Engendering Trade Quy-Toan Do 1 Andrei Levchenko 2 Claudio Raddatz 1 1 Research Department, The World Bank 2 Department of Economics, University of Michigan May 2011
Millennium Development Goals The Third MDG: Promote Gender Equality and Empower Women
The MDGs in a Globalized World Two questions: 1. How does globalization aect the promotion of gender equality and the empowerment of women? 2. How does gender equality and women's empowerment aect the way a country opens to the World?
Globalization as Trade Integration Two storylines: 1. Trade integration changes relative factor prices (hence changes the wages of women relative to men) 2. Women's ability to participate in the labor market is a source of comparative advantage
Outline Motivation and Overview Theory Empirics Conclusion
Overview Theoretical Framework Two-country-two-sector specic-factors model of international trade (Jones, 1971; Mussa, 1974) Roy (1951) model of occupational choice: brain vs. brawn sector with capital owing costlessly between sectors Ricardian source of comparative advantage: relative performance of brain vs. brawn sectors Heckscher-Ohlin source of comparative advantage: female eective participation in labor market
Overview Theoretical Implications Trade integration implies change in relative factor prices: implications on gender gap depends on (Ricardian) comparative advantage Countries that discriminate less against women have a (Heckscher-Ohlin) comparative advantage in female-labor intensive sectors
Overview Empirical test Data Export data from 149 countries and 61 sectors (COMTRADE) Female labor intensity measures (ratio of female to total employment): from UNIDO Construction of female labor needs of exports: female-labor weighted exports
Overview Empirical test Results (i) Sectors that are more female-intensive have larger export shares in countries that discriminate less against women (ii) Countries that have a comparative advantage in female-intensive sectors exhibit higher gender parity Instrumental variables to account for reverse causation using (i) religion to instrument for discrimination, and (ii) predicted export shares to instrument for actual export shares.
Literature Review Brain versus brawn models: Black and Juhn (2000), Qian (2008), Alesina et al. (2011), Pitt et al. (2010), Galor and Weil (1996) Trade liberalization and female outcomes: Rendall (2010), Oostendorp (2009), Aguayo-Tellez et al. (2010), Rees and Riezman (2011), Marchand et al. (2011). Institutions and Trade: Institutions as source of comparative advantage: Beck (2003), Levchenko (2007) Trade as a driver of institutional change: Acemoglu et al. (2005), Braun and Raddatz (2008), Do and Levchenko (2007, 2009)
A Model of Trade with Gender Discrimination Specic-factors model of production Two countries i {X, Y } and two sectors s {F, M} Cobb-Douglas preferences: ( ) ( ) u i C i F, CM i = C i η ( ) F C i 1 η M Specic-factors production structure: Y i F = F i (K F ) α (L F ) 1 α Y i M = Mi (K M ) α (L M ) 1 α
A Model of Trade with Gender Discrimination Modelling gender discrimination Countries exhibit technological dierences (Ricardian comparative advantage): F X M X F Y M Y Countries vary in the extent of gender discrimination: Aggregate capital normalized to K i = 1 Supply of male labor: Li M = 1 Supply of female labor: Li F = 1 δ i Dierences in δ create Hecksher-Ohlin dierences in (eective) female labor endowment
Characterization of Equilibrium Autarky Trade solve for rms' and consumers' optimization market clearing conditions solve for rms' and consumers' optimization world market clearing conditions law of one price Endogenous determination of δ
Autarky Firms' optimization Good M is the numeraire and let f = 1 K M be the inverse of the share of capital that goes to the male sector: K F = 1 1 f M rms' optimization: r = αmf 1 α w M = (1 α) Mf α F rms' optimization: ] 1 α [ f (1 δ) r = αpf f 1 [ f (1 δ) w F = (1 α) pf f 1 ] α
Autarky Consumers' optimization - Budget balance - Market clearing Cobb-Douglas: pc F = ηe C M = (1 η) E Budget balance: expenditure = income E = r + w F (1 δ) + w M Market clearing: consumption = production ( ) f α 1 C F = F (1 δ) 1 α f
Autarky Characterization of Equilibrium After subsituting: (1 η) f = 1 Capital and labor ratios in equilibrium are given by K F = w F (1 δ) = η K M w M 1 η Factor rewards independent of δ (perfect competition and Cobb-Douglas).
Equilibrium prices Trade Firms' and consumers' optimization r i = αm i ( f i) 1 α On the consumption side: wm i = (1 α) ( Mi f i) α [ ( r i = αp i F i f i 1 δ i) ] 1 α f i 1 [ ( wf i = (1 α) pi F i f i 1 δ i) f i 1 ] α p i C i F η = C i M 1 η = r i + w i M + ( 1 δ i) w i F
Trade Law of one price Law of one price p X = p Y or M i ( ) f i 1 α 1 = M i ( f i 1 ) 1 α F i 1 δ i F i 1 δ i
Trade Law of one price Law of one price to express f i as function of f i : ( F f i i M i 1 = M i F i Since f i 1 = K F i, we have KM i ) 1 1 α 1 δ i } {{ 1 δ i } ρ ( f i 1 ) K i F K i M = ρ K i F K i M Result 1: The sector in which the country has a comparative advantage (dened by ρ) will be larger.
Trade Market clearing F good market clearing: world production = world consumption i p i F i ( f i 1 f i ) α ( ) 1 δ i 1 α [ = η r i + ( 1 δ i) wf i + M] w i Substituting the expression of prices: M i ( f i) α [ 1 (1 η) f i ] +M i ( f i) α [ 1 (1 η) f i ] = 0 i
Trade Autarky-trade comparisons Market clearing condition: M i ( f i) α [ 1 (1 η) f i ] +M i ( f i) α [ 1 (1 η) f i ] = 0 Sum of two terms being zero: Result 1 implies and (1 η) f i > 1 ρ i > 1 df i dρ i > 0 Result 2: As a result of trade opening, the sector in which the country has a comparative advantage expands; it expands by more the stronger is comparative advantage.
Trade Summary of results Prediction 1: Gender discrimination is a source of Heckscher-Ohlin comparative advantage: Countries that discriminate more against women have smaller female-labor intensive sectors.
Endogenizing the Gender Gap In the model, the gender gap is δ 1 δ measures female labor force participation (given that male lfp is normalized to 1) 1 δ measures eective female labor force participation when looking at education We propose 3 mechanisms: Opportunity cost of female labor Opportunity cost of female education Political economy
Gender Gap in LFP Occupational choice model: δ is fraction of female time spent at home Formal labor generates income w F (1 δ) Household production generates output v (δ) = δ 1 2 δ2 Households take w F as given and make occupation choices accordingly: Labor Supply Curve Firms take w F as given and make production choices accordingly: Labor Demand Curve
Gender Gap in LFP Labor supply curve Household chooses δ to maximize (1 δ) w F p η 1 (1 δ)2 2 so that f.o.c. denes the labor supply curve 1 δ = w F p η
Gender Gap in LFP Labor demand curve Given rms' rst-order conditions: w F p η = (1 α) F η 1 η (f 1)1 η(1 α) 1 M f α (1 δ) 1 η(1 α) Labor demand curve shifts up as a consequence of trade opening when the country has comparative advantage in the female sector Prediction 2: Countries that have a comparative advantage in the female-labor intensive good have higher female labor force participation (and lower fertility) once they open to trade. The reverse holds for countries with a comparative dis-advantage in female-labor intensive sectors.
Gender Gap in LFP» Ê Ê É
Gender Gap in Education Similar idea: when female wages are higher, higher returns to girls' education Prediction 2bis: Countries that have a comparative advantage in the female-labor intensive good have higher female educational attainment once they open to trade. The reverse holds for countries with a comparative dis-advantage in female-labor intensive sectors.
The Political Economy of Gender Discrimination Household welfare is split between husbands and wives according to u F = (1 ω) U u M = ωu where ω is the share of income brought home by the husband: ω = w M w M + (1 δ) w F Husbands choose δ and trade the size of the pie o against their shares of the pie
Gender Discrimination in Autarky Indepedently of female labor force participation: ω = 1 η Autarky discrimination is minimal (driven by unit elasticity of substitution /Cobb Douglas)
Gender Discrimination under Trade Bargaining power is given by Aggregate utility is given by First-order conditions: ω = 1 f [ f (1 δ) η ] 1 α U (f 1) η ρ df = (1 α) η +(1 α) ρ df (1 η) f 1 f dρ }{{} f dρ f 1 }{{}}{{} direct output loss bargaining gain allocative loss
The political economy of gender discrimination Prediction 2ter: Both countries increase discrimination against women when they open to trade. Yet, the increase in discrimination is more pronounced when the country has a comparative dis-advantage in the female labor intensive sector.
Empirical Strategy Two mechanisms: Prediction 1: sectors that are relatively more female-labor intensive, expand faster in countries that discriminate less against women Predictions 2(s): countries that have a natural comparative advantage in female-labor intensive goods will exhibit lower levels of gender discrimination
Testing Prediction 1 Empirical specication SHARE ic = νfl i Gender c + γ c + γ i + ε ic SHARE ic is the average share of good i in country c total exports Gender c is a measure of the gender gap (female LFP, fertility, education gap) FL i is the share of female workers to total employment in sector i γ c and γ i are country and sector FE IV: instrument FL i Gender c with the interaction FL i Religion c
Testing Predictions 2 Empirical specication Gender c = α + βflnx c + γz c + ε c FLNX c is female labor needs of exports (Almeida and Wolfenzon, 2005) FLNX c = i ω X ic FL i where ω X ic is the share of exports of good i by country c to c s total exports IV: instrument FLNX c using its predicted value in a gravity model (Do and Levchenko, 2007) More on gravity Panel specication: Gender ct = α + βflnx ct + γz ct + γ c + γ t + ε ct
Share of Female Workers Data sources Share of female workers FL i : UNIDO Industrial Statistics Database (IND-STAT4 2009) total employment and female share of employment in 61 distinct sectors FL i is average over 11 countries for which data is available After aggregation: 61 manufacturing sectors High correlation of FL i across countries (.89 Spearman rank correlation) Robustness of results to using US Labor statistics data
Share of Female Workers Descriptive statistics Last 5 FLi Top 5 FLi Bodies for motor vehicles.08 Wearing apparel except fur.71 apparel Railway and tramway.08 Knitted and crocheted fabrics.62 locomotives Motor vehicles.09 Footwear.49 Building and repairing of ships.09 Other textiles.47 and boats Basic iron and steel.10 Electronic valves and tubes and other electronic components.46 Mean Min Max SD N Share of female workers.27.08.71.13 61
Export Shares and Female Labor Needs of Exports Data sources Export shares ωict X : COMTRADE database (1962 onwards) aggregated up ISIC Rev 3 Aggregation to country-level to form Female labor needs of exports: FLNX c = i ω X ic FL i
Female Labor Needs of Exports Summary statistics Last 5 FLNX Top 5 FLNX Lybia 13.3 Lesotho 65.7 Iraq 13.4 Bangladesh 53.1 Saudi Arabia 13.7 Haiti 51.0 Kuwait 13.7 Mauritius 47.9 Gabon 13.8 Mongolia 44.6 Last 5 Changers FLNX Top 5 Changers FLNX Congo -13.6 Cambodia 42.3 Chad -15.4 Honduras 32.6 Niger -16.1 Haiti 27.6 Yemen -16.8 Albania 23.3 Angola -17.9 Sri Lanka 22.6
Female Labor Needs of Exports Summary statistics OECD Non-OECD Mean SD N Mean SD N 1960s 25.1 4.3 20 27.1 7.7 100 1970s 23.9 3.4 20 26.0 7.8 103 1980s 24.4 4.3 20 26.9 9.4 103 1990s 26.2 4.3 20 30.2 11.1 123 2000s 25.6 3.3 21 29.3 12.4 127
Summary Statistics N Mean SD (Log) Female Labor Need of Exports 146 3.29 0.35 (Log) Real GDP per capita 146 8.57 1.14 (Log) Openness 146 4.19 0.56 (Log) Female Years of Schooling 146 1.63 0.69 (Log) Fertility 146 1.15 0.55 (Log) Female Labor Force Participation 146 3.87 0.39 (Log) Population 146 16.05 1.38 (Log) Predicted Female Labor Need of Exports 146-1.43 0.06
How Gender Discrimination Distorts Trade OLS results (1) (2) (3) (4) (5) (6) Dep. Var.: Sectoral Share of Exports FL i * Female Labor Force Participation 0.09** (0.036) 0.08* (0.043) FL i * Fertility -0.08 (0.259) -0.39 (0.322) FL i * Ratio of Female to Total Education 0.29 (4.122) 3.97 (5.487) Capital Intensity * Capital Abundance Yes Yes Yes Skill Intensity * Skill Abundance Yes Yes Yes Country/Sector Dummies Yes Yes Yes Yes Yes Yes R 2 0.246 0.254 0.245 0.253 0.227 0.239 Observations 8,533 6,190 8,533 6,190 7,361 5,732
How Gender Discrimination Distorts Trade 2SLS results (1) (2) (3) (4) (5) (6) Dep. Var.: Sectoral Share of Exports FL i * Female Labor Force Participation 0.23*** (0.072) 0.19** (0.096) FL i * Fertility -2.27*** (0.761) -1.59* (0.761) FL i * Ratio of Female to Total Education 11.07* (6.450) 4.51* (7.125) Capital Intensity * Capital Abundance Yes Yes Yes Skill Intensity * Skill Abundance Yes Yes Yes Country/Sector Dummies Yes Yes Yes Yes Yes Yes Observations 8,416 6,075 8,416 6,075 7,244 5,501 First Stage Diagnostics First Stage Results F-test 359.32 359.32 367.44 252.18 604.16 740.47 Partial R 2 0.947 0.947 0.911 0.920 0.977 0.981
How Comparative Advantage Aects Female Attainment Measures OLS Cross-Country Results (1) (2) (3) (4) Dep. Var.: Fertility Rate Fertility Rate Labor Force Participation Educational Attainment (Log) FLNX -0.30*** (0.082) -0.31*** (0.080) 0.19* (0.112) 0.24* (0.138) Controls Yes Yes Yes Yes R 2 0.636 0.649 0.116 0.592 Observations 146 146 146 126
How Comparative Advantage Aects Female Attainment Measures 2SLS Cross-Country Results (1) (2) (3) (4) Dep. Var.: Fertility Rate Fertility Rate Labor Force Participation Educational Attainment (Log) FLNX -0.49*** (0.136) -0.51*** (0.126) 0.30* (0.160) 0.25 (0.190) Controls Yes Yes Yes Yes First Stage (Log) Predicted FLNX 3.14*** (0.326) 3.14*** (0.320) 3.14*** (0.320) 3.08*** (0.330) F-test 45.38 34.70 34.70 29.73 R 2 0.385 0.387 0.387 0.397 Observations 146 146 146 126
How Comparative Advantage Aects Female Attainment Measures Panel Results (1) (2) (3) (4) (5) (6) (7) Dep. Var.: Fertility Fertility Fertility FLFP FLFP Education Education (Log) FLNX -0.39*** -0.27*** -0.23*** 0.20*** -0.02 0.39*** -0.05 (0.031) (0.030) (0.028) (0.045) (0.026) (0.067) (0.060) Controls Yes Yes Yes Yes Yes Yes Yes Country FE No Yes Yes No Yes No Yes Year FE No No Yes No Yes No Yes R 2 0.609 0.929 0.938 0.113 0.968 0.508 0.949 N 1,247 1,247 1,247 819 819 1,102 1,102
Concluding Remarks Summary of ndings Empirical evidence that gender discrimination aects countries' integration in world markets Reciprocally, evidence that globalization aects the extent to which women are being discriminated against
Concluding Remarks Policy implications - Further issues Heterogeneity of globalization experiences Policies to reduce gender discrimination will face varying constraints depending on countries' specic industrial structures. Globalization increases rather than decreases that heterogeneity. The role of trade in fertility transitions - Industrial policy?
Do-Levchenko (2007) Instrument Use geography to explain trade volumes in each sector In each sector, run the Frankel-Romer (1999) regression: LogX icd = α i + η 1 i ldist cd + η 2 i lpop c + η 3 i larea c + η 4 i lpop d + η 5 i larea d + η 6 i landlocked cd + η 7 i border cd + η 8 i border cd ldist cd + η 9 i border cd pop c + η 10 i border cd area c + η 11 i border cd pop d + η 12 i border cd area d + η 13 i border cd landlocked cd + ɛ icd
Do-Levchenko (2007) Instrument Predict trade in each country pair, sum across all trading partners d = 1,..., C, and exponentiate: ˆX ic = C d = 1 d c e LogX icd. Predicted instrument actual FNLX with predicted FNLX : FLNX c = I ω ic X FL i. i=1 Back to presentation
How Gender Discrimination Distorts Trade (1) (2) (3) (4) (5) (6) First Stage Dep. Var.: FLi * FLFP FLi * FLFP FLi *Fertility FLi *Fertility FLi *Ratio of Female to Total Education FLi *Ratio of Female to Total Education FLi *Muslim Share of Population -30.53*** (1.151) -29.67*** (1.775) 2.71*** (0.111) 2.54*** (0.156) -0.23*** (0.012) -0.20*** (0.014) FLi *Christian Share of Population -12.61*** (0.883) -11.30*** (1.499) 0.68*** (0.098) 0.37*** (0.138) 0.08*** (0.010) 0.12*** (0.013) F-test 359.32 180.25 367.44 252.18 604.16 740.47 Partial R 2 0.947 0.942 0.911 0.920 0.977 0.981 Back to presentation
How Comparative Advantage Aects Gender Discrimination Fertility as a proxy of Gender Discrimination - 2SLS Results 2SLS (4) (5) (6) First Stage Dep. Var.: (Log) FLNX (Log) Predicted FLNX 1.545*** (0.209) 1.548*** (0.212) 1.412*** (0.226) F-test 54.65 20.19 16.20 R 2 0.327 0.328 0.397 Observations 146 146 146 Back to presentation
How Comparative Advantage Aects Gender Discrimination Female LFP as a proxy of Gender Discrimination - 2SLS Results 2SLS (4) (5) (6) First Stage Dep. Var.: (Log) FLNX (Log) Predicted FLNX 1.545*** (0.209) 1.548*** (0.212) 1.412*** (0.226) F-test 54.65 20.19 16.20 R 2 0.327 0.328 0.397 Observations 146 146 146 Back to presentation
How Comparative Advantage Aects Gender Discrimination Female educational attainment as a proxy of gender discrimination - 2SLS Results (4) (5) (6) First Stage Dep. Var.: (Log) FLNX (Log) Predicted FLNX 1.543*** (0.230) 1.538*** (0.221) 1.436*** (0.226) F-test 29.37 20.43 15.82 R 2 0.347 0.373 0.486 Observations 126 126 126 Back to presentation