Introduction Theory Empirics Conclusion Appendix Ethnolinguistic Diversity: Origins and Implications Stelios Michalopoulos Department of Economics Brown University February 27, 2008 University of Houston - Department of Economics Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
129
Embed
Ethnolinguistic Diversity: Origins and Implicationsdevollra/Houston.pdfTheory Empirics Conclusion Appendix Ethnolinguistic Diversity: Origins and Implications Stelios Michalopoulos
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
IntroductionTheoryEmpirics
ConclusionAppendix
Ethnolinguistic Diversity: Origins and Implications
Stelios Michalopoulos
Department of EconomicsBrown University
February 27, 2008
University of Houston - Department of Economics
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Objectives
Provide a theoretical and empirical framework for understanding the
Economic origins of ethnic diversity
Its implications for comparative economic development today
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Objectives
Provide a theoretical and empirical framework for understanding the
Economic origins of ethnic diversity
Its implications for comparative economic development today
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Key Finding
This research shows that:
Diversity in land quality across regions
=)
to the formation and persistence of ethnic diversity
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Additional Findings
Contemporary ethnic diversity displays:
1 A natural component driven by diversity in land quality acrossregions
2 A Man-made component driven by state history
e.g. the European colonization, early economic development etc.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Additional Findings
Contemporary ethnic diversity displays:
1 A natural component driven by diversity in land quality acrossregions
2 A Man-made component driven by state history
e.g. the European colonization, early economic development etc.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Why A Theory of Ethnic Diversity?
Long standing subject of research in the realm of socialsciences
Economists have studied:
the impact of ethnic diversity on economic outcomes
the economic performance of ethnic groups themselves
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Previous Research
Ethnic divisions and economic performance across countries
Easterly, Levine (1997)
Fractionalization and public good provision, civil con�ict etc.
Fearon, Alesina et al (2003), Banerjee el al (2006) among others
Inequality across ethnic groups
Loury (1977), Esteban and Ray (2007)
Optimal State formation
Alesina, Spolaore (1997), Alesina et al (2006)
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Previous Research
Ethnic divisions and economic performance across countries
Easterly, Levine (1997)
Fractionalization and public good provision, civil con�ict etc.
Fearon, Alesina et al (2003), Banerjee el al (2006) among others
Inequality across ethnic groups
Loury (1977), Esteban and Ray (2007)
Optimal State formation
Alesina, Spolaore (1997), Alesina et al (2006)
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Interpretation?
However,
the economic origins of ethnic diversity are poorly understood,both empirically and theoretically
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Forces Behind Ethnic Diversity
Ethnic diversity has deep roots in:
geographic variability
a country�s state history
State history itself may have an independent e¤ect on economicoutcomes
Identifying the foundations of ethnic diversity will improve onthe interpretation of existing �ndings
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Forces Behind Ethnic Diversity
Ethnic diversity has deep roots in:
geographic variability
a country�s state history
State history itself may have an independent e¤ect on economicoutcomes
Identifying the foundations of ethnic diversity will improve onthe interpretation of existing �ndings
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Forces Behind Ethnic Diversity
Ethnic diversity has deep roots in:
geographic variability
a country�s state history
State history itself may have an independent e¤ect on economicoutcomes
Identifying the foundations of ethnic diversity will improve onthe interpretation of existing �ndings
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Idea
Diversity in land quality across regions =) to the formation andpersistence of ethnic diversity
The main elements:
1 Variation in regional land quality )region speci�c human capital
2 Di¤erences in region speci�c human capital )barrier to population mixing
3 Limited population mixing between regions )emergence of di¤erential ethnic traits
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Idea
Diversity in land quality across regions =) to the formation andpersistence of ethnic diversity
The main elements:
1 Variation in regional land quality )region speci�c human capital
2 Di¤erences in region speci�c human capital )barrier to population mixing
3 Limited population mixing between regions )emergence of di¤erential ethnic traits
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Idea
Diversity in land quality across regions =) to the formation andpersistence of ethnic diversity
The main elements:
1 Variation in regional land quality )region speci�c human capital
2 Di¤erences in region speci�c human capital )barrier to population mixing
3 Limited population mixing between regions )emergence of di¤erential ethnic traits
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
The Idea
Diversity in land quality across regions =) to the formation andpersistence of ethnic diversity
The main elements:
1 Variation in regional land quality )region speci�c human capital
2 Di¤erences in region speci�c human capital )barrier to population mixing
3 Limited population mixing between regions )emergence of di¤erential ethnic traits
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Herders
Farmers
AIsland
QualityLand Farmers
Farmers
BIsland
Herders
Farmers
aEthnicity
bEthnicity
AIsland
QualityLand
DiversityEthnic
Farmers
Farmers
BIsland
Herders
Farmers
aEthnicity
bEthnicity
Farmers
Farmers
EthnicityOne
BIsland AIsland
QualityLand
DiversityEthnic
Variation inLand Quality
Linguistic Diversity
Variation inLand Quality
Linguistic Diversity
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Population Movements and the Spread of Languages
(Pre)Historical Evidence
Families of languages spread over areas which wereproductively relevant for the people speaking these languages
Nomad pastoralists spread where herding was economically optimal
Spread of Altaic language family (Turkish and Mongolian) over theGrasslands of Central Eurasia
Farmers expanded towards regions suitable for agriculture.
Indo-European languages spread from Anatolia, Nature (2003)
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Population Movements and the Spread of Languages
(Pre)Historical Evidence
Families of languages spread over areas which wereproductively relevant for the people speaking these languages
Nomad pastoralists spread where herding was economically optimal
Spread of Altaic language family (Turkish and Mongolian) over theGrasslands of Central Eurasia
Farmers expanded towards regions suitable for agriculture.
Indo-European languages spread from Anatolia, Nature (2003)
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Do humans behave like Darwin�s �nches?
Darwin�s �nches and ethnicities:
an ecological niche ) certain shape of beak
ecologically diverse places ) diversity within �nches
a level of land quality ) speci�c human capital
variation in land quality ) ethnic diversity
Channel: Di¤erences in region speci�c human capital)limits population mixing) increasing ethnic diversity.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Do humans behave like Darwin�s �nches?
Darwin�s �nches and ethnicities:
an ecological niche ) certain shape of beak
ecologically diverse places ) diversity within �nches
a level of land quality ) speci�c human capital
variation in land quality ) ethnic diversity
Channel: Di¤erences in region speci�c human capital)limits population mixing) increasing ethnic diversity.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
ObjectiveFindingsEthnic Diversity a focal topicThe IdeaA Parallel
Do humans behave like Darwin�s �nches?
Darwin�s �nches and ethnicities:
an ecological niche ) certain shape of beak
ecologically diverse places ) diversity within �nches
a level of land quality ) speci�c human capital
variation in land quality ) ethnic diversity
Channel: Di¤erences in region speci�c human capital)limits population mixing) increasing ethnic diversity.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
The Basic Structure of the Model
Overlapping-generations economy
t = 0; 1; 2; :::
Single Good
Two regions i and j
Production Inputs: Land, Labor, Region-Speci�c Technology
No population growth
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
The Basic Structure of the Model
Overlapping-generations economy
t = 0; 1; 2; :::
Single Good
Two regions i and j
Production Inputs: Land, Labor, Region-Speci�c Technology
No population growth
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
The Production of Final Output
Output per worker in region i , y it , is
y it =�z ith
it
� �X=Lit
�1��; � 2 (0; 1)
z it � temporary productivity shock
hit � region i speci�c human capital
Lit � labor employed
X � land
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
Households
Individuals are identical and live two-periods
1st Period �Accumulate passively the region speci�c humancapital they are born to.
2nd Period � Inelastically supply the unit time endowment aslabor and consume the earnings
Agents maximize income by working in either region.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
Households
Individuals are identical and live two-periods
1st Period �Accumulate passively the region speci�c humancapital they are born to.
2nd Period � Inelastically supply the unit time endowment aslabor and consume the earnings
Agents maximize income by working in either region.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
Households
Individuals are identical and live two-periods
1st Period �Accumulate passively the region speci�c humancapital they are born to.
2nd Period � Inelastically supply the unit time endowment aslabor and consume the earnings
Agents maximize income by working in either region.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
The Forces at Work
1 Regional productivity shocks generate an incentive to relocate,�t = z it=z
jt :
2 Relocation entails erosion of knowledge due to di¤erences inregion speci�c human capital, ".
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
The Forces at Work
1 Regional productivity shocks generate an incentive to relocate,�t = z it=z
jt :
2 Relocation entails erosion of knowledge due to di¤erences inregion speci�c human capital, ".
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
Formation of Common Ethnicity
Conjecture:
the probability, fT ; that individuals from regions i and j sharecommon traits in period T increases in the intensity ofpopulation mixing between the two regions over time
fT =
TPt=1It
T
It =�1 if migration occurs in period t.0 if otherwise
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
Properties of Migration
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
t λ
s λ
ε1
AreaMigration No
j i MM
i j MM
0>→ i j M
0>→ j i M
0
Stylianos
Text Box
λt: ratio of the regional productivity shocks in period t ε: erosion of regional human capital in case or migration
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
P1 - The Formation of Common Ethnic Traits
1. The probability, fT ; that regions i and j share common ethnictraits in period T increases in the variance of relativeproductivity shock, �t ;
@fT@var (�t)
> 0
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
The Basic StructureHouseholdsLabor Allocation Across RegionsThe Properties of MigrationThe Formation of Common Ethnicity Over Time
P2 - The Formation of Common Ethnic Traits
2. The probability, fT ; that regions i and j share common ethnictraits in period T decreases in the size of the erosion, ";
@fT@"
< 0
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Ideal Data
Need an index of the transferability of region speci�c humancapital, ";
Ideally:
regional distributions of productive activities in a period ofhuman history when the formation of cultural traits was takingplace
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
The Data
Disaggregated level data on the suitability of land foragriculture
Each observation is between 0 and 1, i.e. probability that aparticular grid cell may be cultivated
a) Monthly climate conditions for 1961�1990:
temperature
precipitation
potential sunshine hours.
b) Soil characteristics:
total organic content (soil carbon density)
the nutrient availability.(soil pH)
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
The Data
Disaggregated level data on the suitability of land foragriculture
Each observation is between 0 and 1, i.e. probability that aparticular grid cell may be cultivated
a) Monthly climate conditions for 1961�1990:
temperature
precipitation
potential sunshine hours.
b) Soil characteristics:
total organic content (soil carbon density)
the nutrient availability.(soil pH)
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Data Coverage
Global coverage
The resolution is 0.5 degrees, latitude x longitude, averagesize about 55km by 35km.
58920 observations
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Global land qualityHigh : 0.999
Low : 0.000
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Distribution of land quality across countries: An example0
24
6D
en
sity
0 .2 5 .5 .7 5 1Qu a lity of La nd
Kerne l d e ns ity es tim a tekde ns ity va lu eRed line: Nepal; Blue line: Greece
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Cross-"Arti�cial Country" Analysis
Unit of analysis: arti�cial countries of 4 degrees latitude by 4degrees longitude
Median land quality observations per arti�cial country: 53
Dependent variable: number of languages spoken today
Source: 15th edition of the Ethnologue linguistics database
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Cross-"Arti�cial Country" Analysis
Unit of analysis: arti�cial countries of 4 degrees latitude by 4degrees longitude
Median land quality observations per arti�cial country: 53
Dependent variable: number of languages spoken today
Source: 15th edition of the Ethnologue linguistics database
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Global land qualityHigh : 0.999
Low : 0.000
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72 Number of Languages spoken today in this area: 12
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Arti�cial Countries
Focus on Arti�cial Countries that:
have at least 3 regional observations
each region has population density > 1 person per sq. km
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Results
The support of the distribution of land quality, rangei ;
captures how di¢ cult it is to transfer regional knowledge
Prediction:
Larger spectrum of land qualities ) higher ethnic diversity
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 1a: Summary Statistics for the Cross-Artificial Country Analysis
max 30.00 1.00 0.98 2.17 196.98 1.00 8.00 19.45 2.40
min 1.00 0.00 0.00 0.00 0.61 0.00 1.00 0.00 0.00#_lang : number of languages spoken with at least 1% area coverage within an artificial country;range : spectrum of land qualities within an "artificial country", i.e. the difference in land quality between the region with the highest land quality from that with the lowest; avg : is the average land quality within artificial country;elev_sd : standard deviation of elevation measured in kilometers within artificial country; in_cntry: dummy equals 1 if artificial country's falls as a whole into a single real country; areakm2 : size of each artificial country in thousands of sq. km.#_cntry: number of real countries an artificial country falls into; water_area: area in thousand's of sq km of artificial countryunder water i.e. river or lake; sea_dist: distance of the centroid of an artificial country from the coastline;Data Sources: See Appendix D
Table 1b: The Correlation Matrix for the Cross-Artificial Country Analysis
#_lang range avg elev_sd areakm2 in_country #_cntry water_area sea_dist#_lang 1
sea_dist 0.02 0.25 -0.11 0.09 0.42 -0.20 0.13 0.26 1#_lang : number of languages spoken with at least 1% area coverage within an artificial country;range : spectrum of land qualities within an "artificial country", i.e. the difference in land quality between the region with the highest land quality from that with the lowest; avg : is the average land quality within artificial country;elev_sd : standard deviation of elevation measured in kilometers within artificial country; in_cntry: dummy equals 1 if artificial country's falls as a whole into a single real country; areakm2 : size of each artificial country in thousands of sq. km.#_cntry: number of real countries an artificial country falls into; water_area: area in thousand's of sq km of artificial countryunder water i.e. river or lake; sea_dist: distance of the centroid of an artificial country from the coastline;Data Sources: See Appendix D
Table 2: Main Specification and Robustness in Cross-Artificial Country Regressions
Dependent Variable: Log Number of Languages Spoken
OLS
Baseline
(1)
range 0.952(4.66)***
Observations 548R-squared 0.09OLS regressions with absolute value of t statistics in parentheses;Standard errors are corrected for spatial correlation following Conley (1999);* significant at 10%; ** significant at 5%; *** significant at 1%;
Table 2: Main Specification and Robustness in Cross-Artificial Country Regressions
Dependent Variable: Log Number of Languages Spoken
OLS OLS
BaselineGeographical Controls
(1) (2)
range 0.952 0.488(4.66)*** (2.27)**
avg -0.104(0.50)
areakm2 0.002(1.48)
abs_lat -0.027(5.95)***
elev_sd 0.498(2.18)**
in_country -0.172(1.98)**
#_cntry 0.170(5.33)
sea_dist 0.068(0.64)
water_area 0.017(1.15)
Observations 548 548R-squared 0.09 0.52OLS regressions with absolute value of t statistics in parentheses;Standard errors are corrected for spatial correlation following Conley (1999);* significant at 10%; ** significant at 5%; *** significant at 1%;
Table 2: Main Specification and Robustness in Cross-Artificial Country Regressions
Dependent Variable: Log Number of Languages Spoken
OLS OLS OLS
BaselineGeographical Controls
Continental and Country fixed effects
(1) (2) (3)
range 0.952 0.488 0.524(4.66)*** (2.27)** (3.88)***
avg -0.104 -0.184(0.50) (1.13)
areakm2 0.002 0.002(1.48) (1.83)*
abs_lat -0.027 -0.026(5.95)*** (3.17)***
elev_sd 0.498 0.684(2.18)** (3.07)***
in_country -0.172 -0.094(1.98)** (0.95)
#_cntry 0.170 0.161(5.33) (2.47)**
sea_dist 0.068 -0.038(0.64) (0.42)
water_area 0.017 -0.009(1.15) (0.87)
Observations 548 548 548R-squared 0.09 0.52 0.81OLS regressions with absolute value of t statistics in parentheses;Standard errors are corrected for spatial correlation following Conley (1999);* significant at 10%; ** significant at 5%; *** significant at 1%;Specification (3) includes country and continental fixed effects;
Table 2: Main Specification and Robustness in Cross-Artificial Country Regressions
Dependent Variable: Log Number of Languages Spoken
OLS OLS OLS OLS
BaselineGeographical Controls
Continental and Country fixed effects
Average Land Quality and Interaction
(1) (2) (3) (4)
range 0.952 0.488 0.524 0.545(4.66)*** (2.27)** (3.88)*** (3.03)***
Observations 548 548 548 548R-squared 0.09 0.52 0.81 0.81OLS regressions with absolute value of t statistics in parentheses;Standard errors are corrected for spatial correlation following Conley (1999);* significant at 10%; ** significant at 5%; *** significant at 1%;Specifications (3) and (4) include country and continental fixed effects;
Table 2: Main Specification and Robustness in Cross-Artificial Country Regressions
Dependent Variable: Log Number of Languages Spoken
OLS OLS OLS OLS OLS
BaselineGeographical Controls
Continental and Country fixed effects
Average Land Quality and Interaction
Regions Within Countries
(1) (2) (3) (4) (5)
range 0.952 0.488 0.524 0.545 0.460(4.66)*** (2.27)** (3.88)*** (3.03)*** (2.85)***
Observations 548 548 548 548 263R-squared 0.09 0.52 0.81 0.81 0.78OLS regressions with absolute value of t statistics in parentheses;Standard errors are corrected for spatial correlation following Conley (1999);* significant at 10%; ** significant at 5%; *** significant at 1%;Specifications (3), (4) and (5) include country and continental fixed effects;
Table 2: Main Specification and Robustness in Cross-Artificial Country Regressions
Dependent Variable: Log Number of Languages Spoken
Observations 548 548 548 548 263 548R-squared 0.09 0.52 0.81 0.81 0.78 0.82OLS regressions with absolute value of t statistics in parentheses;Standard errors are corrected for spatial correlation following Conley (1999);* significant at 10%; ** significant at 5%; *** significant at 1%;Specifications (3), (4), (5) and (6) include country and continental fixed effects;
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Pairwise Analysis of Contiguous Regions
Unit of analysis: Pairs of adjacent regions
average distance within a regional pair: 50 km.
Total number of unique regional pairs 159358
Index of Ethnic Similarity: Percentage of common languagesspoken within a pair
i.e. number of common languages spoken divided by the total number
within a regional pair
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Pairwise Analysis of Contiguous Regions
Unit of analysis: Pairs of adjacent regions
average distance within a regional pair: 50 km.
Total number of unique regional pairs 159358
Index of Ethnic Similarity: Percentage of common languagesspoken within a pair
i.e. number of common languages spoken divided by the total number
within a regional pair
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Regional land quality
0.00 - 0 .15
0.16 - 0 .30
0.31 - 0 .45
0.46 - 0 .72
Stylianos
Text Box
The blue arrows point to the neighbors of a region. In this case 8.
Stylianos
Line
Stylianos
Line
Stylianos
Line
Stylianos
Line
Stylianos
Line
Stylianos
Line
Stylianos
Line
Stylianos
Line
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Predictions
The di¤erence in land quality within a pair, lqdi¤ij ; captureshow di¢ cult it is to transfer regional knowledge
Prediction:
Larger di¤erence in land quality ) Lower ethnic similarity
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 3a: Summary Statistics for the Pairwise Analysis of Adjacent Regions
min 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00pct_comlang : number of common languages divided by the total number of unique languages spoken within a pairof adjacent regions; lqdiff : absolute difference in land quality within a pair of adjacent regions; eldiff : absolutedifference in elevation in km's within a pair of adjacent regions; sea_dist: distance from the coastline in 1000s of km of a regional pair; same_country: dummy equals 1 if a regional pair belongs to the same country; #_body_waters: numberof distinct body waters within a regional pair; #_lang_pair: total number of unique languages spoken within a pair;diff_langarea : difference in area of linguistic coverage between the regional neighbors in 1000 of sq kilometers;
Table 3b: The Correlation Matrix for the Pairwise Analysis of Adjacent Regions
diff_langarea -0.09 0.00 0.08 0.03 -0.06 -0.07 -0.02 1.00pct_comlang : number of common languages divided by the total number of unique languages spoken within a pairof adjacent regions; lqdiff : absolute difference in land quality within a pair of adjacent regions; eldiff : absolutedifference in elevation in km's within a pair of adjacent regions; sea_dist: distance from the coastline in 1000s of km of a regional pair; same_country: dummy equals 1 if a regional pair belongs to the same country; #_body_waters: numberof distinct body waters within a regional pair; #_lang_pair: total number of unique languages spoken within a pair;diff_langarea : difference in area of linguistic coverage between the regional neighbors in 1000 of sq kilometers;
Table 4: Main Specification and Robustness in the Pairwise Analysis of Adjacent Regions
Dependent Variable: Percentage of Common Languages
OLS
Baseline
(1)
lqdiff -0.163(7.13)***
eldiff -0.120(9.48)***
abs_lat 0.005(20.20)***
elev 0.002(0.37)
land_quality -0.029(1.80)*
sea_dist 0.019(2.51)**
same_country 0.132(14.96)***
#_body_waters -0.006(0.15)
diff_langarea -0.013(2.41)**
constant 0.475(27.05)
R-squared 0.02Observations 159358OLS regressions with absolute value of t statistics in parentheses.Standard errors are corrected for spatial correlation following Conley (1999)* significant at 10%; ** significant at 5%; *** significant at 1%
Table 4: Main Specification and Robustness in the Pairwise Analysis of Adjacent Regions
Dependent Variable: Percentage of Common Languages
OLS OLS
Baseline
Country and Continental fixed effects
(1) (2)
lqdiff -0.163 -0.122(7.13)*** (15.66)***
eldiff -0.120 0.095(9.48)*** (22.29)***
abs_lat 0.005 0.003(20.20)*** (16.70)***
elev 0.002 0.001(0.37) (0.50)
land_quality -0.029 -0.027(1.80)* (5.97)***
sea_dist 0.019 0.011(2.51)** (4.83)***
same_country 0.132 0.108(14.96)*** (30.37)***
#_body_waters -0.006 0.0002(0.15) (1.95)*
diff_langarea -0.013 -0.009(2.41)** (4.35)***
constant 0.475 0.368(27.05) (10.97)
R-squared 0.18 0.28Observations 159358 159358OLS regressions with absolute value of t statistics in parentheses; In (1), (3) standard errors are correctedfor spatial autocorrelation following Conley (1999); In (2) standard errors are clustered at the level of eachindividual region; * significant at 10%; ** significant at 5%; *** significant at 1%;(2) includes both continental and country and fixed effects for each region within a pair;
Table 4: Main Specification and Robustness in the Pairwise Analysis of Adjacent Regions
Dependent Variable: Percentage of Common Languages
R-squared 0.18 0.28 0.21Observations 159358 159358 4138OLS regressions with absolute value of t statistics in parentheses; In (1), (3) standard errors are correctedfor spatial autocorrelation following Conley (1999); In (2) standard errors are clustered at the level of eachindividual region; * significant at 10%; ** significant at 5%; *** significant at 1%;(2) includes both continental and country and fixed effects for each region within a pair;
Table 4: Main Specification and Robustness in the Pairwise Analysis of Adjacent Regions
Dependent Variable: Percentage of Common Languages
R-squared 0.18 0.28 0.21 0.29Observations 159358 159358 4138 159358OLS regressions with absolute value of t statistics in parentheses; In (1), (3) standard errors are correctedfor spatial autocorrelation following Conley (1999); In (2) and (4) standard errors are clustered at the level of eachindividual region; * significant at 10%; ** significant at 5%; *** significant at 1%;(2), (4) include both continental and country and fixed effects for each region within a pair; (5) allows for differentialmarginal effect of the difference in land quality for pairs of different continents; Africa is the omitted continent;pct_comlang : number of common languages divided by the total number of unique languages spoken within a pairof adjacent regions; lqdiff : absolute difference in land quality within a pair of adjacent regions; eldiff : absolutedifference in elevation in km's within a pair of adjacent regions; sea_dist: distance from the coastline in 1000s of km of a regional pair; same_country: dummy equals 1 if a regional pair belongs to the same country; #_body_waters: numberof distinct body waters within a regional pair;diff_langarea : difference in area of linguistic coverage between the regional neighbors in 1000 of sq kilometers;
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Cross-Real Country Analysis
Dependent variable: ethnolinguistic fractionalization (ELF )come from Fearon and Laitin (2003),
an augmented version of the original ELF measure from the Atlas
Narodov Mira (1964).
This index represents for each country the probability that twoindividuals randomly drawn from the overall population belongto di¤erent ethnolinguistic groups.
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Why the distribution of population across regions matters?
Expected fractionalization, E (ELF ); for a pair of places:
E (ELF ) = (1� fT ) 1�
�LiT
LjT+LiT
�2��
LjTLjT+L
jT
�2!# #
Prob. regions i ; j Probability individuals
di¤erent traits drawn from regions i ; j
Unequal distribution of land quality drives inequality in theregional population densities.
Correlation between the gini of land quality and the gini of regional
population density as of 1990 is 0:59
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
Why the distribution of population across regions matters?
Expected fractionalization, E (ELF ); for a pair of places:
E (ELF ) = (1� fT ) 1�
�LiT
LjT+LiT
�2��
LjTLjT+L
jT
�2!# #
Prob. regions i ; j Probability individuals
di¤erent traits drawn from regions i ; j
Unequal distribution of land quality drives inequality in theregional population densities.
Correlation between the gini of land quality and the gini of regional
population density as of 1990 is 0:59
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 6: Main Specification for Cross-Country Analysis
Adj. R-squared 0.04 0.11Observations 147 147Absolute value of t statistics in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%
Table 6: Main Specification for Cross-Country Analysis
range 0.203 0.455 0.889(2.43)** (3.55)*** (5.73)***
avg 0.076 -0.283(0.42) (1.49)
avgxrange -0.659 -1.21(2.18)** (3.90)***
lqgini -0.874(4.45)***
Adj. R-squared 0.03 0.11 0.21Observations 147 147 147Absolute value of t statistics in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%
Adj. R-squared 0.21 0.42Observations 147 147Absolute value of t statistics in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%
Adj. R-squared 0.21 0.38Observations 147 146Absolute value of t statistics in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%
Adj. R-squared 0.21 0.50Observations 147 143Absolute value of t statistics in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%Continental fixed effects and other geographical characteristics are included
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
The Role of the Colonizers
Historical evidence:
Introduced own ethnicities, manipulated existing ones
=)
distribution of land quality should explain less of the variationin contemporary ethnic diversity in the colonized world
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Observations 93 47Absolute value of t statistics in parentheses* significant at 10%; ** significant at 5%; *** significant at 1%
IntroductionTheoryEmpirics
ConclusionAppendix
Data SourcesCross-Arti�cial Country ResultsPairwise Analysis of Adjacent RegionsCross-Real Country AnalysisThe Colonizing Experience
The Role of the Colonizers
Historical evidence:
Drew borders in an arbitrary manner
=)
distribution of land quality should itself be conduciveto higher levels of fractionalization in colonized world
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 9: Colonization and Natural Fractionalization
nat_ELF
colonized 0.40non-colonized 0.35
Pr(T < t) = 0.04
IntroductionTheoryEmpirics
ConclusionAppendix
Main Contribution
Theoretical insight:
Ethnic groups are bearers of speci�c human capital
Empirical Results
Decompose contemporary ethnic diversity into:
1 Natural component induced by the underlying diversity in landqualities
2 Man-made component re�ecting divergent state histories
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Main Contribution
Theoretical insight:
Ethnic groups are bearers of speci�c human capital
Empirical Results
Decompose contemporary ethnic diversity into:
1 Natural component induced by the underlying diversity in landqualities
2 Man-made component re�ecting divergent state histories
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Applications - Preliminary Findings
Applications:
1. Causal impact of ethnolinguistic diversity on comparativeeconomic development
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: log income per capita,
2002
OLS
(1)
ELF -1.045(0.29)***
avg -0.729(0.42)*
avgxrange 1.299(0.59)**
abs_lat 0.043(0.006)***
eucolony 0.237(0.22)
Observations 137R-squared 0.56Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: log income per capita,
2002
OLS 2SLS
(1) (2)
ELF -1.045 -0.147(0.29)*** (0.77)
avg -0.729 -0.364(0.42)* (0.53)
avgxrange 1.299 1.113(0.59)** (0.63)*
abs_lat 0.043 0.051(0.006)*** (0.009)***
eucolony 0.237 0.331(0.22) (0.24)
F-test of excluded instruments
9.97***
Sargan statistic (p-value)
0.55
Observations 137 137R-squared 0.56Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
Table 11: Instrumental Variable Regressions for Quality of Life Variables
Dep. Var: Access to Clean water 2000
OLS
(1)
ELF -12.204(5.50)**
avg 5.127(8.99)
avgxrange 10.88(12.36)
abs_lat 0.645(0.13)***
eucolony 5.055(4.64)
Observations 134R-squared 0.38Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
Table 11: Instrumental Variable Regressions for Quality of Life Variables
Dep. Var: Access to Clean water 2000
OLS 2SLS
(1) (2)
ELF -12.204 5.279(5.50)** (16.00)
avg 5.127 12.302(8.99) (10.98)
avgxrange 10.88 7.538(12.36) (13.00)
abs_lat 0.645 0.816(0.13)*** (0.20)***
eucolony 5.055 8.097(4.64) (5.40)
F-test of excluded instruments
8.64***
Sargan statistic (p-value)
0.76
Observations 134 134R-squared 0.38Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
IntroductionTheoryEmpirics
ConclusionAppendix
Applications - Future Research
Applications:
2. Inequality across ethnic groups
natural component:
ethnic speci�c human capital di¤erentially complementary with newtechnologies
man-made component:
di¤erential treatment of ethnic groups by the institutions in place
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Applications - Future Research
Applications:
3. The di¤usion of development across countries
genetic distance may proxy for di¤erences in the distribution of land
endowments, thus, di¤erences in society speci�c human capital
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Applications - Future Research
Applications:
4. Geographic Variability: Ethnic and State Formation.
homogeneous land endowments ) common need for a public good )incentive to politically centralize
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
AGO
BDIBEN
BFA
BWA
CAFCIV
CMR
CODCOG
ETH
GAB
GHA
GIN
KENLBR
LSO
MDG
MLI
MOZ
MRT
MWI
NAMNER
NGA
RWA
SDN
SEN
SLESOM
TCD
TGO
TZAUGA
ZMB
ZWE-.4
-.20
.2.4
Cen
traliz
atio
n
-.4 -.2 0 .2 .4Variation in Land Quality
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Soil ph and Precipitation
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Example of Language Divergence
Divergence Within the Indo-European Language Group (Science 2003)
7000 years BP: Greek and Armenian diverge
5000 years BP: Italic, Germanic, Celtic, Indo-Iranian familiesdiverge
1750 years BP: Germanic languages split between WestGermanic and North Germanic
North Germanic: Danish and Swedish
West Germanic: German, Dutch, English
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Example of Language Divergence
Divergence Within the Indo-European Language Group (Science 2003)
7000 years BP: Greek and Armenian diverge
5000 years BP: Italic, Germanic, Celtic, Indo-Iranian familiesdiverge
1750 years BP: Germanic languages split between WestGermanic and North Germanic
North Germanic: Danish and Swedish
West Germanic: German, Dutch, English
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Example of Language Divergence
Divergence Within the Indo-European Language Group (Science 2003)
7000 years BP: Greek and Armenian diverge
5000 years BP: Italic, Germanic, Celtic, Indo-Iranian familiesdiverge
1750 years BP: Germanic languages split between WestGermanic and North Germanic
North Germanic: Danish and Swedish
West Germanic: German, Dutch, English
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Example of Language Divergence
Divergence Within the Indo-European Language Group (Science 2003)
7000 years BP: Greek and Armenian diverge
5000 years BP: Italic, Germanic, Celtic, Indo-Iranian familiesdiverge
1750 years BP: Germanic languages split between WestGermanic and North Germanic
North Germanic: Danish and Swedish
West Germanic: German, Dutch, English
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Arti�cial Country Scatterplot
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
-2-1
01
2Lo
g La
ngua
ges
-1 -.5 0 .5Spectrum of Land Quality
Partial Scatter plot of Spectrum of land quality and log number of languages within “Artificial Countries” controlling for variation in elevation, average land quality and abs latitude.The red dots represent African “Artificial Countries”.
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Real Country Scatterplot
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Net relationship between range and ELF:Cross-Country Analysis
AFG
AGO
ALB
ARE
ARGARM
AUS
AUT
AZEBDI
BEL
BEN
BFA
BGDBGR
BIH
BLR
BOL
BRA
BTN
BWA
CAF
CAN
CHECHLCHN
CIVCMRCOD
COG
COL
CRICUB
CZE
DEU
DJI
DNK
DOMDZA
ECU
EGYERI
ESPEST
ETH
FINFRA
GAB GBRGEO
GHA
GIN
GNBGRC
GTM
GUY
HNDHRV
HTIHUN
IDN
IND
IRL
IRN
IRQ
ISL
ISR
ITAJORJPN
KAZ
KEN
KGZ
KHM
KOR
KW T
LAO
LBR
LBY
LKA
LSO
LTU
LVA
MAR
MDA
MDG
MEX
MKDMLI
MMR
MNG
MOZ
MRT
MW I
MYS
NAM
NER
NGA
NIC
NLD
NOR
NPL
NZL
OMN
PAKPANPER
PHLPNG
POL
PRK
PRT
PRY
ROU
RUS
RWA
SAU
SDNSENSLE
SLV
SOMSVK
SVN
SWE
SWZSYR
TCDTGO
THA
TJK
TKM
TUN
TURTZA
UGA
UKR
URYUSA
UZBVEN
VNM
YEM
ZAFZMB
ZW E
-.4
-.2
0.2
.4R
esid
ual L
angu
ages
Spo
ken
-.6 -.4 -.2 0 .2 .4Residua l Range o f Land Q ua lity
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Examples of Language Groups in Peru
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Ethnic groups in Peruwith an extent of 25 regions
0 . 2 . 4 . 6 . 8 1S U I T
Y a m i n a h u a
S h i p i b o - C o n ib o
P u n o Q u e c h u a
C e n t r a l A y m a r a
A g u a r u n a
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
Examples of Language Groups in Kenya
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
0 .2 .4 .6 .8 1Land Suitability
Samburu
Orma
Maasai
Kamba
Kalenjin
Gikuyu
Garreh-Ajuran
Distribution of Land Quality within ethnic groups in Kenya
A careful inspection of the box plots reveals that ethnic groups are not randomly dis-
persed across regional land qualities within Kenya. In fact, they seem to cluster in territories of
distinct and homogenous land endowments. The Samburu people, the Orma and the Garreh-
Ajuran are all exclusively located at low levels of land quality where agriculture is almost
impossible to maintain.60 The Samburu are semi-nomadic pastoralists who herd mainly cattle
but also keep sheep, goats and camels, see Pavitt (2001). The Orma are semi-nomadic shep-
herds and the Garreh-Ajuran are semi-nomadic pastoralists. These groups have the human
capital to undertake the productive activities which are optimal for the places in which they
are located. On the other hand, the Gikuyu and the Kalenjin are concentrated in territories
of high land quality and they are mainly engaged in agriculture producing: sorghum, millet,
60The description of the main productive activities of each ethnic group, unless otherwise noted, comes fromthe entries found in the Ethnologue website, (http://www.ethnologue.com/).
57
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
IV Regressions for Economic Variables
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: log income per capita growth rate,
1960-2000
OLS
(3)
ELF -1.130(0.58)*
avg 0.145(0.91)
avgxrange -0.095(1.35)
abs_lat 0.04(0.02)***
eucolony -0.902(0.48)*
lngdppc1960 -0.446(0.21)**
Observations 95R-squared 0.36Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: log income per capita growth rate,
1960-2000
OLS 2SLS
(3) (4)
ELF -1.130 0.319(0.58)* (1.99)
avg 0.145 0.82(0.91) (1.26)
avgxrange -0.095 -0.51(1.35) (1.38)
abs_lat 0.04 0.052(0.02)*** (0.02)**
eucolony -0.902 -0.845(0.48)* (0.48)*
lngdppc1960 -0.446 -0.379(0.21)** (0.22)*
F-test of excluded instruments
4.00**
Sargan statistic (p-value)
0.40
Observations 95 95R-squared 0.36Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
IV Regressions for Political Variables
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: Government Effectiveness 1996-2005
OLS
(5)
ELF -0.545(0.27)**
avg -0.61(0.44)
avgxrange 1.086(0.60)*
abs_lat 0.041(0.006)***
eucolony 0.596(0.22)***
Observations 147R-squared 0.38Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: Government Effectiveness 1996-2005
OLS 2SLS
(5) (6)
ELF -0.545 0.051(0.27)** (0.73)
avg -0.61 -0.382(0.44) (0.50)
avgxrange 1.086 0.953(0.60)* (0.62)
abs_lat 0.041 0.046(0.006)*** (0.009)***
eucolony 0.596 0.674(0.22)*** (0.23)***
F-test of excluded instruments
11.52***
Sargan statistic (p-value)
0.56
Observations 147 147R-squared 0.38Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
-
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: Corruption 19962005
OLS
(7)
ELF -0.764(0.27)***
avg -0.675(0.44)
avgxrange 0.719(0.60)
abs_lat 0.041(0.006)***
eucolony 0.635(0.22)***
Observations 147R-squared 0.40Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
-
Table 10: Instrumental Variable Regressions for Economic and Political Variables
Dep. Var: Corruption 19962005
OLS 2SLS
(7) (8)
ELF -0.764 -0.546(0.27)*** (0.70)
avg -0.675 -0.591(0.44) (0.50)
avgxrange 0.719 0.67(0.60) (0.61)
abs_lat 0.041 0.043(0.006)*** (0.009)***
eucolony 0.635 0.664(0.22)*** (0.23)***
F-test of excluded instruments
11.52***
Sargan statistic (p-value)
0.44
Observations 147 147R-squared 0.40Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
IntroductionTheoryEmpirics
ConclusionAppendix
Language Divergence: An ExampleArti�cial Country PlotReal Country PlotGroups in PeruGroups in KenyaIV Econ VarIV Polit VarIV Quality Var
IV Regressions for Quality of Life
Stelios Michalopoulos Ethnolinguistic Diversity: Origins and Implications
Table 11: Instrumental Variable Regressions for Quality of Life Variables
Dep. Var: Infant Mortality 2000
OLS
(3)
ELF 50.157(9.99)***
avg 2.743(16.14)
avgxrange -45.364(21.91)**
abs_lat -1.313(0.24)***
eucolony -8.085(8.09)
Observations 144R-squared 0.54Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%
Table 11: Instrumental Variable Regressions for Quality of Life Variables
Dep. Var: Infant Mortality 2000
OLS 2SLS
(3) (4)
ELF 50.157 30.133(9.99)*** (27.65)
avg 2.743 -4.618(16.14) (18.47)
avgxrange -45.364 -40.759(21.91)** (22.64)*
abs_lat -1.313 -1.51(0.24)*** (0.35)***
eucolony -8.085 -10.965(8.09) (8.84)
F-test of excluded instruments
10.08***
Sargan statistic (p-value)
0.89
Observations 144 144R-squared 0.54Absolute values of standard errors reported in parentheses.* significant at 10%; ** significant at 5%; *** significant at 1%