final significant point is that no group overcomes the initial occupa-tional decline. Those with 2 or 3 occupations notably reduce the size of theiroccupational gap (by 0.15 and 0.20, respectively) obtaining a final occupa-tional gap near or smaller than is the case for immigrants with 1 occupationhave; but no one reduces the occupational gap below -0.24. Unfortunatelythe limited number of years covered by the LSIC precludes interpreting thisresult as a long-run equilibrium outcome or just temporal result on a longerassimilation path.3.1.4 Hypotheses summaryGiven the basic patterns of the data and the literature on U-shaped laborassimilation, one would expect a minimum number of hypotheses to hold.671. Immigrants with high education levels and/or low levels of languageproficiency are likely to experience a greater occupational mismatchas they would face transferability obstacles to their skills in the shortrun.2. The probability of changing occupations should be positively influ-enced by education and language. Highly educated and languageof the wages of the different immigrant groups. Specifically, we would need:W1 = W1occup1 ? (share occup1) + W1occup2 ? (share occup2) + W1occup3 ? (share occup3)W3 = W3occup1 ? (share occup1) + W3occup2 ? (share occup2) + W3occup3 ? (share occup3)The wages of each group could be approximated from the information on the occupationalgap and checked against the averages of the descriptive statistics.66Indeed, the deducted numbers give an average quality for the first occupation con-sistent with the descriptive statistics.67Appendix B.4 presents a motivational model on occupational mismatch and mobilitythat generates similar hypotheses.893.2. Empirical Strategyproficient individuals will have more opportunities to improve theiroccupational status.3. For immigrants who change occupations education should help theoccupational improvement.4. The magnitude of the effect of education on the second occupationalgap remains a matter to be resolved empirically, but its sign shouldstill be negative.3.2 Empirical StrategyI regress immigrants? first (and second) occupational gap on their ed-ucation levels before arriving, language proficiency and socio-demographiccontrols. That is, I estimate the following equation:Y skilliec = ?0 +J=3?j=1?jEducij + ?1Languagei +Xiec?+De? +Gc? + ?iec(3.1)where Y skilliec represents the first (or second) occupational gap of immigranti from ethnicity e who lives in CMA/CA c. A positive coefficient wouldindicate that the variable helps the immigrant improve his occupationalstatus (or at least reduce downward occupational mobility).I incorporate a set of three education dummies (?J=4j=1 Educij): somecollege, Bachelor?s degree and graduate studies (with ?high school or less?the base case). According to our hypothesis and the literature (Chiswick et.al. (2003 and 2005) among others) immigrants with high levels of educationexperience greater difficulty finding an occupation offering the same statusas the ones they had in the country of origin. A dummy for language profi-ciency is included on the assumption that skill transfer would be easier forimmigrants with proficiency in the dominant language spoken in the city. Insome of the estimations I interact a dummy variable indicating whether theimmigrant has a Bachelor?s degree or higher education with the languageproficiency dummy. In this way, I evaluate whether language proficiency hasa different impact on the occupational assimilation of educated immigrants.The set of demographic controls, Xiec, includes dummies for immigrantswho enter the country under the family class program or as a skilled workerspouse or dependant (making ?Skilled worker - Principal Applicant? thebase category). In principle, given that immigrants who enter under the903.2. Empirical Strategyskilled-worker program (Principal Applicant) are evaluated before comingto Canada, one might expect them to be more prepared for the Canadianlabor market or posses distinctive characteristics (such as entrepreneurship)that could help them assimilate into the new environment. Xiec also incor-porates information on the age of the immigrant at arrival and the averageoccupational quality of the immigrant?s ethnicity in the CMA/CA of resi-dence, these two variables are added as to incorporate the greater ability ofyounger immigrants to adapt to new environments and control for possibleethnic network effects on labor market outcomes.To control for structural labor demand differences among cities as wellas possible mean differences of occupational preferences across ethnicities,a set of countries of origin and CMA fixed effects are added. As mentionedabove, country of origin is used as a proxy for ethnicity and some countriesare grouped together (those of Latin-America and Spain) while others aredisaggregated (Hong Kong and Mainland China).3.2.1 City self-selection and the identification methodA common criticism regarding equation (4) is the possibility of endo-geneity in the explanatory variables of interest. The selection of a particularCMA/CA to live in may be linked to the expected return to education orlanguage proficiency, which would bias the coefficients and the interpretationof the results.It could be argued that immigrants CMA/CA selection is conditioned byethnic historical settlement patterns and not by current economic prospects.Census statistics show that the vast majority of immigrants lives in themajor metropolitan areas (namely Toronto, Montreal and Vancouver) andthat their level of geographical concentration greatly exceeds that of natives(see table 3.5). In fact the three largest metropolitan areas represent about34% of the total population yet are home to 63% of all foreign-born. Recentimmigrants (i.e., those who have spent 5 years or less in Canada) show evengreater levels of geographical concentration, with 69% of them residing inToronto, Montreal and Vancouver.68. It should be noted that the geographicconcentration of immigrants in the LSIC in the five largest metropolitanareas (see table 3.4) is very similar to that observed in the Censuses.The summary statistics show that about 50% of immigrants (in the work-ing sample) report that their main reason for choosing a city is: ?family?,68In 2001 Toronto, Montreal and Vancouver represented 73% of the recent immigrants.Even though that number decreased by 2006, it is still higher than the fraction for allimmigrants. Hence, it can only be argued that the rate of concentration is diminishing913.2. Empirical Strategy?friends? or ?ethnicity?. Another 27% chose a city principally for othernon-economic reason (such as ?the weather?, ?taxes? or ?language spokenin the city?). Therefore, it can be argued that most recent immigrants, non-economic reasons constitute the primary variables used to evaluate whereto reside. Nevertheless, a significant fraction (24%) chose a city because ofthe employment prospects (?Possible Jobs? or ?Economic Prospects?).69 Acommon approach to addressing the endogeneity problem cited in the im-migration literature involves instrumenting a particular variable of interest(such as the flow of immigrants to a particular city) by its lagged value (10,20 or 30 years back).In the U.S. this approach has been found to be particularly useful and sig-nificantly strong for immigration aggregates by city (see Card (2001, 2005),Card and DiNardo (2000), Borjas (1998) and Patel and Vella (2007)). How-ever, this method might not be aplicable in the current estimation. Asnoted by Goel and Lang (2009), an immigrant with specific individual fea-tures may be inclined to locate in a given city more than other becauseof long-run specific economic benefits to that city for that particular in-dividual characteristics. Moreover, there exists more than one variable ofinterest (education and language proficiency) that cannot be instrumentedby a historical aggregate value.Instead, I adopt Dahl (2002) self-selection correction procedure to ad-dress the possible non-random selection of CMA/CAs by immigrants. Dahlargues that under a sufficiency assumption the mean error becomes a func-tion of the probability that a person in a particular state (born in a partic-ular country, a specific level of education, etc.) makes the choice actuallyobserved.70 In Dahl?s case, this would refer to an American born in a par-ticular state deciding to which state he should migrate. In our case, itwould involve immigrants born in a particular country deciding to which69The LSIC contains several questions that solicit whether a particular factor (friends,family, language spoken, etc.) was relevant to deciding where to live in. The last questionof the section asks to select the most important reason a specific city was chosen.70The sufficiency assumption implies than in a model of multiple location choices onlythe probability of the (observed) utility maximizing choice of residence matters in deter-mining the selection bias. The particular identities and probabilities of the second bestthrough the Nth best migration choices offer no information about earnings in the city inwhich an individual chooses to live. Adapting the example given by Dahl, we can imaginetwo immigrants born in the same country who chose to move to the same city c becauseit was their expected utility maximizing choice. The fact that one immigrant?s secondchoice of where to live differs from the the other immigrant is not relevant, and cannotaffect the error term for earnings in city c.923.2. Empirical StrategyCMA/CA within Canada they should migrate.71 As explained by Dahl, theutility associated with this correction comes from the possibility that indi-viduals with similar city selection probabilities but different education levels(or other variable of interest) can help identify the effect of education (orthe other variable).My estimation of CMA/CA selection probability uses the Census 2001and employs the non-parametrical specification used by Dahl (2002).72 Irestrict my sample to male immigrants aged 24 to 55. For each countryof origin I create 4 education groups (No High School, High School, SomeUniversity and Bachelor or Graduate), three age groups (24 to 35, 36 to45 and 46 to 55) and four marital status (single, married or common law,divorced or separated and widowed). For each of these groups (ethnic-education-age-marital specific) I calculate the fraction (of the group) thatlocates in a particular CMA/CA (Dahl defines this fraction as ?first-bestprobability?). I use a fourth order polynomial of the first-best selectionprobability as the ?city selection correction?.733.2.2 Probability of occupational changeTo test if the probability of switching occupations is influenced by themain variables of interest I estimate a probit model with a modified versionof equation (4). Iiec is an indicator that the immigrant i from ethnicity e andCMA/CA c changed occupations at least once over the four years after hisarrival. One means that the immigrant had more than one occupation andzero that he remained in the first occupation.74 I also control for CMA/CAand country of origin fixed effects; as well as, for possible ethnic network71See Dahl (2002), pages 2378 - 2384.72To simplify the estimation of probabilities Dahl assumes that people in similar groupsface similar utilities in choosing a city. Hence, the probability choosing a particular city isthe same for all the members of a cell and can be calculated non-parametrically throughthe census.73Dahl suggests the incorporation of additional probabilities to control for city selection.In his paper he controls not only for the first-best probability of Americans migrating toa particular city but also for the probability of Americans deciding to stay in their citiesof origin.74Because of the small number of immigrants who have 3 occupations, I decided notto run a probit model to explain the probability of switching once the second occupationhad been taken up933.2. Empirical Strategyeffects.Pr(Iiec = 1) = ?0 +J=3?j=1?jEducij + ?1Langi +Xiec?+De? +Gc? + ?iec(3.2)3.2.3 Occupational ImprovementTo explore the effect of individual skills on the immigrants who changeoccupations I adapt equation (4) to estimate the occupational improvement.Specifically, I run the change in occupational quality from the first to thesecond occupation on education and language. The rest of controls for age,immigration category, ethnicity, CMA/CA and network effects are incorpo-rated as well.Y skill2iec ? Yskill1iec = ?0 +J=3?j=1?jEducij + ?1Langi +Xiec?+De? +Gc? + ?iec(3.3)3.2.4 Occupational gap through timeRecognizing that the change in occupational codes for immigrants whoreport having only one occupation could reflect occupational improvement, Itake advantage of detailed records of the starting date of the first occupationand calculate the occupational gap immigrants have 0 to 6, 18 to 24 and 30to 36 months after their first occupation started.75 In this way I estimatethe effect of key variables for the whole sample and check for changes in thesize and significance of coefficients through time.I model the occupational gap using the same explanatory variables as inequation (4). The equation to estimate isY skilliect = ?0t +J=3?j=1?jtEducij + ?1tLangi +Xiec?t +De?t +Gc?t + ?iect(3.4)75The starting date of the first occupation was the lowest number of imputed valuesamong all the employment record dates. A range in the order of months was preferred toa particular date to avoid the possibility of temporal unemployment of immigrants in aspecific month.943.3. Findingswhere Y skilliect indicates the occupational gap of immigrant i from ethnicity eand CMA/CA c held a specific number of months (t) after the first occu-pation started. The rest of the variables have the same meaning as before.3.3 Findings3.3.1 First Occupational GapTables 3.6 and 3.7 present regressions for the first occupational gap basedon equation 3.1. The first column of table 3.6 has no interaction variableswhile the second and third add interaction terms of language proficiencyand education, and a dummy variable if the occupation was obtained di-rectly through family or friends (ethnic job). Additionally, the second threecolumns restrict the sample to people who found an occupation in the first12 months after arriving, focusing on immigrants who needed to find anoccupation faster than the rest. All of the regressions have fixed effectsfor CMA/CA and country of origin; and the standard errors presentedare robust and clustered by CMA/CA. Table 3.7 addresses the possibil-ity of city self-selection and re-estimates the first three columns of table3.6. The first estimations restrict the sample to immigrants who reportedmainly choosing a city for reasons other than ?job opportunities? and ?eco-nomic prospects?(columns 1 to 3). The second set of estimations apply Dahl(2003) self-selection correction method (columns 4 to 6), while the last three(columns 7 to 9) apply the self-selection correction to immigrants who chosea city for non-economic reasons.The results confirm our hypothesis regarding the entrance to a new labormarket. Occupational assimilation is harder for people with high educationallevels as they would experience more obstacles to the recognition of theirabilities. The higher the education level an immigrant has the larger the dropon occupational status he experiences. Specifically, people with a bachelor?sdegree or at graduate level suffer a negative impact between 24 to 36 points;which could be roughly understood as a 21% to 30% decrease in weeklywages. The effect is significant at 1% in all columns (and in both tables).People with some university education also suffer a significant decline intheir occupational quality but in a relatively smaller magnitude (16 to 20points, around 19% to 21%).We can not rule out, though, that a lower quality of education abroadcould be driving some of these results. That is, the Canadian labour marketcould be recognizing immigrants? skills, but they are of lower quality thantheir Canadian counterparts (specially, highly educated immigrants).953.3. FindingsLikewise, proficiency on the main language spoken in the city has apositive and significant impact on the occupational gap (around 10 points).These results corroborate the importance of knowledge of the city?s languageto better adapt to the new environment and are consistent with difficultiesin the accreditation of occupational abilities. Immigrants who have a higherunderstanding of the host country language have less problems in the recog-nition and proper identification of their skills. The language variable losessignificance when an interaction with a dummy for educated immigrants(bachelor?s degree or higher) is added; nevertheless, the interaction itselfbecomes significant and with a higher coefficient (around 18 points). Thismight indicate that language proficiency is more relevant for educated im-migrants, in the recognition of their high-skill occupations. It is importantto note that when the interaction is included the coefficients of ?graduate?or ?bachelor? become more negative by about 10 points.Regarding immigration categories, being admitted as a ?Skilled WorkersSpouse and Dependant? has a negative but not robustly significant effecton the initial occupational gap. The coefficients range from -6 to -7 points,but are only marginally significant (10%) in the first three columns of table3.6. On the other hand?Family immigrants?, have a positive coefficientin all the estimations, though they also are not statistically different than?Skilled Workers - Principal Applicants? (the base case). These resultscontrast with the ones observed in Appendix B.2, where family immigrantshad a significant difference from the base case. It could be argued thus thatmost of the difference in immigration categories comes from factors alreadyincorporated in the estimation.It should be noted that although the decision to find a job directlythrough family or friends is a clear endogenous variable, its incorporationhelps us understand the way ethnic networks relate to recent immigrantsemployment outcomes. Columns 3 and 6 of table 3.6 (and columns 3, 6 and9 in table 3.7) show that immigrants who obtained their first occupationthrough their networks had a significant (at 1%) occupational drop (theeffect is as large as a 5% decline in the weekly wage).76 One possible expla-nation is that immigrants who decide to take a ?network? occupation haveunobservable low skills or have greater obstacles to the recognition of theirabilities and hence have a higher occupational mismatch.The results of table 3.7 for the main variables are consistent with the76This results is somewhat contradictory to what Grenier and Xue (2009) found ontheir estimation of the time need to get the intended occupation. There, obtaining a jobthrough friends or family decreases the duration of the search for the intended occupation.963.3. Findingsones of table 3.6. Immigrants with higher education (graduate or bachelor?s)experience a larger occupational mismatch. Language proficiency helps la-bor market assimilation and the interaction with education has a positiveand significant effect (and it increases the negative effect of the educationdummies). People who choose to find a job through their network end upwith a larger (negative) occupational gap.The coefficients and significance of most variables appear to be robustto the city self-selection correction and to limiting the sample to immigrantsthat choose their place of residence mainly for non-economic purposes. Thisis consistent with immigrants choosing residence following historical ethnicsettlement patterns instead of actively self-selecting into particular cities.Also, at the end of columns 4 to 9, I run tests on the validity of the inclusionof the self-selection correction variables. In none of the tests am I able toreject the hypothesis of insignificance. Finally, in addition to the robust(and clustered) standard error, bootstrapped standard errors are calculated(shown in bracket) as the inclusion of generated regressors may affect thevariance-covariance matrix. The bootstrapped standard errors calculatedare very similar to the robust ones.3.3.2 The Second Occupational GapTables 3.8 and 3.9 analyze the second occupational gap with and with-out controlling for CMA/CA self-selection. On table 3.8, columns 1 to 5run regressions for the whole sample, while columns 6 to 10 restrict the es-timations to immigrants employed in the first 12 months after arrival. Inaddition, columns 4, 5, 9 and 10 control for the first occupation. Condition-ing on the first occupation tries to separate the effects of the main variablesat the beginning of the integration process from their effects on the secondoccupation.The results show a decrease in the statistical importance of education inthe second occupational gap. Whereas in table 3.6 the coefficients associ-ated with graduate and bachelor?s education level always had a significanceof 1%, for most of the columns of table 3.8 the variables are not signifi-cant. The size of the educational dummies also changes. Specifically, thecoefficient of the ?graduate? dummy becomes less negative by about 16points (comparing the first columns of the tables). The loss in significanceof the educational variables speaks of a process of assimilation in which ed-ucation has a significant negative effect for the first occupational gap butoccupational improvement makes it a less negative trait and relevant for thesecond occupational gap. However, the fact that even when conditioning973.3. Findingsby the quality of the first occupation educational variables do not becomepositive (or significant), indicates that migrants are still not able to fullyuse their individual skills.Language proficiency becomes not significant, but the interaction be-tween language proficiency and high education has positive and significanteffect. The robustness of the interaction underlines the importance of com-munication for highly educated immigrants and their occupational match.Nonetheless, the inclusion of the interaction increases again the negativesize of the coefficients of the education variables and changes the sign of thelanguage coefficient.The immigration categories show some relatively different patterns thanin table 3.6. The family class category is positive and not significant in allcolumns, but the ?skilled worker S and D? category is negative and signif-icant in some specifications. Nevertheless, although the dummy of findingthe first occupation through the network loses significance, the dummy offinding the second job through family or friends is highly significant andhas a negative coefficient (of around -11 points). Thus, we can sustain theexplanation that the people who decide to obtain a job through this routehave particular characteristics that makes them less able to transfer theirskills.As in table 3.7, table 3.9 re-estimates the equation of the second oc-cupational gap controlling for city self-selection and limiting the sample toimmigrants who chose a city for non-economic reasons. I focus on the speci-fication that already conditions for the first occupation. I find the results tobe fairly similar to the ones of table 3.8. The estimations calculated with theself-selection correction controls do not present major changes from the ini-tial regressions. Also, the bootstrapped standard errors calculated (shownin bracket) are not markedly different from the robust ones. Moreover, thetests on the validity of self-selection correction can not reject the hypothesisof insignificance of the variables added. Thus, the self-selection correctionexercise seems to be not needed for the population analyzed.3.3.3 Occupational ImprovementAs a way to confirm the results of tables 3.8 and 3.9, and following thehypotheses formulated I estimate the occupational improvement from thefirst to the second occupation. That is, I analyze the quality of the secondoccupation minus the quality of the first occupation. Table 3.10 presents983.3. Findingsthe main results.77The first 5 columns of the table do not constraint the sample, whilethe rest restrict it to immigrants whose reasons for choosing a city weremainly drive by non-economic considerations. Also, columns 4,5,9 and 10include the first occupational gap as a conditional variable. Overall, fewvariables are significantly related to occupational improvement. Consistentwith our hypothesis the higher education dummies (?Graduate? and ?Bach-elor?) show a positive sign and significance; which indicates that immigrantswith high levels of education improved the most from the changes in occu-pations. A story of assimilation fits the results as high-skilled immigrantswould have benefitted the most from a better recognition of their abilitiesin their second occupations.It should be noted that the size of the coefficient of the education dum-mies in table 3.10 is very similar to the size of the decrease of the samedummies from tables 3.6 and 3.7 to tables 3.8 and 3.9. That is, the occupa-tional improvement due to education is reflected in new returns to educationfor the second occupation; make in it less of a penalty and not statisticallysignificant (though never a positive attribute).None of the other variables of interest seem to have statistical signifi-cance. Once again immigration categories appear not to have a differenteffect (when observable individual characteristics are included). Curiouslylanguage proficiency and its interaction with education are also not signifi-cant; the interaction has even a negative sign in some specification. A plau-sible explanation, and consistent with the motivation framework, could bethat language proficiency helps immigrants obtain recognition for or betteruse their abilities whereas the education dummies show the increase pay-offfor having those skills.3.3.4 Probability of switching occupationsIn table 3.11 I calculate the probability of a new immigrant having morethan 1 job in the first four years after arrival. The set up is the same asin tables before, a base case, the incorporation of an interaction betweenlanguage proficiency and education, and the inclusion of the endogenousregressor of obtaining the first job directly through family or friends. Inaddition columns 4 to 6 add the first occupational gap. All coefficientsreflect marginal effects.77The key results of table 3.10 do not vary when the self-selection correction procedureis employed; moreover, as in tables 7 and 9 the correction polynomial is not significant.993.3. FindingsIn the first three columns the only explanatory variables that seems ro-bustly significant are the higher education dummies (?Graduate? and ?Bach-elor?). These results are in line with the evidence shown in the previoustables concerning a process of assimilation. If individual skills are progres-sively recognized, immigrants with high levels of education would be morelikely to switch occupations as they would benefit the most from such achange. Oddly, language proficiency has a negative though not consistentlysignificant sign.Despite these encouraging results, it must be noted that the incorpo-ration of the first occupational gap reduces the economic and statisticalsignificance of the educational dummies. The size of the coefficients dropsby 5 to 8 points while their significance disappears completely. The first oc-cupational gap is highly significant and negative by 21 points; which pointsthat immigrants who experience a higher drop in their occupational statusat arrival have a higher probability of changing jobs.In general, this result indicates that the probability of switching is heav-ily related to the occupational status in the home country and the short-run(first) occupational match. This pattern is shown in graph 8, where theaverage occupational gap by occupations is plot. Immigrants with two oc-cupations have a larger drop in the occupational status at arrival but theirsecond occupation makes them obtain an occupational gap similar to theimmigrants with only one occupation. Likewise immigrants with three occu-pations, and the largest initial occupational drop of all, change occupationaluntil their occupational gap (the last one) is on average very close to theoccupational gap of immigrants with only one occupation and immigrantswith two occupation.3.3.5 Occupational Gap through timeTables 3.8 to 3.11 deal with immigrants who change occupations. How-ever, as figure 3.2 shows people with only one job might experience some(though minor) occupational improvement through time. Hence, I re-estimatethe second column of table 3.6 for all immigrants. I modify the data setsuch that I can estimate the occupational gap 0 to 6, 18 to 24 and 30 to 36months after a person started working.78 This set up properly accounts forthe different times immigrant start their first occupations and would allowevaluation of the assimilation process of immigrants after they started work-ing. As in table 3.10, I omit the presentation of the self-selection correction78A larger panel is not possible as some immigrants in the working sample start workingin their 18th month after arrival.1003.4. Conclusionsas the incorporation of the polynomial control function does not affect theestimates and is not significant in all columns.The results concur with the previous findings. The effects of educationand language proficiency (considering the interaction with education) arelargely significant in influencing the occupational gap of a recent immigrantin their first 36 months of employment. People with high levels of educationexperience a large and lasting occupational downturn. Although the signifi-cance or the sign on the education dummies is never reverted, the size of thecoefficients for ?graduate? and ?bachelor? reduces by 10 to 12 points fromthe beginning of employment to 18 to 24 months into the labor market. Thesize of the change is somewhat similar to the one found on the occupationalimprovement table. Hence, it can be argued that occupational change helpsimmigrants assimilate into a new return for their abilities.The interaction between language proficiency and education is positiveand significant; which confirm that knowledge of the city?s language aidseducated immigrants in the recognition of their abilities. The size of thecoefficient has a small changes of around 3 points through the years anddoes not alter its significance.Despite the changes in coefficients, the most remarkable finding is thatthe negative effect of these variables does not change through time. Thechanges in the education variables not quite large enough to make themstatistically different from each other; so we could affirm that the under-valuation of immigrant?s skills is persistent for the first four years. Overall,these results speak of a lengthier process of full integration into the Canadianlabor market.3.4 ConclusionsThe analysis reveals that immigrants with high levels of education (par-ticularly those with a Bachelor?s degree or higher) face a larger negativeoccupational gap in their first occupation. The effects of education in thesecond occupational gap are still negative though of a lesser magnitude andnot robustly significant. The result is consistent with the findings of chapter2. This could suggest that skills are not fully recognized even in the medium-run; though a proper recognition of the lower quality of foreign educationcould also explain the result. Indeed, chapter 2 shows that the location ofstudy matters for the immigrant-native wage gap. It would logically matterfor the quality of occupations.The results are confirmed by analyzing the occupational improvement1013.4. Conclusionsfrom the first to the second occupation. The regressions show that the dif-ferent educational levels significantly influence occupational improvement,with particular robustness for immigrants with a graduate level of educa-tion. The size of the education coefficient in the occupational improvementestimation is similar to the reduction of the size of the education coefficientsin the second occupational gap estimations. This change in size is furtherchecked when analyzing occupational gaps through time (0 to 36 monthsafter starting working). There is a clear reduction in the size of the edu-cational dummies from the beginning of employment to 18 to 24 monthsinto the labor market. However, the negative sign of the coefficients is neverreverted, so it can be argued that the undervaluation of immigrant skillspersists for the first years after arrival.The probability of changing occupations is mainly determined by thefirst occupational gap. Immigrants who experience a greater decline in theiroccupational status have a higher probability of changing jobs. No othervariable shows robust significance once the first occupational gap is added.However, without conditioning for the first occupational gap the educationdummies are significant. Immigrants with high levels of education are morelikely to switch occupations as they would benefit the most from such achange.A quick inspection of figures 3.2, 3.9 and the descriptive statistics showsthat the change in occupations is key for the improvement of occupationalquality. With about 66% of immigrants changing occupations and the im-provement between quality of the first occupation and the quality of theoccupation held by the end of the panel, one can broadly state that between90% to 94% of the improvement is due to changing occupations.Regarding the effect of language, I find that immigrants proficient onthe language of the city tend to have a lower mismatch. The interactionwith a dummy for educated immigrants (with a bachelor?s degree or higher)is positive and significant. The interaction remains economically and sta-tistically significant in the second occupation, pointing to the importanceof communication in high-skill occupations. The same pattern is observedwhen analyzing occupations through time, the interaction of language andeducation is positive and significant for all immigrants 0 to 6, 18 to 24 and30 to 36 months following entry into the labour market.1023.4. ConclusionsTable 3.1: Distribution of Immigrants by Category - Ages 24 to 55Selected SampleImmigration Category Female Male Total (Males Only)Family class - Spouses and Fiancs + Other 63% 37% 12%Family class - Parents and Grandparents 63% 37%Skilled Workers (PA) 23% 77% 76%Skilled Workers (S and D) 81% 19% 12%Business Immigrants (PA and S and D) 56% 44%Government Sponsored Refugees 47% 53%Other Immigrantsa 46% 54%Total 6,019 2359a Includes Provincial Nominees, Privately Sponsored Refugees, Other Refugees and Other Immigrants.1033.4.ConclusionsTable 3.2: Employment by Immigration Category - Ages 24 to 55Participation rate EmployedImmigration Category Wave 1 Wave 2 Wave 3 Wave 1 Wave 2 Wave 3Family class - Spouses and Fiancs + Other 75% 82% 84% 60% 78% 81%Family class - Parents and Grandparents 64% 70% 69% 48% 67% 66%Skilled Workers (PA) 93% 93% 95% 70% 89% 93%Skilled Workers (S and D) 65% 76% 83% 45% 70% 79%Business Immigrants (PA and S and D) 47% 69% 75% 31% 62% 73%Government Sponsored Refugees 32% 69% 75% . 50% 68%Other Immigrantsa 77% 81% 85% 65% 78% 83%Selected Sample (males only) 93% 93% 96% 72% 90% 94%a Includes Provincial Nominees, Privately Sponsored Refugees, Other Refugees and Other Immigrants.1043.4.ConclusionsTable 3.3: Education by Immigration Category - Ages 24 to 55Immigration Category Postgraduate Bachelor Some Univ. High School No High SchoolFamily class - Spouses and Fiancs + Other 14% 30% 26% 18% 11%Family class - Parents and Grandparents . . . 24% 60%Skilled Workers (PA) 33% 54% 11% 1% .Skilled Workers (S and D) 19% 45% 26% 8% 2%Business Immigrants (PA and S and D) 8% 32% 30% 24% 6%Government Sponsored Refugees 4% 13% 29% 32% 21%Other Immigrantsa . 20% 34% 25% 16%Selected Sample (males only) 30% 49% 15% 4% 2%a Includes Provincial Nominees, Privately Sponsored Refugees, Other Refugees and Other Immigrants.1053.4. ConclusionsTable 3.4: Summary StatisticsWorking SampleMeanDependent Variables (calculated values)Ln weekly wage - Home country Occupation 6.80Ln weekly wage - Intended Occupation 6.83Ln weekly wage - Wanted Occupation 6.78Ln weekly wage - First Occupation 6.44Ln weekly wage - Second Occupation 6.52Ln weekly wage - Third Occupation 6.56Independent VariablesAverage Age ( at wave 1) 35.03Graduate Level 0.30Bachelor?s Degree 0.52Some University education 0.13High School or Less 0.05English Proficiency 0.54French Proficiency 0.09Immigration CategoryFamily Class 0.10Skilled Workers (PA) 0.78Skilled Workers (S and D) 0.12How did you find this occupationFirst occupation - through family or friends 0.34Second occupation - through family or friends 0.19Third occupation - through family or friends 0.04Number of Occupations1 Occupation 0.342 Occupations 0.503 Occupations 0.16Missing ValuesMissing Values - Intended Occupation 0.17Missing Values - Wanted Occupation 0.23CMA of ResidenceVancouver 0.13Toronto 0.55Calgary 0.05Ottawa-Hull 0.04Montreal 0.13Others 0.10Why did you choose this City?Family or Friend 0.49Possible Jobs 0.24Others 0.27?Others? include weather, taxes, language spoken in the city, lifestyle and rent.1063.4.ConclusionsTable 3.5: Distribution of the Population by Metropolitan AreasDistribution of Population - Census 2001Total Foreign born Recent ImmigrantsToronto 15.7% 37.3% 43.1%Vancouver 6.6% 13.6% 17.6%Montreal 11.5% 11.4% 11.9%Ottawa 3.6% 3.4% 4.0%Calgary 3.2% 3.6% 3.8%Distribution of Population - Census 2006Total Foreign born Recent ImmigrantsToronto 16.2% 37.5% 40.4%Vancouver 6.7% 13.4% 13.7%Montreal 11.5% 12.0% 14.9%Ottawa 3.6% 3.3% 3.2%Calgary 3.4% 4.1% 5.2%Source: Statistics Canada1073.4.ConclusionsTable 3.6: First Occupational GapBase case Employed first 12 months1 2 3 4 5 6Graduate -0.24 -0.34 -0.35 -0.25 -0.35 -0.36(0.05)** (0.05)** (0.06)** (0.06)** (0.07)** (0.07)**Bachelor -0.26 -0.36 -0.36 -0.26 -0.35 -0.36(0.07)** (0.06)** (0.06)** (0.07)** (0.07)** (0.07)**Some University -0.19 -0.17 -0.17 -0.19 -0.17 -0.17(0.05)** (0.05)** (0.05)** (0.06)** (0.05)** (0.05)**Family Class 0.03 0.01 0.02 0.02 0.01 0.01(0.05) (0.04) (0.04) (0.04) (0.04) (0.04)Skilled Worker (SD) -0.07 -0.07 -0.07 -0.06 -0.06 -0.06(0.03)+ (0.03)+ (0.04)+ (0.03) (0.04) (0.04)Language Proficiency 0.10 -0.05 -0.06 0.10 -0.03 -0.04(0.02)** (0.04) (0.04) (0.02)** (0.04) (0.04)Educated*Language Proficiency 0.18 0.18 0.16 0.17(0.05)** (0.05)** (0.05)** (0.05)**Ethnic Job -0.05 -0.05(0.01)** (0.01)**Constant -1.72 -1.73 -1.41 -1.87 -1.87 -1.58(1.98) (2.01) (2.00) (1.92) (1.93) (1.93)Observations 1579 1579 1579 1484 1484 1484R-square 0.14 0.14 0.14 0.14 0.14 0.14Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Standard errors in parenthesis arerobust to heteroscedasticity and clustered by CMA. The omitted categories in the educational dummies is?HighSchool or less?. The omitted category in the immigration types is ?Skilled Worker - PA?. Additional controls inthe table: Age, Average occupational quality of the ethnicity, CMA/CA fixed effects and Country of origin fixedeffects.1083.4.ConclusionsTable 3.7: First Occupational Gap with Self-selection CorrectionSelf-selection correction +Non-economic reason Self-selection correction Non-economic reason1 2 3 4 5 6 7 8 9Graduate -0.23 -0.33 -0.34 -0.25 -0.35 -0.36 -0.23 -0.34 -0.34(0.06)** (0.06)** (0.06)** (0.05)** (0.05)** (0.05)** (0.05)** (0.06)** (0.06)**[0.07]** [0.06]** [0.07]** [0.06]** [0.07]** [0.07]**Bachelor -0.26 -0.36 -0.37 -0.27 -0.36 -0.37 -0.27 -0.37 -0.37(0.07)** (0.06)** (0.07)** (0.06)** (0.06)** (0.06)** (0.07)** (0.06)** (0.06)**[0.08]** [0.06]** [0.08]** [0.08]** [0.07]** [0.07]**Some University -0.18 -0.16 -0.16 -0.2 -0.17 -0.18 -0.19 -0.16 -0.17(0.05)** (0.05)* (0.05)* (0.05)** (0.04)** (0.05)** (0.05)** (0.05)** (0.05)**[0.06]** [0.04]** [0.06]** [0.06]** [0.05]** [0.07]*Family Class 0.06 0.05 0.05 0.03 0.01 0.02 0.06 0.05 0.05(0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04)[0.05] [0.05] [0.04] [0.05] [0.05] [0.05]Skilled Worker (SD) -0.06 -0.06 -0.05 -0.07 -0.07 -0.06 -0.06 -0.06 -0.05(0.04) (0.04) (0.04) (0.04)+ (0.04)+ (0.04) (0.04) (0.04) (0.04)[0.05] [0.04]+ [0.05] [0.05] [0.05] [0.05]Language Proficiency 0.10 -0.06 -0.06 0.10 -0.05 -0.06 0.1 -0.06 -0.07(0.02)** (0.05) (0.05) (0.02)** (0.04) (0.05) (0.02)** (0.05) (0.05)[0.04]** [0.04] [0.06] [0.04]** [0.05] [0.06]Educated*Lang. Proficiency 0.2 0.2 0.18 0.18 0.2 0.2(0.05)** (0.05)** (0.05)** (0.05)** (0.05)** (0.05)**[0.05]** [0.06]** [0.05]** [0.05]**1st Ethnic Job -0.04 -0.06 -0.04(0.01)* (0.01)** (0.01)*[0.02]** [0.02]*Constant 0.29 0.28 0.57 -1.17 -1.13 -0.77 0.65 0.67 1.01(2.79) (2.83) (2.86) (1.85) (1.83) (1.81) (2.66) (2.69) (2.70)[2.34] [2.01] [2.04] [3.03] [2.91] [2.96]Observations 1204 1204 1204 1579 1579 1579 1204 1204 1204R-squared 0.15 0.16 0.16 0.14 0.14 0.15 0.15 0.16 0.16Chi-square test (4) - Control factorsProb > chi2 0.76 0.66 0.69 0.82 0.87 0.79Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Standard errors in parenthesis are robust to heteroscedasticity and clusteredby CMA. Bootstrap standard errors are in brackets and clustered by CMA. The omitted category in the educational dummies is ?No High School?. Theomitted category in the immigration types is ?Skilled Worker - PA?. Additional controls in the table: Age, Ethnic Occupational Quality, CMA fixed effectsand Country of origin fixed effects.1093.4.ConclusionsTable 3.8: Second Occupational GapBase Case Employed First 12 Months1 2 3 4 5 6 7 8 9 10First Occupation 0.16 0.14 0.15 0.14(0.01)** (0.02)** (0.02)** (0.03)**Graduate -0.07 -0.13 -0.19 -0.08 -0.19 -0.07 -0.15 -0.2 -0.08 -0.2(0.10) (0.10) (0.10)+ (0.09) (0.09)+ (0.09) (0.09) (0.09)* (0.08) (0.08)*Bachelor -0.17 -0.23 -0.28 -0.17 -0.27 -0.17 -0.24 -0.29 -0.17 -0.29(0.11) (0.11)+ (0.11)* (0.10) (0.10)* (0.10) (0.10)* (0.10)* (0.09) (0.09)*Some University -0.11 -0.1 -0.12 -0.11 -0.12 -0.11 -0.1 -0.12 -0.12 -0.12(0.10) (0.10) (0.10) (0.09) (0.09) (0.09) (0.09) (0.09) (0.08) (0.08)Family Class 0.03 0.02 0.02 0.06 0.04 0.02 0.01 0.01 0.05 0.03(0.05) (0.05) (0.05) (0.05) (0.04) (0.05) (0.05) (0.04) (0.04) (0.04)Skilled Worker (SD) -0.05 -0.05 -0.04 -0.04 -0.04 -0.04 -0.04 -0.03 -0.03 -0.03(0.02)* (0.02)* (0.01)* (0.02)+ (0.02)* (0.03) (0.03) (0.04) (0.04) (0.03)Language Proficiency 0.03 -0.07 -0.1 0.01 -0.11 0.03 -0.08 -0.1 0.02 -0.11(0.02) (0.04) (0.05)+ (0.02) (0.05)* (0.02) (0.05) (0.05)+ (0.02) (0.05)*Educated*Lang. Proficiency 0.12 0.14 0.13 0.14 0.16 0.15(0.03)** (0.03)** (0.04)** (0.04)** (0.04)** (0.04)**1st Ethnic Job -0.03 -0.03 -0.02 -0.02(0.02)+ (0.02) (0.02) (0.02)2nd Ethnic Job -0.11 -0.1 -0.11 -0.1(0.01)** (0.01)** (0.01)** (0.01)**Constant -3.73 -3.73 -3.51 -4.42 -4.18 -3.84 -3.83 -3.66 -4.4 -4.21(3.75) (3.73) (3.70) (3.84) (3.78) (3.66) (3.64) (3.61) (3.78) (3.71)Sample 1043 1043 1043 1043 1043 996 996 996 996 996R-squared 0.12 0.13 0.14 0.14 0.15 0.12 0.12 0.14 0.13 0.15Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Standard errors in parenthesis are robust to heteroscedasticityand clustered by CMA. The omitted categories in the educational dummies is?No High School?.The omitted category in the immigration types is?Skilled Worker - PA?. Additional controls in the table: Age, Ethnic Occupational Quality, CMA fixed effects and Country of origin fixed effects.1103.4.ConclusionsTable 3.9: Second Occupational Gap with Self-selection CorrectionSelf-selection correction +Non-economic reason Self-selection correction Non-economic reason1 2 3 4 5 6 7 8 91st Occupation 0.16 0.16 0.15 0.15 0.15 0.14 0.16 0.16 0.14(0.03)** (0.03)** (0.03)** (0.02)** (0.02)** (0.02)** (0.02)** (0.02)** (0.03)**[0.02]** [0.02]** [0.03]** [0.03]** [0.03]** [0.03]**Graduate -0.07 -0.13 -0.17 -0.09 -0.15 -0.20 -0.06 -0.12 -0.17(0.11) (0.11) (0.11) (0.09) (0.09) (0.09)+ (0.10) (0.11) (0.11)[0.08] [0.07]* [0.09]* [0.11] [0.11] [0.11]Bachelor -0.18 -0.23 -0.27 -0.18 -0.23 -0.28 -0.17 -0.23 -0.26(0.10) (0.11)+ (0.11)* (0.10) (0.10)* (0.10)* (0.11) (0.11)+ (0.11)*[0.10]+ [0.09]* [0.11]** [0.11] [0.10]* [0.11]*Some University -0.12 -0.10 -0.12 -0.12 -0.10 -0.13 -0.11 -0.10 -0.11(0.10) (0.09) (0.09) (0.10) (0.10) (0.10) (0.10) (0.10) (0.10)[0.10] [0.09] [0.09] [0.11] [0.09] [0.10]Family Class 0.05 0.04 0.04 0.06 0.05 0.04 0.05 0.04 0.03(0.04) (0.04) (0.04) (0.05) (0.05) (0.05) (0.04) (0.04) (0.04)[0.05] [0.05] [0.05] [0.06] [0.05] [0.05]Skilled Worker (SD) -0.02 -0.02 -0.02 -0.04 -0.04 -0.03 -0.02 -0.02 -0.02(0.02) (0.02) (0.02) (0.02)* (0.01)* (0.01)* (0.02) (0.02) (0.02)[0.02] [0.02] [0.02] [0.02] [0.02] [0.03]Language Proficiency 0.00 -0.09 -0.11 0.01 -0.08 -0.11 0.00 -0.09 -0.11(0.03) (0.04)+ (0.05)* (0.02) (0.05) (0.05)+ (0.03) (0.04)+ (0.04)*[0.04] [0.05]+ [0.07] [0.05] [0.07] [0.07]Educated*Lang. Proficiency 0.11 0.12 0.12 0.13 0.11 0.12(0.03)** (0.03)** (0.04)** (0.04)** (0.03)** (0.03)**[0.04]** [0.05]** [0.05]* [0.05]**1st Ethnic Job -0.01 -0.03 -0.02(0.02) (0.01)+ (0.02)[0.02] [0.02]2nd Ethnic Job -0.09 -0.10 -0.09(0.01)** (0.01)** (0.01)**(0.02)** (0.02)**Constant 0.15 0.17 0.33 -3.34 -3.30 -3.08 0.54 0.58 0.74(3.47) (3.45) (3.41) (3.69) (3.65) (3.65) (3.32) (3.29) (3.30)Observations 807 807 807 1043 1043 1043 807 807 807R-squared 0.16 0.16 0.17 0.14 0.14 0.15 0.16 0.17 0.18Chi-square test (4)Prob > chi2 0.92 0.81 0.93 0.96 0.98 0.99Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively.Standard errors in parenthesis are robust to heteroscedasticityand clustered by CMA. Bootstrap standard errors are in brackets and clustered by CMA. The omitted category in the educational dummiesis ?No High School?. The omitted category in the immigration types is ?Skilled Worker - PA?. Additional controls in the table: Age,Ethnic Occupational Quality, CMA fixed effects and Country of origin fixed effects.1113.4.ConclusionsTable 3.10: Occupational Improvement - First to Second OccupationBase Case Non-economic Reasons1 2 3 4 5 6Graduate 0.15 0.16 0.16 0.15 0.17 0.17(0.05)* (0.04)** (0.04)** (0.05)** (0.04)** (0.04)**Bachelor 0.08 0.09 0.09 0.09 0.11 0.11(0.04)+ (0.03)* (0.03)* (0.04)* (0.03)** (0.03)**Some University 0.04 0.04 0.04 0.05 0.04 0.04(0.04) (0.04) (0.04) (0.04) (0.04) (0.05)Family Class 0.00 0.01 0.01 -0.01 -0.01 -0.01(0.02) (0.02) (0.02) (0.02) (0.02) (0.02)Skilled Worker (SD) -0.02 -0.02 -0.02 -0.01 -0.02 -0.01(0.05) (0.05) (0.05) (0.06) (0.06) (0.06)Language Proficiency -0.03 -0.01 -0.01 -0.05 -0.02 -0.02(0.02) (0.05) (0.05) (0.02)+ (0.05) (0.05)Educated*Lang. Proficiency -0.03 -0.03 -0.03 -0.03(0.06) (0.07) (0.05) (0.05)1st Ethnic Job -0.01 0.00(0.02) (0.02)Constant -0.72 -0.72 -0.68 1.91 1.9 1.92(3.10) (3.11) (3.09) (2.68) (2.69) (2.70)Sample 1043 1043 1043 807 807 807R-squared 0.04 0.04 0.04 0.06 0.06 0.06Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Standard errors inparenthesis are robust to heteroscedasticity and clustered by CMA. The omitted categories in theeducational dummies is?No High School ?. The omitted category in the immigration types is ?SkilledWorker - PA?. Additional controls in the table: Age, Ethnic Occupational Quality, CMA fixed effectsand Country of origin fixed effects.1123.4.ConclusionsTable 3.11: Probability of Having More than One Occupation (Marginal Effects)1 2 3 4 5 6 7 8 9First Occup. -0.28 -0.28 -0.28(0.02)** (0.02)** (0.02)**First Occup. Gap -0.21 -0.21 -0.21(0.02)** (0.02)** (0.02)**Graduate 0.16 0.15 0.15 0.19 0.17 0.18 0.11 0.07 0.08(0.06)* (0.05)** (0.05)** (0.07)** (0.06)** (0.06)** (0.07) (0.07) (0.06)Bachelor 0.15 0.14 0.15 0.18 0.16 0.16 0.10 0.06 0.07(0.06)* (0.05)** (0.05)** (0.07)* (0.06)** (0.06)** (0.07) (0.06) (0.06)Some University 0.15 0.15 0.16 0.16 0.16 0.16 0.11 0.12 0.12(0.06)** (0.06)** (0.05)** (0.06)** (0.06)** (0.05)* (0.07)+ (0.07)+ (0.06)*Family Class 0.05 0.05 0.04 0.01 0.01 0.00 0.06 0.05 0.05(0.05) (0.04) (0.04) (0.05) (0.05) (0.05) (0.04) (0.04) (0.04)Skilled Worker (SD) 0.08 0.08 0.08 0.05 0.05 0.05 0.06 0.06 0.06(0.05) (0.05) (0.05) (0.05) (0.05) (0.05) (0.06) (0.06) (0.06)Language Proficiency -0.04 -0.06 -0.05 -0.02 -0.05 -0.04 -0.02 -0.08 -0.07(0.02)* (0.04)+ (0.04) (0.02) (0.03) (0.03) (0.02) (0.03)* (0.04)*Educated*Lang. Proficiency 0.03 0.02 0.04 0.04 0.07 0.07(0.04) (0.04) (0.03) (0.03) (0.04)+ (0.04)+1st Ethnic Job 0.04 0.02 0.03(0.02)* (0.02) (0.02)Sample 1579 1579 1579 1579 1579 1579 1579 1579 1579Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively. Standard errors in parenthesis are robust to het-eroscedasticity and clustered by CMA.The omitted categories in the educational dummies is ?No High School?.The omitted category inthe immigration types is ?Skilled Worker - PA?. Additional controls in the table: Age, Ethnic Occupational Quality, CMA fixed effectsand Country of origin fixed effects.1133.4.ConclusionsTable 3.12: Occupational Gap Through Time - Months After First Occupations0 to 6 months 18 to 24 months 30 to 36 months 0 to 6 months 18 to 24 months 30 to 36 monthsGraduate -0.36 -0.24 -0.23 -0.37 -0.25 -0.24(0.05)** (0.06)** (0.04)** (0.05)** (0.06)** (0.04)**Bachelor -0.38 -0.28 -0.29 -0.38 -0.29 -0.3(0.06)** (0.07)** (0.06)** (0.06)** (0.08)** (0.06)**Some University -0.19 -0.14 -0.16 -0.19 -0.15 -0.17(0.06)** (0.05)* (0.05)** (0.06)** (0.05)* (0.05)**Family Class -0.01 0.02 0.01 -0 0.03 0.01(0.04) (0.04) (0.04) (0.04) (0.04) (0.04)Skilled Worker (SD) -0.07 -0.05 -0.04 -0.06 -0.05 -0.04(0.04)+ (0.02)+ (0.02) (0.04) (0.02)+ (0.02)Language Proficiency -0.05 -0.05 -0.04 -0.06 -0.06 -0.05(0.04) (0.03) (0.04) (0.04) (0.03) (0.04)Educated*Lang. Proficiency 0.17 0.14 0.14 0.18 0.15 0.14(0.05)** (0.03)** (0.03)** (0.05)** (0.03)** (0.03)**1st Ethnic Job -0.05 -0.05 -0.04(0.01)** (0.01)** (0.01)**Constant -2.85 -1.92 -1.47 -2.56 -1.63 -1.23(1.85) (2.18) (2.35) (1.87) (2.19) (2.39)Sample 1483 1483 1483 1483 1483 1483R-squared 0.14 0.11 0.12 0.14 0.11 0.12Note: *, ** and *** denote significance at 10%, 5% and 1% levels respectively.Standard errors in parenthesis are robust to heteroscedasticityand clustered by CMA. The omitted categories in the educational dummies is ?No High School?. The omitted category in the immigrationtypes is ?Skilled Worker - PA?. Additional controls in the table: Age, Ethnic Occupational Quality, CMA fixed effects and Country of originfixed effects.1143.4.ConclusionsFigure 3.1: Kernel Density - Ln weekly wages by Occupation for native Canadians1153.4.ConclusionsFigure 3.2: Kernel Density - Ln Weekly Wages Across Waves by Number of Occupations1163.4.ConclusionsFigure 3.3: Kernel Density - Occupational Distribution of the Intended and Wanted Occupations1173.4.ConclusionsFigure 3.4: Occupations held before and after arriving in Canada1183.4.ConclusionsFigure 3.5: Occupational Gap by Education and Number of Occupations1193.4.ConclusionsFigure 3.6: Occupational Gap by Language Proficiency and Number of Occupations1203.4.ConclusionsFigure 3.7: Occupational Gap by Immigration Category and Number of Occupations1213.4.ConclusionsFigure 3.8: Occupational Gap by Network Job and Number of Occupations1223.4.ConclusionsFigure 3.9: Occupational Gap by number of Occupations1233.4.ConclusionsFigure 3.10: Kernel Density - Ln Weekly Wages Across Occupations124Chapter 4Immigrants EnglishProficiency ImprovementFour Years After ArrivalThis chapter analyzes the changes in the English proficiency of recentimmigrants to Canada. It continues using the LSIC to document newcomersintegration in their first four years. Proficiency in the host-country languageis an important assimilation dimension for immigrants. It defines the extentof their interactions with the new environment and determines the optimaluse of their skills (e.g., access to medical services and employment opportu-nities).Probit and ordered probit estimations are used to identify individualcharacteristics related to language proficiency improvement or decline. Speak-ing abilities are chosen as an overall indicator of language proficiency and thesample is separated according to immigrants? initial level: basic, intermedi-ate or advanced. The main focus lies on the relation between immigrationcategories and language proficiency improvement.The chapter is organized as follows. Section 4.1 presents the measuresof language proficiency available in the LSIC. Section 4.2 discusses the re-quirements family immigrants and skilled workers have to meet in order tobe accepted as permanent residents. Section 4.3 describes the characteris-tics of the sample used. Section 4.4 introduces econometric considerationsfor the estimations proposed. Section 4.5 presents the results. Section 4.6concludes.4.1 Language proficiency meassures in the LSICThe LSIC was conducted by Statistics Canada and Citizenship and Im-migration Canada (CIC) on immigrants arriving between October 1st, 2000and September 30th, 2001. The newcomers were interviewed three times: 6months, 24 months and 48 months after arrival. Only immigrants who ap-1254.1. Language proficiency meassures in the LSICplied from outside the country and were 15 years old or older at the time ofarrival were followed; even then only those who responded to the first wavewere traced for the second, and only those who responded to the secondwave were traced for the third.79 The numbers of observations per wave are12040, 9322 and 7716, respectively.80Survey interviews were conducted face-to-face or by phone in one of 15languages (including English and French) chosen by Statistics Canada toinclude about 93% of newcomers.81 Phone interviews were conducted whena face-to-face interview was not possible (LSIC microdata user guide, pg.37)and interviews lasted between 65 and 90 minutes.The LSIC provides information on a variety of immigrant characteris-tics, including demographics (sex, age, country of origin, etc.), education,employment, social networks and language proficiency.82 The section onlanguage proficiency is of particular interest. It includes 6 self-assessed ques-tions addressing the ability to speak, write and read in the official languages(3 for English and 3 for French), each having five possible answers: fairlywell, well, intermediate, bad and not at all. The question on speaking pro-ficiency is the primary dependant variables of this chapter. Nevertheless,the LSIC has 10 additional self-assessed questions (5 for English and 5 forFrench) addressing how easy it is for immigrants to communicate duringthe cours of day-to-day activities; such as telling their address, indicatingtheir occupation before arriving, understanding a message over the phone,speaking with a doctor and re-scheduling an appointment. Each questionhas four possible answers: easy, with some help, with a lot of help and cannot do it.The study focuses on immigrants between 25 and 55 years old living inEnglish Canada, whose mother tongue is not English and were admittedunder one of three immigration categories: Skilled Immigrant - PrincipalApplicant, Skilled Immigrant - Spouse and Dependant, and Family Immi-79In an study of the health evolution of recent immigrants to Canada Fuller-Thomsonet al. (2011) state that about half of the immigrants who did not complete the second orthird LSIC survey were reached but did not or could not complete the interview.80Concerns about attrition problems are minimized as previous research has not foundan important selection process between the waves. For instance, comparing sampled vslost observations Fuller-Thomson et al. (2011) find no mark difference in age, gender,immigration class or place of birth.81The 15 languages are: English, French, Punjabi, Spanish, Arabic, Tagalog, Tamil,Cantonese, Mandarin, Farsi, Russian, Urdu, Korean, Serbo-Croatian and Gujarati82The LSIC also contains information on foreign credentials, health, income and per-ception of settlement.1264.1. Language proficiency meassures in the LSICgrants (Spouse, fiance, sons, daughters, parents, grandparents and others).83The age range chosen reflects the intent to have groups with relatively sim-ilar number of observations across ages. Analysis of the full LSIC sampleshows that family immigrants have a disproportional presence after the ageof 55, while skilled workers (Principal Applicants) represent a relativelysmall proportion of those below 25 (see table 4.1). From 25 to 55 the se-lected immigrant categories have a more even distribution (23% of the wholedatabase is younger than 25 or older than 55). The categories selected rep-resent the majority of immigrants to Canada (according to CIC 81% of allpermanent residents accepted in 2001 belong to these categories).My focus on English Canada is based on the higher number of obser-vations available and the low levels of bilingualism.84 Most immigrants toFrench Canada reside in Montreal, a city with a large bilingual population.85One could claim that learning French in Montreal differs significantly fromlearning it elsewhere in Quebec or learning English in provinces other thanQuebec. Only a small number of observations would remain if Montrealwas excluded from French Canada. In the end, all immigrants to Quebecwere eliminated from the sample.86 The final number of observations totaled3346.87The other two important immigrant groups represented in the LSIC,though with a smaller number of observations, are Business Immigrants andRefugees. The distinctive characteristics of these groups make it difficult tocompare them to the selected sample. Table 4.2 shows that the average busi-ness immigrant accrued savings for more than $ 93,000 (Cdn.) six monthsafter arrival, five times more than skilled workers (Principal Applicants) andaround twelve times more than family immigrants. It is fair to say that a83This group can also incorporate brothers or sisters, nephews or nieces, granddaughtersor grandsons as long as they are orphaned, under 18 years of age and not married or in acommon-law relationship.84In addition, immigrants to Quebec follow the Quebec-immigration rules that empha-size initial French proficiency and abilities to assimilate into the ?Quebec culture?.85Statistics Canada reports that in 2001 the share of the population in Montreal whospoke only French at home was 62.4% while the share who spoke English (either alone orin combination with other languages) was 24.3%. In Toronto and Vancouver the share ofpopulation who spoke only English was 62.5% and 65.3% respectively. However, the sharewho spoke French (either alone or in combination with other languages) in each of thesecities was less than 1.5%.86The few immigrants who move out or move into Quebec are also eliminated from thesample.87Additional minor restrictions regarding non-missing values for key variables wereincluded. For instance, people claiming to be working but who didn?t have an answer forthe ethnicity of their co-workers were eliminated.1274.1. Language proficiency meassures in the LSICsubstantial level of wealth would change the influence key variables (such aseducation and age) might have on the language assimilation process. Incor-porating savings as a control variable may not be sufficient to disentanglethe non-linearities as no other immigrant category resides in their financialneighbourhood.The same reasoning applies to government refugees. The amount ofsavings they have six months after immigrating is minimal (C$ 240). About70% of them received social assistance in those six months (see table 4.3);at least ten times more than skilled workers and family immigrants. Theamount of assistance received is also greater for refugees ($ 7,731 versus lessthan $ 4,000 for skilled workers and family immigrants recipients). Aydemir(2009) explains that most refugees are eligible for federal income supportand training in their first year and for provincial support later on providedthey search for work or take classes.88In addition to the LSIC, information from the 20% sample of the 2001Canadian Census is used to construct a measure of the importance of theethnic community in each CMA/CA. Following the literature (see for in-stance, Munshi (2003), Goel and Lang (2007) and Patel and Vella (2007))I calculate the number of immigrants from a particular country of originas a fraction of the total population in a given CMA/CA. The exceptionare Spanish speaking countries (namely Latin America and Spain), which Icluster into a single region given their common language and culture. Thismetric tries to capture the correlation between the presence of an ethniccommunity and the improvement in language proficiency.89 Although theconstructed measure might suffered from double causality (as language pro-ficient immigrants would choose not to leave in a ethnic enclave), I use itmainly as a control variable.88There is another immigration category in the LSIC, Provincial Nominees. Howeverthe Provincial Nominee program was fairly new at the time of the LSIC first wave. Thenumber of immigrants in this category is negligible.89A similar measured is constructed by Dustmann and Fabbri (2003) in their studyof language fluency of non-white immigrants in the UK and its effect on earnings andemployment.1284.2. Requirements for Becoming a Landed Immigrant4.2 Requirements for Becoming a LandedImmigrant: Family Immigrant vs SkilledWorkerThe foreign born face different requirements for obtaining landed immi-grant status according to their category of immigration. Some are requiredto invest in the country, others to demonstrate desirable professional skills,while others might be accepted if facing persecution (on account of race, re-ligion, nationality or political opinion). These requirements follow Canada?simmigration policy components: a social component, a humanitarian com-ponent and an economic component (see Young (1998)). This study focuseson the social and economic components as the humanitarian componentpertains to the acceptance of refugees.The social component relates to the reunification of Canadian citizensand permanent residents with their closest relatives. Under the immigrationrules in place when LSIC immigrants applied, sponsors of family immigrantsmust be at least 19 years old and be Canadians or permanent residents.90They must be able and commit themselves to provide financial support fortheir relatives for a ten-year period.91 This financial condition was notrequired when sponsoring a spouse or children younger than 19 years of age,single and without children of their own.Family immigrants must have a close relationship with the sponsor. Forinstance, they can be a spouse (in a heterosexual marriage) or a fiance, par-ents or grandparents, children 19 years old or younger, children older than19 but dependent on their parents or children under 19 intended for adop-tion. They can also be brothers, sisters, nephews, nieces or grandchildrenprovided they are single, under 19 years of age and orphaned.92 Family im-migrants have to meet the criteria relating to health and good character, butthey don?t have to meet any other criteria, such as education level, languagefluency or employment skills.The economic component of the immigration policy is designed to fosterthe development of Canada by selecting immigrants based on their capacity90All of the immigrants interviewed by the LSIC were accepted under the 1976 Immi-gration Act. In 2002 a new set of rules for accepting immigrants was implemented underthe Immigrant and Refugee Protection Act.91Sponsors should meet Statistics Canada Low Income Cut-Off (LICO) one year priorto the application and sign, together with their relatives, an agreement concerning theirfinancial obligations92If The sponsor is alone in Canada and has none of the relatives mentioned, he or shecan sponsored any relative.1294.3. Descriptive statisticsto invest or create jobs, or on their occupational skills 93 The first two referto the investor and entrepreneur categories (grouped under the title businessimmigrants), while the third refers to the skilled worker category. Again, Ishall focus only on the last category.A nine-factor points system was used to evaluate skilled workers. Thesystem assigns points by age, education, vocational preparation, experi-ence, occupational demand, arranged employment, knowledge of English orFrench, demography, personal suitability and a demographic control fac-tor.94 Applicants required 70 points to be considered but the presence of arelative in the country reduces the threshold by 5 points. Spouse and de-pendants accompanying applicants were admitted without being evaluated.Once we condition for observable factors (such as age, gender, education,initial level of language fluency among others), improvements in English flu-ency upon arrival indicate either different environments and incentives forlanguage assimilation (related possibly to immigration categories) or dif-ferent unobservable abilities to improve language fluency. Neither familyimmigrants nor companions of skilled workers were required to pass passan evaluation to enter Canada (besides health and good character consid-erations). They are both close family dependants and, given our sampleage restriction, of a working age.95 The comparison between family immi-grants and skilled worker applicants is less straightforward. Skilled workerapplicants were individually evaluated. Control variables may capture someof the intrinsic differences across immigration categories, still their differ-ent language performance will reflect unobservable abilities and dissimilarassimilation environments.4.3 Descriptive statisticsTable 4.4 presents the descriptive statistics of the selected sample. Av-erage, standard deviations and fractions are calculated using the surveyweights of the LSIC, as required by the Research Data Center informa-tion release policy.96 The table shows groupings by immigration category,93 Green and Green (1999) describe the economic goals of Canada?s immigration policyfrom 1870 to 1997. Ferrer et al. (2012) asses the evolution of Canada?s immigration policy;particularly since the introduction of the points system (late 1960s) up to late 2000s.94The demographic control factor was not based on the applicant?s characteristics buton the level of immigrants accepted every year.95The age restriction addresses the main distinction when thinking about family im-migrants; namely that they are considerably older than the rest.96Maximum and minimum values are prohibited to be released.1304.3. Descriptive statisticsCMA/CA of residence and country of origin.Regarding immigration categories, we see that about 82% of immigrantsin the sample are skilled workers (counting principal applicants as well asspouses and dependants). According to CIC statistics the ratio of skilledworkers to family immigrants is close to 2 to 1. This is owing to the imposedage restrictions. The family immigrant category has a significant presencein less than 25 and more than 55 age range for newcomers. The focus onmigrants between 25 and 55 distorts the initial distribution.Immigrants also cluster geographically. Vancouver and Toronto receivedabout three quarters of the sampled immigrants. The distribution is sim-ilar to the one observed by the 2001 Canadian Census. According to it,between 1991 and 2001 of all foreign-born arrivals living in English Canada,around 50% resided in Toronto and 20% in Vancouver. Moreover, few coun-tries dominate the immigration inflow. In the sample China, India and thePhilippines provide more than half of all immigrants. The 2001 census showsthat these three countries were the main source of immigrants for the 1991to 2001 period.Is important to note that between 31% and 50% of the interviews wereconducted by phone (31% in wave 1, 50% in wave 2 and 40% in wave 3).97It could be reasoned that the people giving the interview in English (alwaysmore than 60% of the population), might have found difficult to misclassifytheir (English) speaking proficiency.Indeed, Dustmann and Van Soest (2001, 2002) argue that self-reportedlanguage proficiency is likely to suffer from misclassification errors. It issuspected that undervaluation occurs at the top and overvaluation at thebottom.98 I address these concerns by grouping the responses into threecategories: basic, intermediate and advanced English proficiency. The basiccategory has the two lowest levels of proficiency and the advanced categorythe two highest. I do this for the speaking, writing and reading proficiencyquestions.99 This approach will not overcome language proficiency misclas-sifications but it could minimize them.In fact, only about one fifth of immigrants grouped in the basic speakingcategory decided to have the first interview in English (see table 4.5). Forimmigrants in the advanced category the opposite is true; 79% chose to havethe first interview in English. Immigrants in the intermediate category were97The distribution is similar to the one observed for the whole survey: 32%, around50% and 37% - LSIC microdata user guide, pg.35.98It should be noted that Dustmann and Van Soest focus is on presenting a methodologyto overcome inference problems when using language ability as a explanatory variable.99A similar procedure is done by Dustmann (1994, pg. 136).1314.3. Descriptive statisticssplit 40% and 60% between English and not English. These number supportthe validity of the broad grouping of self-reported proficiency. Measurementerror in the self-reported questions is a present problem in this research,but the consistency of the share of interviews done in English by categoriesacross all waves endorse the grouping made. Immigrants with low Englishproficiency would not be able to give an interview in English (particularly ifconducted over the phone), while immigrants with high proficiency would.The evolution of language proficiency over the years presents a commonpattern (see table 4.6). The majority of immigrants start with an advancedlevel of English proficiency. With the exception of speaking proficiency, morethan 70% claim to have advanced English proficiency (65% in speaking,78% in reading and 72% in writing). The proportion of those with basic orintermediate proficiency declines between the first and the second interview.The changes between the second and third waves are small, suggesting thatimmigrants decide to adapt to the new dominant language soon after arrival.In the last interview between 74% and 82% of immigrants claim to haveadvanced English proficiency.The alternative five self-assessed measures show the same trend and haveeven higher proportions of those claiming English proficiency by the thirdwave. The responses to some of the additional questions would imply a veryhigh level of English proficiency. For instance, providing home addresses anddescribing past occupations is easily done by 92% and 81% of immigrants.Such high numbers likely don?t reflect the fluency of newcomers but thestraightforwardness of the questions.When transition matrices are constructed, immigrants improving from abasic category spread almost evenly across the intermediate and advancedcategories in all three measures (speaking, writing and reading). Moreover,immigrants exhibit minor transitions from the higher proficiency categoriesto the lower ones (see table 4.7). At most 10% of those in the advancedcategory in the first interview report having an intermediate or basic profi-ciency in the second interview, regardless of the fluency metric. Still, it isnecessary to consider that the reported reduction in English fluency mightbe driven by adjustments in the the perceptions of fluency. A person mightconsider herself proficient in the first months after arrival but fully realizeotherwise after a year.English proficiency improvement (and decline) from the first to the sec-ond interview is very similar to the cumulative change from the first to thethird interview, which was conducted around four years after arrival. Itappears that most of the language proficiency improvement occurs in the1324.4. Econometric Considerationsfirst years.100 The alternative proficiency measures show a similar trend(see table 4.8).4.4 Econometric ConsiderationsThe lack of adequate data on host-country language proficiency has pre-cluded the development of a standard way to approach the subject. Themost common and intuitive methodology involves the use of an ordered pro-bit model. Dustmann (1994, pg. 139) presents a detailed application of theordered probit model to language proficiency. The maximum likelihood esti-mation of a non-linear model with 3 possible levels (basic, intermediate andadvanced) seems suitable to estimate language proficiency. Yet, the use ofcross-section data casts doubts on the interpretation of the results. Insteadof determinants of host-country language proficiency one may be describingthe characteristics of those currently fluent. More important for this study,the variables related to life long host-country language proficiency may notbe the same as those related to improvement upon arrival (nor have thesame importance).The LSIC allows for a measurement of English proficiency improvement.Linking the change in English proficiency from the first to the third waveto the initial characteristics of immigrants would identify the key variablesrelated to improvement. Still, there are features of the proposed approachthat need to be considered. A linear set up helps present the possible en-dogeneity problems. Assume that the latent variable, language proficiency(yit), is fully measurable and can be represented as a linear combination ofvariables. Xi incorporates constant individual characteristics (such as ed-ucation, age, gender, etc), immigration categories (e.g., family immigrants,skilled workers - S and F, etc.) and ethnic network characteristics (e.g., theinitial number of co-workers of the same ethnicity). Consider the individualfactors ?i and ?it reflecting unobserved constant and time-variant languageabilities, respectively. We can then represent language proficiency as:yit = ?t +Xi?t + ?i + ?it + ?it (4.1)Where ?it is a normal distributed error. The fact that the coefficients as-sociated with individual characteristics change over time (?t) indicates thata particular variable may have a different effect over the years. The coef-ficients of some variables may be close to zero at the time of arrival but100The short span of the panel prevents the evaluation of other moments of Englishfluency improvements later in the life of immigrants.1334.4. Econometric Considerationsbecome important years later. Now, if we take the difference between aninitial and a final level of language proficiency, assuming a panel of threeperiods, we would drop the time-invariant individual language ability ?i andget:yi3 ? yi1 = ?3 ? ?1 +Xi(?3 ? ?1) + (?i3 ? ?i1) + (?i3 ? ?i1)?yi = ??+Xi(??) + ?iy?i = ??+Xi?? + ?iwhere ?i is equal to ?i3 ? ?i1 + ?ig3 ? ?ig1. If we assume that ?it and ?itare independent across individuals then the covariance of the errors acrossobservations (cov(?i, ?j)) should be negligible. Nevertheless, the covariancebetween the new error term and the independent variables Xi might notbe zero. The time-variant individual factors (?it) might linger on the dif-ference equation and relate to some of the regressors. For example, theethnic network (co-workers and friends of the same ethnicity and the size ofthe ethnic community) could have been deliberatively chosen based on thepresent and future individual language abilities (?i, ?i1 and ?i3). In thatcase, by construction, ?i would be related to the ethnic network part of Xi.The probable absence of strict exogeneity needs to be kept in mind since itcould render the results descriptive in nature.The problem becomes more complex once we take into considerationthat language proficiency is not easily measurable and that the initial levelof proficiency may influence the cost of improvement. That is, improvementsat the initial stages might be easier than improvements at intermediate oradvanced levels.In considering these possible problems, I use two types of maximum like-lihood estimations. The first is a simple probit model (shown together witha linear probability model), where the dependent variable is an indicator oflanguage improvement, one if the immigrant?s language proficiency at wave 3is higher than the language proficiency at wave 1 and zero otherwise. Clearlythis estimation is not applicable to immigrants with high English proficiencyin their first wave as the indicator would always be zero. Because improve-ments from a basic English level could be more likely than improvementsfrom an intermediate level, I separate immigrants accordingly to their initiallanguage proficiency. I estimate one probit for immigrants with basic En-glish proficiency and another for those with intermediate proficiency. I alsoestimate a probit model for losing language proficiency, where the dependentvariable is one if the language proficiency at wave 3 is lower than the lan-1344.4. Econometric Considerationsguage proficiency at wave 1 (zero otherwise). Again, I exclude immigrantswith an initial basic English level and run separate estimations for thosewith advanced and intermediate English proficiency. The second model isan ordered probit where I continue separating the samples according to theinitial language proficiency or including the initial level as a dummy.There are 4 categories of independent variables: demographics, type ofimmigration, family variables and choice variables. Demographics includeage at the time of the first interview, gender and years of education. I in-clude dummies for two immigration categories; i.e., skilled worker - spouseand dependant and family immigrants; making skilled worker - principalapplicant the base category. Family variables incorporate the number ofmembers in a household, a dummy if the immigrant is married, a dummyif the declared address in the first wave is near to an English as a SecondLanguage Assessment Center, and savings in Canadian dollars declared inthe first interview.101 The choice variables have one set of dummies describ-ing ethnic friendships, another set representing the proportion of co-workersfrom the same ethnicity and a calculation of the portion of all the peoplehaving same ethnicity in a given metropolitan area (CMA/CA). Specifically,I include 3 dummies representing developed friendships upon arrival, one fornew friendships with few or no friends of the same ethnicity, one for friend-ships with half of new friends of the same ethnicity and one for most orall of those new friends of the same ethnicity (the base case are migrantswithout new friends). I also have 3 dummies representing ethnic co-workersat wave 1: one for few or no co-workers of the same ethnicity, one for half ofco-workers of the same ethnicity and one for most or all ethnic co-workers(the base case is not employed immigrants). In addition, I include a set ofregions of origin and CMA/province dummies.102 Appendix C.1 provides adictionary of variables.I choose the speaking level as a measure of overall English proficiency.Though achieving a very basic speaking proficiency might be undemand-ing, intermediate and advanced levels are difficult to attain. Recall thata comparatively low proportion of immigrants is able to speak English at101In order to start classes in the English as a Second Language program the immigrant?sproficiency has to be evaluated in an Assessment Center.102The regions included are: Central America, South America, The West ( U.S., U.K.,West Europe and Oceania), East Europe, South Europe, Africa, West and Central Asia,Eastern Asia, Southeast Asia and Southern Asia. The CMA/province list includes:Toronto, Vancouver, Calgary, Edmonton and Ottawa as CMAs; and groups the provincesby: rest of British Columbia, rest of Ontario; Manitoba, Saskatchewan and rest of Albertain one group; and Newfoundland and Labrador, Prince Edward Island, Nova Scotia andNew Brunswick in another.1354.5. Findingsan advanced level. Only 65% claim to have advanced speaking proficiencyupon arrival (compared to 72% and 78% for writing and reading abilities,respectively). Also, once the basic structures of English and a basic vo-cabulary are mastered reading is substantially simplified. Higher standardsof writing can be achieved by people with intermediate proficiency whenenough time is given. Verbal communication is telling. It requires on thespot fluency. Moreover, 60% or more of the interviews in the first and lastwaves were face-to-face, giving less opportunity to misclassify speaking flu-ency. The three measures have a positive and somewhat high correlationthough (higher than 0.70).4.5 Findings4.5.1 Levels of English ProficiencyTable 4.9 presents an ordered probit of the level of speaking proficiencyfor the third wave of the surveys. The base specification has demographicvariables, immigrant categories and family variables. The five specificationsinclude information on the similarity of the ethnicity of co-workers or friends.The results show the characteristics of individuals with a high level of com-munication skills. Some demographic variables are quite significant. Youngand educated immigrants tend to have a high level of English proficiency(gender does not seem to play a role). Immigrant categories are also quiterelevant. Family immigrants and spouses or dependants of skilled work-ers have a lower level of English speaking proficiency, family immigrantsbeing at the greatest disadvantage.103 In constrat, family variables, suchas marriage and household size, do not show robust significance throughthe specifications. Lastly, having few ethnic friends or co-workers workers(in the first six months after arrival) is associated with a a higher level ofspeaking proficiency. This is likely a self-selection driven result. Those withhigher communication abilities would be able to have friends or work withothers outside their own ethnic group. Just as well the sign and signifi-cance of the ethnic concentration variable indicate that those who live inmetropolitan areas with a larger ethnic enclave have a lower speaking pro-ficiency by the third wave. I replicate the analysis for reading and writingproficiency with similar results: age and education have the same sign andsignificance. Immigration categories also show the same correlation with103Immigrant categories are incorporated as dummies hence the magnitudes of thechange between categories is the same and the otherwise not comparable order probitcoefficients can be compared.1364.5. Findingslanguage proficiency, and the coefficient of the family immigrant dummywas more negative than the spouses or dependants of skill workers in allcases (see tables4.10 and 4.11). The main difference lies in the relation be-tween ethnic acquaintances (friend and co-workers) and English proficiency,which is not robust for the new measures of communication skills.4.5.2 Language Proficiency Improvement: Basic andIntermediate LevelsSeparating the sample by the initial level of language proficiency allowsfor the identification of possible non-linearities in the explanatory variables.I find that for people with an initial basic level of language proficiency onlyage and education appear robustly significant. Consistent with human cap-ital theory, younger and more educated immigrants are more likely to im-prove their English proficiency. The additional time that young age confersprovides an incentive to improve upon the initial proficiency level. Likewise,educated immigrants have more to gain from improving their communica-tion skills. It is curious that no other variable shows statistical robustness.Immigrant category dummies have a negative sign but weak significance,while the share of ethnic acquaintances (in and outside of work) show nosignificance.Immigrants arriving with an intermediate level of English proficiencyshow again the importance of education and age for language improvement.For them, however, immigration categories also play a role. Immigrantsarriving as spouses or dependants of skill workers or as family immigrantsare less likely to improve their language proficiency than skilled workers.The result could be driven by the need skilled workers have to enter theCanadian labour market and the returns to English proficiency. Of thetwo immigrant categories, family immigrants seem to be the less likely toimprove, suggesting that they may live in an environment that does notrequire much communication in English.To get a better picture of the language assimilation process and gaindegrees of freedom table 4.14 aggregates the samples of immigrants withbasic and intermediate skills. The first three columns run linear probabil-ity models, the second three run probit models and the last three add adummy for basic language proficiency at arrival. The results are consistentwith the previous findings. Age and years of education are crucial factorsin determining English proficiency improvement. Regarding immigrant cat-egories, only family immigrant appear statistically relevant with a negativecoefficient. It should be noted that for all these variables the size of the1374.5. Findingscoefficients is similar to that found in previous tables.The last three columns though show that immigrants with basic Englishon arrival are more likely to improve their language proficiency than im-migrants arriving with an intermediate proficiency level. The result is con-sistent with an increasing marginal cost of improving language proficiency.Immigrants at lower proficiency levels would have less difficulty improvingthan immigrants at higher levels.4.5.3 Decline in Language Proficiency: Intermediate andAdvanced LevelsThe probability of immigrants losing their English proficiency is alsorelated to demographic variables and immigration categories regardless ofthe initial level of proficiency. For immigrants arriving with an intermediateor advanced level of English proficiency education and age have the expectedeffect (see tables 4.15 and 4.16). The initial proficiency of older immigrantsis more likely to decline by the third wave, while the opposite is true foreducated immigrants. The results mirror those found in tables 4.12 and4.13Regarding immigrant categories one sees that newcomers arriving as fam-ily are more likely to lose their English proficiency as time passes. I arguethat the high significance of the coefficients indicate that family immigrantsface a different environment than skilled workers or their dependants. Anenvironment that does not require English communication on a regular basiswould predispose migrants to lose their proficiency. It is interesting to notethough that the coefficient of the family immigrant dummy is larger (morethan double) and more significant (0.1% versus around 2%) for immigrantsarriving with an intermediate level of English. That is, arriving as a fam-ily immigrant has a larger effect on people with an intermediate level thanthose with an advanced English proficiency. No other variables show robustsignificance.I merge both samples (intermediate and advanced) as before and analyzeif the findings hold. Table 4.17 has six columns. The first three run linearregressions while the second three run probit models. In all the columns adummy for the initial level of English is added. Overall, the findings hold.The sign, size and significance of age, education and the family immigrantdummy are consistent with tables 4.15 and 4.16. Also consistent with theresults of table 4.14, the initial level of English proficiency affects the prob-ability of losing it. Immigrants with a high level of proficiency have a higherprobability of decreasing their level than immigrants arriving with an in-1384.6. Conclusionstermediate level. Non-linearities in the change of language proficiency arepresent as much for improving it as for losing it.4.5.4 Ordered Probit Conditioning on Initial EnglishProficiencyI replicate the estimations of the level of English proficiency on the thirdwave (see table 4.9) but include the initial level of proficiency as a control.Table 4.18 presents the results. I run three specifications for those with abasic level first, three for those with a basic or intermediate level, addinga dummy for initial basic proficiency, and finally three for all the samplecontrolling for initial basic and intermediate levels.Once again age and education are statistically relevant to determiningEnglish proficiency. Given the lack of statistical robustness of other vari-ables, I would argue that human capital factors are relevant to immigrantEnglish proficiency improvement, and as such they provide a non-economicdimension of societal assimilation.The family immigrant dummy shows a negative coefficient throughoutthe table but its significance is not quite robust for people with an initialbasic proficiency when controlling for ethnic concentration (third column).The result is similar to the findings in table 4.12. When those with interme-diate or advanced initial proficiency are added, there is no decline in familyimmigrant significance. Though the sample size more than quadruples whenthese two groups are included, the evidence suggests that immigration cat-egories don?t play a significant role in English improvement for immigrantswith basic proficiency.4.6 ConclusionsThis chapter examines the changes in immigrants? English proficiencyin their first four years in Canada. The data show that immigrants do notchange their English proficiency to a considerable extent and that most ofthe change happens in the first two years after arrival. The proportion ofimmigrants at an advanced level of English speaking proficiency rises from65% six months after arrival to 74% two years after arrival. There is noincrease from the second to the fourth year. Still, I analyze the probabilityof improving or losing English speaking proficiency.Two demographic variables show robustly consistent and significant co-efficients in all the specifications: age and years of education. Regardlessof their initial level, younger and more educated immigrants are more likely1394.6. Conclusionsto improve their English proficiency (if they start at a basic or intermediatelevel) and less likely to lose it (if they start at an intermediate or advancedlevel). These two variables highlight human capital considerations. Thebenefits of having higher English proficiency are enjoyed to greater extentby educated migrants and for a longer time by younger migrants.Regarding immigration categories, those arriving as family immigrantswith an intermediate level are less likely to improve their proficiency (thanskilled workers). This result might be driven by the environment this typeof immigrant faces; which may not require continuous communication inEnglish. However, an explanation based on unobservable time-variant in-dividual abilities for family immigrants can not be disregarded. The effectof the family immigrants? dummy is null for immigrants arriving with ba-sic English proficiency. With a beginner?s level, arriving as a particulartype of immigrant might not significantly influence the required changes toEnglish proficiency. Family immigrants are also more likely to lose theirEnglish proficiency. The coefficients found are positive and significant inall the specifications and robust for the initial level of English proficiency(intermediate or advanced).1404.6.ConclusionsTable 4.1: Age Distribution by Immigration CategoryLess than 25 25 to 34 35 to 44 45 to 55 More than 55Family Immigrants 45.34 19.02 8.60 32.57 91.92Skilled Workers (PA) 2.23 45.66 48.24 30.05 .Skilled Workers (S and D) 28.73 28.21 29.96 13.95 .Business Immigrants 10.05 1.60 6.34 13.94 .Govt. Refugees 6.49 2.84 2.94 3.85 .Others 7.16 2.68 3.96 5.64 .100 100 100 100 100Less than 25 25 to 34 35 to 44 45 to 55 More than 55Dist. by Age Group 16 29 36 11 71414.6. ConclusionsTable 4.2: Average Savings 6 Months After Arrival by Immigration GroupAverage SavingsFamily Immigrants $ 7,852Skilled Workers (PA) $ 18,248Skilled Workers (S and D) $ 20,005Business Immigrants $ 93,365Govt. Refugees $ 240Others $ 2,703Table 4.3: Government Social Assistance by Immigration GroupShare Who Receives Average AmountSocial Assistance for RecipientsFamily Immigrants 2% 3,297Skilled Workers (PA) 7% 2,936Skilled Workers (S and D) 7% 3,750Business Immigrants . .Govt. Refugees 70% 7,731Others 16% 6,0031424.6. ConclusionsTable 4.4: Descriptive StatisticsAverage Standard DeviationAge 35.4 7.1Gender 0.5Total Years of Education 15.5 3.3Education in English (last degree) 0.4Skilled Worker Principal Applicant 0.49Skilled Worker Spouse and Dependants 0.31Family Immigrant Spouse and Fiance 0.20ESL Centre near 0.12Number of Members 0 to 4 0.24 0.50Number of Members 5 to 14 0.66 0.85Number of Members higher than 18 2.35 1.06Share of Ethnic Concentration (*100) 1.83 1.54Interview by phone Wave 1 a 0.31Interview by phone Wave 2 0.50Interview by phone Wave 3 0.40CMA/CA of ResidenceToronto 0.57Vancouver 0.17Calgary 0.06Edmonton 0.04Ottawa 0.03Rest (48 CMA/CAs) 0.14Country of OriginChina 0.25India 0.18Phillipines 0.10Pakistan 0.06South Korea 0.05Rest (90 countries) 0.36Number of Observations 3466a Interviews are done by phone or face-to-face1434.6.ConclusionsTable 4.5: Share of Interviews in EnglishInterview in wave 1 Interview in wave 2 Interview in wave 3English Not English English Not English English Not EnglishBasic 18% 82%Wave 1 Intermediate 40% 60%Advanced 79% 21%Total 62% 38%Basic 19% 81%Wave 2 Intermediate 43% 57%Advanced 80% 20%Total 69% 32%Basic 19% 81%Wave 3 Intermediate 47% 53%Advanced 82% 18%Total 70% 30%1444.6. ConclusionsTable 4.6: Evolution of English ProficiencyWave 1 Wave 2 Wave 3(6 months) (24 months) (48 months)Main MeasurementsBasic 15 9 9Speaking Abilities Intermediate 20 17 17Advanced 65 74 74Basic 8 7 7Reading Abilities Intermediate 13 10 11Advanced 78 84 82Basic 11 8 9Writing Abilities Intermediate 17 14 15Advanced 72 77 76Alternative MeasurementsCan not / With a Lot of Help 5 3 3Giving your Address With Some Help 4 2 1Easily 92 95 96Can not / With a Lot of Help 8 5 5Explaining your Past Occup. With Some Help 12 9 6Easily 81 86 89Can not / With a Lot of Help 10 7 6Taking a Message With Some Help 20 16 12Easily 69 78 82Can not / With a Lot of Help 17 14 10Explaining Symptoms to With Some Help 23 20 17a Doctor Easily 60 66 73Can not / With a Lot of Help 10 7 6Setting up a Meeting With Some Help 14 10 7Easily 76 83 871454.6.ConclusionsTable 4.7: Changes in English ProficiencySpeaking AbilitiesWave 2 Wave 3Basic Intermediate Advanced Basic Intermediate AdvancedBasic 47 28 25 100 Basic 46 26 28 100Wave 1 Intermediate 8 33 59 100 Wave 1 Intermediate 8 34 58 100Advanced 1 9 90 100 Advanced 1 10 89 100Writing AbilitiesWave 2 Wave 3Basic Intermediate Advanced Basic Intermediate AdvancedBasic 51 24 25 100 Basic 51 24 26 100Wave 1 Intermediate 9 32 58 100 Wave 1 Intermediate 11 30 59 100Advanced 1 9 90 100 Advanced 2 10 88 100Reading AbilitiesWave 2 Wave 3Basic Intermediate Advanced Basic Intermediate AdvancedBasic 58 16 26 100 Basic 53 20 26 100Wave 1 Intermediate 10 27 63 100 Wave 1 Intermediate 9 27 64 100Advanced 1 6 93 100 Advanced 2 8 90 1001464.6.ConclusionsTable 4.8: Communication Abilities ChangesUnderstanding a Phone MessageWave 2 Wave 3Can not With help Easily Can not With help EasilyCan not 46 31 23 100 Can not 41 29 30 100Wave 1 With help 7 34 59 100 Wave 1 With help 6 26 68 100Easily 1 8 91 100 Easily 1 5 94 100Speaking with a DoctorWave 2 Wave 3Can not With help Easily Can not With help EasilyCan not 53 26 21 100 Can not 40 34 27 100Wave 1 With help 18 38 44 100 Wave 1 With help 11 33 56 100Easily 2 11 87 100 Easily 2 6 92 100Re-arraging an AppointmentWave 2 Wave 3Can not With help Easily Can not With help EasilyCan not 47 23 30 100 Can not 45 19 36 100Wave 1 With help 10 26 64 100 Wave 1 With help 9 17 74 100Easily 1 6 93 100 Easily 1 4 95 100Explaining a former OccupationWave 2 Wave 3Can not With help Easily Can not With help EasilyCan not 51 17 32 100 Can not 46 16 38 100Wave 1 With help 8 21 71 100 Wave 1 With help 5 17 78 100Easily 1 6 93 100 Easily 1 4 95 1001474.6. ConclusionsTable 4.9: Ordered Probit - Level of Speaking at Wave 3 (Coefficients)(1) (2) (3) (4) (5)Age -0.038 -0.038 -0.038 -0.038 -0.039(0.004)??? (0.004)??? (0.004)??? (0.004)??? (0.003)???Gender 0.047 0.057 0.052 0.065 0.068(0.102) (0.099) (0.093) (0.093) (0.094)Yrs of Educ 0.160 0.157 0.157 0.152 0.148(0.015)??? (0.015)??? (0.015)??? (0.015)??? (0.012)???Immigration Categories:Skilled Worker (S and D) -0.358 -0.337 -0.337 -0.322 -0.312(0.076)??? (0.072)??? (0.082)??? (0.078)??? (0.076)???Family Inmg. -0.852 -0.833 -0.790 -0.777 -0.770(0.075)??? (0.077)??? (0.081)??? (0.083)??? (0.083)???ESL Centre 0.094 0.092 0.094 0.092 0.097(0.130) (0.128) (0.137) (0.135) (0.138)Married -0.050 -0.046 -0.049 -0.048 -0.040(0.066) (0.067) (0.068) (0.069) (0.074)Household size -0.021 -0.015 -0.014 -0.009 -0.016(0.007)??? (0.008)?? (0.009)? (0.009) (0.011)?Savings wave 1 (in 10,000) 0.014 0.013 0.013 0.012 0.011(0.008)? (0.008)? (0.008)? (0.008)? (0.008)?Few ethn Friends 0.408 0.377 0.349(0.076)??? (0.074)??? (0.081)???Half Ethn Friends 0.401 0.392 0.364(0.065)??? (0.061)??? (0.062)???All Ethn. Friends -0.010 0.005 0.005(0.086) (0.069) (0.071)Some Ethn. Coworkers 0.215 0.167 0.197(0.073)??? (0.079)?? (0.085)???Half Ethn. Coworkers 0.113 0.0881 0.124(0.041)??? (0.043)?? (0.042)???All Ethn. Coworkers -0.314 -0.303 -0.243(0.052)??? (0.045)??? (0.048)???Ethnic Concent. -0.124(0.050)???Obs 3466 3466 3466 3466 3466R 0.2782 0.2858 0.2856 0.2919 0.2977Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote signif-icance at 10%, 5% and 1% levels respectively. All regressions include 9 source region-of-origin dum-mies and 8 area-of-residence dummies. The omitted case in the immigration categories is ?SkilledWorker - Principal applicant?.1484.6. ConclusionsTable 4.10: Ordered Probit - Level of Reading at Wave 3 (Coefficients)(1) (2) (3) (4) (5)Age -0.038 -0.038 -0.039 -0.039 -0.040(0.002)??? (0.002)??? (0.002)??? (0.002)??? (0.002)???Gender 0.083 0.093 0.102 0.112 0.114(0.065) (0.064) (0.066) (0.068)? (0.069)?Yrs of Educ 0.178 0.1754 0.1748 0.1732 0.1697(0.017)??? (0.017)??? (0.016)??? (0.017)??? (0.016)???Immigration Categories:Skilled Worker (S and D) -0.261 -0.244 -0.257 -0.244 -0.240(0.056)??? (0.056)??? (0.056)??? (0.057)??? (0.058)???Family Inmg. -0.806 -0.806 -0.782 -0.787 -0.777(0.050)??? (0.056)??? (0.060)??? (0.067)??? (0.067)???ESL Centre 0.023 0.027 0.018 0.022 0.019(0.101) (0.010) (0.101) (0.100) (0.113)Married -0.057 -0.048 -0.060 -0.052 -0.042(0.099) (0.099) (0.097) (0.097) (0.089)Householld size -0.007 -0.003 -0.003 0.000 -0.007(0.010) (0.010) (0.011) (0.011) (0.001)Savings wave 1 (in 10,000) 0.008 0.007 0.007 0.006 0.005(0.003)?? (0.003)?? (0.003)?? (0.003)?? (0.003)??Few ethn Friends 0.238 0.233 0.196(0.095)?? (0.095)?? (0.105)?Half Ethn Friends 0.118 0.126 0.083(0.089) (0.089) (0.095)All Ethn. Friends -0.101 -0.087 -0.091(0.106) (0.101) (0.108)Some Ethn. Coworkers 0.080 0.043 0.077(0.089) (0.098) (0.098)Half Ethn. Coworkers -0.041 -0.052 -0.015(0.048) (0.056) (0.065)All Ethn. Coworkers -0.174 -0.157 -0.088(0.041)??? (0.046)??? (0.059)Ethnic Concent. -0.136(0.040)???Obs 3466 3466 3466 3466 3466R - square 0.3069 0.3106 0.3084 0.3117 0.3188Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote signif-icance at 10%, 5% and 1% levels respectively. All regressions include 9 source region-of-origin dum-mies and 8 area-of-residence dummies. The omitted case in the immigration categories is ?SkilledWorker - Principal applicant?.1494.6. ConclusionsTable 4.11: Ordered Probit - Level of Writing at Wave 3 (Coefficients)(1) (2) (3) (4) (5)Age -0.036 -0.036 -0.036 -0.036 -0.037(0.005)??? (0.005)??? (0.005)??? (0.005)??? (0.004)???Gender 0.135 0.143 0.153 0.162 0.165(0.078)? (0.078)? (0.075)?? (0.077)?? (0.078)??Yrs of Educ 0.176 0.174 0.172 0.170 0.167(0.016)??? (0.017)??? (0.016)??? (0.016)??? (0.016)???Immigration Categories:Skilled Worker (S and D) -0.274 -0.261 -0.266 -0.257 -0.257(0.063)??? (0.063)??? (0.063)??? (0.063)??? (0.060)???Family Inmg. -0.741 -0.739 -0.702 -0.704 -0.700(0.088)??? (0.088)??? (0.100)??? (0.101)??? (0.100)???ESL Centre 0.001 0.004 -0.004 -0.002 0.003(0.098) (0.095) (0.101) (0.099) (0.103)Married -0.045 -0.037 -0.049 -0.043 -0.030(0.073) (0.076) (0.072) (0.075) (0.073)Householld size -0.003 0.000 0.002 0.004 -0.002(0.013) (0.014) (0.013) (0.014) (0.014)Savings wave 1 (in 10,000) 0.005 0.005 0.005 0.004 0.003(0.005) (0.004) (0.004) (0.004) (0.004)Few ethn Friends 0.171 0.158 0.129(0.134) (0.128) (0.140)Half Ethn Friends 0.189 0.191 0.162(0.089)?? (0.088)?? (0.097)?All Ethn. Friends -0.099 -0.083 -0.081(0.093) (0.084) (0.089)Some Ethn. Coworkers 0.114 0.085 0.111(0.071) (0.080) (0.079)Half Ethn. Coworkers -0.001 -0.011 0.022(0.042) (0.042) (0.048)All Ethn. Coworkers -0.269 -0.257 -0.195(0.038)??? (0.042)??? (0.046)???Ethnic Concent. -0.124(0.032)???Obs 3466 3466 3466 3466 3466R - square 0.2676 0.2710 0.2714 0.2742 0.2801Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote sig-nificance at 10%, 5% and 1% levels respectively. All regressions include 9 source region-of-origindummies and 8 area-of-residence dummies. The omitted case in the immigration categories is?Skilled Worker - Principal applicant?.1504.6.ConclusionsTable 4.12: Improvement for People Beginning at a Basic Level (wave 1 to wave 3)Linear Model Probit Model (Marginal Effects)(1) (2) (3) (4) (5) (6) (7) (8)Age -0.011 -0.011 -0.011 -0.012 -0.016 -0.015 -0.017 -0.017(0.004)??? (0.004)??? (0.004)??? (0.003)??? (0.004)??? (0.004)??? (0.004)??? (0.004)???Gender 0.015 0.016 0.031 0.028 0.012 0.013 0.039 0.036(0.053) (0.053) (0.052) (0.050) (0.080) (0.080) (0.079) (0.075)Yrs of Educ 0.028 0.027 0.025 0.022 0.051 0.051 0.049 0.044(0.006)??? (0.007)??? (0.007)??? (0.007)??? (0.015)??? (0.016)??? (0.015)??? (0.015)???Skill Worker (S and D) -0.058 -0.057 -0.058 -0.037 -0.077 -0.076 -0.076 -0.042(0.101) (0.099) (0.098) (0.094) (0.148) (0.142) (0.143) (0.148)Family Inmg. -0.210 -0.212 -0.198 -0.158 -0.212 -0.216 -0.189 -0.134(0.104)?? (0.104)?? (0.098)?? (0.092)? (0.145) (0.140) (0.135) (0.145)ESL Centre 0.022 0.023 0.018 -0.004 0.009 0.012 0.002 -0.015(0.031) (0.034) (0.032) (0.032) (0.051) (0.053) (0.051) (0.049)Married 0.033 0.031 0.051 0.042 0.017 0.022 0.049 0.043(0.062) (0.055) (0.054) (0.052) (0.087) (0.075) (0.075) (0.069)Householld size -0.009 -0.009 -0.009 -0.011 -0.022 -0.021 -0.021 -0.026(0.010) (0.010) (0.010) (0.008) (0.012)? (0.011)? (0.010)?? (0.008)???Savings wave 1 (in 10,000) 0.005 0.005 0.005 0.004 0.008 0.007 0.006 0.004(0.002)??? (0.002)??? (0.002)?? (0.002)? (0.004)? (0.004)? (0.003)? (0.003)?Few Ethn Friends 0.074 0.088 0.068 0.084 0.079 0.046(0.110) (0.107) (0.102) (0.188) (0.180) (0.180)Half Ethn Friends -0.033 -0.035 -0.063 -0.032 -0.041 -0.084(0.063) (0.062) (0.052) (0.081) (0.073) (0.060)All Ethn. Friends 0.000 0.006 0.013 -0.016 -0.011 -0.002(0.055) (0.052) (0.053) (0.084) (0.076) (0.085)Some Ethn. Coworkers -0.097 -0.061 -0.112 -0.082(0.056)? (0.053) (0.068) (0.067)Half Ethn. Coworkers 0.017 0.040 0.007 0.048(0.043) (0.043) (0.074) (0.076)All Ethn. Coworkers -0.099 -0.066 -0.156 -0.105(0.071) (0.069) (0.098) (0.098)Ethnic Concent. -0.054 -0.090(0.009)??? (0.006)???Obs 591 591 591 591 589 589 589 589R - square 0.4076 0.4089 0.4169 0.4346 0.3623 0.3633 0.3721 0.3923Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significance at 10%, 5% and 1%levels respectively. All regressions include 9 source region-of-origin dummies and 8 area-of-residence dummies. The omitted casein the immigration categories is ?Skilled Worker - Principal applicant?.1514.6.ConclusionsTable 4.13: Improvement for People Beginning at an Intermediate Level (wave 1 to wave 3)Linear Model Probit Model (Marginal Effects)(1) (2) (3) (4) (5) (6) (7) (8)Age -0.011 -0.011 -0.011 -0.012 -0.013 -0.013 -0.014 -0.015(0.004)??? (0.043) (0.004)??? (0.004)??? (0.005)??? (0.005)??? (0.005)??? (0.004)???Gender 0.044 0.045 0.044 0.049 0.052 0.054 0.052 0.058(0.003) (0.033) (0.030) (0.026)? (0.032) (0.036) (0.033) (0.033)Yrs of Educ 0.026 0.025 0.025 0.022 0.030 0.029 0.028 0.026(0.005)??? (0.005)??? (0.005)??? (0.005)??? (0.006)??? (0.006)??? (0.007)??? (0.006)???Skill Worker (S and D) -0.096 -0.095 -0.091 -0.083 -0.102 -0.100 -0.096 -0.087(0.038)?? (0.038)?? (0.037)?? (0.038)?? (0.040)?? (0.041)?? (0.039)?? (0.039)??Family Immig -0.179 -0.177 -0.171 -0.179 -0.212 -0.210 -0.204 -0.219(0.044)??? (0.004)??? (0.004)??? (0.039)??? (0.055)??? (0.050)??? (0.049)??? (0.045)???ESL Centre 0.062 0.061 0.059 0.044 0.075 0.074 0.071 0.055(0.064) (0.062) (0.062) (0.073) (0.070) (0.068) (0.067) (0.083)Married -0.002 -0.005 -0.001 -0.016 -0.010 -0.016 -0.013 -0.033(0.108) (0.110) (0.111) (0.099) (0.0132) (0.133) (0.133) (0.119)Household size 0.015 0.015 0.015 0.009 0.016 0.016 0.016 0.008(0.005)??? (0.005)??? (0.005)??? (0.005)? (0.006)??? (0.006)??? (0.006)??? (0.006)???Savings wave1 (in 10,000) 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001(0.004) (0.004) (0.004) (0.003) (0.005) (0.005) (0.005) (0.004)Few Ethn. Friends 0.045 0.039 0.028 0.070 0.062 0.053(0.066) (0.066) (0.079) (0.069) (0.069) (0.083)Half Ethn. Friends 0.051 0.050 0.028 0.086 0.085 0.061(0.071) (0.069) (0.087) (0.073) (0.007) (0.094)All Ethn. Friends 0.014 0.015 0.013 0.025 0.026 0.023(0.080) (0.078) (0.079) (0.085) (0.084) (0.086)Some Ethn. Coworkers 0.030 0.051 0.035 0.073(0.041) (0.004) (0.053) (0.051)Half Ethn. Coworkers 0.024 0.037 0.027 0.046(0.021) (0.032) (0.021) (0.034)All Ethn. Coworkers -0.012 0.024 -0.010 0.028(0.035) (0.032) (0.039) (0.036)Ethnic Concent. -0.074 -0.089(0.037)?? (0.045)??Obs 694 694 694 694 684 684 684 684R - square 0.1630 0.1638 0.1645 0.1917 0.1252 0.1264 0.1270 0.1501Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significance at 10%, 5% and 1%levels respectively. All regressions include 9 source region-of-origin dummies and 8 area-of-residence dummies. The omitted casein the immigration categories is ?Skilled Worker - Principal applicant?.1524.6.ConclusionsTable 4.14: Improvement for People Beginning at Intermediate or Basic Level (wave 1 to wave 3)Linear Model Probit Model(Marginal Effects) Probit Model (Marginal Effects)(1) (2) (3) (4) (5) (6) (7) (8) (9)Age -0.12 -0.01 -0.01 -0.02 -0.02 -0.02 -0.02 -0.02 -0.02(0.004)??? (0.004)??? (0.004)??? (0.005)??? (0.005)??? (0.004)??? (0.005)??? (0.004)??? (0.004)???Gender 0.03 0.04 0.04 0.04 0.05 0.05 0.04 0.05 0.04(0.013)? (0.014)??? (0.016)?? (0.017)?? (0.018)??? (0.019)??? (0.017)?? (0.018)??? (0.019)??Yrs of Educ 0.03 0.03 0.02 0.04 0.04 0.04 0.04 0.4 0.03(0.005)??? (0.006)??? (0.005)??? (0.001)??? (0.001)??? (0.009)??? (0.001)??? (0.001)??? (0.009)???Skill Worker (S and D) -0.08 -0.07 -0.07 -0.08 -0.08 -0.07 -0.08 -0.7 -0.07(0.059) (0.056) (0.054) (0.071) (0.067) (0.067) (0.071) (0.067) (0.067)Family Inmg. -0.20 -0.19 -0.18 -0.21 -0.20 -0.19 -0.21 -0.2 -0.19(0.056)??? (0.050)??? (0.052)??? (0.075)??? (0.066)??? (0.072)??? (0.075)??? (0.066) (0.072)???ESL Centre 0.03 0.03 0.01 0.03 0.02 0.01 0.03 0.02 0.01(0.025) (0.024) (0.033) (0.032) (0.031) (0.039) (0.032) (0.030) (0.039)Married 0.01 0.02 0.00 0.00 0.00 -0.02 0.0001 0.004 -0.02(0.059) (0.064) (0.057) (0.082) (0.088) (0.082) (0.082) (0.088) (0.082)Household size 0.003 0.004 0.000 0.000 0.001 -0.01 0.0002 0.001 -0.01(0.005) (0.005) (0.003) (0.006) (0.005) (0.004) (0.006) (0.005) (0.004)Savings wave1 (in 10,000) 0.003 0.003 0.002 0.004 0.003 0.002 0.004 0.003 0.002(0.003) (0.003) (0.003) (0.004) (0.004) (0.003) (0.004) (0.004) (0.003)Few Ethn Friends 0.04 0.03 0.05 0.03 0.05 0.03(0.005) (0.0053) (0.065) (0.074) (0.064) (0.074)Half Ethn Friends 0.02 0.01 0.04 0.00 0.03 0.004(0.044) (0.052) (0.057) (0.074) (0.057) (0.074)All Ethn. Friends 0.01 0.01 0.003 0.01 0.002 0.006(0.055) (0.053) (0.073) (0.075) (0.072) (0.075)Some Ethn. Coworkers 0.01 0.03 0.01 0.06 0.01 0.06(0.038) (0.037) (0.046) (0.043) (0.047) (0.043)Half Ethn. Coworkers 0.03 0.04 0.03 0.05 0.02 0.05(0.019) (0.026)? (0.024) (0.003)? (0.023) (0.003)?All Ethn. Coworkers -0.06 -0.02 -0.08 -0.03 -0.08 -0.03(0.036) (0.038) (0.044) (0.048) (0.044)? (0.048)Ethnic Concent. -0.06 -0.09 -0.09(0.021)??? (0.024)??? (0.024)???Basic level 0.23 0.23 0.24(0.024)??? (0.027)??? (0.023)???Obs 1284 1284 1284 1284 1284 1284 1284 1284 1284R-square 0.2695 0.2732 0.2954 0.2287 0.2325 0.2564 0.2287 0.2325 0.2564Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significance at 10%, 5% and 1% levels respectively.All regressions include 9 source region-of-origin dummies and 8 area-of-residence dummies. The omitted case in the immigration categories is?Skilled Worker - Principal applicant?.1534.6.ConclusionsTable 4.15: Language Proficiency Decrease for People Beginning at an Intermediate Level (wave 1 to wave 3)Linear Probability Model Probit Model: Marginal Effects(1) (2) (3) (4) (5) (6) (7) (8)Age 0.007 0.01 0.01 0.01 0.004 0.004 0.004 0.004(0.001)??? (0.001)??? (0.002)??? (0.002)??? (0.001)??? (0.001)??? (0.001)??? (0.001)???Gender -0.010 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01 -0.01(0.006) (0.007) (0.007)? (0.007)? (0.005) (0.006) (0.007) (0.007)Yrs of Educ -0.017 -0.02 -0.02 -0.01 -0.01 -0.01 -0.01 -0.01(0.004)??? (0.004)??? (0.004)??? (0.004)??? (0.003)??? (0.003)??? (0.003)??? (0.002)???Skill Worker (S and D) 0.016 0.01 0.01 0.01 0.02 0.02 0.02 0.02(0.024) (0.021) (0.021) (0.021) (0.024) (0.020) (0.018) (0.018)Family Inmg. 0.107 0.11 0.11 0.11 0.11 0.10 0.10 0.10(0.027)??? (0.027)??? (0.029)??? (0.029)??? (0.035)??? (0.034)??? (0.033)??? (0.031)???ESL Centre -0.025 -0.02 -0.02 -0.02 -0.03 -0.03 -0.03 -0.03(0.022) (0.021) (0.023) (0.025) (0.020) (0.019) (0.018) (0.021)Married 0.024 0.03 0.03 0.03 0.03 0.03 0.03 0.03(0.027) (0.027) (0.026) (0.029) (0.015)?? (0.013)?? (0.013)?? (0.013)??Household size -0.008 -0.01 -0.01 -0.01 -0.01 -0.004 -0.005 0.00(0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005) (0.005)Savings wave1 (in 10,000) -0.001 -0.001 -0.001 -0.001 -0.002 -0.002 -0.002 -0.001(0.001) (0.0005)?? (0.0004)??? (0.001) (0.002) (0.001)? (0.001)? (0.001)Few Ethn Friends -0.06 -0.06 -0.05 -0.04 -0.04 -0.04(0.041) (0.039) (0.040) (0.026) (0.023) (0.024)Half Ethn Friends -0.03 -0.03 -0.03 -0.03 -0.03 -0.03(0.021) (0.021) (0.022) (0.008)??? (0.008)??? (0.008)???All Ethn. Friends -0.01 -0.01 -0.01 -0.004 -0.003 -0.001Some Ethn. Coworkers 0.01 0.01 0.05 0.04(0.026) (0.028) (0.024)?? (0.027)Half Ethn. Coworkers -0.03 -0.03 -0.02 -0.02(0.021) (0.022) (0.021) (0.020)All Ethn. Coworkers 0.01 0.01 0.01 0.002(0.035) (0.032) (0.026) (0.021)Ethnic Concent. 0.01 0.01(0.006)?? (0.06)Obs 694 694 694 694 613 613 613 613R-square 0.1694 0.1724 0.1753 0.18 0.2305 0.2363 0.2425 0.24Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significance at 10%, 5% and 1% levelsrespectively. All regressions include 9 source region-of-origin dummies and 8 area-of-residence dummies. The omitted case in theimmigration categories is ?Skilled Worker - Principal applicant?.1544.6.ConclusionsTable 4.16: Language Proficiency Decrease for People Beginning at an Advanced Level (wave 1 to wave 3)Linear Probability Model Probit Model (Marginal Effects)(1) (2) (3) (4) (5) (6) (7) (8)Age 0.004 0.004 0.004 0.005 0.003 0.003 0.003 0.003(0.001)??? (0.001)??? (0.001)??? (0.001)??? (0.001)??? (0.001)??? (0.001)??? (0.001)???Gender 0.001 0.000 -0.002 -0.001 0.004 0.003 0.002 0.002(0.024) (0.024) (0.022) (0.023) (0.018) (0.019) (0.018) (0.018)Yrs of Educ -0.014 -0.014 -0.014 -0.014 -0.013 -0.012 -0.012 -0.012(0.002)??? (0.002)??? (0.002)??? (0.001)??? (0.002)??? (0.002)??? (0.002)??? (0.002)???Skill Worker (S and D) 0.003 0.002 0.003 0.002 0.004 0.003 0.004 0.003(0.015) (0.015) (0.015) (0.015) (0.012) (0.012) (0.012) (0.012)Family Inmg. 0.050 0.049 0.046 0.046 0.044 0.043 0.039 0.039(0.020)?? (0.019)??? (0.018)?? (0.019)?? (0.022)?? (0.020)?? (0.019)?? (0.021)?ESL Centre -0.004 -0.005 -0.005 -0.009 -0.002 -0.002 -0.002 -0.004(0.013) (0.013) (0.013) (0.013) (0.014) (0.015) (0.014) (0.014)Married 0.013 0.010 0.014 0.011 0.016 0.015 0.017 0.015(0.012) (0.012) (0.012) (0.013) (0.009)? (0.009)? (0.009)?? (0.010)Household size -0.004 -0.004 -0.005 -0.004 -0.002 -0.003 -0.004 -0.003(0.002)?? (0.001)??? (0.002)?? (0.002)? (0.001)? (0.001)??? (0.002)?? (0.002)Savings wave1 (in 10,000) -0.002 -0.002 -0.002 -0.001 -0.001 -0.001 -0.001 -0.001(0.001)? (0.001)? (0.001)? (0.001) (0.001) (0.001) (0.001) (0.001)Few Ethn Friends -0.029 -0.027 -0.027 -0.025 -0.024 -0.023(0.019) (0.021) (0.020) (0.012)?? (0.012)? (0.012)??Half Ethn Friends -0.035 -0.037 -0.037 -0.027 -0.029 -0.028(0.027) (0.027) (0.028) (0.016)? (0.017)? (0.017)?All Ethn. Friends 0.000 -0.004 -0.007 -0.001 -0.004 -0.005(0.019) (0.019) (0.021) (0.014) (0.015) (0.015)Some Ethn. Coworkers -0.010 -0.012 -0.010 -0.011(0.007) (0.007)? (0.006) (0.007)?Half Ethn. Coworkers 0.006 0.001 0.005 0.003(0.010) (0.009) (0.009) (0.008)All Ethn. Coworkers 0.063 0.054 0.047 0.041(0.019)??? (0.021)?? (0.014)??? (0.014)???Ethnic Concent. 0.021 0.010(0.010)?? (0.005)??Obs 2181 2181 2181 2181 2181 2181 2181 2181R-Square 0.0742 0.0764 0.0807 0.0852 0.1177 0.1218 0.1274 0.1305Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significance at 10%, 5% and 1% lev-els respectively. All regressions include 9 source region-of-origin dummies and 8 area-of-residence dummies. The omitted case in theimmigration categories is ?Skilled Worker - Principal applicant?.1554.6. ConclusionsTable 4.17: Lang. Prof. Decrease for People Beginning at an Intermediateor Advanced LevelLinear Probability Model Probit Model (Marginal Effects)(1) (2) (3) (4) (5) (6)Age 0.005 0.005 0.005 0.004 0.004 0.004(0.001)??? (0.001)??? (0.001)??? (0.000)??? (0.000)??? (0.000)???Gender -0.003 -0.004 -0.005 0.001 -0.001 -0.001(0.018) (0.015) (0.016) (0.015) (0.014) (0.014)Yrs of Educ -0.016 -0.015 -0.015 -0.013 -0.013 -0.012(0.002)??? (0.002)??? (0.001)??? (0.002)??? (0.002)??? (0.002)???Skill Worker (S and D) 0.002 0.000 -0.001 0.003 0.002 0.001(0.012) (0.012) (0.011) (0.011) (0.010) (0.010)Family Inmg. 0.081 0.076 0.076 0.073 0.066 0.066(0.020)??? (0.019)??? (0.019)??? (0.022)??? (0.020)??? (0.021)???ESL Centre -0.015 -0.015 -0.018 -0.012 -0.012 -0.013(0.013) (0.014) (0.014) (0.013) (0.014) (0.014)Married 0.008 0.008 0.006 0.015 0.015 0.014(0.009) (0.010) (0.010) (0.008)? (0.008)?? (0.008)Household size -0.003 -0.004 -0.003 -0.003 -0.004 -0.003(0.003) (0.003)? (0.003)??? (0.002) (0.002) (0.003)Savings wave1 (in 10,000) -0.002 -0.002 -0.002 -0.002 -0.001 -0.001(0.003)??? (0.001)??? (0.001)??? (0.001) (0.001) (0.001)Few Ethn Friends -0.039 -0.038 -0.030 -0.028(0.015)?? (0.014)??? (0.008)??? (0.008)???Half Ethn Friends -0.046 -0.044 -0.033 -0.032(0.022)?? (0.020)?? (0.012)??? (0.011)???All Ethn. Friends -0.011 -0.013 -0.007 -0.007(0.013) (0.014) (0.009) (0.009)Some Ethn. Coworkers -0.010 -0.013 -0.007 -0.009(0.007) (0.006)?? (0.006) (0.005)Half Ethn. Coworkers -0.004 -0.009 -0.001 -0.004(0.014) (0.012) (0.012) (0.011)All Ethn. Coworkers 0.047 0.038 0.033 0.028(0.012)??? (0.009)??? (0.009)??? (0.007)???Ethnic Concent. 0.020 0.009(0.009)?? (0.005)?Intermediate Level -0.078 -0.086 -0.85 -0.052 -0.056(0.014)??? (0.015)??? (0.015)??? (0.005)??? (0.006)???Advanced Level 0.056(0.006)???Obs 2875 2875 2875 2875 2875 2875R-Square 0.0695 0.0751 0.0801 0.1139 0.1228 0.1257Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significanceat 10%, 5% and 1% levels respectively. All regressions include 9 source region-of-origin dummies and 8area-of-residence dummies. The omitted case in the immigration categories is ?Skilled Worker - Principalapplicant?.1564.6.ConclusionsTable 4.18: Ordered Probit (coefficients) - Language Proficiency at Wave 3 Conditioning on Initial LevelInitial Level: Basic Initial Level: Basic + Interm. Initial Level: All(1) (2) (3) (4) (5) (6) (7) (8) (9)Age -0.035 -0.037 -0.040 -0.039 -0.040 -0.042 -0.033 -0.034 -0.035(0.008)??? (0.008)??? (0.007)??? (0.009)??? (0.009)??? (0.006)??? (0.004)??? (0.003)??? (0.002)???Gender -0.026 0.078 0.071 0.063 0.103 0.111 0.029 0.055 0.060(0.175) (0.171) (0.163) (0.051) (0.053)? (0.054)?? (0.090) (0.080) (0.082)Yrs of Educ 0.132 0.124 0.109 0.116 0.111 0.099 0.118 0.113 0.107(0.029)??? (0.029)??? (0.030)??? (0.022)??? (0.022)??? (0.019)??? (0.017)??? (0.017)??? (0.014)???Skill Worker (S and D) -0.148 -0.133 -0.036 -0.166 -0.154 -0.133 -0.050 -0.046 -0.034(0.250) (0.250) (0.262) (0.169) (0.161) (0.160) (0.101) (0.099) (0.096)Family Inmg. -0.605 -0.554 -0.402 -0.599 -0.571 -0.544 -0.488 -0.457 -0.439(0.211)??? (0.218)?? (0.239)? (0.146)??? (0.139)??? (0.150)??? (0.079)??? (0.083)??? (0.081)???ESL Centre 0.011 -0.018 -0.047 0.069 0.064 0.023 0.080 0.078 0.083(0.070) (0.071) (0.081) (0.118) (0.119) (0.150) (0.104) (0.109) (0.114)Married -0.104 0.000 -0.041 -0.076 -0.068 -0.119 -0.071 -0.069 -0.061(0.296) (0.293) (0.274) (0.120) (0.125) (0.116) (0.069) (0.071) (0.077)Household size -0.021 -0.017 -0.040 0.017 0.020 0.003 0.009 0.017 0.009(0.023) (0.021) (0.021)? (0.011) (0.010)?? (0.012) (0.007) (0.009)? (0.012)Savings wave1 (in 10,000) 0.019 0.016 0.011 0.012 0.011 0.008 0.013 0.012 0.010(0.003)??? (0.003)??? (0.004)??? (0.007)? (0.007) (0.005)? (0.008) (0.008) (0.007)Few Ethn Friends 0.338 0.233 0.257 0.187 0.295 0.254(0.359) (0.371) (0.169) (0.192) (0.079)??? (0.094)???Half Ethn Friends -0.218 -0.386 0.095 -0.004 0.246 0.206(0.242) (0.169)? (0.113) (0.150) (0.064)??? (0.063)???All Ethn. Friends 0.004 0.026 0.022 0.017 0.038 0.039(0.166) (0.193) (0.152) (0.164) (0.080) (0.083)Some Ethn. Coworkers -0.335 -0.209 -0.075 0.030 0.033 0.072(0.207) (0.213) (0.080) (0.106) (0.059) (0.063)Half Ethn. Coworkers -0.101 0.021 0.027 0.097 0.006 0.050(0.155) (0.158) (0.045) (0.059) (0.051) (0.041)All Ethn. Coworkers -0.484 -0.302 -0.260 -0.132 -0.278 -0.200(0.224)? (0.227) (0.100)??? (0.101) (0.050)??? (0.050)???Ethnic Concent. -0.306 -0.236 -0.155(0.056)??? (0.075)??? (0.049)???Basic Level -0.499 -0.494 -0.527 -1.322 -1.266 -1.309(0.043)??? (0.047)??? (0.045)??? (0.051)??? (0.056)??? (0.046)???Intermediate Level -0.739 -0.698 -0.724(0.033)??? (0.043)??? (0.032)???Obs 591 591 591 1285 1285 1285 3466 3466 3466Note: Standard Errors are robust and clustered at the CMA/CA level. *, ** and *** denote significance at 10%, 5% and 1% levels respectively.All regressions include 9 source region-of-origin dummies and 8 area-of-residence dummies. The omitted case in the immigration categories is?Skilled Worker - Principal applicant?.157Chapter 5ConclusionsThis dissertation studies immigrants? assimilation into the Canadianeconomy. It examines the key reasons behind their growing earnings gapwith respect to natives. It also compares the quality of their occupationsin the first four years after arrival with the one they had in their homecountries and evaluates their English proficiency improvements during thoseyears. Taken as a whole the results present a picture of human capital trans-ferability problems.Chapter 2 exploits the new information available in the 2006 Canadiancensus regarding the location where the highest degree of education wasattained. Our estimations are able to reduce the native-immigrant wage gapfrom around 11% to close to 3%. The separation of both education and workexperience by source (Canadian versus foreign) as well as the identification ofthe location of study are deciding factors for the reduction of the wage gap.The Canadian labour market shows a negative wage premium for educationacquired abroad. In general, education acquired in Asia is the least valued,while education obtained in the US, Oceania, west Europe or the UK is asvalued as Canadian Education (valued more in the case of the UK).Chapter 3 shifts the focus of the analysis from the earnings to the qualityof occupations immigrants attain. Using the LSIC, the analysis reveals thatimmigrants with high levels of education face a larger occupational gap;that is, they experience a deeper decline in occupational status in their firstoccupation in Canada (compare to their home country occupation). Theeffects of education in the second occupational gap (for those with a secondoccupation) are still negative though of a lesser magnitude and not robustlysignificant. The pattern is confirmed when analyzing occupational statusthrough time. These results are consistent with long lasting human capitaltransferability barriers (driven by unrecognized skills or recognized lowerabilities). Interestingly, language proficiency helps reduce the occupationalgap specially for highly educated immigrants.Precisely, Chapter 4 uses the LSIC to study the changes in immigrants?English proficiency. The data show that immigrants do not change theirEnglish proficiency in a considerable manner and that most of the change158Chapter 5. Conclusionshappens in the first two years after arrival. Still, human capital considera-tions are present. Younger and more educated immigrants are consistentlymore likely to improve their English proficiency and less likely to lose it.Regarding immigration categories, those arriving as family immigrantsare less likely to improve their English proficiency (than skilled workers)and more likely to lose it (explained by either a different environment andincentives or by unobservable time-variant individual abilities). The effect isnull though for immigrants arriving with basic English proficiency. Arrivingas a particular type of immigrant may not matter in that case.Over the course of this dissertation, we see the importance of humancapital considerations on the integration of immigrants into the Canadianeconomy. Either through the imperfect transfer of skills that affects theearnings and quality of occupations of immigrants attains or through demo-graphic variables that affect the incentives to improve language proficiency.159BibliographyAkresh, I.R. (2008) ?Occupational trajectories of legal us immigrants: Down-grading and recovery.? Population and Development Review 34(3), 435?456Aydemir, A. (2009) ?Ethnic enclaves and human capital investments: Evi-dence from a natural experiment.? Working Paper, Institute for the Studyof Labor - IZA,Aydemir, A., and M. Skuterud (2005) ?Explaining the deteriorating entryearnings of canada?s immigrant cohorts, 1966?2000.? Canadian Journal ofEconomics/Revue canadienne d?e?conomique 38(2), 641?672Borjas, G.J. (1985) ?Assimilation, changes in cohort quality, and the earningsof immigrants.? Journal of labor Economics pp. 463?489(1994) ?The economics of immigration.? Journal of economic literature32(4), 1667?1717(1995) ?Assimilation and changes in cohort quality revisited: what hap-pened to immigrant earnings in the 1980s?? Technical Report, aBratsberg, B., and J.F. Ragan Jr (2002) ?The impact of host-country school-ing on earnings: A study of male immigrants in the united states.? Journalof Human resources pp. 63?105Card, D. (2001) ?Immigrant inflows, native outflows, and the local labormarket impacts of hiogher immigration.? Journal of Labor Economics(2005) ?Is the new immigration really so bad?*.? The Economic Journal115(507), F300?F323Card, D., and J.E. DiNardo (2000) ?Do immigrant inflows lead to nativeoutflows?? Technical Report, National Bureau of Economic ResearchCarliner, G. (1981) ?Wage differences by language group and the market forlanguage skills in canada.? Journal of Human Resources pp. 384?399160BibliographyChiswick, Barry R, and Paul W Miller (2010) ?Does the choice of referencelevels of education matter in the oru earnings equation?? Economics ofEducation Review 29(6), 1076?1085Chiswick, B.R. (1977) ?A longitudinal analysis of the occupational mobilityof immigrants.? In ?Proceedings of the 30 th Annual Winter Meetings,Industrial Relations Research Association, Madison, IRRA? pp. 20?27(1978) ?The effect of americanization on the earnings of foreign-born men.?The Journal of Political Economy 86, 897?921(1991) ?Speaking, reading, and earnings among low-skilled immigrants.?Journal of Labor Economics pp. 149?170Chiswick, B.R., and P.W. Miller (1995) ?The endogeneity between languageand earnings: International analyses.? Journal of labor economics pp. 246?288(1996) ?Ethnic networks and language proficiency among immigrants.?Journal of Population Economics 9(1), 19?35(2001) ?A Model of Destination-Language Acquisition: Application toMale Immigrants in Canada.? Demography 38(3), 391?409(2002) ?Immigrant earnings: Language skills, linguistic concentrations andthe business cycle.? Journal of Population Economics 15(1), 31?57Chiswick, B.R., Y.L. Lee, and P.W. Miller (2003) ?Patterns of immigrantoccupational attainment in a longitudinal survey.? International Migration41(4), 47?69(2005) ?A longitudinal analysis of immigrant occupational mobility: Atest of the immigrant assimilation hypothesis1.? International MigrationReview 39(2), 332?353Chui, T., and K. Tran (2005) ?Longitudinal survey of immigrants to canada:progress and challenges of new immigrants in the workforce?(2006) ?Longitudinal survey of immigrants to canada: A regional perspec-tive of the labour market experiences.? Statistics CanadaCitizenship, and Immigration Canada (2009) ?Facts and figures 2009: Immi-gration overviewpermanent and temporary residents.? Technical Report,Research and Evaluation Branch161BibliographyClark, Melissa A, and David A Jaeger (2006) ?Natives, the foreign-born andhigh school equivalents: New evidence on the returns to the ged.? Journalof Population Economics 19(4), 769?793Dahl, G. (2002) ?Mobility and the returns to education: testing a roy modelwith multiple markets.? Econometrica 70, 2367?2420Dustmann, C. (1994) ?Speaking fluency, writing fluency and earnings ofmigrants.? Journal of Population economics 7(2), 133?156Dustmann, C., and A. Soest (2001) ?Language fluency and earnings: esti-mation with misclassified language indicators.? Review of Economics andStatistics 83(4), 663?674Dustmann, C., and A. Van Soest (2002) ?Language and the earnings ofimmigrants.? Industrial and Labor Relations Review pp. 473?492Dustmann, C., and F. Fabbri (2003) ?Language proficiency and labourmarket performance of immigrants in the uk*.? The Economic Journal113(489), 695?717Ferrer, A., and W.C. Riddell (2008) ?Education, credentials, and im-migrant earnings.? Canadian Journal of Economics/Revue canadienned?e?conomique 41(1), 186?216Ferrer, A., D.A. Green, and W.C. Riddell (2006) ?The effect of literacy onimmigrant earnings.? Journal of Human Resources 41(2), 380?410Ferrer, A., Garnett Picot, and W. Craig. Riddell (2012) ?New directionsin immigration policy: Canada?s evolving approach to immigration selec-tion.? Canadian Labour Market and Skills Researcher NetworkFriedberg, R.M. (2000) ?You can?t take it with you? immigrant assimila-tion and the portability of human capital.? Journal of Labor Economics18(2), 221Fuller-Thomson, E., A.M. Noack, and U. George (2011) ?Health de-cline among recent immigrants to canada: Findings from a nationally-representative longitudinal survey.? Canadian Journal of Public Health102(4), 273?80Goel, D., and K. Lang (2007) ?Effects of social networks on labour marketoutcomes of recent immigrants.? manuscript, Boston University162Bibliography(2009) ?Social ties and the job search of recent immigrants.? TechnicalReport, National Bureau of Economic ResearchGreen, A.G., and D.A. Green (1999) ?The economic goals of canada?s im-migration policy: Past and present.? Canadian Public Policy/Analyse depolitiques pp. 425?451Green, D., and C. Worswick (2004) ?Immigrant earnings profiles in the pres-ence of human capital investment: Measuring cohort and macro effects.?Institute for Fiscal Studies Working Paper 04Green, D.A. (1999) ?Immigrant occupational attainment: Assimilation andmobility over time.? Journal of Labor Economics 17(1), 49?79Grenier, G. (1984) ?The effects of language characteristics on the wages ofhispanic-american males.? Journal of Human Resources pp. 35?52Grenier, G., and L. Xue (2009) Duration of Access of Canadian Immigrantsto the First Job in Intended Occupation (Department of Economics, Uni-versity of Ottawa= De?p. de science e?conomique, Universite? d?Ottawa)Grondin, C. (2005) ?Knowledge of official languages among new immigrants:How important is it in the labour market.? Statistics Canada,CatalogueHartog, Joop, and Aslan Zorlu (2009) ?How important is homeland edu-cation for refugees economic position in the netherlands?? Journal ofPopulation Economics 22(1), 219?246Hou, F., and M. Beiser (2006) ?Learning the language of a new country:A ten-year study of english acquisition by south-east asian refugees incanada.? International Migration 44(1), 135?165Imai, Susumu, Derek Stacey, and Casey Warman (2011) ?From engineerto taxi driver: Occupational skills and the economic outcomes of immi-grants.? Technical Report, Queen?s Economics Department Working PaperPatel, K., and F. Vella (2007) ?Immigrant networks and their implicationsfor occupational choice and wages.? IZA Working Paper No. 3217Rivera-Batiz, F.L. (1990) ?English language proficiency and the economicprogress of immigrants.? Economics Letters 34(3), 295?300Sullivan, P. (2009) ?Estimation of an occupational choice model when occu-pations are misclassified.? Journal of Human Resources 44(2), 495?535163Young, Margaret (1998) ?Canada?s immigration program.? ParliamentaryResearch Branch - Law and Government Division164Appendix AAppendix to Chapter 2A.1 Sample RestrictionsMajor Restrictions1 Age between 20 and 64 years old2 Education level higher than high school3 Wage income for 2005 higher than zero but lower than 88888884 Number of weeks worked in 2005 higher (or equal) than one and smallerthan 525 Full-time workers only6 No non-permanent residents7 Immigrants age at arrival has to be between 15 and 29 years old.8 Weekly wage higher than 15 times the provincial minimum hourly wage.Secondary Restrictions9 Canadian-born should have missing values for ?age at immigration?10 Location of study clearly identified (drop if location of study=?OutsideCanada?or ?Distance learning?)11 Country of origin clearly specified (drop if country of origin = ?Other?)12 Drop if years of education plus 6 is greater than age13 Drop if years of education plus 6 is greater than age at arrival for immi-grants withtheir highest degree obtained abroadInconsistencies ChecksMany of these restrictions are redudant given the above conditions14 Drop observations if logarithm of the weekly wages is less or equal tozero15 Year of immigration must be a non-missing value for immigrants165A.2 Minimum Hourly Wage by ProvinceProvince Minimum Wage Effective DateAlberta 7.00 September 2005British Columbia 8.00 November 2001Manitoba 7.25 April 2005New Brunswick 6.30 January 2005Newfoundland and labrador 6.25 June 2005Northwest Territories 8.25 December 2003Nova Scotia 6.80 October 2005Nunavut 8.50 March 2003Ontario 7.45 February 2005Prince Edward Island 6.80 January 2005Quebec 7.60 May 2005Saskatchewan 7.05 September 2005Yukon 7.20 October 1998166A.3 Years of Education by Highest DegreeAttainedHighest Degree Achieved Years of Education AssignedOther trades certificate or diploma 13Registered apprenticeship certificate 13College, CEGEP or other non-university certificate (3 to 12months)13College, CEGEP or other non-university certificate (13 to24 months)14College, CEGEP or other non-university certificate (morethan 2 years)14University certificate or diploma below bachelor level 15Bachelor?s degree 16University certificate or diploma above bachelor level 17Degree in medicine, dentistry, veterinary or optometry 18Master?s Degree 18Earned doctorate degree 21167A.4 Region of Origin and Location of StudyLocation of Study Country of Origin Groupings Location of Study1 Canada 1 Canada 1 Canada2 USA 2 US 2 US, UK, France, Oceania and rest of Europe3 South America 3 Jamaica 3 East Europe, Romania and Poland4 Rest of America 4 South America 4 West and Central Asia and China5 UK 5 Rest of America 5 India, Pakistan and the rest of Asia6 France 6 UK 6 South America and the rest of America7 Poland 7 France 7 Philippines and South-East Asia8 Romania 8 Poland 8 Africa9 East Europe 9 Romania10 Rest Europe 10 East Europe11 Africa 11 Rest Europe12 Western and Central Asia 12 Africa13 China 13 Western and Central Asia14 Phillipines 14 China15 India 15 Hong Kong16 Pakistan 16 Phillipines17 South East Asia 17 Vietnam18 Rest Asia 18 India19 Oceania 19 Pakistan20 South East Asia21 Rest Asia22 Oceania168Appendix BAppendix to Chapter 3B.1 Sample AttritionTable B.1: Probit model marginal effectProbitHigh level of speaking -0.00(0.01)High level of writing CMA -0.00(0.01)Ethnic Concentration at work -0.02(0.01)Age CMA -0.00(0.00)?Sex 0.01(0.01)Land month 0.04(0.06)Land month2 -0.00(0.00)Graduate -0.02(0.01)Bachelor 0.01(0.01)Some University -0.02(0.01)High School -0.01(0.01)Family Class 0.03(0.04)Skilled worker -0.01(0.04)Business Class 0.01(0.04)Refugee -0.03(0.04)Region: U.S.A. -0.01(0.07)Region: Central America -0.01(0.07)Region: Caribbean and Bermuda 0.05(0.08)Region: South America 0.03Continued on next page169B.1. Sample AttritionTable B.1 ? continued from previous pageProbit(0.07)Region: Western Europe -0.01(0.06)Region: Eastern Europe 0.03(0.07)Region: United Kingdom 0.01(0.07)Region: Other Northern Europe 0.27(0.20)Region: Southern Europe -0.02(0.06)Region: Western Africa 0.08(0.08)Region: Eastern Africa 0.06(0.08)Region: Northern Africa 0.04(0.07)Region: Southern Africa -0.03(0.07)Region: West Central Asia - Middle East 0.03(0.06)Region: Eastern Asia 0.07(0.07)Region: Southeast Asia 0.02(0.06)Region: Southern Asia 0.07(0.07)Region: Oceania -0.04(0.06)Sample 9,305* significant at 5 percent; ** significant at 1 percent170B.2 Ranking of Occupations a SampleDESCRIPTION INDEXE011 Judges 7.89C013 Geologists, geochemists and geophysicists 7.13B012 Financial and investment analysts 7.04C175 Railway and Marine traffic controllers 6.98J015 Supervisors, Forest Products Processing 6.9J111 Central Control and Process Operators, Mineral and Metal Processing 6.83H016 Contractors and Supervisors, Mechanic Trades 6.78B411 Supervisors, general office and administrative support clerks 6.74C054 Land surveyors 6.7C063 Computer Programmers 6.65H019 Contractors and Supervisors, Other Construction Trades 6.6G924 Other Personal Service Occupations 6.55J016 Supervisors, Textile Processing 6.5F143 Theatre, Fashion, Exhibit and Other Creative Designers 6.46I017 Aquaculture Operators and Managers 6.4F111 Library and Archive Technicians and Assistants 6.36J123 Glass Forming and Finishing Machine Operators and Glass Cutters 6.33J213 Electronics Assemblers, Fabricators, Inspectors and Testers 6.25J225 Plastic Products Assemblers, Finishers and Inspectors 6.18H514 Jewellers, Watch Repairers and Related Occupations 6.11J222 Furniture and Fixture Assemblers and Inspectors 5.98G814 Babysitters, Nannies and Parents? Helpers 5.36+ Significant at 10%; * at 5%; ** at 1%Note: This index is constructed using the average logarithm of weekly wagesof full-time Canadian paid workers. There are 22 or 23 occupations betweeneach category presented.171B.3 Individual Characteristics and the Occupational GapEducation Immig. Category Lang. Prof. Network Job1st Gap 2nd Gap 1st Gap 2nd Gap 1st Gap 2nd Gap 1st Gap 2nd GapGraduate -0.28 -0.15(0.05)** (0.06)*Bachelor -0.30 -0.22(0.06)** (0.08)*Some University -0.17 -0.12(0.04)* (0.07)Lang. Prof. 0.14 0.07(0.02)** (0.02)**Family Class 0.13 0.11(0.03)** (0.03)**Skilled Worker (S & D) -0.08 -0.07(0.04)+ (0.04)Network Job -0.07 -0.05(0.01)** (0.01)**Constant -0.10 -0.10 -0.36 -0.28 -0.45 -0.32 -0.34 -0.26(0.05)+ (0.07) (0.02)** (0.01)** (0.01)** (0.02)** (0.02)** (0.01)**Sample 1579 1043 1579 1043 1579 1043 1579 1043+ Significant at 10%; * significant at 5%; ** significant at 1% multicolumn 5c172B.4 Motivational ModelI use a simple approach to model the initial occupational mismatch and occupational mobilityof recent immigrants. Consider a (logarithm) wage equation for the home country based onindividual skills (Si) and with a return to skills equal to ?1.whomei = ?0 + ?1Si (B.1)Consider further a scenario in which immigrants are employed at arrival and there are two typesof occupational recognition: Type H (high) and Type L (low).104 The wage equation for type Hhas a return to education equal or less than the one at home (?1 ? ?1). The wage equation fortype L occupation has a lower return to education than the type H, and definitely lower than theone at the home country (?1 ? ?1 > ?2). The probability of getting a type H recognition uponarrival is equal to PH and it could be thought to be positively influenced by language proficiency.PH : wHi = ?0 + ?1Si (B.2)1 ? PH : wLi = ?0 + ?2Si (B.3)From here we can calculate the expected wage an immigrant would receive and the expected firstdifference between the wages in the home country and host country.E(wi/Si) = ?0 + PH(?1Si) + (1 ? PH)(?2Si) (B.4)Given that education does not affect PH , high educated immigrants would face a greater declinein their expected first wages than their poorly educated counterparts. To the extend that ?1 >?2, language proficiency, through its effect on the probability of receiving a type H recognition,diminishes the wage drop. Moreover, the cross derivative of education and language should bepositive.E(Occupational Gap/Si) = (PH)(?1 ? ?1)Si + (1 ? PH)(?2 ? ?1)Si?Occup. Gap?S= (PH)(?1 ? ?1) + (1 ? PH)(?2 ? ?1) < 0?Occup. Gap?Lang.= (?1 ? ?2)Si(?PH?Lang.) > 0We could think of a two-period model in which the only way to improve skill recognition is tochange occupations.105 However, changing occupations is costly. An individual would have tobear a cost of Ci, which comes from the known cumulative distribution F (c). F (c) is strictlyincreasing and differentiable. An immigrant in a low skill recognition occupation would onlychange occupations if the benefit from doing so is sufficiently large (PH(?1 ? ?2)Si) ? Ci > 0).Immigrants in occupation type H do not switch jobs. Thus the probability an immigrant changingoccupations is equal to:=???0 if wH or occupational gap (?1 ? ?1)Si ;F ((PH)(?1 ? ?2)Si) if wL or occupational gap (?2 ? ?1)Si.104Given the high labor market participation rate for the selected sample, the absenceof unemployment is not an extreme assumption.105In principle, there could exist different probabilities for receiving the high type skillrecognition following the first occupation (e.g. P ?H). The implications of the model wouldnot change.173B.4. Motivational ModelConditioned on being in a type L recognition, the probability of switching jobs is positivelyinfluenced by having a high level of education and being proficient on the host country language.?Prob.?S= F ?()(PH(?1 ? ?2)) > 0?Prob.?Lang.= F ?()(?PH?Lang.)(?1 ? ?2)Si > 0If we assume that an immigrant only takes a new job if it offers higher returns, the expectedwage improvement from the first to the second period is only relevant for immigrants with a lowrecognition category (type L). The improvement becomes a function of the probability of gettinga high recognition category (type H), the education return differentials (?1 ? ?1) and education(Si).Occup. Improvement = PH(?1 ? ?2)SiFinally, we can analyze immigrants? expected wage gap for their second occupations. Factoringin their occupations in the first period we get:=???. if wH in the first occupation;(1 ? PH)(?2 ? ?1)Si + PH(?1 ? ?1)Si if wL in the first occupation.To the extent that even after switching some immigrants would not get the same returns toeducation as in the home country, we should expect education to still have a negative effect onthe second occupational gap. Its magnitude though will depend on the fraction of immigrants inthe high recognition category (PH). If PH is large enough and the difference between the returnto education in the home country (?1) and the return to education in a type H occupation (?1)is small, then the effect of education on the second occupational gap would be small.?E(2nd Occupational Gap)?Si= (1 ? PH)(?2 ? ?1) + (PH)(?1 ? ?1)174B.5 Dictionary of VariablesDependent VariablesFirst Occupational Gap Quality of the first occupation held in Canada minusquality of the occupation heldin the home countrySecond Occupational Gap Quality of the second occupation held in Canada mi-nus the quality of theoccupation held in the home countryOccupational Improvement Second Occupational Gap minus first OccupationalGapOccupational Quality Average Logarithm of weekly wages by the StandardOccupational Classification (1991)Independent VariablesDemographicsAge Age at wave 1English 1 if the immigrant has knowledge of English (writingvery well), 0 otherwiseFrench 1 if the immigrant has knowledge of French (writingvery well), 0 otherwiseLanguage Proficiency 1 if person has knowledge of English and lives inEnglish Canada (or Montreal)or has knowledge of French and lives in FrenchCanada, 0 otherwiseEducationGraduate A person with an education level higher than a Bach-elor?s degreeBachelor A person with a Bachelor?s degreeSome University studies A person with an education level higher than Highschool butwithout a Bachelor?s degreeHigh School A person with a High School diplomaNo High School A person with no High school diploma (Referencecase)Educated 1 if person has an education equal or high than aBachelor?s degreeContinued on next page175B.5. Dictionary of VariablesTable B.2 ? continued from previous pageLang. Proficiency*Educated Interaction of ?Language Proficiency? and ?Edu-cated?Immigration categoryFamily 1 if respondent is in Family class, 0 otherwise.Skilled Workers S and D 1 if respondent landed as a Skilled Worker Spouse orDependant , 0 otherwiseSkilled Workers PA 1 if respondent landed as a Skilled Worker PrincipalApplicant, 0 otherwise (Reference Category)Ethnic distributionAverg. Ethnic Wage Average of the occupational quality of the immi-grant?s ethnicity in a given cityEthnic Job - 1st 1 if person found his first occupation directly throughfamily or friend, 0 otherwiseEthnic Job - 2nd 1 if person found his second occupation directlythrough family or friend, 0 otherwise176B.6 Wage ImprovementChanges from Wage 1 to Wage 2 Changes from Wage 2 to Wage 3(1) (2) (3) (4) (5) (6) (7) (8)Constant 0.16 0.12 0.08 0.05 0.11 0.15 0.14 0.14(0.02)?? (0.00)?? (0.03)?? (0.05) (0.005)?? (0.00)?? (0.08)+ (0.08)+CMA/CA fixed effects No Yes No Yes No Yes Yes YesCountry of Origin fixed effects No No Yes Yes No No Yes YesObservations 398 398 398 398 320 320 320 320R-square 0.00 0.03 0.09 0.12 0.00 0.00 0.07 0.07Note: + Significant at 10%, * at 5%, ** 1%.177Appendix CAppendix to Chapter 4C.1 Dictionary of VariablesAge Declared age in years at wave 1Gender Dummy variable - 1 if male, 0 if femaleYrs of Educ. Number of years of Education obtained before arrivalSkilled Worker (S and D) Dummy variable - Spouse and Dependants of anSkilled Worker ImmigrantFamily Immg. Dummy variable - Family ImmigrantESL Centre Dummy variable - 1 if the first three digits of thezipcode coincide with the presenceof an ESL evaluation centre; zero otherwiseMarried Dummy variable - 1 if the person has an spouse or isin a common-law union,zero otherwiseHousehold size Declared number of people living in the household inwave 1Savings Wave1 (in 10,000) Declared savings in Canada or abroad in wave 1Few Ethn Friends Dummy variable - 1 if at wave 1 the immigrant madenew friends - a few from thesame ethnicity, 0 otherwiseSome Ethn Friends Dummy variable - 1 if at wave 1 the immigrant madenew friends - Half from thesame ethnicity, 0 otherwiseContinued on next page178C.1. Dictionary of VariablesTable C.1 ? continued from previous pageAll Ethn Friends Dummy variable - 1 if at wave 1 the immigrant madenew friends - All or most fromthe same ethnicity, 0 otherwiseSome Ethn. Coworkers Dummy variable - 1 if at wave 1 the immigrant isworking and some of his/hercoworker are from the same ethnicity, 0 otherwiseHalf Ethn Coworkers Dummy variable - 1 if at wave 1 the immigrant isworking and half of his/hercoworker are from the same ethnicity, 0 otherwiseAll Ethn. Coworkers Dummy variable - 1 if at wave 1 the immigrant isworking and all or most of his/hercoworker are from the same ethnicity, 0 otherwise179

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0074045/manifest

Comment

Related Items