Export markets and labor allocation in a low-income country

Theo tin IMF

Workshop on Macroeconomic Policy and Income Inequality IMF
February 10, 2017
Brian McCaig Wilfrid Laurier University
Nina Pavcnik Dartmouth College
• Low-income countries have drastically increased their engagement in international trade (Hanson (2012))
○ Low and middle income country share of world exports grows from 21 to 43% between 1992 and 2008
○ Their share of exports in GDP more than doubles from 26 to 55% between 1992 and 2008

• 70 to 80% of employment in low-income countries in small, informal, household-run businesses
○ Gollin (2002, 2008), Nataraj (2011)
• Informal businesses are less productive than formal counterparts ○ La Porta and Shleifer (2008, 2014), Nataraj (2011)
• Systematic negative relationship between prevalence of employment in informal, household businesses and economic development
○ Aggregate income differences across countries reflect inefficient allocation of resources across sectors and firms (Hsieh and Klenow (2009), Restuccia and Rogerson (2008))
○ Reallocation of workers from informal businesses to more productive establishments can increase aggregate productivity

• International trade can contribute to economic development through reallocation of workers out of household businesses toward more productive establishments
○ Large literature on trade and labor markets in less developed countries (Goldberg and Pavcnik (2007), Harrison, McMillan, and McLaren (2011))
• An unexplored margin of labor reallocation in response to trade due to a combination of data limits:
○ Only data on formal sector workers,
○ Only data on urban workers, or
○ Unable to observe this margin in the data

This paper
• Examines the relationship between international trade and allocation of labor across the informal and formal sectors in a low-income country
• The labor force module of the Vietnam Household Living Standards Survey includes information on the household business sector
○ The survey records whether a person works for an employer in household business or registered enterprise sector
○ Focus on reallocation across this margin
○ 85% of workers economy wide (66% in manufacturing) employed in household business sector
• Large export shock due to the U.S.-Vietnam Bilateral Trade Agreement (BTA) implemented in December 2001
○ Main policy change is a decline in U.S. tariffs on VN exports
• Evaluate the impact of this reallocation on aggregate labor productivity in manufacturing

Related Literature
• International trade and labor markets
○ Goldberg and Pavcnik (2007), Harrison, et. al. (2011), Kovak (2010),
Topalova (2010), McCaig (2011), Menezes-Filho and Muendler (2011)
• Policy distortions and aggregate productivity in developing countries
○ Restuccia and Rogerson (2008), Hsieh and Klenow (2009), Bloom and Van
Reenan (2007), McMillan and Rodrik (2011)
○ Prevalence of microenterprises notable component of firm distribution in LDCs
○ Gollin (2002, 2008), Hsieh and Olken (2014), La Porta and Shleifer (2014)
○ Many studies focus on the effects of supply-side constraints on the growth and formalization of microenterprises (Banerjee and Duflo (2005), de Mel, McKenzie, Woodruff (2013))
○ We examine the role of trade distortions, which disproportionally lower profitability of more productive establishments (Melitz (2003))

• Conceptual framework
• The nuts and bolts of the study
○ The U.S-Vietnam Bilateral Trade Agreement
○ Household business definition
○ Data description
○ Empirical methodology
• Findings
• Aggregate productivity gains from reallocation
• Conclusions

Export Costs and Household Business Employment
• Expansion in exports due to declines in US tariffs on Vietnamese exports
• Firms differ in profitability due to heterogeneity in marginal cost of production
○ Marginal costs differences could reflect entrepreneurial ability (Lucas (1978)) or productivity (Melitz (2003))
○ Hh businesses relatively less productive (Gollin (2008), La Porta and Shleifer (2014))
• Fixed cost of exporting
○ Declines in export costs disproportionally benefit more productive firms
• Labor reallocates from household businesses toward more productive firms

Export Costs and Household Business Employment
• Household businesses do not need to directly compete with exporting firms in product markets
• General equilibrium effects through labor market
○ Declines in export cost increase aggregate labor demand
○ This increases the opportunity costs of working in a household business (Lucas (1978))
○ Labor reallocates away from household businesses toward other employers
• This channel consistent with increased wages in areas of VN with greater pre- BTA concentration of exporting industries (McCaig (2011))

U.S.-Vietnam Bilateral Trade Agreement
• Implemented in December 2001
• Negligible changes in VN import tariff commitments to US
○ MFN tariffs to the US prior to the agreement
○ Long term commitment to improve market access in services, IPR, FDI
• US grants Vietnam “Most Favored Nation” status
○ Vietnam moved from Column 2 to MFN tariff schedule
• Tariffs decline on average by 21.1 percentage points
○ from 23.4 to 2.4%
○ declines vary across industries (St. dev. is 17.9 ppt)
○ largest in manufacturing

• rapid expansion of trade, concentrated in manufacturing • value of exports to the US more than triples 2001-2004

U.S. Tariff declines and export expansion
US imports from Vietnam and US tariff changes, 2001-04
All traded industries
33 32
17 352134
36 25 31 28
20 27 14
29 22 24 19
15 5
-.6 -.4
BTA change in US tariff
• Industries with bigger tariff declines experience bigger export growth
• This pattern not pronounced for exports to the EU, nor in pre-reform data

US tariffs and exports: other high-income market
Destination market Tariff change
Observations R-squared
Tariff change
Observations R-squared
Traded Manufacturing Traded
Panel A: Change in ln exports, 2001 to 2004
0.142 (1.070) 19 0.001
0.823 (0.904) 19 0.035
-5.677*** -4.331* (1.474) (2.111)
24 19
0.283 0.119
0.372 (0.675) 24
Panel B: Change in ln exports, 1997 to 2000
-0.808 0.181 (1.896) (1.722) 24 19 0.011 0.001
0.362 (0.599) 24 0.011

Worker-level Data
• Vietnam Household Living Standards Surveys
○ Nationally representative hh survey with labor market module
○ Two cross sections 2002, 2004
○ 12 month recall (cover 2001 and 2003)
○ Panel subcomponent
○ 1993 and 1998 household surveys for falsification checks
• Usual information on worker demographic characteristics, education, location, industry affiliation
• Information on the industry of employment
○ 60 overall, 34 traded industries, 22 manufacturing
○ Link to industry level measures of tariffs

Household businesses
• Informality is a multidimensional concept
• Vietnam differentiates between household businesses vs. registered enterprises
○ All foreign, state, and collective businesses have to legally register as an enterprise
○ Private businesses can legally operate as a
○ household business
○ registered private enterprise
• Consistent with definition in other countries based on registration of firms

Household business employment
• Information on employer type
○ Self-employment (household farm or a household businesses)
○ Work for another household’s business
○ Enterprise sector (private, foreign, state, collectives)
• Indicator for working in a household business
○ 1 if self-employed or work for another household’s business
○ 0 if other employer (i.e., enterprise sector)
• Abstract from reallocation of workers within employer groups

Vietnamese non-farm HH businesses
• Predominantly operate in services (70%)
○ 40% in retail, 10% in hotels and restaurants
• 30% operate in manufacturing
• The most common activities within the manufacturing sector
○ Food and beverages (42.6%)
○ Wood processing (17.9%)
○ Clothing and apparel; dressing and dyeing of fur (11.0%)
○ Manufacture of furniture; manufacturing n.e.c. (8.4%)
○ Textiles (7.9%)

Household business employment and tariffs
• The relationship between employment in a household business and industry tariffs:
Hijt = Xijt D + tariffjt + j +tt + pp + ijt
○ H is an indicator for employment in hh business
○ X is a vector of individual characteristics (age, age squared, education indicators, gender, minority, rural indicator)
○ tariff is the U.S. tariff in industry j at time t
○ λ is an industry fixed effect
○ t is a time fixed effect
○ p is a province fixed effect
Hh business employment and tariffs: manufacturing
22 21

3024 25 28
15 19
36 26
31 27
-.6 -.4 Change in US tariff -.2 0
-.2 -.1
Household business employment and tariffs
Industry tariff
Worker characteristics Year fixed effects Industry fixed effects? Province fixed effects Number of industries Observations R-squared
Traded All industries 0.209*** 0.127***
(0.014) (0.032)
Manufacturing 0.156*** (0.020)
(1) (2) (3)
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes 34 60 22
176,546 248,793 0.415 0.591
27,072 0.293
• Traded: Avg. tariff declines 21.1 pct. point4.4 pct. point decline in hh business employment
• Manufacturing: Avg. tariff decline 30.3 pct. point4.7 pct point decline

Additional Results
• Falsification Check: No relationship prior to reform
• Greater effects for individuals in internationally integrated provinces
• Greater effects for younger cohorts
• Greater effects for the least and for the most educated
• Results robust to controls for unobserved individual heterogeneity (in a panel subsample)
• Movement across industries in response to tariffs:
○ Structure of industry employment changing within the enterprise sector
○ Overall though, little change in industry structure of employment

Potential Aggregate Labor Productivity Gain
• Labor allocation across heterogeneous economic units (sectors, employers) has consequences for aggregate productivity
• Policy distortions and aggregate productivity in developing countries
○ Restuccia and Rogerson (2008), Hsieh and Klenow (2009), Bloom and
Van Reenan (2007), McMillan and Rodrik (2011)
• BTA might increase aggregate labor productivity through the reallocation of workers into the enterprise sector

Aggregate labor productivity gain in manufacturing
• Focus on manufacturing
• Use standard development accounting approach to compute associated productivity change (Caselli (2005), Gollin, Lagakos, Waugh (2014))
○ Previously applied to productivity gaps between agriculture and non- agriculture
P  sBTA P  P  eh
○ sBTA is the share of manufacturing workers reallocated to the enterprise sector due to the BTA
○ Pe is productivity in the enterprise sector
○ Ph is productivity in the household business sector

Aggregate labor productivity gain in manufacturing
• We start with the “standard” Cobb-Douglas production function for thinking about how to measure productivity differences across sectors
Y  A K  L 1 iiii
• Under assumptions of perfect competition and markets, this leads to the
familiar result:
w MRPL 1ARPL iii
• This suggests two approaches for measuring labor productivity gap:
○ ARPL (Gollin, Lagakos, and Waugh (2014))
○ Wages (Vollrath (2014), Herrendorf and Schoellman (2015))
w  MRPL  ARPL eee

Aggregate labor productivity gain in manufacturing
Average revenue product of labor
Productivity ratio
Adjusted for hours & human capital
Importance of worker heterogeneity
• Productivity is higher in the enterprise than in the household business sector
○ Consistent with La Porta and Schleifer (2008, 2014), Nataraj (2011), Hsieh and
Olken (2014)
• Reallocation of workers would be productivity-enhancing
• Adjust the gap for worker heterogeneity across the two sectors
○ Hours worked
○ Education
○ Accounting for worker heterogeneity matters

Aggregate labor productivity gain in manufacturing
• How can one reconcile the estimates of the gap based on ARPL vs. earnings?
• Distortions facing the firms (Hsieh and Klenow (2009))
○ Profit function:
 1 PY1 wL1 rK i Yi i i Li i i Ki i i
○ Relationship between ARPL and wage gaps:
w  MRPL ARPL e1Le1Ye e1e e
w1 1  MRPL 1ARPL h Lh Yh h h h

Aggregate labor productivity gain in manufacturing
Average revenue product of labor
Productivity ratio
Adjusted for hours & human capital
Share of hours reallocated
Annual productivity change
• After adjusting for heterogeneity, the ratio is 6.0
• The associated percentage change in productivity due to BTA
reallocation from household businesses to enterprises is then 3.5%
• Similar results within textiles and apparel as well as within one major manufacturing centre

Aggregate labor productivity gains
• Allow output-labor elasticity to differ across the two sectors
○ Output-elasticity of 1 and 0.68 in informal and formal sector (Restrepo-
Echavarria (2014))
○ Elasticity ratio of about 1.5, MRPL gap drops from 6 to 4
• Measurement concerns:
○ Use of two different surveys:
○ Earnings per worker similar in enterprise sector
○ Gap exists even in data from 2006 from one survey
○ Measurement concerns more severe for microenterprises
○ Understate revenue up to 30% (de Mel et al. (2009)): 6 drops to 4.2
○ Overstate effective hours work (Fafchamps et al. (2014))
• Gap drops from 6 to 3.7 if adjusted for both and to 2.5 if adjusted for output- labor elasticity as well (1.1% annual gain)

• We find that labor reallocates from hh businesses to the enterprise sector with increased export opportunities
○ Effects bigger for individuals in more internationally integrated provinces and for younger workers
• Demand-side policies, which disproportionally benefit initially more productive firms, aid relative expansion of jobs in the registered enterprise sector in low income countries
• Employment shifts toward industries with greater declines in export costs among VN employers in the enterprise sector
○ Employers in the enterprise sector most directly impacted
• Removal of export distortions predicted to improve aggregate labor productivity through the reallocation to the enterprise sector

Thank you

Note: We are not responsible for typo error, price change, substitute item, or available of item from online ads.


Leave a Reply