gÁbor antal institute of economics - has central european university
DESCRIPTION
Foreign Ownership and the Distribution of Wages in Hungary, 1992-2000: An Unconditional Quantile Decomposition Approach. SEBA – IE/CASS – IE/HAS Conference June 30 , 201 1. GÁBOR ANTAL Institute of Economics - HAS Central European University. Introduction • •••. - PowerPoint PPT PresentationTRANSCRIPT
Foreign Ownership and the Distribution of Wages in Hungary, 1992-2000: An Unconditional Quantile Decomposition Approach
GÁBOR ANTALInstitute of Economics - HASCentral European University
SEBA – IE/CASS – IE/HAS ConferenceJune 30, 2011
Motivation I• Transition provides fruitful setting to investigate changes in
wage distribution▫Wage determination became decentralized within a
couple of years▫Changes affecting both supply and demand side of labor
market•Hungary displayed largest level of earnings inequality before
transition (Rutkowski 1996) AND largest growth in earnings inequality between 1994 and 2005 (OECD 2007)
• Special data source▫Firm-level data on ≈400,000 business units▫ Linked employer-employee dataset of 2.9 million worker-year
observations on workers employed by ≈40,000 business units▫Spanning 1986-2008
Long spells for diff-in-diff analysis and matching
Introduction ••••
Motivation II
• Largest volume of FDI in region during nineties (OECD 2000) remaining high later▫More ownership switches for identification than any other
study in literature
• Foreign owners may differ more from domestic ones than in developed economies
•Only FDI’s effect on conditional average wages analyzed▫Unconditional wages?▫Differences across the distribution?
Introduction ••••
Research Question•What would have happened to the unconditional wage
distribution (wage inequality) in 2000, had the share of foreign employment remained at its 1992 level?▫ Is FDI’s effect the same across the distribution? ▫ Is it rather a composition effect or a wage structure effect?
• Literature context▫Effect of (de)unionization on wage inequality in the US
DiNardo et al. (1996), DiNardo and Lemieux (1997), Firpo et al. (2007, 2008)
▫Wage arrears and wage inequality in Russia Lehmann and Wadsworth (2007)
Introduction ••••
Contribution•No study yet to explicitly analyze FDI’s effects on
unconditional wage distribution• Application of a newly developed decomposition method in
this context• Typical paper in literature on FDI and wages:▫FDI’s effect on conditional mean wages
Firm-level: Conyon et al. (2002), Lipsey and Sjöholm (2004), Feliciano and Lipsey (2006), Girma and Görg (2007), Brown et al. (2010)
LEED: Martins (2004), Almeida (2007), Heyman et al. (2007), Huttunen (2007), Earle and Telegdy (2008)
▫Some analysis of effect on wage structure in a few studies Huttunen (2007), Almeida (2007), Eriksson and Pytliková (2011),
Heyman et al. (2011)
Introduction ••••
Employee Information•Hungarian Wage Survey ▫ Conducted in 1986, 1989, and then yearly 1992-2008▫ Includes all firms with >20 employees plus random
sample of small (11-20 employees in 1996-99, 5-20 in 2000-08)
▫Workers sampled randomly based on birth date in medium and large firms (5th and 15th for production workers, also 25th for nonproduction)
▫ All workers in small firms (<20 employees in 1996-2001, <50 since 2002)
▫ Earnings, gender, age, education, occupation, date of hiring, location of plant
Data •••••
Employer Information
•Hungarian Tax Authority Data▫ 1992-2008:
All legal entities using double-entry bookkeeping Total employment in data ≈ All business sector employees in
Hungary▫ 1986-1992:
Sample of firms from HWS▫ Balance sheet and income statement items, employment,
legal form, industry, county of HQ
• LEED: HWS and HTA data linked through firm identifier
Data •••••
Key Variables: Wages and Ownership•Monthly gross earnings▫As reported by the employer (contrast with HH surveys, e.g.
CPS)▫Monthly base salary
+ Overtime pay+ Regular bonuses and premia, commissions, allowances…+ Tenure-proportional extraordinary bonuses based on previous year’s records
• Foreign ownership status▫If >50% share of total equity▫Large number of ownership switches▫Can distinguish types of ownership histories
Data •••••
Weighting and Longitudinal Links
• Three set of weights▫Worker weights within firm
to account for different sampling schemes of BC and WC workers
▫ Firm weights in LEED to weight up to business sector employment
▫ Firm weights in HTA data to account for differences in firms size and for pre-1992
sample size• Firms are linked over time• ≈50% of workers linked within firm based on birth date and
other individual characteristics
Data •••••
• Selected from LEED; years 1986, 1989, 1992-2008 (current focus: 1992-2000)
• For-profit firms in business sector▫with more than 20 employees▫with not more than 2 ownership switches▫ in industries with any foreign presence
• Full-time workers aged 15-74• 25,031 companies (16,790 in 1992-2000)• 2,498,412 worker-years (797,250 in 1992-2000)
Sample
Data •••••
Estimation Method•Detailed decomposition of unconditional wage changes by
quantile, based on recentered influence functions (RIF)
• RIF: Measures the effect of a perturbation in a distribution on some distributional statistic (Hampel 1974)
• Key idea: Effect of changes in distribution of covariates on wage distribution captured by RIF regression (Firpo et al. 2009)
• A decomposition analogous to O-B decomposition of changes in mean can be performed with help of RIF regressions (Firpo et al. 2007)
Methodology •••••
Estimated Effects of FDI on Unconditional Quantiles of Wage Distribution
Results ••••••
0.1
5.3
.45
.6
0 .2 .4 .6 .8 1Quantile
1992
2000
Foreign
0.1
.2.3
.4
0 .2 .4 .6 .8 1Quantile
1992
2000
Foreign
Men Women
Results of Aggregate Decomposition - Men
-.2
0
.2
.4
Log
Wag
e C
hang
e
0 .2 .4 .6 .8 1Quantile
Total change Composition effect
Wage structure effect
Approximation error
Reweighting error
Results ••••••
Results of Detailed Decomposition - Men
-.3
-.2
-.1
0
.1
.2
.3
Log
Wag
e C
hang
e
0 .2 .4 .6 .8 1Quantile
Foreign Education
Experience
Occupation
Region
Industry
Other
Results ••••••
Composition Effects - Men
-.05
0
.05
.1
.15
Log
Wag
e C
hang
e
0 .2 .4 .6 .8 1Quantile
Foreign Education
Experience
Occupation
Region
Industry
Results ••••••
Wage Structure Effects - Men
-.15
0
.15
.3
Log
Wag
e C
hang
e
0 .2 .4 .6 .8 1Quantile
Foreign Education
Experience
Occupation
Region
Industry
Constant
Results ••••••
Contribution of FDI to Changes in Log Wage Differentials
90-10 90-50 50-10
Men Total Change 0.376 0.187 0.189 FDI Composition Effect 0.021 0.034 -0.013 FDI Wage Structure Effect -0.001 0.003 -0.004
Women Total Change 0.350 0.170 0.180 FDI Composition Effect 0.010 0.018 -0.008 FDI Wage Structure Effect 0.013 0.001 0.003
Results ••••••
Distribution of Foreign Ownership Share in 2000
0
10
20
30
40
50
60
Per
cent
of F
irms
0 20 40 60 80 100Foreign Ownership Share
•Only firms with positive foreign share:
Within-Firm Representation of Workers
0
2
4
6
8
10
Den
sity
0 .2 .4 .6 .8 1Share of workers observed
Mean = 0.255, Median = 0.085
All firms
0
5
10
15
Den
sity
0 .2 .4 .6 .8 1Share of workers observed
Mean = 0.132, Median = 0.074
Emp>20
0
5
10
15
Den
sity
0 .2 .4 .6 .8 1Share of workers observed
Mean = 0.088, Median = 0.070
Emp>100
0
10
20
30
40
1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008Year
Foreign share in business sector employmentPercent of foreign firms in business sectorPercent of workers employed by foreign firms in LEEDPercent of foreign firms in LEED
Foreign Penetration in Sample and in Business Sector•Only firms with more than 20 employees
Descriptives
1992 2000
Foreign Employment Share (%) 4.6 30.5
Domestic Foreign Domestic Foreign
Monthly Earnings 116.1 152.4 131.5 202.4 (71.1) (104.7) (134.7) (225.2) Female (%) 37.4 46.7 37.0 46.0 Education (%)
Elementary 32.9 33.8 23.6 19.6 Vocational 32.7 33.5 38.4 33.7 High school 27.1 23.5 29.9 32.7 University 7.3 9.2 8.2 14.0
Experience 22.1 20.4 23.1 19.3 (10.6) (10.5) (10.9) (10.9)
Descriptives – cont.Occupation (%)
Elementary Occupations 11.1 10.8 9.6 5.4 Skilled Manual Workers 48.3 58.2 50.5 53.3 Service Workers 9.2 5.0 10.9 7.2 Clerks 6.8 5.4 5.9 6.0 Associate Professionals 12.7 9.6 12.1 14.5 Professionals 6.2 6.9 2.9 6.2 Managers 5.7 4.0 8.2 7.3
Industry (%) Agriculture 18.3 0.2 12.3 0.6 Mining 0.2 0.0 0.3 0.0 Food&Beverages 6.2 11.1 6.5 7.6 Textile 5.4 12.3 6.8 9.8 Wood&Paper 2.6 2.6 3.1 2.5 Chemicals 4.8 3.7 2.7 9.5 Minerals&Water 5.3 4.9 6.7 7.5 Machines&Equipment 8.8 43.1 9.8 26.2 Utilities 3.0 0.0 2.8 5.1 Construction 6.1 8.8 6.3 1.8 Retail Trade 9.5 7.2 7.3 7.4 Wholesale Trade 4.0 4.0 4.0 5.2 F.I.R.E. 1.5 0.1 4.5 5.9 Business Services 2.6 1.1 4.9 3.6 Other Services 21.7 0.8 22.0 7.3
N 74,724 3,869 59,987 29,932
(Recentered) Influence Functions• Consider a perturbation in wage distribution :
Then IF and RIF of the distributional statistic :
• If “moves” towards :Change in given by:
▫where
Methodology •••••
• Consider the (unconditional) wage distributions as:
▫where is a vector of covariates distributed as
• Then the IIF becomes:
• Ceteris paribus effect of location shift in distr. of covariate , so that is given by
RIF Regression I
Methodology •••••
• Functional form assumption:
• For the τth quantile, the estimated RIF is equal to
▫where is the sample quantile and is a kernel density estimate
and the data generating process in year is given by
RIF Regression II
Methodology •••••
•Decompose mean overall change in unconditional quantiles between end and base period:
• Aggregate decomposition with DFL (1996) reweighting
•Detailed decomposition with DFL (1996) reweighting
Unconditional Quantile Decomposition
Methodology •••••
Foreign Effects by Quantile in RIF Regressions Men
Women
1992 2000 1992 2000
1st Decile 0.152** 0.366** 0.188** 0.297** (0.021) (0.033) (0.021) (0.030) 2nd Decile 0.190** 0.326** 0.250** 0.338** (0.023) (0.022) (0.029) (0.028) 3rd Decile 0.204** 0.313** 0.287** 0.336** (0.028) (0.018) (0.036) (0.026) 4th Decile 0.229** 0.307** 0.312** 0.303** (0.035) (0.018) (0.046) (0.022) Median 0.262** 0.310** 0.304** 0.271** (0.041) (0.020) (0.045) (0.020) 6th Decile 0.292** 0.331** 0.277** 0.264** (0.044) (0.025) (0.047) (0.023) 7th Decile 0.347** 0.349** 0.255** 0.244** (0.051) (0.029) (0.036) (0.027) 8th Decile 0.386** 0.364** 0.246** 0.261** (0.058) (0.032) (0.032) (0.033) 9th Decile 0.424** 0.451** 0.242** 0.331** (0.057) (0.045) (0.035) (0.039) N 44,072 50,495 31,887 37,235
Foreign Presence by Quantiles of Firm-Level Average Wages (2005)
0
10
20
30
40
50
Num
ber o
f for
eign
firm
s
0 20 40 60 80 100Average wage percentile
Foreign Presence by Quantiles of Within-Firm Variances of Log Wages (2005)
0
10
20
30
40
Num
ber o
f for
eign
firm
s
0 20 40 60 80 100VLOG percentile