Trade Reform and Worker Flows in Brazil Marc-Andreas Muendler UC - - PowerPoint PPT Presentation

LOCAL IMPACTS OF GLOBAL MARKETS

Trade Reform and Worker Flows in Brazil

Marc-Andreas Muendler UC San Diego partly joint with Na´ ercio Menezes, U and IBMEC S˜ ao Paulo ILO-WTO Workshop on Global Trade and Employment: Sept 1, 2009

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Trade, Employment and Earnings

∙ The real wage is 푊푖 푃푖 for income group 푖 ∙ Most of the literature is concerned with 푊푖 conditional on employment ∙ But there are spells of unemployment with 푊푖 = 0 after trade reform ∙ Trade reform also alters 푃푖, expectedly lowering it to different degrees for different income groups

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Reform and Growth

∙ Despite vigorous pro-competitive and trade reforms, growth in Latin America throughout the 1990s remained relatively slow ∙ Lacking resource reallocation following pro-competitive reform may be a cause of sluggish performance ∙ Brazil’s trade reform triggers worker displacements particularly from protected industries, as trade theory predicts and welcomes ∙ But neither comparative-advantage industries nor exporters absorb trade-displaced workers for years

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Empirical Objective and Findings

∙ Assess labor reallocation after Brazil’s large-scale trade reform ∙ Linked employer-employee data permit tracking of individuals between identified establishments over 1986-2001 period ∙ Shifts in product-market shares to more advanced firms and exporters commonly interpreted as evidence for successful resource realloca- tion ∙ But labor flows away from export sectors and exporters because labor productivity increases faster than production so that output shifts to more productive firms but labor does not

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Tariffs and Reallocations at Four-year Horizon

Failed reallocations Reallocation durations

0.63 0.63 0.54 0.43 0.37 0.28 0.21 0.17 0.15 0.13 0.15 0.15 0.20 0.22 0.24

.1 .15 .2 .25 Workers Without Reallocation (share) Product Market Tariff (ad valorem) 86 87 88 89 90 91 92 93 94 95 96 97 Calendar year Product Market Tariff Displaced Workers Without Reallocation

0.63 0.63 0.54 0.43 0.37 0.28 0.21 0.17 0.15 0.13 0.15 0.15 6.84 10.06 10.25

6 7 8 9 10 Reallocation Duration (months) Product Market Tariff (ad valorem) 86 87 88 89 90 91 92 93 94 95 96 97 Calendar year Product Market Tariff Reallocation Duration

Sources: RAIS 1986-2001 (1% random sample), workers nationwide of any age and gender, dis- placed from a formal-sector job; rehired into a formal-sector job within 48 months. Product tariffs from Kume et al. (2000), employment weighted at N´ ıvel 50 sector level.

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

A Literature View

∙ Haltiwanger, Kugler, Kugler, Micco & Pag´ es (2004): Sectors in six Latin American countries; tariff reductions and appreciations raise churning, reduce employment (also Wacziarg & Wallack 2004) ∙ Goldberg & Pavcnik (2003): Industry-level two-step approach; tariff declines associated with informality in Colombia, not in Brazil ∙ Saint-Paul (1997), Cunat & Melitz (2006): Countries with flexible fac- tor markets export products from industries with high factor turnover ∙ Melitz (2003), Bernard, Redding & Schott (2007): Exporting firms within sectors; Raith (2003): Competition and performance

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Labor Market Performance and Economic Outcomes

1986 1990 1992 1994 1998 FAILED REALLOCATIONS WITHIN A YEAR Mean failure rate (share of displaced) .285 .354 .441 .391 .474

female workers .387 .427 .500 .451 .517 young workers .297 .361 .445 .384 .446 high-school or college educ. workers .305 .350 .416 .366 .435

Idle labor (foregone share of GDP) .000 .024 .009 .037 DURATIONS OF SUCCESSFUL REALLOCATIONS WITHIN A YEAR Mean duration (in months) 2.918 3.927 4.280 4.125 4.253

female workers 3.157 3.965 4.097 4.017 4.097 young workers 2.896 3.909 4.184 3.969 4.105 high-school or college educ. workers 2.558 3.397 3.622 3.458 3.633

Idle labor (foregone share of GDP) .000 .008 .004 .008

Sources: RAIS 1986-1999 (1% random sample), workers nationwide of any age and gender, displaced from a formal-sector job; not rehired into a formal-sector job within 12 months (upper panel) or rehired into a formal-sector job within 12 months (lower panel). PME 1986-1999, share of idle workers (unemployed

• r withdrawn from labor force), and Banco Central do Brasil, GDP

.

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Productivity Change and Market Shares, Olley & Pakes (1996)

TFP and Output shares Labor Prod. and Employment shares Cross section

• Ann. chg.

Cross section

• Ann. chg.

wgtd. unwgtd. cov. raw cov.∗ wgtd. unwgtd. cov. raw cov.푎 (1) (2) (3) (4) (5) (6) (7) (8) 1986 1.018 .924 .095 1.011 1.019

• .008

1990 1.000 .899 .101 .065 1.000 .997 .003

• .029

1994 1.013 .918 .096 .067 1.023 1.019 .005

• .043

1998 1.035 .910 .125 .047 1.073 1.043 .030

• .039

푎Four-year average of the raw covariance between annual share changes and outcome changes.

Source: PIA firms 1986-98 (1991 missing); log total factor productivity based on Olley & Pakes (1996) estimation (at N´ ıvel 50), inferring labor productivity at changing capital stocks. Note: Cross-sectional productivity decomposition as in Olley & Pakes (1996): 푦푡 = ¯ 푦푡 + ∑

푖 ∆휃푖푡∆푦푖푡,

where 푦푡 is weighted and ¯ 푦푡 is unweighted mean log productivity and ∆ denotes deviations from cross- section means (rebased to unity in 1990). Annual productivity change correlation ∑

푖∈퐶 ∆휃푖,푡∆푦푖,푡 (raw

covariance) from Haltiwanger (1997) decomposition, where ∆ denotes annual change (not rebased).

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Workforce Changeover

∙ Two salient workforce changeovers evident from labor-demand de- composition based on Katz-Murphy (1992) framework – Within traded-goods sector, marked occupation downgrading and simultaneous education upgrading – Between sectors, labor demand shift towards least and most skilled (traceable to weaker declines of low-skill intensive traded-goods industries and stronger expansions of high-skill intensive nontraded-

• utput industries)

∙ Observations broadly consistent with predictions of Heckscher-Ohlin trade theory for low-skill abundant economy

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Occupational Workforce Recomposition

Traded-goods sectors Nontraded-output sectors

14.1 60.2 7.1 8.8 9.7 13.8 59.9 7.5 9.0 9.8 13.4 59.5 7.5 9.6 10.0 13.6 59.3 7.6 9.3 10.1 13.3 59.5 7.6 9.3 10.3 12.7 59.7 7.8 9.3 10.5 10.0 62.9 7.5 9.2 10.3 10.3 63.8 7.4 8.7 9.9 8.2 67.9 7.0 8.2 8.7 8.4 68.7 6.4 8.0 8.4 8.3 69.2 6.0 8.0 8.5 8.5 70.2 5.6 7.6 8.0 8.4 70.5 5.7 7.5 7.9 8.7 70.2 5.5 7.8 7.8 8.8 70.6 5.3 7.7 7.6 8.7 70.9 5.2 7.6 7.5

20 40 60 80 100 Share 1986 87 88 89 90 91 92 93 94 95 96 97 98 99 2000 01 Unskilled Bl. Collar Skilled Bl. Collar Unskilled Wh. Collar Tech’l. or Superv.

• Prof. or Manag’l.

14.0 34.4 15.3 20.1 16.3 13.5 34.9 15.4 20.0 16.1 13.5 35.0 15.4 19.7 16.4 13.3 34.6 15.5 20.0 16.6 13.2 34.2 15.7 20.1 16.7 13.8 33.6 16.0 20.0 16.6 12.9 32.4 16.6 21.8 16.4 13.3 31.5 16.7 22.3 16.2 14.2 29.4 17.5 22.5 16.4 15.2 29.2 17.7 21.2 16.7 15.4 28.8 18.1 21.2 16.5 15.8 29.2 17.8 21.0 16.2 16.0 29.3 17.2 21.4 16.0 15.8 28.9 17.3 21.6 16.5 15.7 29.2 17.0 21.7 16.4 15.8 29.5 16.8 21.6 16.3

20 40 60 80 100 Share 1986 87 88 89 90 91 92 93 94 95 96 97 98 99 2000 01 Unskilled Bl. Collar Skilled Bl. Collar Unskilled Wh. Collar Tech’l. or Superv.

• Prof. or Manag’l.

Source: RAIS 1986-2001 (1-percent random sample), male workers nationwide, 25 to 64 years

• ld, with employment on December 31st.

Note: Traded-goods sectors are agriculture, mining and manufacturing (subsectors IBGE 1-13 and 25), nontraded-output industries are all other sectors. Shares based on worker numbers.

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Difference between Schooling within Occupations and Mean Schooling

Traded-goods sectors Nontraded-output sectors

−1.7 −1.6 −1.1 −0.9 −1.2 −1.1 −1.0 −1.0 1.0 0.7 1.5 1.7 4.0 3.9 4.3 4.3 5.6 4.9 5.4 5.4

−4 −2 2 4 6 Years of Schooling 1986 1988 1990 1992 1994 1996 1998 2001 Calendar year Unskilled Bl. Collar Skilled Bl. Collar Unskilled Wh. Collar Tech’l. or Superv.

• Prof. or Manag’l.

−3.0 −2.6 −3.2 −3.1 −2.4 −2.4 −2.0 −1.8 −0.3 −0.4 −0.3 −0.2 2.8 2.7 2.5 2.3 4.7 4.2 3.8 3.4

−4 −2 2 4 6 Years of Schooling 1986 1988 1990 1992 1994 1996 1998 2001 Calendar year Unskilled Bl. Collar Skilled Bl. Collar Unskilled Wh. Collar Tech’l. or Superv.

• Prof. or Manag’l.

Source: RAIS 1986-2001 (1-percent random sample), male workers nationwide, 25 to 64 years

• ld, with employment on December 31st.

Note: Traded-goods sectors are agriculture, mining and manufacturing (subsectors IBGE 1-13 and 25), nontraded-output industries are all other sectors. Mean years of schooling weighted by worker numbers within occupations, less mean years of schooling weighted by worker numbers across all

• ccupations.

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Separations and Accessions

∙ Foreign competition significantly raises separation rates at formal- sector manufacturing firms ∙ Sectors with revealed comparative advantage and exporters exhibit significantly higher separation and significantly lower accession rates ∙ There is a significant monotonic increase in worker separations from manufacturing over time, and a significant (almost) monotonic drop in worker accessions

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Worker-fixed Effect Estimation: Separations & Accessions 1990-98

Separations Accessions cLogit cLogit

OLS-FE

cLogit cLogit

OLS-FE

• Prod. Mkt. Tariff
• 1.129
• 1.845
• .269

1.767 2.232 .283

(.292)∗∗∗ (.333)∗∗∗ (.034)∗∗∗ (.261)∗∗∗ (.324)∗∗∗ (.034)∗∗∗

• Intm. Input Tariff

3.237 3.378 .354

• 3.217
• 3.479
• .405

(.452)∗∗∗ (.508)∗∗∗ (.050)∗∗∗ (.395)∗∗∗ (.482)∗∗∗ (.050)∗∗∗

Exporter Status .291 .285 .044

• .416
• .414
• .054

(.019)∗∗∗ (.019)∗∗∗ (.002)∗∗∗ (.017)∗∗∗ (.017)∗∗∗ (.002)∗∗∗

BADV 1990 .139

• .016

(.015)∗∗∗ (.013)

Tenure at plant 1.460 1.460 .080

(.028)∗∗∗ (.028)∗∗∗ (.001)∗∗∗ Year effects yes yes yes yes yes yes Sector effects yes yes yes yes Obs. 306,907 306,907 567,650 283,665 283,665 562,410 Pseudo 푅2 .126 .127 .048 .062 .062 .030 Source: RAIS 1990-98 (one-percent random sample), workers nationwide of any gender or age, sepa- rated from or acceding into manufacturing job; and complementary data. Note: Sector, plant and worker controls not reported. Separations exclude transfers, deaths, and retire- ments; accessions exclude transfers. Reference obs. are employments with no separation or accession. Robust standard errors in parentheses: ∗ significance at ten, ∗∗ five, ∗∗∗ one percent.

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Empirical Concerns

∙ Labor-market institutions: Severance pay increases with tenure ∙ Firms choose exporting status simultaneously with employment IV: Imports from other countries into Brazil’s export destinations ∙ Tariff reductions are targeted at low-efficiency sectors; sector characteristics correlate with labor turnover IV: Components of the sectoral real exchange rate ∙ Linear worker-FE IV regressions; higher-order interactions

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

Trade Exposure and Predicted Labor Market Outcomes

1990 1992 1994 1998 Trade Exposure Product Market Tariff .384 .210 .144 .170 Intermediate Input Tariff .298 .159 .111 .133 Change in Separation rates predicted by tariff drops for all sectors and firms

• .011
• .011
• .007

for top comparative-advantage quintile

• .028
• .026
• .017

for exporters .005 .012 .013 Change in Accession rates predicted by tariff drops for all sectors and firms

• .066
• .093
• .084

for top comparative-advantage quintile

• .140
• .200
• .182

for exporters

• .058
• .083
• .076

Sources: RAIS 1990-98, workers nationwide of any gender or age, with manufacturing job; and comple- mentary data. Note: Sector information at subsector IBGE level. Predicted changes in separation and accession rates based on marginal effects of product and intermediate-input tariff changes implied by column (2) and (4) estimates in trade-interaction table ( ˆ 푃(1− ˆ 푃) is .175 for separations and .179 for accessions; interaction terms evaluated at comparative advantage in 1990 and share of exporters in current year).

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

The Evidence So Far

∙ Trade reform triggers multiple changes to separations and accessions ∙ Comparative-advantage industries and exporters do not expand em- ployment in the short term ∙ Falling tariffs depress accessions at manufacturers, especially in com- parative-advantage industries and at exporters; falling tariffs markedly raise separations from exporters ∙ Failed reallocations and longer durations of reallocations result

ILO-WTO Workshop Geneva 2009: Trade Reform and Worker Flows in Brazil c

⃝ Marc Muendler

What Is Amiss

∙ Distributional effects of trade for the employed are “established” within classic frameworks but little understood in novel settings (Helpman, Itskhoki & Redding 2009) ∙ Employment effects of trade are beginning to become understood but wage effects upon re-employment remain amiss (cf. Jacobson, LaLonde & Sullivan 1993) ∙ The incidence of real wage changes also depends on the consump- tion basket for different income groups (cf. Porta 2006; Broda & Ro- malis 2009)