This paper explores the relationship between ethnic competition and complementarity in returns to self-employment. On the one hand, theory suggests that competition should be detrimental to profits, thus lowering returns to self-employment. On the other hand, self-employment within an ethnic economy could lead to economic gains. These opposing forces blur the potential relationship between competition and profits, which calls for an empirical investigation. We use detailed individual data from the 2010 American Community Survey and the 2000 US Census. We find that while business competition is detrimental to profits, higher ethnic competition is associated with higher returns and so is larger and wealthier pool of coethnics in the location of the business. The growing importance of ethnic minorities lends policy-relevance to our study.
6. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
5
7. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
5
8. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
5
9. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
But ethnic competition may be different: ethnic enterpreneurship and ethnic enclaves
• Coethnics on the demand side
Auster & Aldrich, 1984; Sanders & Nee, 1987; Rath 2000
5
10. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
But ethnic competition may be different: ethnic enterpreneurship and ethnic enclaves
• Coethnics on the demand side
Auster & Aldrich, 1984; Sanders & Nee, 1987; Rath 2000
• Coethnics on the supply side
Wilson & Portes, 1980; Waldinger et al., 1985; Waldinger, 1989; Rath & Kloosterman, 2000; Kloosterman, 2004
5
11. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
But ethnic competition may be different: ethnic enterpreneurship and ethnic enclaves
• Coethnics on the demand side
Auster & Aldrich, 1984; Sanders & Nee, 1987; Rath 2000
• Coethnics on the supply side
Wilson & Portes, 1980; Waldinger et al., 1985; Waldinger, 1989; Rath & Kloosterman, 2000; Kloosterman, 2004
• Business spillovers
Wilson & Martin, 1982
5
12. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
But ethnic competition may be different: ethnic enterpreneurship and ethnic enclaves
• Coethnics on the demand side
Auster & Aldrich, 1984; Sanders & Nee, 1987; Rath 2000
• Coethnics on the supply side
Wilson & Portes, 1980; Waldinger et al., 1985; Waldinger, 1989; Rath & Kloosterman, 2000; Kloosterman, 2004
• Business spillovers
Wilson & Martin, 1982
5
13. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
But ethnic competition may be different: ethnic enterpreneurship and ethnic enclaves
• Coethnics on the demand side
Auster & Aldrich, 1984; Sanders & Nee, 1987; Rath 2000
• Coethnics on the supply side
Wilson & Portes, 1980; Waldinger et al., 1985; Waldinger, 1989; Rath & Kloosterman, 2000; Kloosterman, 2004
• Business spillovers
Wilson & Martin, 1982
−→ Complementarities for ethnic self-employed
5
14. Motivation
Competition is bad for business
• Generally profits are driven to lower levels
Bain, 1956; Demsetz, 1973; Cowling and Waterson, 1976
• More dispersion in market shares yields greater inequality in profits
Tirole, 1988
But ethnic competition may be different: ethnic enterpreneurship and ethnic enclaves
• Coethnics on the demand side
Auster & Aldrich, 1984; Sanders & Nee, 1987; Rath 2000
• Coethnics on the supply side
Wilson & Portes, 1980; Waldinger et al., 1985; Waldinger, 1989; Rath & Kloosterman, 2000; Kloosterman, 2004
• Business spillovers
Wilson & Martin, 1982
−→ Complementarities for ethnic self-employed
Question: is ethnic competition detrimental to migrant self-employment?
5
16. We need a big dataset
• Competition is best measured at local level −→ market definition
6
17. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
6
18. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
6
19. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
• Ethnicities may differ in the scope of networking −→ rule out “specific cases”
6
20. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
• Ethnicities may differ in the scope of networking −→ rule out “specific cases”
6
21. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
• Ethnicities may differ in the scope of networking −→ rule out “specific cases”
Work with ACS data (2000 and 2010)
• Huge data set
6
22. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
• Ethnicities may differ in the scope of networking −→ rule out “specific cases”
Work with ACS data (2000 and 2010)
• Huge data set
• Large and diverse pool of ethnicities
6
23. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
• Ethnicities may differ in the scope of networking −→ rule out “specific cases”
Work with ACS data (2000 and 2010)
• Huge data set
• Large and diverse pool of ethnicities
• Incomes of self-employed available, for all individuals
6
24. We need a big dataset
• Competition is best measured at local level −→ market definition
• Data has to be on all self-employed
Implicit assumption: competition from non-personal market players the same for every ethnicity
• Ethnic competition can only be measured if we enough immigrant self-employed at a
given market
−→ Market defined by industry and geography
• Ethnicities may differ in the scope of networking −→ rule out “specific cases”
Work with ACS data (2000 and 2010)
• Huge data set
• Large and diverse pool of ethnicities
• Incomes of self-employed available, for all individuals
• PUMA units for geography and SIC for sector and still enough granularity
6
26. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
7
27. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
• Ethnicity is proxied by ancestry (ANCESTR1G, 120+ origins, recoded)
7
28. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
• Ethnicity is proxied by ancestry (ANCESTR1G, 120+ origins, recoded)
• Migration defined by self-reported status
7
29. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
• Ethnicity is proxied by ancestry (ANCESTR1G, 120+ origins, recoded)
• Migration defined by self-reported status
• Self-employed report income, not actually profits
7
30. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
• Ethnicity is proxied by ancestry (ANCESTR1G, 120+ origins, recoded)
• Migration defined by self-reported status
• Self-employed report income, not actually profits
• right-censored at $126 000
7
31. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
• Ethnicity is proxied by ancestry (ANCESTR1G, 120+ origins, recoded)
• Migration defined by self-reported status
• Self-employed report income, not actually profits
• right-censored at $126 000
• observe many negative incomes (∼ 2% of observations)
7
32. The measures
• Locality is proxied by public use microdata areas (PUMA, ∼2100 units)
• Sector is reported at 3-digit SIC (IND1990, 196 categories, 113 with sufficient representation)
−→ market j = PUMA + SIC
• Ethnicity is proxied by ancestry (ANCESTR1G, 120+ origins, recoded)
• Migration defined by self-reported status
• Self-employed report income, not actually profits
• right-censored at $126 000
• observe many negative incomes (∼ 2% of observations)
• cannot separate own wage from business returns −→ market controls needed
7
35. The measures
Measuring competition: Herfindahl-Hirschmann Index (HHI)
HHIj =
J
j=1
s2
j and HHIj,e =
J
j=1
E
e=1
s2
e,j plus HHIj,eimi =
J
j=1
E
e=1
s2
j,e if e ∈ imi
8
36. The measures
Measuring competition: Herfindahl-Hirschmann Index (HHI)
HHIj =
J
j=1
s2
j and HHIj,e =
J
j=1
E
e=1
s2
e,j plus HHIj,eimi =
J
j=1
E
e=1
s2
j,e if e ∈ imi
Measuring complementarity: homophily
NUMBERe,j =
e
/
N
8
37. The measures
Measuring competition: Herfindahl-Hirschmann Index (HHI)
HHIj =
J
j=1
s2
j and HHIj,e =
J
j=1
E
e=1
s2
e,j plus HHIj,eimi =
J
j=1
E
e=1
s2
j,e if e ∈ imi
Measuring complementarity: homophily
NUMBERe,j =
e
/
N
and WEALTHe,j =
e
incomes/
N
incomes
8
38. The model
Hourly income from SE = β0 + βX + δMills Ratio
+γI Industry Competition
+γE Ethnic Competition
+γC Complementarity
• Heckman correction for selection into employment at all
9
39. The model
Hourly income from SE = β0 + βX + δMills Ratio
+γI Industry Competition
+γE Ethnic Competition
+γC Complementarity
• Heckman correction for selection into employment at all
• Control for individual characteristics (X): age, gender, education, years in the US,
command of English
9
40. The model
Hourly income from SE = β0 + βX + δMills Ratio
+γI Industry Competition
+γE Ethnic Competition
+γC Complementarity
• Heckman correction for selection into employment at all
• Control for individual characteristics (X): age, gender, education, years in the US,
command of English
• We cannot know if the business is actually “ethnic” Waldinger et al. (1990)
9
41. The model
Hourly income from SE = β0 + βX + δMills Ratio
+γI Industry Competition
+γE Ethnic Competition
+γC Complementarity
• Heckman correction for selection into employment at all
• Control for individual characteristics (X): age, gender, education, years in the US,
command of English
• We cannot know if the business is actually “ethnic” Waldinger et al. (1990)
• All variables standardized (coefficients inform about the strength of a relationship)
9
42. The results
1 2 3 4 5 6 7
HHI 0.023* 0.003* 0.003* 0.002* 0.006* 0.005* -0.009*
NUMBER -0.033* -0.007* -0.009* -0.014*
WEALTH 0.053* 0.051* 0.064*
HHIeth -0.005* -0.005*
HHIethimi 0.035*
ethnicity yes
ethnicity x HHI yes
PUMA yes yes yes yes yes yes yes
SIC yes yes yes yes yes yes yes
* denotes significance at 1% level
All regressions control for individual characteristics: age, age squared, gender (and interaction with age),
education (and interaction with gender), years in the US, command of English
0.005 ∼ 10 cents per hour and 0.064 ∼ 1.5 dollars per hour −→ considerably less than education or
gender
10
43. The results - summary and robustness
Overall
• Ethnic competition is good for business
11
44. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
11
45. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
• Complementarity is good for business, but demand and access to capital appear more
general than access to labor
11
46. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
• Complementarity is good for business, but demand and access to capital appear more
general than access to labor
11
47. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
• Complementarity is good for business, but demand and access to capital appear more
general than access to labor
This result is robust to
• Excluding the dominant ethnicity
11
48. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
• Complementarity is good for business, but demand and access to capital appear more
general than access to labor
This result is robust to
• Excluding the dominant ethnicity
• Controllng for the duration of stay in the US
11
49. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
• Complementarity is good for business, but demand and access to capital appear more
general than access to labor
This result is robust to
• Excluding the dominant ethnicity
• Controllng for the duration of stay in the US
• Alternative model specifications
11
50. The results - summary and robustness
Overall
• Ethnic competition is good for business
• Though not among immigrants
• Complementarity is good for business, but demand and access to capital appear more
general than access to labor
This result is robust to
• Excluding the dominant ethnicity
• Controllng for the duration of stay in the US
• Alternative model specifications
• “Leave one out”
11
54. Conclusions
• Although competition is bad for business, ethnic competition is not
• Ethnic complementarity can also be helpful
• This phenomenon may explain the prevalence if immigrant/ethnic self-employment
12
55. Conclusions
• Although competition is bad for business, ethnic competition is not
• Ethnic complementarity can also be helpful
• This phenomenon may explain the prevalence if immigrant/ethnic self-employment
12
56. Conclusions
• Although competition is bad for business, ethnic competition is not
• Ethnic complementarity can also be helpful
• This phenomenon may explain the prevalence if immigrant/ethnic self-employment
What to do next?
• Analyze other countries?
• More robustness checks
12