Essays on the Economics of Language and Language Policy by Alex Armstrong A thesis submitted to the Graduate Program in Economics in conformity with the requirements for the Degree of Doctor of Philosophy Queen(cid:146)s University Kingston, Ontario, Canada June 2013 Copyright c Alex Armstrong 2013 (cid:13) Abstract This thesis concerns the economic dimensions of second language knowledge and acquisition and the economic implications of language policies. The value of the ability to speak and understand a second language depends on the extent to which it enlarges one(cid:146)s communicative sphere which, in turn, depends on the language abilities of others. This implies that second language acquisition decisions are associated with strategic considerations and spillover e⁄ects. Consequently, the equilibrium distribution of language skills may not be socially e¢ cient and policy remedies may be called for. The (cid:133)rst essay of the thesis investigates the relationship between earnings, second language knowledgeandthedistributionoflanguageskillsinlocallabourmarketsinCanadausingcensus data. We estimate the elasticity of local language complementarity in earnings: a parameter that measures the importance of the linguistic environment in the earnings of the individual as well as the importance of language in the economy generally. The second essay addresses the e¢ ciency of second language acquisition decisions in a theo- reticalmodelwherebilingualismisrewardedwithahigherwagefortworeasons. First,language skills constitute a form of human capital in the sense that a worker(cid:146)s productivity is positively related to the proportion of the population with whom she shares a language. Second, language skills serve as a signal of productivity to employers. In general, the private and social bene(cid:133)ts of bilingualism do not align due to counteracting network and signalling welfare e⁄ects. The third essay concerns the role of language policy in improving social outcomes. A tax- subsidy system is considered under various assumptions about the ability of the government or planner to discriminate between individuals and groups. A Pareto improvement is possible if the government can condition the tax-subsidy system on language acquisition costs but not i otherwise. Thefourthessayconsiderstheoptimalprovisionofpublicserviceswhenindividuals(cid:146)e⁄ective consumption of the services depends on their pro(cid:133)ciency in the language they are provided in. The planner faces a trade-o⁄ between compensating minority language speakers for their lower wages and encouraging their integration by rewarding higher levels of dominant language pro(cid:133)ciency. ii Acknowledgements I am highly indebted to my supervisors, Robin Boadway and Charles Beach, for their advice and encouragement in preparing this thesis. I also bene(cid:133)ted from discussions with various other faculty members at Queen(cid:146)s University including Frank Lewis, James Bergin and Sumon Majumdar. Thanks are due to my colleagues in the Ph.D. program at Queen(cid:146)s, especially Louis Perrault, Matt Webb, Steve Kivinen, Michel Cloutier, Jean-Denis Garon and Karl Skogstad all ofwhomtookaninterestinmyresearchinvariousways. Lastly,butmostimportantly,Iwishto acknowledgethecontributionofmypartner, Sarah, withoutwhosepatienceandunderstanding, this thesis would not exist. iii Table of Contents Abstract i Acknowledgements iii Table of Contents iv Table of Tables vii Table of Figures viii Chapter 1: Introduction 1 1:1: Related Literature 3 1:1:1 Empirical Literature 4 1:1:2:Theoretical Literature 6 1:2 Outline of the Thesis 8 Chapter 2: Local Complementarities and the Returns to Language Skills 11 2:1: The Bilingual Wage Premium in Canada 13 2:2: Earnings and Local Complementarities in Language Knowledge 18 2.2.1 The Elasticity of Local Language Complementarity 21 2:2:2: The Elasticity of Local Language Complementarity over Time 24 2:2:1: The Elasticity of Local Language Complementarity over the Earnings Distribution 28 2:2: Conclusion 37 Chapter 3: Equilibria and E¢ ciency in Bilingual Labour Markets 39 3:1: Model 41 3:1:1: Pure Human Capital Model 46 iv 3:1:2: Pure Signalling Model 53 3:1:3: Mixed Human Capital - Signalling Model 57 3:1:4 Summary 62 3:2: Second Language Acquisition in Canada 65 3:3: Conclusion 70 Chapter 4: Second Language Acquisition and Optimal Language Policy 72 4:1: Model 74 4:2: Welfare and E¢ ciency 78 4:3: Policy 83 4:3:1: First Best 83 4:3:2: Second Best 85 4:3:3: Third Best 88 4:4: Conclusion 89 Chapter 5: The Provision of Language-Speci(cid:133)c Public Services 90 5:1: Model 95 5:2: Dominant Language Fluency of the Minority 97 5:2: The Provision of Language-Speci(cid:133)c Public Services 102 5:2:1: Utilitarian Objective 103 5:2:2: Maximin Objective 107 5:2:2: Majority Voting 111 5:3: Heterogeneous Learning Costs 113 5:3:1: Dominant Language Fluency of the Minority 114 5:3:2: The Provision of Language-Speci(cid:133)c Public Services 115 v 5:4: Conclusion 119 Chapter 6: Conclusion 122 Bibliography 125 Appendix A: Chapter 2 Regression Results 134 Appendix B: Chapter 3 Proofs 168 Appendix C: Chapter 4 Proofs 175 Appendix D: Chapter 5 Comparative Statics and Derivations 183 vi Table of Tables 2:1: Selected Estimates of the Bilingual Wage Premium 15 2:2: The Distribution of Language Skills and Bilingual Wage Premia in Canada 2005 16 2:3: Estimates of the Elasticity of Local Language Skill Complementarity in Earnings: Canada 2005 22 3:1: Summary and Comparison of the Results from the Theoretical Model 64 3:2: SUR Estimates of Rates of Bilingual Knowledge in Canada 2005 68 4:1: Linguistic Equilibria 78 4:2: Optimal Linguistic Distributions 81 A:1: Log Wage Regressions with Language Group Dummy Variables as Regressors: Males 134 A:2: Log Wage Regressions with Language Group Dummy Variables as Regressors: Females 138 A:3: Log Wage Regressions with lnp as a Regressor: Males 142 A:4: Log Wage Regressions with lnp as a Regressor: Females 146 A:5: Log Wage Regressions 1981 150 A:6: Log Wage Regressions 1986 153 A:7: Log Wage Regressions 1991 156 A:8: Log Wage Regressions 1996 160 A:9: Log Wage Regressions 2001 164 vii Table of Figures 2 1: Distributions of the Local Population with whom (cid:0) Individuals Share a Language 2005 20 2 2: Estimates of the Elasticity of Local Language Skill (cid:0) Complementarity in Earnings: Canada 1980 to 2005 - Industry and Occupational Controls Included 26 2 3: Estimates of the Elasticity of Local Language Skill (cid:0) Complementarity in Earnings: Canada 1980 to 2005 - Industry and Occupational Controls Excluded 27 2 4: Unconditional Quantile Estimates of the Elasticity of Local (cid:0) Language Skill Complementarity in Earnings - Industry and Occupational Controls Included: 2005 30 2 5: Unconditional Quantile Estimates of the Elasticity of Local (cid:0) Language Skill Complementarity in Earnings - Industry and Occupational Controls Excluded: 2005 31 2 6: Unconditional Quantile Estimates of the Bilingual Wage Premium (cid:0) for Francophones in Quebec: 2005 32 2 7: Conditional Quantile Estimates of the Elasticity of Local (cid:0) Language Skill Complementarity in Earnings - Industry and Occupational Controls Included: 2005 34 viii 2 8: Conditional Quantile Estimates of the Elasticity of Local (cid:0) Language Skill Complementarity in Earnings - Industry and Occupational Controls Excluded: 2005 35 2 9: Conditional Quantile Estimates of the Bilingual Wage Premium (cid:0) for Francophones in Quebec: 2005 36 3 1: The Equilibrium Distribution of Language Skills 47 (cid:0) 3 2: Di⁄erences in the Bilingual Wage Premia 50 (cid:0) 3 3: Knowledge of Both O¢ cial Languages in Canadian Communities (cid:0) by Mother Tongue 2005 66 4 1: Optimal and Equilibrium Distributions when n < n 82 x y (cid:0) 6 2: The Pareto Sets for Each Language Group are Weakly Disjoint 179 (cid:0) ix