GÁBOR ANTAL Central European University Institute of Economics - HAS JOHN S. EARLE Central European University W.E. Upjohn Institute ÁLMOS TELEGDY Central European University Institute of Economics - HAS EACES Workshop April 8, 2010 FDI and Wages: Evidence from LEED in Hungary
45
Embed
GÁBOR ANTAL Central European University Institute of Economics - HAS JOHN S. EARLE Central European University W.E. Upjohn Institute ÁLMOS TELEGDY Central.
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
GÁBOR ANTALCentral European UniversityInstitute of Economics - HAS
JOHN S. EARLECentral European University
W.E. Upjohn Institute
ÁLMOS TELEGDYCentral European UniversityInstitute of Economics - HAS
EACES WorkshopApril 8, 2010
CEU, BudapestSeptember 24, 2009
FDI and Wages:Evidence from LEED in Hungary
Motivation: Employer Wage Effects
Employer effects on wages (Abowd et al., 1999; Haltiwanger et al. 2007)
Questions: What firm characteristics associated with high/low
wage? Neutral or biased across types of workers? What explains?
(1) No additional controls(2) Gender, education category, potential
experience(3) + interactions(4) + manager, new hire dummies
Dynamics: Ownership interacted with event time
Specifications II
Error term (uijt): OLS Firm fixed effects (FE) ~29,000 FE combined with narrowly defined worker groups (GFE)
~400,000
NN PS matching (e, lp, w, expshare 1 and 2 years before acqusition; quadratic polynom.) 325 acqd, 279 control firms; 330,510 obs. PS: normalize around acquisition year, weight controls Exact matching on 2-digit industry and year OLS, FE, GFE Good covariate balance
Stock of university graduates and young workers increases after acquisition
LPMs with individual characteristics on LHS, acquisition dummy on RHS; FE estimation
More hiring after acquisition (mostly one year after), in favor of young high-skilled
LPMs with new hire dummy on LHS, acquisition dummy interacted with individual characteristics on RHS; FE estimation
Separation rates: to be done
Composition of Firms
Acquisitions weakly correlated with wages and firm exit
Probit with firm-level exit on LHS, acquisition dummy interacted with log wagebill on RHS
Foreign Acquisitions and Wage Structure
Fixed Effects Matching (FE)
Group Effects Matching (GFE)
Female 0.009 0.013 0.004 0.004
Vocational 0.013 0.014 -0.018 -0.009
High school 0.037** 0.052** 0.003 0.023
University 0.161** 0.136** 0.082** 0.115**
Experience -0.001 -0.005** -0.002 -0.002
Exp2 * 100 -0.000 0.007* -0.003 -0.001
New Hire -0.009 0.014 0.013 0.008
Manager 0.115** 0.032 0.120** 0.059
Foreman 0.073** -0.005 0.060* 0.002
R2 0.405 0.490 0.164 0.199
Measurement I
Hypothesis: Higher working hours at acquired firms
Monthly paid hours for 1999-2005Tests:
Monthly vs hourly earnings Same effect
Hours as a dependent variable No foreign effect
Hours as a covariate Leaves foreign effect unchanged
Caveat: Overtime probably mismeasured for non-production workers, and hard to test for production separately, since no wage effect
Measurement II
Hypothesis: Domestic firms are more likely to underreport wages Aux. hypotheses: Probability of cheating is lower in big
enterprises and in industries with a low cheating index (Elek and Szabó 2008)
Tests: LPM for 1[w < minw + 3%]
Negative foreign effect (not high enough to explain total wage difference)
Foreign interacted with size Zero/positive effect (reject hypothesis)
Foreign interacted with industry cheating index Zero/negative correlation (reject hypothesis)
Conclusions
OLS: foreign wage premium is 36 percentFE, GFE, matching premium is 9–17 percentDivestment effect is 40-50% of acquisition effectAll worker types benefit; high educated the most5% premium for incumbent workers, composition
change in favor of young high-skilledResults not driven by measurement errorProductivity best candidate for explaining the gap
Previous Studies I
Firm-level data: Positive, sometimes large foreign wage premium
Controls for employment composition or LEED:Smaller effects, sometimes insignificant
The premium varies by skill groupTreatment of selection bias is important
Previous Studies II
Many datasets are not real LEED, but firm-level data with information on composition
Short time series (usually ≤ 5 years) Matching only on immediate pre-acquisition
yearFew ownership changes with enough pre- and
post treatment observationsMost studies from developed countries
exposed to FDI for a long timeWage structure: mostly skilled-unskilled
Firm Characteristics by Ownership II
Domestic Foreign
Industry
Agriculture 8.4 1.9
Industry 25.7 43.4
Construction 10.6 2.2
Trade 28.7 27.0
FIRE 4.2 8.1
Business Services 9.2 8.8
Other Services 13.0 8.5
Tests of Covariate Balance
Normalized Difference
Treated-Controls Probability of Rejecting
Inequality of Means
One Year Before Acquisition
Average Earnings 0.042 0.473 Sales 0.058 0.317 Employment 0.068 0.247 Capital 0.016 0.787 Export Share 0.003 0.954
Two Years Before Acquisition
Average Earnings 0.036 0.520 Sales 0.053 0.362 Employment 0.058 0.312 Capital -0.060 0.291