Fake News and Cultural Identity: Evidence from South Tyrol in 1939 

Get Complete Project Material File(s) Now! »

The Economics of Diversity

Next, I give an overview on the economics of diversity, in general, and with a focus on cultural diversity. When speaking about diversity in the context of economic outcomes, we generally refer to diversity as a trait of a population. Any given pop-ulation can be diverse in many ways: gender, age composition, language, religion, ethnicity and so on. The interplay between economics and the diverse characteris-tics of a population is non negligible. On one hand, economic policies tend to affect different parts of the population differently. In economics, we call this analysis of the heterogeneity of effects by specific characteristics. In the work that represents the first chapter of this thesis, for example, we find that language training for im-migrants has a different effect for men than for women, for elder people relative to younger people, for different groups of immigrants (labor migrants, family migrants and refugees), and for immigrants whose native language is more or less linguisti-cally distant to the destination country’s main language.
While the analysis of the heterogeneity of effects is a well-established and common part of any economic analysis, the reverse question gives rise to a relatively recent strand of literature: How does diversity affect economics outcomes?
Relying on a range of different types of diversity (e.g. ethnic, religious, linguistic, ge-netic, birthplace or cultural diversity), the current body of research finds theoretical and empirical results that point in opposite directions. On one hand, the combina-tion of diversity in skills, experience and ideas across groups and individuals (their knowhow), tend to lead to positive economic outcomes: increasing productivity and spurring innovation. On the other hand, higher degrees of diversity can also lead to negative economic outcomes, caused by misunderstanding and hostility across diverse groups or individuals, who each believe that their way of doing things is the only right way: decreasing productivity and increasing conflicts.
Even though the findings on whether diversity is « good » or « bad » for the economy diverge, scholars agree upon the fact, that 1) it is difficult to measure diversity, 2) outcomes differ according to the measurement of diversity.
Common measures of diversity are ethnic fractionalization (Easterly and Levine 1997; Alesina et al. 2003; Fearon 2003, Alesina and Zhuravskaya, 2011; Alesina et al., 2016), linguistic diversity (Desmet et al., 2012), genetic diversity (Ashraf and Galor, 2013) birthplace diversity (Alesina, Harnoss and Rapoport, 2016), and other measures of social diversity (Bisin and Verdier, 2000 and 2001; and Bisin et al. 2011; Fouka, 2020; Laitin, 1998; Carvalho and Koyama, 2017; Fryer Jr et al., 2004; Rubinstein et al. , 2013; Biavaschi et al., 2017).
Research relying on measures of ethnic fractionalization, by the aforementioned au-thors, finds a negative effect between ethnic/linguistic diversity and a number of outcome variables: economic growth, quality of government, inequality, and, and an increase in conflicts. Looking at genetic diversity, Ashraf and Galor (2013), find an inverted u-shape relationship between diversity and development. In the discussion around migration economics, we have already briefly discussed birthplace diversity – a measure which repaints the picture about diversity and economic outcomes. First, the authors show that ethno-linguistic, genetic and birthplace diversity are mostly uncorrelated with each other. Second, they differ economically in that ethno-linguistic diversity has negative or insignificant results, while birthplace diversity has a robust positive effect on income even in the long run, and controlling for a large set of control variables. Economic analyses at the firm level largely confirm the macro results: negative effects for ethnic diversity, positive effects for birthplace diversity (Hjort, 2004; Brunow et al. 2015; Parrotta et al., 2014; Ozgen et al., 2013; Boeheim et al., 2012).

The Effect of Language Training on Immigrants’ Economic Integration: Empirical Evidence from France

We examine the impact of language training on the economic integration of immigrants in France. The assignment to this training, offered by the French Ministry of the Interior, depends mainly on a precise rule: the training is pro-vided if the test score of an initial language exam is below a certain threshold. This eligibility rule creates a discontinuity in the relation between the test result and the variables of interest, which is used to estimate the causal effect of language training, through the method of Regression Discontinuity Design. We find that the number of assigned hours of training significantly increases labor force participation of the treated individuals. The language classes ap-pear to have a larger effect for individuals with higher levels of education, while there is no robust differential effect by type of migration, gender or age. Our estimated coefficients are remarkably similar when we rely on local lin-ear regressions using the optimal bandwidth with few observations around the threshold and when we control parametrically for a polynomial of the forcing variable and use the whole estimation sample. We conclude with a discussion of the candidate mechanisms for the improved labor market participation of immigrants.

The French « Contrat d’Accueil et d’Intégration » (CAI)

As previously explained, every new legal immigrant to France who is older than 16 and is coming from a country outside the EU, has to sign a contract, which imposes a civil training (e.g., on French institutions and on the values of the French Repub-lic), a language training, an information session on life in France and a statement of professional competence. The language training is the only part that is conditional upon the passing or failing of an entry exam – all other parts are mandatory for ev-ery CAI signatory. Until recently this integration contract was the so-called Contrat d’accueil et d’intégration (CAI), introduced in 2007. In July 2016, the French Gov-ernment launched an updated version of this program, under the name of Contrat d’intégration republicaine (CIR). In this paper, we evaluate the language training component of the CAI.
Following the signature of the CAI, all individuals had to take an evaluation of lan-guage skills which is designed and carried out by « FLE » (Français langue étrangère or French as a Foreign Language) instructors. The exam was carried out individu-ally for every CAI signatory, during a general interview at the OFII (Office Français de l’Immigration et de l’Intégration or French Immigration and Integration Office) with an auditor of OFII who is « FLE » certified. This language exam evaluated im-migrants’ actual language skills relative to an A1.1 (a “survival”) level. Individuals with language skills above this level were likely to succeed at the highest grade, even though their French language skills might have been very basic. For the individuals who were assigned this training – designed by « FLE » instructors as well – the num-ber of assigned hours ranged from 60 to 400, depending on how much training was needed to achieve an A1.1 level. The scope was, indeed, to bring all individuals’ language skills to the same level: A1.1. The number of hours was decided by the or-ganization that carried out the language classes and depended on guidelines written by the language class providers (e.g., language schools) before submitting their offer and being chosen through government procurement contracts.5 The content of the training was structured in three blocks: public life (e.g., institutions, values, secu-larism), practical life (e.g., health, education) and professional life (e.g., job search, employment contract types). Once the training was over, an evaluation was made and the individuals got a diploma of basic French skills, which was an important requirement to renew the residence permit. After the first renewal, this one-time language course allowed to extend the residence permit as often as the individuals wished. Only in case the immigrants wanted to naturalize, they had to present a B1 certificate level. In sum, the CAI system foresaw no system of progressive learning of the new language: once the level A1.1 was attained, individuals did not have to further improve their French – unless they wanted to apply for naturalization. Thus the offered training stopped at levels that were generally too low for economic integration. The individuals did not speak French well enough to be considered employable.

Regression discontinuity design (RDD)

As explained, each new legal immigrant to France has to take a language exam assessing her knowledge of the French language, written and spoken. If the result is insufficient, the person is likely to be assigned to a language program. The rule is the following: the treatment (i.e., the linguistic training) is likely to be offered when the test results are inferior or equal to a certain threshold. More precisely, the exam is composed of five parts. The first part is an oral examination, where the individual can obtain a score between zero and 70 points. The four additional written tests give a maximum of 30 possible points. The total score ranges from zero to 100 points and is measured in intervals of five points. 45 is the last failing grade and 50 is the first passing grade. This eligibility rule – which is not necessarily known by the immigrants – creates a discontinuity in the relation between the test result and the variables of interest. The discontinuity will be used to estimate the causal effect of the linguistic training, through the method of Regression Discontinuity Design.

Fuzzy RDD

The treatment is heterogeneous in the sense that different individuals can be as-signed a different number of hours of language training. The range goes from 0 to 400 hours, in steps of 10. The assignment to the language training and the number of hours do however not solely depend on the fact of passing or not the language exam. The government may take into account other personal and socio-demographic characteristics of the immigrant as well. So, for the same test score, an immigrant coming from a non-francophone country may have a higher probability of being assigned language training (and more hours of classes) than an immigrant from a francophone country (Le Quentrec-Creven, 2014). Nonetheless, the language exami-nation is the most important variable taken into consideration. As Figure 1 shows – considering the total test score of the entry exam as horizontal axis and the number of hours of language training as vertical axis – there is a clear cutoff at the passing threshold.
Given that our running variable is not the only determinant of the treatment status, we employ the method of fuzzy RD design, using the test result to build an instru-ment for the number of hours of language classes the immigrant was assigned.
Our first and second stage regressions are set up the following way. Equation 1 represents the second stage of the 2SLS specification: Yi = 0 + 1LCi + 2Xi + « i (1).
Yi is the outcome of interest for individual i (e.g., labor force participation, employ-ment status, objective language skills, etc.) and Xi is a vector of control variables. Our main explanatory variable is LCi, which is equal to 0 if no language classes were assigned or the number of prescribed hours if the language training was assigned. We are, thus, interested in the coefficient 1. Equation 2 is the first stage of our specification: LCi = 0 + 1Ti + 2Xi + i (2).

Table of contents :

1 General Introduction
2 The Economics of Migration and Diversity
2.1 The Economics of Migration
2.2 The Economics of Diversity
3 Contribution and Outline of the dissertation
2 The Effect of Language Training on Immigrants’ Economic Integration: Empirical Evidence from France 
1 Introduction
2 The French « Contrat d’Accueil et d’Intégration » (CAI)
3 Data
3.1 ELIPA
3.2 Data on Test Results and Outcome Variables
3.3 Descriptive Statistics
4 Regression discontinuity design (RDD)
4.1 Fuzzy RDD
4.2 Two approaches: local linear regressions and parametric estimates using the whole sample
4.3 Validity of the RD design
5 Results
5.1 Mechanisms
6 Concluding remarks
7 Appendix
7.1 A.1 Selection into or out of training: Robustness checks for possible manipulation of the assignment variable
3 Culture and Emigration: Evidence from the Hitler-Mussolini Migration Option Agreement. 
1 Introduction
2 Historical context
2.1 Migration decision: The procedure
3 Data
4 Empirical Strategy
5 Results
5.1 Heterogeneity of effects
5.2 Alternative explanations: Selection-into-migration by the Governments
5.3 Empirical determination of the emigration threshold e
6 Conclusion
7 Appendix
7.1 A1: Theoretical frame for the determination of the emigration threshold
4 Fake News and Cultural Identity: Evidence from South Tyrol in 1939 
1 Introduction
2 Historical Background
2.1 Fake news « Sicilian Legend »: July 3, 1939
3 Data
3.1 Nibelungen Index
4 Methodology
4.1 Validity of the methodology
5 Main results
5.1 Channels: Change in preferences or change in incentives?
5.2 Heterogeneity of effects
5.3 Persistence of the effect
5.4 Robustness Tests
6 Conclusion
7 Appendix
5 Résumé 
Summary

GET THE COMPLETE PROJECT