Language of Instruction as Education Policyand Its Effects on the Socio-economic Outcomes of Young People

Dieudonné Tamfutu Munsi

doi:10.21996/wnfv-vq65

Language of Instruction as Education Policyand Its Effects on the Socio-economic Outcomes of Young People

Dieudonné Tamfutu Munsi

Publikation: Afhandling › Ph.d.-afhandling

234 Downloads (Pure)

Abstract

Denne afhandling indeholder fire tæt relaterede artikler. Det første kapitel beskriverden oprindelige ‘syntetisk kontrol’-metode, de forhold hvorunder den kan benyttes overbeviseligt og nogen af de nyeste udviklinger indenfor metoden. Jeg afdækker udvidelser afden oprindelige metode tilpasset til situationer, hvor donor-enheder kan bidrage til den kontrafaktiske forudsigelse med negative vægte eller vægte, der ikke summerer til én, såvel somudvidelser hvor bias er reduceret med regressionsmetoder. Jeg konkluderer med en kortdiskussion af tilgange til inferens i kontekst af syntetisk kontrol.
I det andet kapitel undersøger vi effekterne på uddannelsesmæssige resultater som følgeaf et abrupt skifte i undervisningssproget i Rwanda, fra ét ikke-indfødt sprog (Fransk) tilet andet ikke-indfødt sprog (Engelsk). Dette skifte, indført i 2008, påvirkede alle niveaueraf uddannelse uden nogen form for sprogtræning af lærere og studerende. Ved hjælp af ensyntetisk kontrol metode, kommer vi frem til, at ændringen i undervisningssproget medførteet straks og vedholdende fald i indskrivelser på grundskoler (primary schools). Ydermerefinder vi et fald i indskrivelser på gymnasier (secondary schools), på det tidspunkt hvorden første kohorte påvirket af politikken ville have påbegyndt gymnasier. Som en direktekonsekvens af et lavere antal indskrivninger på folkeskoler finder vi også en reducering afdet forventede antal år i skole. Vi leverer antydende beviser for, at faldet i indskrivelser kantilskrives højere frafaldssatser, mest blandt elever der anses for at være for gamle for deresrespektive klasser.
Et højere uddannelsesniveau er som regel associeret med en udskydelse af ægteskab ogfødsler, i det tredje papir undersøger vi det kausale link mellem en politik som plausibeltreducerede uddannelsesniveauet i Rwanda og teenage-fødsler. Rwanda ændrede pludseligtderes undervisningssprog i 2008 på alle uddannelsesniveauer fra Fransk til Engelsk, selvom enmeget lille del af befolkningen var i stand til at tale Engelsk. Dette medførte et signifikantfald i indskrivelser på grundskoler og gymnasier. Ved brug af både en syntetisk kontrolmetode med aggregeret data samt en difference-in-difference metode med data på individueltniveau, dokumenterer vi en forøgelse af teenage-fødsler som et resultat af denne politik. Dogmere upræcist, antyder vores resultater også at denne effekt var størst i byområder.
På trods af, at det er blevet vist at bedre sundhed, fører til højere uddannelse, bådei udviklede lande samt udviklingslande, så er det kausale link fra uddannelse til sundhedmindre klart. Derfor undersøger vi i det fjerde kapitel, hvordan uddannelsespolitik, potentielt, kan reducere uddannelsesniveauet og dermed også de berørte unge voksnes helbredsamt helbredet af deres børn, der var født efter, at interventionen fandt sted. Ved brug afaggregeret data samt en syntetisk kontrol metode, finder vi begrænset bevis for en effekt påvoksnes sundhed og nogle beviser for effekter på børns sundhed. Vores analyser bidrager tilforståelsen af faktorerne bag transmissionen af fattigdom og ulighed mellem generationer, ikonteksten af et udviklingsland.

Originalsprog	Engelsk
Bevilgende institution	Syddansk Universitet
Vejledere/rådgivere	Trandafir, Mircea, Hovedvejleder Daysal, N. Meltem, Bivejleder
Dato for forsvar	25. aug. 2023
Udgiver	Syddansk Universitet. Det Samfundsvidenskabelige Fakultet
DOI	https://doi.org/10.21996/wnfv-vq65
Status	Udgivet - 27. jun. 2023

Dokumenter og links

10.21996/wnfv-vq65

Open Access VersionForlagets udgivne version, 21,8 MB

SDU link

Tjek adgangen til materialet i Syddansk Universitetsbiblioteks søgemaskine

Citationsformater

@misc{fb6cb7bb53f34839a5edc1d2a213bfde,

title = "Language of Instruction as Education Policyand Its Effects on the Socio-economic Outcomes of Young People",

abstract = "This dissertation consists of four closely related papers. The first chapter describesthe original version of the synthetic control method, the conditions under which it can beconfidently used, and some of the most important recent developments on the topic. Icover extensions of the original method to situations when donor units can contribute to thecounterfactual with negative weights or with weights that do not sum up to one, as well asvarious regression-based modifications that attempt to reduce or eliminate bias. I concludewith a brief discussion of recent approaches to inference in the context of synthetic controls. The second chapter investigates the effects on educational outcomes of a sudden changein the language of instruction in Rwanda from one non-native language (French) to another(English). This change, enacted in 2008, affected all levels of education without any languagetraining for students or teachers. Using a synthetic control approach, we find that the changein the language of instruction led to an immediate and persistent decline in primary schoolenrollment and to a decline in secondary school enrollment at the time when the first cohortfully affected by the policy would have started secondary school. As a direct consequenceof lower school enrollment, we also find a reduction in the expected number of years ofschooling. We provide suggestive evidence that the decline in enrollment is due to higherdrop-out rates, mostly among students who are old-for-grade. Given that more education is generally associated with delayed marriage and fertility, thethird chapter explores the causal link between a policy that plausibly reduced educationalattainment in Rwanda and teen fertility. In 2008, Rwanda suddenly changed the languageof instruction at all levels of schooling from French to English, despite the fact that avery small fraction of the population was able to speak English. This led to a significantreduction in primary and secondary school enrollment. Using both a synthetic controlapproach with aggregate data and a difference-in-differences approach with individual-leveldata, we document an increase in teen fertility as a result of this policy. Although moreimprecise, our results also suggest that the effects are larger in urban areas. While better health has been shown to lead to more education, both in the developedand in the developing world, the causal link from education to health is less clear. Thus,the fourth chapter of this dissertation examines the role of an education policy that presumably reduced educational attainment on the health of young adults affected by the policyand on children born after the intervention. Using aggregate data and a synthetic controlmethod, we find limited evidence of an effect on the health of adults and some evidence ofeffects on the health of children. Our analyses help better understand the determinants ofthe intergenerational transmission of poverty and inequality in the context of a developingcountry.",

keywords = "Language Change, Language of Instruction, Education Policy, Educational outcomes, Teen fertility, Intergenerational transmission of Health, Synthetic Control Methods",

author = "Munsi, \{Dieudonn{\'e} Tamfutu\}",

year = "2023",

month = jun,

day = "27",

doi = "10.21996/wnfv-vq65",

language = "English",

publisher = "Syddansk Universitet. Det Samfundsvidenskabelige Fakultet",

school = "SDU",

}

TY - GEN

T1 - Language of Instruction as Education Policyand Its Effects on the Socio-economic Outcomes of Young People

AU - Munsi, Dieudonné Tamfutu

PY - 2023/6/27

Y1 - 2023/6/27

N2 - This dissertation consists of four closely related papers. The first chapter describesthe original version of the synthetic control method, the conditions under which it can beconfidently used, and some of the most important recent developments on the topic. Icover extensions of the original method to situations when donor units can contribute to thecounterfactual with negative weights or with weights that do not sum up to one, as well asvarious regression-based modifications that attempt to reduce or eliminate bias. I concludewith a brief discussion of recent approaches to inference in the context of synthetic controls. The second chapter investigates the effects on educational outcomes of a sudden changein the language of instruction in Rwanda from one non-native language (French) to another(English). This change, enacted in 2008, affected all levels of education without any languagetraining for students or teachers. Using a synthetic control approach, we find that the changein the language of instruction led to an immediate and persistent decline in primary schoolenrollment and to a decline in secondary school enrollment at the time when the first cohortfully affected by the policy would have started secondary school. As a direct consequenceof lower school enrollment, we also find a reduction in the expected number of years ofschooling. We provide suggestive evidence that the decline in enrollment is due to higherdrop-out rates, mostly among students who are old-for-grade. Given that more education is generally associated with delayed marriage and fertility, thethird chapter explores the causal link between a policy that plausibly reduced educationalattainment in Rwanda and teen fertility. In 2008, Rwanda suddenly changed the languageof instruction at all levels of schooling from French to English, despite the fact that avery small fraction of the population was able to speak English. This led to a significantreduction in primary and secondary school enrollment. Using both a synthetic controlapproach with aggregate data and a difference-in-differences approach with individual-leveldata, we document an increase in teen fertility as a result of this policy. Although moreimprecise, our results also suggest that the effects are larger in urban areas. While better health has been shown to lead to more education, both in the developedand in the developing world, the causal link from education to health is less clear. Thus,the fourth chapter of this dissertation examines the role of an education policy that presumably reduced educational attainment on the health of young adults affected by the policyand on children born after the intervention. Using aggregate data and a synthetic controlmethod, we find limited evidence of an effect on the health of adults and some evidence ofeffects on the health of children. Our analyses help better understand the determinants ofthe intergenerational transmission of poverty and inequality in the context of a developingcountry.

AB - This dissertation consists of four closely related papers. The first chapter describesthe original version of the synthetic control method, the conditions under which it can beconfidently used, and some of the most important recent developments on the topic. Icover extensions of the original method to situations when donor units can contribute to thecounterfactual with negative weights or with weights that do not sum up to one, as well asvarious regression-based modifications that attempt to reduce or eliminate bias. I concludewith a brief discussion of recent approaches to inference in the context of synthetic controls. The second chapter investigates the effects on educational outcomes of a sudden changein the language of instruction in Rwanda from one non-native language (French) to another(English). This change, enacted in 2008, affected all levels of education without any languagetraining for students or teachers. Using a synthetic control approach, we find that the changein the language of instruction led to an immediate and persistent decline in primary schoolenrollment and to a decline in secondary school enrollment at the time when the first cohortfully affected by the policy would have started secondary school. As a direct consequenceof lower school enrollment, we also find a reduction in the expected number of years ofschooling. We provide suggestive evidence that the decline in enrollment is due to higherdrop-out rates, mostly among students who are old-for-grade. Given that more education is generally associated with delayed marriage and fertility, thethird chapter explores the causal link between a policy that plausibly reduced educationalattainment in Rwanda and teen fertility. In 2008, Rwanda suddenly changed the languageof instruction at all levels of schooling from French to English, despite the fact that avery small fraction of the population was able to speak English. This led to a significantreduction in primary and secondary school enrollment. Using both a synthetic controlapproach with aggregate data and a difference-in-differences approach with individual-leveldata, we document an increase in teen fertility as a result of this policy. Although moreimprecise, our results also suggest that the effects are larger in urban areas. While better health has been shown to lead to more education, both in the developedand in the developing world, the causal link from education to health is less clear. Thus,the fourth chapter of this dissertation examines the role of an education policy that presumably reduced educational attainment on the health of young adults affected by the policyand on children born after the intervention. Using aggregate data and a synthetic controlmethod, we find limited evidence of an effect on the health of adults and some evidence ofeffects on the health of children. Our analyses help better understand the determinants ofthe intergenerational transmission of poverty and inequality in the context of a developingcountry.

KW - Language Change

KW - Language of Instruction

KW - Education Policy

KW - Educational outcomes

KW - Teen fertility

KW - Intergenerational transmission of Health

KW - Synthetic Control Methods

U2 - 10.21996/wnfv-vq65

DO - 10.21996/wnfv-vq65

M3 - Ph.D. thesis

PB - Syddansk Universitet. Det Samfundsvidenskabelige Fakultet

ER -

Language of Instruction as Education Policyand Its Effects on the Socio-economic Outcomes of Young People

Abstract

Dokumenter og links

SDU link

Fingeraftryk

Citationsformater