Does migration to Thailand reduce Thai wages?

1
Does migration to Thailand reduce Thai wages?

• John Bryant and Pungpond Rukumnuaykit
• Institute for Population and Social Research
• Mahidol University, Thailand
• Presentation to Migration and Development BBL, 6
  December 2007

2
Labor market impact of migration to developing
countries

• Some developing countries receive significant
  immigration
• eg South Africa, Argentina, Tajikistan, Thailand
• Most econometric studies in developed countries
  suggest effect of immigration is small
• But hazardous to extrapolate to developing
  countries

3
Thailand is a good place to study labor market
effects of immigration

• Significant immigration
• Roughly 2 million immigrants, mostly from
  Myanmar, though some from Cambodia, Laos, other
• Almost all entered illegally
• Almost all do unskilled work
• Relatively good data on immigrants
• 2004 amnesty/registration probably captured about
  half of all migrants
• Ministry of Labor provided access to data
• Extensive, relatively good data on Thais

4
Results from poll of 4,184 Thais, 2006
If the government allowed more migrants to
enter Thailand, would this harm the wages of Thai
workers?
Response Percent
Yes 83
No 5
No response 12
Total 100
5
Analytical strategy for this study

• Use geographical variation
• Test whether, all else equal, Thais in districts
  with more migrants have lower wages
• Most common strategy in developed-country studies
• Allow for endogenous migration
• Migrants move to places with high wages
• Include control variables
• Instrument migrant intensity
• (Also allow for endogenous under-counting of
  migration)

6
Dependent variable Wages of Thais

• District average for (log of) hourly wages
• Only private employees (41 of labor force)
• Use regression to remove the effect of
  differences in human capital

7
Main explanatory variable Migrant intensity

• Burmese migrants as a proportion of the district
  population (expressed in log terms)

8
Raw relationship between wages of Thais and
migrant intensity

• Positive slope because migrants go to places with
  high wages

9
Allowing for endogeneity (i) Add control
variables
Statistical model of effect of migrant intensity on Thai wages Coefficient on migrant intensity
No controls 0.090
With controls 0.026

• Control variables
• Distance from Bangkok
• Border district
• Gross Provincial Product per capita
• Percent of households that are poor
• Employment structure

10
Allowing for endogeneity (ii) Instrument on
distance to border
Migrant intensity vs distance to Burmese border

• Very strong relationship between migrant
  intensity and distance to border, even with
  controls for demand for labor
• Reason
• Transport costs
• Friends and neighbors effects
• Source of exogenous variation in migrant intensity

11
Allowing for endogeneity (ii) Instrument on
distance to border
Statistical model of effect of migrant intensity on Thai wages Coefficient on migrant intensity
No controls 0.113
With controls 0.026
With controls, and instrumenting on distance to border -0.023
12
Extensions and robustness tests - Wages
Issue Finding
Effect stronger for low-skilled Thais? Inconclusive
Effect stronger for Thai males? Weak confirmation
Spatial dependency Not important
Effect on wages of government employees None
13
Extensions - Employment
Issue Finding
Does immigration reduce the hours worked by Thai private employees? No
Does immigration reduce the proportion of Thais who are employed? No
Does migration reduce the proportion of Thais working as private employees? No
14
Limitations

• Compensatory migration by Thais may be masking
  labor market impact of immigration
• But cannot test because data on internal
  migration by Thais are unreliable
• Even with controls, distance to the Burmese
  border may be correlated with demand for labor or
  under-reporting
• Coefficient on migrant intensity in main wage
  model only just significant (p0.041)

15
Conclusions

• Wages A 10 increase in migrants reduces Thai
  wages by roughly 0.2
• Employment No effect on Thai employment

16
Comparison with previous estimates

• Most previous studies report coefficients from a
  regression of log wages against migrant share
• We regress log wages against log migrant share
• To compare, we need to divide our estimate by the
  district average for migrant share
• -0.023 0.013 -1.77
• According to a meta-analysis, the median value
  for developed countries is -0.12

17
Coverage of migrants in the 2000 Thai Census

• Burmese
• Number of people of Burmese nationality, born
  outside Thailand, aged 15, according to 2000
  Census 63,229
• Number of Burmese who registered to work in
  Thailand in 2001 451,335
• Japanese
• Number of people of Japanese nationality, born
  outside Thailand, aged 15, and employed,
  according to 2000 Census 2,582
• Number of work permits issued to Japanese
  nationals, 2000 13,355