Trade and Inequality: From Theory to Estimation

Transcription

1 Trade and Inequality: From Theory to Estimation Elhanan Helpman, Harvard and CIFAR Oleg Itskhoki, Princeton Marc Muendler, UCSD Stephen Redding, Princeton December 2012 HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

2 Motivation The relationship between wage inequality and international trade has been widely debated In recent decades, developed countries such as the United States have experienced both rising wage inequality and trade expansion A number of developing countries have experienced increased wage inequality in the aftermath of trade liberalization Goldberg and Pavcnik (2007) These findings are hard to reconcile with the traditional Heckscher-Ohlin framework, which is focused on: 1 Between-group wage inequality for skilled and unskilled workers 2 Reallocations across sectors with different factor intensity and changes in factor intensity within sectors 3 Wage inequality rises in skill-abundant countries and declines in skill-scarce countries HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

3 More Recent In the data most inequality increase is within-groups of workers with similar characteristics New theoretical models have emphasized novel mechanisms for trade to affect wage inequality 1 Wage dispersion across firms within industries 2 Within-group wage inequality for workers with the same observed characteristics We propose a theory-grounded methodology for estimating the link between trade and firm-driven wage inequality The methodology is illustrated with Brazilian data HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

4 This Paper Uses linked employee-employer data for Brazil from to provide reduced-form evidence on the distribution of wages across workers and firms Establishes stylized facts about Brazilian wage inequality within sector-occupations for workers with similar observables (residual inequality) between firms Develops a structural model to quantify the role of firm heterogeneity in wage inequality extension of HIR ( Econometrica, 2010) a model of within-sector, between-firm residual inequality wages and employment vary with firm productivity and trade participation Estimates this model and assess the role of trade in wage inequality HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

5 Related Literature Theories of firm heterogeneity and trade following Melitz (2003) with wage differences across firms Search and matching frictions: Cosar, Guner & Tybout (2010), Helpman, Itskhoki & Redding (2010) Effi ciency wages: Davis & Harrigan (2011) Fair wages: Amiti & Davis (2012), Egger & Kreickemeier (2009) Evidence on the employer-size wage premium and wage dispersion across plants Davis & Haltiwanger (1991), Abowd & Kramarz (1999), Oi & Idson (1999), Abowd et al. (2001) Literature on firm exporting and wages using plant-level data and matched employer-employee data Bernard & Jensen (1995, 1997), Amiti & Cameron (2011), Schank et al. (2007), Davidson et al. (2010), Krishna et al. (2010), Eaton et al. (2011) HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

6 Brazilian Data Matched employer-employee data from All workers employed in the formal sector Focus on the manufacturing sector Observe firm, industry and occupation Observe worker education (high school, college degree), demographics (age, sex) and experience (employment history) Our sample includes around 7 million workers and 100,000 firms in each year Trade transactions data from Merged with the linked employee-employer data Observe whether a firm exports HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

7 Sectors Twelve aggregate sectors (IBGE) More than 250 disaggregated industries (CNAE) HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

8 Occupations Five aggregate occupations More than 300 disaggregated occupations (CBO) HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

9 Within Sector-Occupation Inequality Decompose overall wage inequality into within and between sector-occupation components 1 N t T = W + B N t (w it w t ) 2 = 1 N i=1 t k K N kt (w it w kt ) N i=1 t N kt ( w kt w t ) 2 k K where workers are indexed by i and time by t k denotes sector, occupation or sector-occupation cells N t is the number of workers w it is the log wage and a bar denotes a mean HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

10 Within Sector-Occupation Inequality Level Growth Panel A (1990) ( ) Within occupation 80% 92% Within sector 83% 73% Within sector-occupation 67% 67% Within detailed-occupation 58% 60% Within sector- 52% 54% detailed-occupation Panel B (1994) ( ) Within sector-occupation 68% 125% Within detailed-sector- 47% 140% detailed-occupation HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

11 Within-Group Wage Inequality Estimate Mincerian log wage equation for worker i for each year t: w it = z it ϑ t + ν it Observables (z it ): indicator variables for Education (high-school, college degree) Age bins and male/female Experience quintiles Decompose wage dispersion into observables and within-group wage inequality var (w it ) = var ( z it ˆϑ t ) + var (ˆνit ) Decompose within-group wage inequality into within and between sector-occupation components HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

12 Within-Group Wage Inequality Level Growth A. Overall wage inequality (1990) ( ) Observables inequality 43% 52% Residual wage inequality 57% 48% B. Residual wage inequality (1990) ( ) Within sector-occupation 88% 91% (1994) ( ) Within sector-occupation 89% 103% Within detailed-sector- 83% 110% detailed-occupation HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

13 Within-Group Wage Inequality 3 HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

14 Between-Firm Inequality For each sector-occupation (k) and year (t) separately, estimate a Mincerian log wage equation: w it = z it ϑ lt + ψ jlt + ν it where i indicates worker and j denotes firm Observables (z it ): indicator variables for education, age bins, sex, and experience bins Firm-occupation-year fixed effects (ψ jlt ) control for overall wage differences across firms, including Differences in wage premia for the same worker characteristics Differences in unobserved workforce composition Decompose wage dispersion within sector-occupations into: Observables component (dispersion of z it ˆϑ kt ) Between-firm component (dispersion of firm fixed effects ˆψ jlt ) Covariance of observables and firm fixed effects components Within-firm component (dispersion of ˆν it ) HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

15 Between-Firm Inequality Within Sector-Occupation Bins Level Growth A. Unconditional (1990) ( ) Between-firm wage inequality 55% 115% Within-firm wage inequality 45% 15% B. Controlling for Worker Observables Worker observables 17% 2% Covariance observables and firm effects 11% 24% Between-firm wage inequality 38% 86% Within-firm wage inequality 34% 11% HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

16 Theory Helpman, Itskhoki and Redding (2010) Brands are produced by heterogeneous firms, with productivity e θ Fixed and variable trade costs Monopolistic competition in the product market Search and matching in the labor market Workers draw a match-specific ability a from a Pareto distribution Multilateral wage bargaining CES preferences with an elasticity of substitution 1/ (1 β) HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

17 Theory Production: the production function is: y = e θ H γ ā, 0 < γ < 1 Screening: a firm can identify workers with productivity above a c at cost e η Ca δ c /δ ā = k k 1 a c As an identifying restriction we assume η θ Potential justification: θ driven primarily by technology, η by organizational features Search and matching: DMP labor market, multilateral wage bargaining W = Fixed export costs: heterogeneous βγ R 1 + βγ H e ε F x = bak /δ c HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

18 Model Predictions A firm with idiosyncratic shock {θ, η, ε}: R = κ r Υ 1 β Γ (e θ) β ( β(1 γk) Γ e η) δγ (1 β)(1 k /δ) H = κ h Υ Γ (e θ) β(1 k /δ) β(1 γk)(1 k /δ) Γ (e η) δγ k δ W = κ w Υ k(1 β) δγ (e θ) βk ( ( δγ e η) k δ 1+ β(1 γk) δγ ) Market access variable Υ = 1 + I x ( Υx 1 ), Υ x = 1 + τ β 1 β A x A d Selection into exporting ( ) 1 β ( Γ I x = {κ π Υx 1 e θ) β ( Γ e β(1 γk) η) δγ F x e ε } HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

19 Econometric Model Empirical model of X j = {h j, w j, ι j } j : h = α h + µ h ι + u, w = α w + µ w ι + ζu + v, ι = I{z f }, ( u, v, z ) N 0, σ 2 u 0 ρ u σ u 0 σ 2 v ρ v σ v ρ u σ u ρ v σ v 1 where ( ( ) ) 1 β u = 1+χ Γ θ + β(1 γk) δγ χ η, v = χ ( η E{η u} ) = χ ( η πu ), z = σ 1 χ = k /δ 1 k /δ ( β Γ θ + β(1 γk) δγ ) η ε = σ 1 ( ) (1 + ζ)u + v ε, > 0 and ζ = χ(1 + π). µ h = (1 k/δ) log Υ x (1 β)/γ, µ w = k /δ 1 k /δ µ h, ( f = σ 1 α π + log F x log [ ]) Υ x (1 β)/γ 1. We estimate the parameter vector Θ = {α h, α w, ζ, σ u, σ v, ρ u, ρ v, µ h, µ w, f }. HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

20 Maximum Likelihood Estimation L ( Θ X ) = j P{x j Θ}, P{x j Θ}σ u = φ ( û j ) 1σv φ ( ˆv j ) Φ ( f ρ u û j ρ v ˆv j 1 ρ 2 u ρ 2 v ) 1 ιj ( ( 1 Φ û j = ( h j α h µ h ι j ) /σu, ˆv j = [ (w j α w µ w ι j ) ζ(h j α h µ h ι j ) ] /σ v. f ρ u û j ρ v ˆv j 1 ρ 2 u ρ 2 v )) ιj HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

21 Estimates HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

22 Likelihood Function HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

23 Firm-Level Moments HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

24 Worker-Level Moments HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

25 Firm Distributions HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

26 Worker Distributions Data Model HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

27 Dynamic Fit Change in the Variance of Log Worker Wages Data Structural HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

28 Counterfactual Reduction in Variable Trade Costs HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29

29 Conclusions Neoclassical trade theory emphasizes wage inequality between occupations and industries In contrast, our theory points to wage dispersion within occupations and industries Using matched employer-employee data for Brazil, we show that Around 2/3 of wage inequality is within sector-occupations Residual wage inequality is as important as worker observables Betwen-firm component accounts for much of within sector-occupation inequality Between-firm wage dispersion related to trade participation Develop a framework for estimation of a model with firm heterogeneity and wage dispersion across firms Use this framework to quantify the contribution of changes in trade openness to wage inequality and to undertake counterfactuals HIMR (Harvard, Princeton, UCSD and Princeton) Trade and Inequality December / 29