From Rags to Riches – How Robust is the Influence of Culture on Entrepreneurial Activity? Christian Busch, Andrea Lassmann November 19th 2009
Entrepreneurship and its Determinants
Determinants of entrepreneurship Micro-level: age, gender, education, education, etc. Aggregate level : access to funding, taxes, unemployment, enclaves Fundamental determinants: Institutions and Culture
Culture is essentially unobservable
„Culture as those customary beliefs and values that ethnic, religious, and social groups transmit fairly unchanged from generation to generation“ (Guiso, Sapienza, Zingales, 2006, JEP)
Survey-based measures difficult to separate from (usually perceptionbased) measures of institutions
Culture assumed to show some persistence November 19th 2009
2
Motivation
Quasi-experiment of immigration: Differences in self-employment between immigrants from different countries
(Institutional) environment is the same for all immigrants but has changed for each individual
Controlling for self-selection by looking at individual characteristics
Remaining behavioral differences due to individualized and internalized values of their native cultures
November 19th 2009
3
Existing Literature
For culture to be relevant, self-employment of immigrants should be related to self-employment shares in immigrants’ home countries.
Yuengert (1995): positive relationship
Fairlie/Meyer (1996): correct for U.S. natives-immigrant differences, and find not effect
Akee et al. (2007): positive effect of previous self-employment, but large negative effect of home-region self-empl.
Oyelere/Belton (2009): Developing countries have higher average self-employment Immigrants from developed countries have higher probability of being self-employed November 19th 2009
4
Approach
Puzzling finding of non-existing / negative relationship of immigrant selfemployment with home country self-employment
Differences in immigrant self-employment cannot entirely be explained by observable differences between immigrants
Suggested modifications Alternative proxy for entrepreneurial activity Account for adjustment over time Control for determinants of entrepreneurship in home countries – Proximate causes: unemployment, etc. – Ultimate causes: institutions
November 19th 2009
5
Data
Logit using immigrant self-employment as dependent variable
U.S. Integrated Public Use Microdata Series (IPUMS 5%) from the year 2000 separate all self-employed and incorporated self-employed exclude persons 69 years, unemployed restrict number of US natives to 100’000 1.3 million observations; 108’000 self-employed (~8%); 36’000 inc. (~3%) include large number of demographic control variables
Home country entrepreneurial activity (ILO) Share of self-employed on active population Share of employers on active population Use different years in order to increase sample
Home country determinants of entrepreneurial activity (WDI) Home country institutions: ICRG, AGI
November 19th 2009
6
.8
Self-Employment as a Proxy for Entrepreneurship…
CMR HTI
.6
RWA MDG MDV ETH
GAB BOL PER HND IRNDOM JAM PHL ECU DMA ZMB COL GRC YEM LSO SLV LCA NAM BGD MNG THA LKA CPV PAN MAR BLZ KOR CHL EGY WBGMHL BRA STP MDA GEO ARG TON CRI MEX URY PRT GRD POL TUN TUR ITA ROM MYS TTO CYP NZL IRL ESP HRV SUR LTU NCL CANISL MUS PRIAUS MKD BRB SGP GBR CHE JPN SVN BWA UKR LVA HUNCZE OMN ISR NLD AUT HKG MAC DEU USA DNK NOR RUS SVK EST PAK
.2
.4
KHM VNM
0
ERI
4
6
8 log Income per Capita
10
November 19th 2009
7
.8
… may be Determined by Market Distortions
CMR
.6
HTI
.4
MDV BOL PER
.2 0
DOM JAM IRN
ECU
DMA GRC SLV LCA PAN BLZ KOR CHL WBG ARG BRA URY CRI PRT MEX GRD POL ITA MYS TTO CYP IRL HRV NZL ESP ISL LTU SUR NCL CAN MUS AUS PRI CHE BRB SGP GBR JPN OMN SVN BWA CZE UKR ISR LVA NLD AUT HKG HUN DEU MAC ABW ANT USA DNK NOR SVK RUS EST QAT COL
0
ETH
PAK
20
VNM KHM
HND PHL
MNG NAM LKA
YEM LSO THA
ZMB BGD
EGY MDA GEO TON TUR ROM
40 60 Agricultural share of employment
80
100
November 19th 2009
8
.15
Modification I: Employer Shares as an Alternative Proxy BEL FIN
.1
SWE
NAM
NZL GRC KOR CHE ISL PRT PER SGP CYP WBG IRL HKG AUT HRV PHL SLV MKD ARG ESP ISRDEU AUS COL HND TTO POL CZE URY MDV IRNDOM LVA MUS CMR SVN MAC CHL MAR TUN JPN BRA PAN JAM MYS TUR HUNMEX CPV BOL THA SVK YEM LKA OMN SUR EST ECU
.05
ETH
0
Share of Employers
ITA
4
MNG GEO ZMB PAK LSO MDG MDA STP VNM BGD KHM RWA
6
DMA LTU LCA BLZ GRD CRI
RUS ROM MHL TON
GAB
8 log Income per Capita
BRB
10
November 19th 2009
9
U.S. versus Home Country Self-Employment (b) Employers only
.8
.15
(a) Self-Employment
BEL FIN
HTI
CMR
.1
.6
ITA
NZL
IRN
ECU GRC
.05
KOR
0 0
.05
.1 selfempshare_us Fitted values
.15
selfempshare
GRC
CHE KOR
CRI
0
.2
.4
ETH PAK BOL PER HNDKHMVNM DOM JAM PHL ECU DMA COL SLV LCA BGD THA CPV LKA PAN BLZ MDA MAR EGY CHL GEO ARG TON BRA CRI URY MEX PRT GRD POL ITA TUR ROM MYS TTO CYP IRL NZL ERIESP HRV LTU CAN AUS PRIBRB CHE SGP JPN MKD GBR UKR CZE NLD AUT HKG DEULVA HUN DNK RUSNOR SVK
SWE
.2
PER PRT HKG AUT HRV IRL PHL SLV DEUESP ARG AUS COL HND TTO POL CZE URY DOM CHL MAR JPN BRA PAN JAM MYS TUR HUN MEX BOL THA LKA RUS ROM PAK ETH VNM KHM BGD
.02
.04
.06 incshare_us Fitted values
IRN
.08
.1
employershare
November 19th 2009
10
Modification II: Institutional Adjustment versus Cultural Persistence
November 19th 2009
11
Modification III: Account for Cross-Country Determinants of Entrepreneurship Approach: Multilevel Analysis; Clustering at the country-level
Proximate Causes Income per capita, unemployment, demographic structure, agricultural labour share, distance from equator Ultimate Cause: Institutions Complex relationship: Any significant variable indicates cultural (behavioural) persistence Expect institutions to primarily determine (persistent) behaviour
November 19th 2009
12
Results I: Accounting for Convergence & Home Country Determinants of Entrepreneurship (1) All
(2) < 10 Years
(3) > 10 Years
Unemployment Rate
(4) < 5 Years -0.002
(5) All
(6) < 10 Years
(7) > 10 Years
(0.013)
Share of Employment in Agriculture
Share of Old Population (> Age 65) Share of Young Population (Age 0-14) Latitude (Distance from Equator) Share of Urban Population Log GDP per capita Self-employment share
-0.004
-0.002 (0.008)
-0.039**
-0.037
(0.020)
(0.028)
-0.045***
-0.041**
(0.013)
(0.016)
0.544
0.609
(0.559)
(0.645)
0.001
-0.000
(0.006)
(0.006)
-0.021
-0.107
(0.072)
(0.099)
-0.589
-0.885
-0.54
1.547**
(-1.02)
(1.89)*
(-0.85)
(0.692)
0.15 65 952019
0.13 65 319692
0.16 65 632327
(0.015)
(0.007)
Employer share Pseudo R-squared Number of Clusters Number of Observations
(8) < 5 Years 0.005
0.132 65 159512
6.095
4.417
6.269
3.985***
(2.70)***
(1.92)*
(2.77)***
(1.508)
0.12 60 813559
0.08 60 281738
0.12 60 531821
0.076 60 139148 November 19th 2009
14
Results II: Self-Employment and Institutions International Country Risk Guide
Aggregate Governance Indicators
Expropriation
Rule of Law
Bureaucracy
Contract
Government
Corruption
Regulation
Rule of Law
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
-0.832* (0.458)
-0.826* (0.458)
-0.826* (0.458)
1.725** (0.709)
1.715** (0.709)
1.725** (0.709)
1.725** (0.709)
(A) no home-country control variables
Self-Employment Share
-1.312*** (0.507)
-1.003** (0.457)
-1.003** (0.457)
-1.312*** (0.507)
-0.826* (0.458)
(B) Including home-country control variables Self-Employment Share
0.358 (0.761)
1.429** (0.703)
1.429** (0.703)
0.358 (0.761)
(C) Including Institutions only Self-Employment Share
1.355 (0.943)
0.961 (0.600)
0.144 (0.603)
0.231 (1.308)
0.938 (0.618)
0.913 (0.741)
1.032* (0.621)
0.818 (0.688)
Institution
0.180* (0.109)
0.136** (0.066)
0.326*** (0.097)
-0.126 (0.086)
-0.542*** (0.122)
0.134 (0.087)
-0.156 (0.095)
0.128 (0.107)
(D) Including home-country control variables and Institutions
Self-Employment Share
0.932 (0.738)
1.158* (0.627)
0.776 (0.491)
0.084 (0.884)
1.736** (0.710)
1.499** (0.762)
1.731** (0.737)
1.539** (0.783)
Institution
0.140 (0.101)
0.117* (0.065)
0.311*** (0.080)
-0.071 (0.078)
-0.556*** (0.122)
0.099 (0.070)
-0.036 (0.133)
0.072 (0.088)
Pseudo R-squared Number of Observations Number of Clusters
0.133 147606 54
0.132 153823 60
0.133 153823 60
0.133 147606 54
0.131 159512 65
0.131 159403 64
0.131 159512 65
0.131 159512 65 November 19th 2009
15
Results III: Employers and Institutions International Country Risk Guide
Aggregate Governance Indicators
Expropriation
Rule of Law
Bureaucracy
Contract
Government
Corruption
Regulation
Rule of Law
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
5.071** (2.199)
5.122** (2.196)
5.122** (2.196)
3.891** (1.514)
3.771** (1.517)
3.891** (1.514)
3.891** (1.514)
(A) no home-country control variables Share of Employers
5.540** (2.313)
5.233** (2.212)
5.233** (2.212)
5.540** (2.313)
5.122** (2.196)
(B) Including home-country control variables Share of Employers
2.902* (1.536)
3.853** (1.518)
3.853** (1.518)
2.902* (1.536)
(C) Including Institutions only Share of Employers
4.784* (2.454)
3.609* (1.941)
4.004** (2.004)
4.591** (2.203)
4.293* (2.267)
3.337 (2.194)
4.810** (2.113)
3.403 (2.245)
Institution
0.174 (0.139)
0.161* (0.087)
0.385** (0.150)
-0.160 (0.102)
-0.967*** (0.224)
0.179 (0.135)
-0.267** (0.122)
0.148 (0.163)
(D) Including home-country control variables and Institutions Share of Employers
3.015** (1.503)
3.081** (1.565)
3.348** (1.454)
3.571** (1.621)
3.923** (1.529)
2.962* (1.611)
4.281*** (1.471)
3.172* (1.700)
Institution
0.067 (0.128)
0.131 (0.090)
0.349*** (0.103)
-0.167* (0.091)
-0.990*** (0.222)
0.147 (0.106)
-0.147 (0.134)
0.103 (0.125)
Pseudo R-squared Number of Observations Number of Clusters
0.074 134992 51
0.075 138692 56
0.076 138692 56
0.075 134992 51
0.076 139148 60
0.074 139041 59
0.075 139148 60
0.074 139148 60 November 19th 2009
16
Conclusions
Crucial differences between proxies
Home country determinants important Explain why previous studies did not find positive relationship Change the sign on home country self-employment Reconcile different influence of both proxies Home country employers share more robust Institutions not the main determinant
Unable to reject cultural hypothesis
Remaining problems: Parameter heterogeneity with respect to income Lack of explanation about channels of influence November 19th 2009
17