Comments by Chinhui Juhn University of Houston Did Immigration Contribute to Wage Stagnation of Unskilled Workers? By Givoanni Peri Comments by Chinhui Juhn University of Houston
Summary Immigration accounts for only a small fraction of the rise in wage inequality and decline in wages of less skilled workers Immigrants (particularly highly skilled) may have contributed to wage growth via innovation and entrepreneurship (Peri, Shih, and Sparber, 2015; Gauthier-Loiselle and Hunt, 2010; Kerr and Lincoln, 2009; Fairlie, 2012)
Two questions Given an elasticity of substitution between HS Dropouts and HS grads, σHD, do immigrant induced shocks in relative supplies account for declines in relative wages of HS dropouts? What is a reasonable estimate of 1/σHD ?
Borjas, Freeman and Katz (1997) Table 14. Immigrant Contribution to Labor Supply HS dropouts vs. HS graduates Sample and Year Dropouts Graduates Log gap 1960 0.088 0.051 0.035 1970 0.069 0.046 0.022 1980 0.109 0.058 0.047 1990 0.242 0.079 0.141 1995 0.383 0.083 0.244
Borjas, Freeman and Katz (1997) From 1980-1995, immigrant induced shock in supply of HS dropouts increases .197 Predicted relative wage change = .197*(-.344) = -.063 out of -.109 actual change in relative wage Explains more than half of the relative decline in wages of dropouts Useful to extend to 2010 and beyond using similar methods They use 1/σ=.344 , which is large, so good to assess using more recent estimates
Peri (2016), Table 2
Peri (2006), Table 2 From 1980-2000, immigrant induced shock in supply of HS dropouts increases 6 percentage points Predicted relative wage change = .06*(-.571)= -.034 out of -.119 actual change in relative wage Explains less than one-third Immigrant contribution to relative supply increase of HS dropouts in 1980-1990 seems particularly small
Suggested robustness checks Use the same methodology as BFK (1997) and see what has happened in the more recent period Sample selection criteria for “wage sample” and “quantity sample” a la Katz and Murphy (1992) Criteria for “wage sample” is to minimize composition changes Select full-time workers with strong labor market attachment Hold composition (across sex, age, possibly immigrant nationality) fixed in calculating relative wage changes
Suggested robustness checks Criteria for “quantity sample” is to be inclusive All workers including part-time and self-employed Hours instead of bodies Efficiency units which would entail making some assumptions about substitutability of labor types How to aggregate young and old workers? Men and women? Treat them equally or weight by their relative wage? Implicitly assuming perfect substitutes
Estimate of σHD Most studies use spatial variation (across states, cities, commuting zones) Critical issue is unobserved demand shocks (which may be location X skill type specific) Instrument immigrant-induced supply shocks with initial stock of immigrants by nationality in location interacted with national inflow (Altonji and Card, 1991; Card, 2001; Card, 2009)
Estimate of 1/σHD BFK (1997) .344 Card (2009), Table 2, .02 (IV version) Borjas, Grogger, Hanson (2012), Table 2, report a range from .024 to .152 depending on MSA/state level data and different controls for MSA/state-specific trends An estimate of .02 implies that HS dropouts and HS graduates are perfect substitutes so immigrants did not contribute to inequality Even the highest estimate in BGH suggests that the elasticity is smaller than .344
Estimate of 1/σHD Is there a way to improve the networks instrument? The instrument addresses where the immigrants locate but does not address the timing of immigrant shock One possibility is to exploit the “push” vs. “pull” types of immigration
Estimate of 1/σHD Monras (2015) Uses immigration shock induced by Mexican Peso Crisis, together with the network instrument, to examine impact on relative wages of high school and less workers SR (1994-1995) elasticity is .874 (Table C.5) LR (1990-2000) elasticity is .383 (Table 6) A number of reasons why the elasticities are not directly comparable to those reported in the paper here but one reason for the difference may be the utilization of a “push” type of immigration episode
To summarize Peri (2016) offers a useful overview which is in agreement with now consensus view that immigration played a minor role in the rise in wage inequality and relative wage declines of less skilled A direct comparison with BFK (1997) using more recent data and similar methodologies would be useful In exploring alternatives to estimate the key elasticity parameter, utilizing “push” type of immigration episodes may be fruitful