Policy Research Working Paper 9065

The Effects of International Scrutiny on Manufacturing Workers

Evidence from the Rana Plaza Collapse in Bangladesh

Laurent BossavieYoonyoung ChoRachel Heath

Social Protection and Jobs Global PracticeNovember 2019

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Produced by the Research Support Team

Abstract

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Policy Research Working Paper 9065

After the tragic factory collapse of Rana Plaza in 2013, the direct reforms and indirect responses of retailers have both plausibly affected workers in the ready-made garment sector in Bangladesh. These reforms include a minimum wage increase, high profile but voluntary audits, and an increased reluctance to subcontract to smaller factories. This paper uses six rounds of the Labor Force Survey and adopts a difference-in-difference approach to evaluate the net effects of these changes on garment workers, compared with workers in other manufacturing industries and other

plausible control groups. Although employers appear to have increased sick leave and some measures of safety at work, they simultaneously have reduced job security in the form of written contracts. The study also finds that, a few years after Rana Plaza, average hourly wages have fallen significantly for female workers. The results suggest that regulations that are initially aimed at helping workers can have unintended adverse effects on several dimensions of workers’ outcomes.

This paper is a product of the Social Protection and Jobs Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at lbossavie@worldbank.org or ycho1@worldbank.org.

The Effects of International Scrutiny on Manufacturing Workers:

Evidence from the Rana Plaza Collapse in Bangladesh

Laurent Bossavie∗, Yoonyoung Cho†, Rachel Heath‡

November 20, 2019§

Keywords: Garment sector reforms, minimum wage, working conditions, gender

JEL classification: J16, J31, J32, J81, O12

1 Introduction

The collapse of the Rana Plaza factory building in Bangladesh in April of 2013 killed over a

thousand workers and injured about 2,500 others. It is widely considered the worst accident in

the history of the global garment industry, and the world’s worst industrial disaster since the 1984

Bhopal gas tragedy in India (Reuters, 2017). In its aftermath, local and international attention on

the garment sector intensified, resulting in a series of new policies and changes to the behavior of

retailers. Direct reforms included an increase in the minimum wage and high-profile but voluntary

audits for safety conditions. While the goal was to raise salaries and improve working conditions, a

concern arose that reforms that raise costs among garment factories could ultimately hurt workers.

In this paper, we evaluate the net effect of these post-Rana Plaza changes on workers in the

ready-made garment (RMG) sector in Bangladesh. Indirectly, retailers may have become hesitant

∗World Bank; lbossavie@worldbank.org†World Bank; ycho1@worldbank.org‡Department of Economics, University of Washington; rmheath@uw.edu§We thank S Anukriti, Andrew Foster, Melanie Khamis, Adriana Kugler, Annemie Maertens, Tyler McCormick,

David Newhouse, Laura Schechter, Shing-Yi Wang, Chris Woodruff, and attendees of the 2018 World Bank/IZAJobs & Development Conference, the University of Washington Labor-Development Brownbag, the World BankJobs Group Brownbag Lunch Series, the Georgetown Center for Economic Research 2019 Biennial Conference, theKorea Development Institute School of Public Policy Seminar, and the University of Washington Forum for PoliticalEconomy and Economics for helpful comments. This research was made possible by the generous support from theKorea World Bank Partnership Facility Trust Fund for Bangladesh. The views expressed in this paper are entirelythose of the authors; they do not necessarily represent the views of the International Bank for Reconstruction andDevelopment/World Bank and its affiliated organizations or those of the Executive Directors of the World Bank orthe governments they represent.

1

to buy from factories that subcontract or seem unlikely to abide by labor regulations. We use six

rounds of the Bangladesh Labor Force Survey, spanning from 2003 to 2016, and employ a difference-

in-difference estimation strategy that compares workers in the garment sector to workers in other

manufacturing industries. Each survey was collected in all four quarters of the year, allowing us to

separately examine short-term effects (in the remainder of 2013) and longer-term effects (in 2015

and 2016). We show that treatment effects are very similar when we use alternative control groups

or a synthetic control instead of other manufacturing industries.

We find that, despite a minimum-wage increase, average hourly wages have fallen for female

garment workers a few years after the reforms. In contrast, we report no significant effect for male

workers. When we look at the time path of wage effects, we find that wage decreases compound

over time; women’s hourly wage had significantly fallen by 2015 and 2016. Quantile regressions

confirm that while low earners indeed appear to have benefited from the reforms, consistent with

a role for the minimum wage increase, higher wage earners were particularly adversely impacted.

We go on to examine measures of job security, non-wage benefits, and health/safety. Workers

were 14 percentage points less likely to have a written contract after Rana Plaza. This suggests

additional welfare losses for workers, as hypothetical choice experiments have indicated that job

stability is the most valued aspect of formal work among Bangladeshi workers (Mahmud et al.,

2019). At the same time, some measures of working conditions increased post Rana Plaza; workers

were 5.0 percentage points more likely to have paid sick leave, and female workers were 17 percentage

points less likely to report abusive behavior, and a marginal statistically significant decrease of 11

percentage points in dangerous conditions.

We explore several potential explanations for why women in particular saw large wage decreases

as a result of Rana Plaza. We first show that it appears to be gender per se, rather than some

observable characteristic correlated with gender, that drives the decrease in women’s wages: in-

teracting the treatment with age, education, and number of children in a respondent’s household

does little to change the treatment effect on women. Instead, we provide suggestive evidence that,

as increased costs decreased demand for labor, women’s labor supply is more inelastic, so they

stay in the industry even with lower wages. This phenomenon is potentially particularly strong in

Bangladesh, where women historically have not worked outside the home, and can help explain the

divergence between our results and several studies that find that women and men were equally hit

2

during macroeconomic shocks (Smith et al. 2002; Hallward-Driemeier, Rijkers and Waxman 2011;

Zhi et al. 2013).

These changes affect many workers and their families, given that the RMG sector in Bangladesh

is widely considered to be a key contributor to the country’s robust economic growth, poverty

reduction, and women’s empowerment. The sector’s share of total exports increased from 53

percent in 1995 to more than 80 percent in 2015, with its exports reaching about USD 24.5 billion

(Farole et al., 2017). The export share of GDP tripled between 2003 and 2015, and Bangladesh is

now the world’s second largest garment exporter after China. The sector provides jobs and earnings

opportunities to over 4 million low and semi-skilled workers, and represents over 40 percent of the

total industrial employment in the country. Moreover, the large expansion of the garment industry

has provided employment opportunities to women, changing their lives and decisions on schooling,

marriage, and childbearing (Heath and Mobarak, 2015).

Our results join a literature that points out that regulation intended to help workers may have

unintended negative consequences. For instance, Besley and Burgess (2004) find that pro-worker

labor regulations in India decreased productivity and ultimately increased urban poverty. Botero

et al. (2004) show similar results hold in a sample of 85 countries; heavier regulation of labor is

associated with lower employment rates. Parker, Foltz and Elsea (2016) find that policies targeting

human rights abuses can have similar effects: legislation aimed at stopping the export of conflict

minerals hurt infant health in the Democratic Republic of the Congo. While these results are

plausibly driven by employers’ response to increased labor costs, Adhvaryu, Nyshadham and Xu

(2018) and Boudreau (2019) highlight a complementary mechanism in Indian and Bangladeshi

garment factories, respectively. These papers argue that reforms that raise workers’ expectations

for job quality can decrease job satisfaction if these expectations are not met.

If it is effective, scrutiny from retailers and other watchdogs may have similar effects as formal

regulation, and Harrison and Scorse (2010) find that international attention to the wages and work-

ing conditions in textiles and footwear factories in Indonesia in the 1990s raised wages. Amengual

and Distelhorst (2019) find that a management change at Gap Inc. increased labor compliance

in sourcing factories. Our results suggest that international pressure post Rana Plaza, which was

focused on working conditions, was also successful in improving working conditions, but factories

may do so at the expense of employees’ wages.

3

Our paper also joins a large literature on the relationship between wages and working conditions.

Standard theories of compensating differentials posit that firms choose wages and working condi-

tions to equalize the marginal utility of an additional dollar investment in wages versus working

conditions for the marginal worker. We join Summers (1989) and Mitchell (1990) in emphasizing

that wages fall when firms are compelled by external forces to improve their working conditions.

Finally, our work is also related to a literature in trade and development on the ways in which

destination countries affect workers in export industries in low-income countries. The existence

of a wage premium for working in an export factory suggests that access to international markets

helps workers in low income countries, as their firms make increased profits from selling to wealthier

consumers in high-income countries (Verhoogen, 2008, Frıas, Kaplan and Verhoogen, 2009, Amiti

and Davis, 2011). The sudden opportunity to export that came with Myanmar’s trade liberaliza-

tion also improved working conditions (Tanaka, 2017). However, exposure to trade can also hurt

workers in import-competing industries, and Topalova (2010) and Kovak (2013) find evidence that

reductions in trade barriers increased poverty in India and Brazil, respectively. Other work finds

that consumer sentiment may also matter. Dragusanu and Nunn (2018) find that fair trade certi-

fication in Costa Rica helps coffee farmers, though there are distributional consequences: unskilled

farmers are unaffected. Our quantile regressions show that distributional effects go in the opposite

direction in our study – higher wage workers appear to lose more – suggesting that the effects of

international pressure may reach lower paid workers in ways that certification may not, if it results

in minimum wage increases.

The rest of the paper is organized as follows. Section 2 presents the context of the post-Rana

Plaza reforms as well as the policy changes that affected the RMG sector following the collapse. In

section 3, we describe our data and our identification strategy. Section 4 discusses our main results.

Section 5 discusses the identifying assumptions and tests the robustness of the results to alternative

control groups. In section 6, we examine hypotheses for why we find differentially stronger wage

effects on women. Section 7 concludes.

4

2 Context

2.1 Employment in the Garment Sector

The garment sector currently employs approximately 4 million workers in Bangladesh in over 5,000

factories, and accounts for over 40 percent of the country’s total industrial employment (Farole et al.,

2017). Employment in the RMG sector has grown at an explosive yearly rate of 15.5 percent since

its inception in the early 1980s, as indicated by data from the Bangladesh Garment Manufacturers

Exporting Association (BGMEA) in figure 1. This growth has been fueled by migration from

rural areas, whose lack of information appears to contribute to poor working conditions (Boudreau,

Heath and McCormick, 2017). Workers are fairly mobile; by 12 months after the time of hiring, for

instance, only 62 percent of all hired workers who are still working in the garment industry remain

in their original factory (Heath, 2018).

2.2 Minimum Wage Increase

The minimum wage is set at the industry level in Bangladesh, with no formal overall minimum

wage. The garment industry is one of the few sectors that has statutory minimum wages; these have

been in place since 1985. One minimum is set for each of the seven grade levels, corresponding to

different levels of skills and experience in the industry. After subsequent periodic revisions to reflect

inflation and cost of living, it was set at BDT 3,000 (USD 38) a month as of 2010 for low-skilled

workers. After the Rana Plaza collapse, domestic and international communities lobbied to raise

the wage, seeking a “living wage” of BDT 9,500 (USD 120). Factory owners argued that such a

large increase would hamper competitiveness, and in December 2013, the sector’s minimum wage

was raised, but only to a compromise of BDT 5,300 (USD 68) a month for low-skilled workers.1

Figure 2 provides a histogram of the monthly wage of workers in the RMG sector in 2013 quarter

1 (before the Rana Plaza collapse) and 2015 (after the minimum increase had been in effect for

over a year) using the Bangladeshi Labor Force Survey data described in section 3.1. While there

is some evidence of lumping at the minimum wage – only 1.7 percent of workers earn between BDT

4,000 and BDT 5,000 in 2015, compared to the 9.3 percent of workers who earn between BDT 5,000

and BDT 6,000 – the mode is considerably higher: the most common wage interval is between BDT

1In December 2018, after the period studied by this paper, the minimum wage was further raised to BDT 8,000.

5

8,000 and BDT 9,000, earned by 26 percent of workers.

2.3 Safety Standards in the RMG Sector, before and after Rana Plaza

Before the Rana Plaza collapse, workers had protested, and some safety measures had been put

in place, such as the Fire Prevention and Fire Fighting Law of 2003, the Fatal Accidents Act of

1855, the Bangladesh National Building Code of 2006, and the Labour Act of 2006. However, the

effectiveness of these acts was limited; they were overall vague, narrowly defined, and lacked focus

(Rahman and Moazzem, 2017). Moreover, authorities lacked technical expertise, and enforcement

was weak. For instance, factory visits were pre-announced, and factories with violations found only

needed to be revisited within a span of three years.

The fast growth of the industry led large factories to subcontract to small factories in order

to remain cost competitive and meet deadlines. While the principal factories are typically highly

compliant to worker and safety norms/codes, their subcontractor factories are small, often huddled

into buildings that are not designed for industrial purposes, but house five or more small factories.2

The general consensus among industry experts is thus that working conditions in these factories

are worse than those in the principal factories that sell to retailers (Rahman and Moazzem 2017;

Naved et al. 2018).

Following the Rana Plaza incident, the representatives of the Government of Bangladesh, the

European Union (EU) represented by the European Commission, and the International Labour

Organization (ILO) supported the Bangladesh Sustainability Compact to promote improved labor

standards and responsible business conduct in the RMG sector in Bangladesh. Representatives

from industry (including brands, retailers and small and medium enterprises), employers, trade

unions and other key stakeholders also supported the initiative. They put in place the National

Tripartite Action Plan on Fire Safety and Structural Integrity, which consisted of a set of detailed

measures to improve safety in RMG factories.

In parallel, although major retailers initially discussed a single coordinated industry response,

two different industry agreements emerged. European buyers joined to create the Accord on Fire

and Building Safety in Bangladesh (AFBSB), and a group of American buyers joined to create the

2The Rana Plaza factory was also housed in a building that was permitted for commercial purposes, but it wasillegally renovated to have more units for industrial purposes.

6

Alliance for Bangladesh Worker Safety (ABWS) in 2013. Both initiatives were voluntary measures

meant to assess RMG factories’ compliance to building and fire safety. The Accord is a legally-

binding agreement that commits signatories to a five-year program of safety audits and remediation

investments in their Bangladeshi supplier bases. The Alliance positioned has similar stated goals

as the Accord but with more limited binding commitments, including for financing remediation.

As a result of these measures, the majority of export oriented RMG factories have been assessed

and those that did not qualify have been flagged for remediation. While the Accord and Alliance

have provided some help with remediation financing, the reforms prescribed by the Accord and

Alliance have been primarily financed by the factories themselves, under buyer pressure. This led

to an important rise in compliance related expenditures, and therefore to an increase in the cost of

doing business for RMG entrepreneurs. This increase in expenses was simultaneous to an increased

pressure from buyers for garment factories to bring down their prices. Some had to reduce their

production capacity to make up for additional expenditures on compliance requirements, while some

could simply not cope with the additional expenses, which led to business closures. Therefore, the

sustainability of such reforms is doubted by industry experts.

3 Empirical Strategy

3.1 Data

We use the Labor Force Surveys (LFS) conducted in 2003, 2005, 2010, 2013, 2015, and 2016 by the

Bangladesh Bureau of Statistics. The Bangladesh LFS was designed to be representative at the

national level – for both rural and urban areas – as well as for each of Bangladesh’s seven divisions.

Each working age individual from the selected households is interviewed for detailed labor market

outcomes including labor market status, type of employment, monthly wages for those that are

paid employees, as well as a detailed occupation and sector of activity at the four-digit level. The

sample sizes are large, with over 100,000 individuals of working age (15 to 64) interviewed in years

prior to 2015, and above 300,000 in 2015 and 2016. The 2003-2013 surveys took place over the

course of a year, while the 2015 and 2016 surveys represented four distinct quarterly surveys that

we merged together. Thus, any seasonality in wages (bonuses or raises) should be averaged out

over the course of each survey. In each survey round, workers were asked the number of hours they

7

had worked over the past seven days. By contrast, the measures of earnings varied from year to

year, so to construct our measure of hourly wages:

• In 2003, workers were asked monthly earnings from their main activity in the past year. We

calculate average hourly earnings by dividing this amount by the hours worked in the past

week multiplied by 4.33 (the average number of weeks in a month, across the year).

• In 2005, 2010 and 2013, workers were asked about wages from the previous week in their

primary job, so we divide this by the hours worked in the past week.

• In 2015 and 2016, workers were asked take home income for the past one month (both cash

and in-kind). We calculate average hourly earnings by dividing this amount by the hours

worked in the past week multiplied by 4.33.

In each round, earnings were asked to paid employees: both regular wage workers and well as

casual/piece rate workers. While self-employed workers were asked about their earnings prior to

2013, for consistency over our period of study, we restrict our sample to paid employees.

There may be differences in the wages calculated across the different survey rounds. Most

obviously, wages paid in 2015 and 2016 may be higher because they reflect in-kind payment as well.

The fact that 2003 asked about monthly earnings over the past year presumably reduces variance

between workers, if workers take the average of their income over the past year. Similarly, the

average weekly pay over the past month in 2015 and 2016 could display lower variance than in

rounds 2005-2013, when workers reported weekly pay directly. However, these differences would be

absorbed in the year fixed effects in the regression analysis, and would not be a leading concern

unless they systematically vary between garment and non-garment sectors, conditional on the

control variables in our main specification (equation 1), which allows for effects like lower in-kind

payment to educated workers.

The data also contain several measures of non-wage benefits and other aspects of job quality.

Some of these variables are only available in later rounds, as summarized in the table below:

8

Measure Years available Details

Sick Leave all Paid sick leave provided by employer

Maternity Leave 2010 onwards Maternity leave provided by employer (does not specify paid)

Contract 2005 onwards Written contract (oral contract coded as zero)

Injured 2013 onwards Hurt in any accident at work

Dangerous 2013 onwards At least one reported danger (see below)

Abused 2013 onwards At least one report of abuse (see below)height

Specifically, the reported dangers could include dust, fumes, noise or vibration; fire, gas, flames;

extreme cold or heat; dangerous tools; work underground or at heights; work in water/pond/river;

workplace too dark or confined; chemicals/explosives; other things (specify). Reported abuse could

include: constantly shouted/insulted; beaten /physically hurt; sexually abused; others. Table A3

gives the frequency of specific abuses and dangerous conditions.

3.2 Timing of the Treatment

The collapse of Rana Plaza occurred on April 24, 2013. Rescue efforts continued until May 13,

including the high profile rescue of a woman on May 10. Workers began to riot, starting two

days after the collapse, and were joined the next day by left-leaning political parties, including

the Bangladesh National Party, demanding the arrest of those whose negligence they believed

contributed to the collapse and an independent commission to identify safety threats. Protests

continued off and on for many months.

Response from retailers also began shortly after the collapse. Specifically, retailers and NGO’s

met one week after the collapse and created the Accord on Factory and Building Safety. The

Alliance came in place in July 2013. Activists also pushed retailers in the months following the

collapse, with some calling for boycotts, but others calling for brands to stay and “clean up the

industry” (Judy Gearheart, executive director of the International Labor Rights Forum on April

29, 2013).

Overall, wage and working conditions responses to the Rana Plaza incident and subsequent

reforms likely unfolded gradually across the months following the incident. Some responses – such

as compositional changes in the industry due to factory shutdowns – could have taken even longer.

9

We consider the first quarter of 2013 to be pre-treatment, drop the second quarter (as it is a mix

of pre-treatment survey periods and those surveyed in the immediate aftermath), and consider the

third and fourth quarters to be post-treatment. Thus, while our baseline specification averages the

effects on workers surveyed from 5 weeks to 2.5 years after the Rana plaza collapse, we also run

specifications where we explicitly consider treatment effect heterogeneity by time since the collapse.

3.3 Identification Strategy

Our treatment group consists primarily of workers employed in the garment industry according to

the ISIC-3 classification, as the post-Rana reforms were targeted to that specific industry. However,

we also include textile workers in the treatment group of our baseline estimation, because the textile

industry is closely tied to the garment industry in the value chain and thus reforms that affect the

garment industry may also have impacted the textile industry. Moreover, given that the activities

of the textile and garment industries are very similar, it is possible that some garment workers were

reported as textile workers and vice versa during survey data collection.

As a control group for treated garment sector workers, we first consider other workers in man-

ufacturing, defined by ISIC-3 four digit industry codes between 1500 and 3720. Table A2 gives a

list of these industries, which primarily include food products (rice milling, sugar confectionery,

etc.) and wooden furniture and handicrafts. The summary statistics for the treatment and control

groups are reported in table 1.3 The groups are not perfectly similar in levels, motivating our ex-

ploration of parallel trends and consideration of alternate control groups in section 5. For instance,

workers in other manufacturing are older, less educated, and less likely to be located in Dhaka

than garment sector workers. Garment sectors workers also earn more and are more likely to be in

factories.

Then, for an outcome Y for worker i in geographic area j (specifically, the division) at time t,

we estimate the following equation:

Yijt =β1Garmentijt + β2Garmentijt × PostRanaP lazat (1)

+ Femaleijt × λjt + γtFemaleijt ×Xijt × λt + εijt

3Table A1 provides summary statistics for the alternate control groups that we consider in section 5.

10

The estimated β2 gives the net causal effect of the post-Rana Plaza reforms on the outcome of

garment workers, under the assumption of parallel trends between the garment sector and other

manufacturing. We include gender interacted with both division by year fixed effects and year

fixed effects interacted with a vector of control variables Xijt: namely age, education, and an

urban dummy. These controls flexibly account for changes in labor outcomes over time that may

vary by location or a worker’s human capital, and possibly differentially so by gender. We discuss

further threats to identification in section 5, including a test for parallel trends and discussion of

the possibility that the control group was indirectly treated by the RMG reforms.

4 Results

4.1 Treatment Effects on Wages

Table 2 reports the results of our difference-in-difference estimation on workers’ outcomes. We

first examine hourly wages. Estimates reported in column 1 show that, on average, the post-Rana

Plaza reforms reduced workers’ wages by 8.4 percent. This estimated effect is approximately the

same order of magnitude as the positive coefficient on the main effect of working in the garment

industry (7.9 percent), suggesting that the prior wage premium for working in the garment industry

went away after Rana Plaza. This average wage decrease comes even in light of an increase in the

minimum wage for wage observations taken after 2013, suggesting the increase in labor costs was

even more important.

Estimates in columns 2 and 3 highlight striking differences in the wage impact of the reforms by

gender and time horizon. Column 2 indicates that while males on average lost 4.9 percent of their

wages after Rana Plaza, the wages of female garment workers fell by 20 percent. This adverse effect

on female workers’ wages appears to have further increased several years after the reforms. Column

3 indicates that women’s wages fell by 11 percent in 2013, but 24 percent in 2015 and 26 percent

in 2016. While these estimates are quite large, as with the overall wage effect, they are roughly the

same order of magnitude at the pre-Rana Plaza main effect of working in the garment industry on

women’s wages (24 percent). That is, the garment industry used to pay considerably more than

other employment opportunities for women (conditional on the control variables in equation 1),

but after Rana Plaza, it paid about the same. Men’s wages, by contrast, do not show evidence

11

of a differential time pattern after Rana Plaza. In section 6, we explore different reasons for the

larger wage loss among women and find suggestive evidence in support of a model in which women

– who have worse outside options – stay in the garment industry, even after a shift in labor demand

reduces wages.

To shed light on potential mechanisms that could drive the wage decline, we allow estimate

quantile regressions that allow for heterogenous effects along the wage distribution. These results,

displayed in figure 3, show a clear downward trend in the effect of the post-Rana Plaza reforms on

earnings per hour throughout the distribution of male wages. The right hand side panel of figure 3

reports the additional effect of the reforms by quintile for female workers. The pattern is the same

as for the full sample, suggesting that the negative effect on high earners is even stronger in the

female sample.

This finding is compatible with the existence of two opposite effects of the post-Rana Plaza

reforms on wages, whose magnitude depends on workers’ location along the wage distribution.

The minimum wage increase would have a positive effect on wages of low-paid workers for which

the minimum wage is binding. On the other hand, an effect of the opposite sign would appear if

employers respond to increased costs by cutting labor costs among workers for which the minimum

wage is not binding. The magnitude of this negative wage effect would increase along the wage

distribution, as employers have more room to cut wages the further away the workers are from the

new minimum wage. The existence of these two opposite effects is compatible with the net effects

of the post-Rana Plaza on wages progressively declining and becoming increasingly negative along

the wage distribution.

4.2 Treatment Effects on Non-pecuniary Aspects of Job Quality

Columns 4 to 6 of table 2 (panel A) explore the effects of the post-Rana Plaza reforms on hours

worked. Across all pre-treatment time periods, there is no strong evidence of changes in hours of

work. Column 6 indicates a moderate decrease of 2.3 weekly hours for women in the short run,

which disappears by 2015. Overall, we find little evidence that employers responded to Rana Plaza

by changing hours.

In panel B of table 2, we next examine two important non-monetary job benefits: sick leave

and maternity leave. Both male and female workers were 5.0 percentage points more likely to have

12

sick leave after Rana Plaza. This effect was particularly strong in 2016, suggesting that some firms

took time to begin these benefits. By contrast, we find little evidence that maternity leave was

affected by Rana Plaza.

In contrast, Panel C of table 2 shows that Rana Plaza reduced the likelihood that garment

workers held a written contract by 14 percentage points. This effect is slightly smaller for fe-

male workers (4.8 percentage points), although the interaction between female and treatment is

not statistically significant. The reduction in formal contracts suggests that employers may have

responded to the increased costs imposed by the reforms by offering more precarious employment

contracts that allow firms greater flexibility to adjust the size of the labor force.

Finally, we examine workers’ reports of injuries on the job, having been abused, and dangerous

conditions. While these variables should be interpreted with caution because they are only available

from 2013 onwards – so that the main effect of being in the garment industry is identified only

from the first quarter of 2013 – we believe they are nonetheless important to examine. While there

is little evidence that the rate at which workers suffer injuries changed after Rana Plaza, abusive

behavior does appear to have fallen among women. In particular, women in garments are 17 percent

less likely to report being abused after Rana Plaza. This is a very large effect, relative to the mean

rate of abusive behavior of 33 percent, pre Rana Plaza. Female workers are also 11 percentage

points less likely to report dangerous conditions; this effect is marginally statistically significant (P

= 0.144).

5 Discussion of Identification Assumptions and Robustness to Al-

ternate Control Groups

5.1 Test for Parallel Trend and Alternate Control Groups

Equation 1 estimates the causal impact of the Rana Plaza reforms on garment workers, if the control

group is similar enough to the treatment group that control workers would exhibit parallel trends

in key outcomes in the absence of the treatment. We use other manufacturing as our preferred

control group, under the assumption that other manufacturing industries have labor forces whose

outside options evolve similarly over time and face similar macroeconomic shocks. We begin by

13

assessing whether trends are parallel between the garment sector and control group leading up to

the reform.

Specifically, we examine a Garment dummy interacted with year dummies over pretreatment

time periods (2003, 2005, 2010 and 2013 quarter 1), to test whether there is a non-linear pre-

treatment trend in garment areas. We estimate the following equation separately for men and

women:

Yijt = β0Garmentijt + βtGarmentijt × λt + λjt + γtXijt × λt + εijt (2)

We then test whether the estimated βt coefficients are jointly significant, which would indicate

that the difference between the garment sector and other industries varies from period to period

in the pre-treatment period. The estimation results are given in table 3 and visually in figure

4. The primary control group of manufacturing, while not perfect, appears parallel in raw wages.

However, since the trends do diverge to some extent when we look at wages conditional on the

control variables in equation 1 – as evidenced by the rejection of joint insignificance of the year

dummies for women in column 1 – we consider several alternate control groups. First, we broaden

the control group to include all non-agricultural wage workers and then all wage workers. While

we still reject joint insignificance of the year dummies for both cases (columns 2 and 3), figure 4

illustrates that the differential trend is not the same as with all manufacturing. That is, while

each control group is imperfect, they are imperfect in different ways, and so we interpret similar

estimates in these groups as some evidence that our treatment effect is not driven by differential

trend in the garment industry relative to one given control group.

Second, we provide synthetic control estimates (Abadie and Gardeazabal, 2003). Because out-

side the garment industry, men and women do very different kinds of jobs, we select separate

control groups for men and women. To avoid concerns about overfitting (given that we only have

four pre-treatment time periods), we restrict our donor pool for each gender to industries with at

least 500 pre-treatment observations of that gender. Table A4 shows the industries selected.

Table 4 repeats the main specification on wages for these alternate control groups. In the first

two columns, we find a similar pattern of results for wages using either all workers, or all non-

agricultural workers. In particular, while both specifications – in contrast to the manufacturing

control group – actually estimate a positive treatment effect on women’s wages in 2013, we find

14

the same pattern of negative effects on women’s wages by 2015 and 2016. Similarly, the synthetic

control specification displays a positive treatment effect on women’s wages in 2013, but it has

turned negative by 2015. The magnitude of the negative hourly wage effect for females by 2016 is

large in all specifications. Thus, while the extent to which the Rana Plaza reforms had an effect

on women’s wages in 2013 appears somewhat sensitive to specification, the wage decline for women

by 2015 and 2016 is quite robust to specification.

5.2 Other Threats to Identification

Even if trends are parallel leading up to Rana Plaza – as occurs by construction for the synthetic

control group shown in tables 3 and 4 – a violation of the parallel trend assumption required for

difference-in-difference would occur if the control group was indirectly affected by the treatment.

For instance, other similar industries could be hurt if the reputation of all Bangladeshi industries

– not just the garment sector – were hurt, and thus firms in these other industries were also

prompted to improve working conditions. While we cannot totally rule out these effects, we know

of no anecdotal evidence that other industries were affected by the Rana Plaza collapse. The

garment industry is very high-profile, and discussion tends to center on it per se.

Alternatively, if workers pushed out of the garment industry joined other industries, outcomes

in these hours could be affected by an outward shift in labor supply. We examine the potential for

such shifts in figure 5. It shows a short-term dip in garment sector employment, which appears to

dissipate by 2015. Thus, we conclude that there were not large overall extensive margin effects.

Even if there are not large effects on the size of the garment sector workforce, there could be a

compositional change in the workers that remain, post Rana Plaza. We investigate the potential for

this latter hypothesis by examining compositional changes along three key dimensions, as displayed

in table 5: workers’ years of education, age and location in an urban or rural area, to potentially

account for rural to urban migration that may have occurred our period of study. We examine these

compositional effects separately by gender (Panel A and B) as well as in the full sample (Panel C).

In the full sample, we also check for changes in gender composition of the workforce. We find no

evidence of changes in education or share urban post Rana-Plaza. We do find evidence that workers

get younger, driven by males, who are on average 1.5 years younger post Rana Plaza. Moreover,

the labor force becomes 3.3 percentage points more female post Rana Plaza, although the increase

15

is not statistically significant at conventional levels (P = 0.134). We will return to the evidence on

gender in section 6.2, when we explore how potential differential labor supply elasticities between

men and women could explain differential wage effects. Overall, it does appear that there is some

potential for compositional changes to underly the estimated effects from equation 1, in addition to

causal impacts on a fixed type of worker. Moreover, if workers who left garments went to industries

in the control group, it is possible the composition of the control group is affected as well.

Finally, one may be concerned that the policy changes that took place due to Rana Plaza

might be contemporaneous to other shocks affecting the garment industry, which could also affect

workers’ outcomes. For example, the garment sector could have also been affected by increased

global competition or other shifts in the global economy that may have taken place over the same

period. While we include fixed effects interacted with gender and other control variables to capture

macroeconomic shocks, it is still possible that the macroeconomic variables affect the garment

industry, say because it more extensively exports, as discussed in section 2. To investigate this

possibility, we look at the evolution of exports from Bangladesh over time, as shown in figure

6. Time series do not show a noticeable break in export trends around the Rana Plaza event,

alleviating concerns that our estimated treatment effects may capture the confounding influence of

contemporaneous macroeconomic shocks.

6 Why Did Female Wages Fall?

A common and striking theme through the main specification and the robustness checks in the

previous section is that the post-Rana Plaza responses reduced female workers’ wages, especially

in 2015 and 2016. In this section, we delve deeper into potential reasons why the wages of female

workers were more disproportionately affected by such reforms. We begin by looking for evidence

that it is not gender per se that drives the disproportionate effect on women, but rather, observable

characteristics correlated with gender. Given that we find little evidence that observable charac-

teristics of women explain the larger decrease in wages that they face, we turn to explanations that

highlight a more direct role of gender in determining employers’ responses to the collapse.

16

6.1 Observable Characteristics Correlated with Gender

Gender might be correlated with other observable characteristics of garment workers or job types

that may be driving heterogeneous responses to the reforms. For instance, if the post-Rana Plaza

reforms had an impact mostly on less educated workers and female workers have less education

than male workers, the disproportionate gender impact may be due to firm size rather than gender

itself. This is possible, given that table 1 shows some differences between male and female garment

workers, including education.

To test the extent to which observable characteristics correlated with gender can explain the

stronger wage effects on women, we add to our baseline regression a set of interactions between

the treatment of interest and a set of observable characteristics of workers that may be correlated

with gender: indicators for above median level of schooling (seven years) of workers in our sample,

below median age, the number of children in the household, living in Dhaka division and living in

an urban area. We first add the interaction of each variable with treatment and assess the extent

to which these additional covariates diminish the estimated treatment effect on women. We also

include an interaction of the variable with a female dummy to assess whether there is a differential

effect on women of that variable. We also include interactions with a large firm dummy, defined

by a firm size of over 10 workers.4 While working in a large firm may very well be endogenous to

the Rana Plaza reforms – so that it should not be read as the effect of Rana Plaza among workers

pre-determined to be in a large firm – we nonetheless think that it is a useful descriptive way to

assess the set of firms that drive the treatment effects on wages, and how these firms may vary

between male and female workers.

The results are displayed in table 6. Panel A shows that the larger wage effects among women do

not seem to be driven by their observable characteristics of women. The effects of the post-Rana

Plaza reforms on garment wages remains negative and statistically significant, and of a similar

magnitude in all specifications. There is evidence that younger workers of both genders are less

hurt by the responses; their wages are 9.3 percent higher after Rana Plaza than workers above age

25. This is consistent with the quantile regression results in figure 3 that show that higher-paid

workers have been more adversely impacted than lower-paid workers.

4This is the only information on firm size that is consistently available across waves of the Labor Force Survey.

17

Panel B of table 6 examines location and firm size. There is some evidence that the post-

Rana Plaza reduction in women’s wages has been mostly driven by workers in the Dhaka division.

Namely, a female worker in Dhaka suffered an average 27 percent wage drop after Rana Plaza,

while a female worker outside Dhaka suffered a 12 percent decrease (P = 0.130). One potential

explanation for this finding could be that enforcement and inspection efforts following the post-Rana

Plaza reforms have been disproportionately concentrated in Dhaka, where the main export-oriented

factories are located. In this case, one would expect a disproportionate increase in costs associated

with compliance with the new regulations for firms located in the area. The point estimate on

the female treatment effect interacted with an urban location is also negative, though statistically

insignificant.

There is similarly evidence in column 6 that women in large firms are less affected by Rana

Plaza; their wages fall by 8.8 percent less than women in small firms. This suggests that larger

firms, despite pressure for labor inspection and international scrutiny, were better able to cope

with financial burden and plausibly are more likely to abide by the minimum wage regulations. To

the contrary, small firms appear to bear the brunt of burden of the reforms and to pass on the

increasing costs on workers’ wages, especially women.

6.2 Direct Effects of Gender

Rana plaza likely decreased demand for labor in the garment industry by increasing the cost of

each worker hired, now that the factory is compelled to provide more benefits and better working

conditions. Additionally, the diminished reputation of clothes sourced from Bangladesh may have

also decreased demand for attire purchased there, as retailers likely substituted toward countries

with better safety records. Even though this decrease applied to all workers, it could particularly

affect women if their labor supply is plausibly more inelastic than that of male workers, and thus,

their wages would likely fall more. This hypothesis is consistent with recent evidence from Menzel

and Woodruff (2019), who find that the wage gap between male and female garment workers is

driven by lower mobility within the garment sector; similar frictions might also impede their job

opportunities in other sectors.

One possible driver of a more inelastic labor supply among women in the garment industry

is the lack of employment opportunities for women outside this industry. Descriptive statistics in

18

our sample indeed show a higher concentration of female manufacturing workers in the garment

industry compared to males. While about 48 percent of male workers employed in manufacturing

are in the garment industry, over 78 percent of female manufacturing workers are employed in the

garment sector. This suggests the existence of barriers to working outside the garment industry

among women. As a result, female workers may not respond to negative shocks affecting the

garment industry to the same extent as male workers, who would switch to a more attractive job

in a different sector.

Alternatively, the demand for female labor could have differentially fallen. Section 4.2 found

evidence of increased sick leaves, which are likely associated with disproportionately greater costs

of female labor (Gruber 1994; Ruhm 1998). While employers in a competitive labor market would

ideally prefer to pass on the extra costs only to the women that they know will take advantage of

these benefits, they likely cannot tell upon wage-setting which women will do so. So, instead, they

statistically discriminate against all women and lower their wages accordingly.

While both an overall demand decrease in an environment in which female labor is more inelastic

than males and a differential increase in the cost of women’s labor can explain why women’s wages

fell more than men’s, these theories differ in their prediction on the effect of Rana Plaza on the

gender composition of the labor force. In particular, if women’s labor supply is more inelastic than

men’s, then more of them will stay in the labor force after the demand decrease. By contrast, a

differential decrease in demand for women’s labor would increase the share of males among the

workforce.

Table 5 thus provides suggestive evidence differentiating these theories. The labor force became

3.3 percentage points disproportionately more female after Rana Plaza, though the effect is not

statistically significant at conventional levels (P = 0.134). Thus, we view this result as suggestive

evidence that the wage results are consistent with the theory that women are more likely to stay

in the labor force, even in the face of reduced labor demand.

7 Conclusion

This paper studies the effects of the post-Rana Plaza reforms on labor market outcomes of garment

workers compared to other manufacturing sector workers. We find that increased international

19

scrutiny has improved some measures of working conditions as they intended to, such as paid sick

leave or safety at work.

We also report, however, that garment workers have been negatively impacted by the reforms

along other important dimensions. First, despite an increase in the minimum wage, the post-Rana

Plaza reforms negatively affected female workers’ hourly earnings a couple of years after the reform.

Given minimal changes in hours of work, this decrease translates into reduced take-home earnings.

One possible explanation for this is that employers compensated for increasing safety compliance

costs by lowering the wages of the workers for whom the new minimum wage was not binding.

Female workers bore the brunt of the reduction in wage rates due to the post-Rana Plaza reforms.

Except for a small share of low-earning young females, who appear to have been protected by

the minimum wage policy, female workers in the rest of the wage distribution experienced wage

decreases. Such negative impacts were driven by small firms especially in Dhaka, on which pressures

to improve working conditions and thus costs associated with them were likely high.

Additionally, the reforms appear to have significantly reduced job security among garment

workers. This indicates that employers introduced more flexibility in hiring and firing decisions, as

a response of increasing non-labor costs and a more uncertain economic environment induced by

the reforms.

Taken together, our results suggest that the push to improve working conditions after Rana

Plaza may have brought unintended consequences, which were particularly felt by female workers.

Our findings suggest that policy-makers should take into account such consequences when imple-

menting reforms, and consider pairing them with policies that mitigate potential adverse effects on

workers’ job security and wages.

20

References

Abadie, Alberto, and Javier Gardeazabal. 2003. “The economic costs of conflict: A case study of theBasque country.” The American Economic Review, 93(1): 113–132.

Adhvaryu, Achyuta, Anant Nyshadham, and Huayu Xu. 2018. “Hostel takeover: Living conditions,reference dependence, and the well-being of migrant workers.”

Amengual, Matthew, and Greg Distelhorst. 2019. “Can Sourcing Help Enforce Global Labor Stan-dards? Evidence from the Gap Inc Supply Chain.”

Amiti, Mary, and Donald R Davis. 2011. “Trade, firms, and wages: Theory and evidence.” The Reviewof Economic Studies, 79(1): 1–36.

Besley, Timothy, and Robin Burgess. 2004. “Can labor regulation hinder economic performance? Evi-dence from India.” The Quarterly Journal of Economics, 119(1): 91–134.

Botero, Juan C, Simeon Djankov, Rafael La Porta, Florencio Lopez-de Silanes, and AndreiShleifer. 2004. “The regulation of labor.” The Quarterly Journal of Economics, 119(4): 1339–1382.

Boudreau, Laura. 2019. “Multinational enforcement of labor law: Experimental evidence from Bangladeshsapparel sector.” mimeo.

Boudreau, Laura, Rachel Heath, and Tyler McCormick. 2017. “Migrants, information, and workingconditions in Bangladeshi garment factories.” mimeo.

Dragusanu, Raluca, and Nathan Nunn. 2018. “The effects of Fair Trade certification: evidence fromcoffee producers in Costa Rica.” National Bureau of Economic Research Working Paper 24260.

Farole, Thomas, Yoonyoung Cho, Laurent Bossavie, and Reyes Aterido. 2017. “Bangladesh jobsdiagnostic.” Jobs working paper 9. Washington, D.C. : World Bank Group.

Frıas, Judith A, David S Kaplan, and Eric A Verhoogen. 2009. “Exports and wage premia: Evidencefrom Mexican employer-employee data.” Columbia University mimeo.

Gruber, Jonathan. 1994. “The incidence of mandated maternity benefits.” The American Economic Re-view, 84(3): 622–641.

Hallward-Driemeier, Mary, Bob Rijkers, and Andrew Waxman. 2011. Ladies first? Firm-levelevidence on the labor impacts of the East Asian crisis. The World Bank.

Harrison, Ann, and Jason Scorse. 2010. “Multinationals and anti-sweatshop activism.” The AmericanEconomic Review, 100(1): 247–273.

Heath, Rachel. 2018. “Why do firms hire using referrals? Evidence from Bangladeshi garment factories.”Journal of Political Economy, 126(4): 1691–1747.

Heath, Rachel, and A Mushfiq Mobarak. 2015. “Manufacturing growth and the lives of Bangladeshiwomen.” Journal of Development Economics, 115: 1–15.

Kovak, Brian K. 2013. “Regional effects of trade reform: What is the correct measure of liberalization?”American Economic Review, 103(5): 1960–76.

Mahmud, Minhaj, Italo A Gutierrez, Krishna Kumar, and Shanthi Nataraj. 2019. “What aspectsof formality do Workers value? Evidence from a choice experiment in Bangladesh.” IZA Discussion Paper12121.

Menzel, Andreas, and Christopher Woodruff. 2019. “Gender wage gaps and worker mobility: Evidencefrom the garment sector in Bangladesh.” National Bureau of Economic Research Working Paper 25982.

21

Mitchell, Olivia S. 1990. “The effects of mandating benefits packages.” National Bureau of EconomicResearch Working Paper 3260.

Naved, Ruchira, Tabassum Rahman, Samantha Willan, Rachel Jewkes, and Andrew Gibbs.2018. “Female garment workers experiences of violence in their homes and workplaces in Bangladesh: Aqualitative study.” Social Science & Medicine, 196: 150–157.

Parker, Dominic P, Jeremy D Foltz, and David Elsea. 2016. “Unintended consequences of sanctionsfor human rights: Conflict minerals and infant mortality.” The Journal of Law and Economics, 59(4): 731–774.

Rahman, Mustafizur, and Khondaker Golam Moazzem. 2017. “The legacy of Rana Plaza: Improvinglabour and social standards in Bangladeshs apparel industry.” In Governing Corporate Social Responsibilityin the Apparel Industry after Rana Plaza. 81–109. Springer.

Reuters. 2017. “Factbox: Major industrial accidents in Bangladesh in recent years.” July 3.

Ruhm, Christopher J. 1998. “The economic consequences of parental leave mandates: Lessons fromEurope.” The Quarterly Journal of Economics, 113(1): 285–317.

Smith, James P, Duncan Thomas, Elizabeth Frankenberg, Kathleen Beegle, and GracielaTeruel. 2002. “Wages, employment and economic shocks: Evidence from Indonesia.” Journal of Popula-tion Economics, 15(1): 161–193.

Summers, Lawrence H. 1989. “Some simple economics of mandated benefits.” The American EconomicReview, 79(2): 177–183.

Tanaka, Mari. 2017. “Exporting Sweatshops? Evidence from Myanmar.” Review of Economics and Statis-tics, 1–44.

Topalova, Petia. 2010. “Factor immobility and regional impacts of trade liberalization: Evidence on povertyfrom India.” American Economic Journal: Applied Economics, 2(4): 1–41.

Verhoogen, Eric A. 2008. “Trade, quality upgrading, and wage inequality in the Mexican manufacturingsector.” The Quarterly Journal of Economics, 123(2): 489–530.

Zhi, Huayong, Zhurong Huang, Jikun Huang, Scott D Rozelle, and Andrew D Mason. 2013.“Impact of the global financial crisis in rural China: Gender, off-farm employment, and wages.” FeministEconomics, 19(3): 238–266.

22

Figure 1: Total employment in the garment industry over time

01

23

4m

illio

ns o

f wor

kers

1985 1990 1995 2000 2005 2010 2015Year

Source: Bangladesh Garment Manufacturers’ Export Association (BGMEA)

Nation−wide Employment in the Garment Industry

23

Figure 2: Wages in the garment industry before and after Rana Plaza, by gender

010

2030

4050

050

0010

00015

00020

00025

000

3000

0+ 050

0010

00015

00020

00025

000

3000

0+

Male Female

Perc

ent

nominal monthly wage in taka

Quarter 1 of 2013

010

2030

4050

050

0010

00015

00020

00025

000

3000

0+ 050

0010

00015

00020

00025

000

3000

0+

Male Female

Perc

ent

nominal monthly wage in taka

Quarter 1 of 2015

24

Figure 3: Difference-in-differences estimate of the post Rana Plaza reforms, by quantile of the wagedistribution

-0.4

0-0

.20

0.00

0.20

Gar

men

t X P

ost

0 .2 .4 .6 .8 1Quantile

-0.2

0-0

.15

-0.1

0-0

.05

0.00

0.05

Gar

men

t X P

ost X

Fem

ale

0 .2 .4 .6 .8 1Quantile

Notes: Quantiles of the distribution of hourly wages are reported on the x-axis. Difference-in-difference coefficients

estimated from quantile regression estimates of equation 1 are reported in the y-axis.

25

Figure 4: Difference-in-differences, various control groups

22.

53

3.5

4lo

g(w

age)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.3-.2

-.10

.1w

age

resi

dual

2003

2005

2010

201

320

1520

16

year

Conditional

Females

33.

23.

43.

63.

8lo

g(w

age)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.15

-.1-.0

50

.05

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Males

Manufacturing

Control Garment

2.5

33.

54

log(

wag

e)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.2-.1

0.1

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Females

33.

23.

43.

63.

84

log(

wag

e)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.15

-.1-.0

50

.05

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Males

All non-agricultural wage earners

Control Garment26

Difference-in-differences, various control groups (continued)

2.5

33.

54

log(

wag

e)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.2-.1

0.1

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Females

33.

23.

43.

63.

8lo

g(w

age)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.15

-.1-.0

50

.05

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Males

All wage earners

Control Garment

2.5

33.

54

log(

wag

e)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.2-.1

0.1

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Females

33.

23.

43.

63.

8lo

g(w

age)

2003

2005

2010

201

320

1520

16

year

Raw wages

-.15

-.1-.0

50

.05

wag

e re

sidu

al

2003

2005

2010

201

320

1520

16

year

Conditional

Males

Synthetic Control

Control Garment27

Figure 5: Percentage of workers in the garment industry

0.2

.4.6

.8fra

ctio

n in

gar

men

ts

2003

2005

2010

201

320

1520

16

year

Women

0.2

.4.6

.8fra

ctio

n in

gar

men

ts

2003

2005

2010

201

320

1520

16

year

Men

Percent of workers in garments over time

Population of Employed Workers Pop. of Wage WorkersPopulation of Light Manufacturing Workers

28

Figure 6: Export growth: 2003-2015

Source: World Bank World Development Indicators database.

29

Table 1: Summary statistics in estimation sample

Garments Other P-values for test of equalityManufacturing F vs M Gar vs Other

Female Male Female Male Gar Other F M

Age 26.430 29.400 31.065 31.976 0.000 0.000 0.000 0.000Years of education 5.065 6.792 4.211 6.003 0.000 0.000 0.000 0.000Is currently married 0.680 0.643 0.646 0.674 0.000 0.007 0.002 0.000Urban 0.575 0.465 0.520 0.446 0.000 0.000 0.000 0.000Age 25 or under 0.555 0.422 0.389 0.349 0.000 0.000 0.000 0.000Dhaka district 0.653 0.621 0.532 0.435 0.000 0.000 0.000 0.000

Hourly wage) 32.248 43.113 26.125 38.747 0.000 0.000 0.000 0.000Hours last week 52.345 55.532 48.805 53.558 0.000 0.000 0.000 0.0001(Sick leave) 0.327 0.356 0.224 0.248 0.000 0.003 0.000 0.0001(Maternity leave) 0.451 0.284 0.246 0.100 0.000 0.000 0.000 0.0001(Contract) 0.592 0.628 0.422 0.385 0.000 0.000 0.000 0.0001(Injured) 0.015 0.022 0.022 0.059 0.000 0.000 0.027 0.0001(Abused) 0.164 0.121 0.165 0.105 0.000 0.000 0.874 0.0001(Dangerous) 0.284 0.298 0.413 0.417 0.029 0.670 0.000 0.000Firm size 10+ 0.670 0.763 0.522 0.444 0.000 0.000 0.000 0.000

Year = 2003 0.099 0.073 0.101 0.136 0.000 0.000 0.639 0.000Year = 2005 0.083 0.147 0.169 0.185 0.000 0.029 0.000 0.000Year = 2010 0.105 0.144 0.093 0.104 0.000 0.054 0.035 0.000Year = 2013 0.291 0.187 0.261 0.176 0.000 0.000 0.001 0.011Year = 2015 0.194 0.207 0.180 0.190 0.004 0.159 0.063 0.000Year = 2016 0.228 0.242 0.195 0.209 0.004 0.063 0.000 0.000N 14,444 18,465 3,224 17,973

Notes: Firm size available only 2010 onwards. Contract available 2005 onwards. Injured, Dangerous, and Abused

available only 2013 onwards.

30

Table 2: Difference in Difference estimates of the Net effect of post Rana Plaza reforms on garmentWorkers

Dependent Variable Log(hourly wage) Hours last week(1) (2) (3) (4) (5) (6)

Garment 0.0787*** 0.0249 0.0249 1.622*** 1.279*** 1.279***(0.0250) (0.0241) (0.0241) (0.451) (0.425) (0.425)

Garment × Post -0.0840*** -0.0493* -0.0597* -0.337 -0.123 1.836**(0.0273) (0.0263) (0.0346) (0.551) (0.522) (0.896)

Garment × Female 0.237*** 0.237*** 1.514 1.514(0.0523) (0.0523) (1.004) (1.004)

Garment × Female × Post -0.154*** -0.0525 -0.951 -4.144***(0.0576) (0.0664) (1.189) (1.349)

Garment × Post × 2015 0.0456 -3.399***(0.0290) (0.903)

Garment × Post × 2016 -0.0182 -1.464(0.0302) (0.922)

Garment× Female × -0.172*** 6.254***Post × 2015 (0.0606) (1.480)

Garment× Female × -0.126** 1.976Post × 2016 (0.0549) (1.493)

Marginal Effects on a:Male in 2015 -0.0140 -1.563P-value 0.619 0.0100Male in 2016 -0.0780 0.372P-value 0.00900 0.560Female in 2013 -0.112 -2.308P-value 0.0770 0.0780Female in 2015 -0.239 0.547P-value 0.00100 0.731Female in 2016 -0.257 -1.796P-value 0 0.279R2 0.428 0.430 0.430 0.155 0.156 0.157N 47386 47386 47386 47386 47386 47386

31

Table 2 B: Difference-in-difference estimates of the net effect of post Rana Plaza reforms on garmentworkers

Dependent Variable 1(Sick Leave) 1(Maternity Leave)(1) (2) (3) (4) (5) (6)

Garment 0.0494*** 0.0486*** 0.0486*** 0.138*** 0.149*** 0.149***(0.0155) (0.0160) (0.0160) (0.0268) (0.0259) (0.0260)

Garment × Post 0.0500** 0.0505** 0.00126 0.0240 0.0205 0.0448(0.0215) (0.0222) (0.0499) (0.0315) (0.0302) (0.0615)

Garment × Female 0.00362 0.00362 -0.0434 -0.0434(0.0244) (0.0244) (0.0595) (0.0595)

Garment × Female × Post -0.00205 -0.0252 0.0113 -0.0802(0.0337) (0.0514) (0.0655) (0.0895)

Garment × Post × 2015 -0.0152 -0.140**(0.0481) (0.0563)

Garment × Post × 2016 0.131** 0.0737(0.0551) (0.0612)

Garment× Female × 0.0663 0.156**Post × 2015 (0.0494) (0.0686)

Garment× Female 0.0482 0.133× Post × 2016 (0.0693) (0.0835)

Marginal Effects on a:Male in 2015 -0.0140 -0.0950P-value 0.444 0Male in 2016 0.133 0.118P-value 0 0.00100Female in 2013 -0.0240 -0.0350P-value 0.652 0.716Female in 2015 0.0270 -0.0190P-value 0.396 0.756Female in 2016 0.156 0.171P-value 0.0110 0.0330R2 0.295 0.295 0.298 0.218 0.218 0.225N 47386 47386 47386 39107 39107 39107

32

Table 2C: Difference-in-difference estimates of the net effect of post Rana Plaza reforms on garmentworkers

Dependent Variable 1(Contract) 1(Injured)(1) (2) (3) (4) (5) (6)

Garment 0.294*** 0.315*** 0.315*** -0.0396** -0.0642** -0.0642**(0.0247) (0.0218) (0.0218) (0.0182) (0.0262) (0.0262)

Garment × Post -0.140*** -0.150*** -0.0898 0.0150 0.0322 0.0619**(0.0311) (0.0284) (0.0603) (0.0189) (0.0268) (0.0289)

Garment × Female -0.0993** -0.0993** 0.0684** 0.0684**(0.0504) (0.0504) (0.0306) (0.0306)

Garment × Female × Post 0.0484 0.0437 -0.0360 -0.0566*(0.0577) (0.0742) (0.0315) (0.0337)

Garment × Post × 2015 -0.0790 -0.0410**(0.0612) (0.0160)

Garment × Post × 2016 -0.0683 -0.0401***(0.0624) (0.0138)

Garment× Female × 0.0293 0.0573***Post × 2015 (0.0704) (0.0201)

Garment× Female × -0.0549 0.0130Post × 2016 (0.0686) (0.0176)

Marginal Effects on a:Male in 2015 -0.169 0.0210P-value 0 0.459Male in 2016 -0.158 0.0220P-value 0 0.419Female in 2013 -0.0460 0.00500P-value 0.601 0.802Female in 2015 -0.0960 0.0220P-value 0.199 0.277Female in 2016 -0.169 -0.0220P-value 0.0190 0.257R2 0.183 0.183 0.184 0.0292 0.0310 0.0326N 43187 43187 43187 24048 24048 24048

Notes: Sample includes individuals in garments/textiles or other manufacturing (ISIC-3 codes 1500-3720). All

regressions include controls for worker’s age, gender, level of schooling, region dummies, and dummy for urban

location, interacted with year dummies and gender dummies, and a triple interaction with year and gender.

Sampling weights included. Standard errors in parentheses; clustered at the primary sampling unit. * p < 0.05, **

p < 0.01, *** p < 0.001

33

Table 2D: Difference-in-difference estimates of the net effect of post Rana Plaza reforms on garmentworkers

Dependent Variable 1(Abused) 1(Dangerous)(1) (2) (3) (4) (5) (6)

Garment 0.0565 0.00751 0.00751 -0.0683 -0.0932* -0.0932*(0.0507) (0.0592) (0.0592) (0.0448) (0.0515) (0.0515)

Garment × Post -0.0410 0.0194 0.0183 -0.0473 -0.0171 -0.0698(0.0519) (0.0600) (0.0725) (0.0467) (0.0529) (0.0653)

Garment × Female 0.136* 0.136* 0.0692 0.0692(0.0739) (0.0739) (0.0776) (0.0776)

Garment × Female × Post -0.186** -0.286*** -0.0923 -0.286***(0.0775) (0.0932) (0.0827) (0.0947)

Garment × Post × 2015 -0.00380 0.127***(0.0425) (0.0409)

Garment × Post × 2016 0.00527 0.0200(0.0427) (0.0440)

Garment× Female × Post × 2015 0.146** 0.266***(0.0590) (0.0555)

Garment× Female × Post × 2016 0.165*** 0.353***(0.0627) (0.0647)

Marginal Effects on a:Male in 2015 0.0150 0.0570P-value 0.808 0.272Male in 2016 0.0240 -0.0500P-value 0.694 0.361Female in 2013 -0.267 -0.356P-value 0.00300 0Female in 2015 -0.126 0.0370P-value 0.0580 0.589Female in 2016 -0.0970 0.0170P-value 0.166 0.822R2 0.116 0.118 0.119 0.342 0.342 0.350N 32833 32833 32833 32451 32451 32451

Notes: Sample includes individuals in garments/textiles or other manufacturing (ISIC-3 codes 1500-3720). All

regressions include controls for worker’s age, gender, level of schooling, region dummies, and dummy for urban

location, interacted with year dummies and gender dummies, and a triple interaction with year and gender.

Sampling weights included. Standard errors in parentheses; clustered at the primary sampling unit. * p < 0.05, **

p < 0.01, *** p < 0.001

34

Tab

le3:

Tes

tof

par

alle

ltr

end

inw

ages

,p

retr

eatm

ent

Dep

end

ent

Var

iab

le:

log

ofh

ourl

yw

age

Fem

ales

Male

s(1

)(2

)(3

)(4

)(5

)(6

)(7

)(8

)C

ontr

olgr

oup

Oth

erm

anu

f.A

llM

inu

sA

gA

llS

ynth

Con

trol

Oth

erm

anu

f.A

llM

inu

sA

gA

llS

ynth

Contr

ol

Gar

men

t0.

317*

**0.

0068

9-0

.036

2-0

.015

7-0

.049

2*-0

.132***

-0.1

35***

0.0

410

(0.0

813)

(0.0

435)

(0.0

434)

(0.0

481)

(0.0

290)

(0.0

259)

(0.0

264)

(0.0

341)

Yea

r=

2005

XG

arm

ent

-0.0

895

0.09

680.

0727

-0.0

408

0.10

6*0.0

617

0.0

898

0.0

135

(0.1

57)

(0.0

815)

(0.0

828)

(0.0

893)

(0.0

588)

(0.0

573)

(0.0

592)

(0.0

603)

Yea

r=

2010

XG

arm

ent

-0.1

11-0

.101

-0.0

544

-0.0

0897

-0.0

155

0.0

656**

0.0

876***

-0.0

371

(0.1

21)

(0.0

671)

(0.0

667)

(0.0

671)

(0.0

439)

(0.0

334)

(0.0

337)

(0.0

518)

Yea

r=

2013

XG

arm

ent

-0.3

91**

*0.

0904

0.08

430.

0208

0.05

560.0

676*

0.0

702*

-0.2

06

(0.0

980)

(0.0

595)

(0.0

586)

(0.0

706)

(0.0

465)

(0.0

359)

(0.0

359)

(0.1

63)

F-s

tat

for

join

tsi

gof

6.50

13.

368

1.83

40.

161

1.81

81.6

04

2.4

59

0.7

54

garm

ent

Xye

ard

um

mie

sP

-val

ue

0.00

0228

0.01

780.

139

0.92

20.

142

0.1

42

0.0

609

0.5

20

R2

0.43

20.

394

0.37

00.

391

0.26

40.2

81

0.2

72

0.2

76

N41

4012

375

1471

096

3411

340

62753

96395

9982

No

tes:

Sa

mp

lein

clu

des

obs

erva

tio

ns

fro

msu

rvey

rou

nd

s2

00

3,

20

05

,2

01

0,

an

d2

01

3(q

ua

rter

1).

All

regr

essi

on

sin

clu

de

con

tro

lsfo

rw

ork

er’s

age

,ge

nd

er,

leve

l

of

sch

ooli

ng,

regi

on

du

mm

ies,

an

dd

um

my

for

urb

an

loca

tio

n,

inte

ract

edw

ith

yea

rd

um

mie

sa

nd

gen

der

du

mm

ies,

an

da

trip

lein

tera

ctio

nw

ith

yea

ra

nd

gen

der

.

Sa

mp

lin

gw

eigh

tsin

clu

ded

.S

tan

da

rder

rors

clu

ster

eda

tth

ep

rim

ary

sam

pli

ng

un

it:

**

*p<

0.0

1,

**

p<

0.0

5,

*p<

0.1

35

Table 4: Difference-in-difference estimates with alternate control groups

Dependent Variable: log of hourly wageControl group All All minus Ag

(1) (2) (3) (4)

Panel A: Broader set of industries

Garment -0.0633*** -0.0633*** -0.0784*** -0.0784***(0.0182) (0.0182) (0.0176) (0.0176)

Garment × Post -0.0339* -0.0444* -0.0240 -0.0314(0.0205) (0.0266) (0.0198) (0.0263)

Garment × Female 0.0404 0.0404 0.0898*** 0.0898***(0.0283) (0.0283) (0.0284) (0.0284)

Garment × Female × Post -0.0896*** 0.190*** -0.135*** 0.147***(0.0328) (0.0418) (0.0331) (0.0428)

Garment × Post × 2015 0.0662*** 0.0600**(0.0234) (0.0234)

Garment × Post × 2016 -0.0349 -0.0364(0.0248) (0.0249)

Garment× Female × Post × 2015 -0.312*** -0.298***(0.0408) (0.0423)

Garment× Female × Post × 2016 -0.453*** -0.475***(0.0383) (0.0394)

Marginal Effects on a:Male in 2015 0.0220 0.0290P-value 0.330 0.187Male in 2016 -0.0790 -0.0680P-value 0.00100 0.00400Female in 2013 0.146 0.116P-value 0 0.00300Female in 2015 -0.100 -0.122P-value 0.00600 0.00100Female in 2016 -0.342 -0.395P-value 0 0

R2 0.363 0.365 0.377 0.379N 272933 272933 211739 211739

36

Table 4 (continued): Difference-in-difference estimates with alternate control groups

Dependent Variable: log of hourly wageSample Female Male

(1) (2) (3) (4)

Panel B: Synthetic ControlGarment × Female -0.0209 -0.0209

(0.0270) (0.0270)

Garment × Female × Post -0.0365 0.126***(0.0350) (0.0450)

Garment× Female × Post × 2015 -0.279***(0.0497)

Garment× Female × Post × 2016 -0.370***(0.0527)

Garment 0.00441 0.00441(0.0299) (0.0299)

Garment × Post -0.00930 -0.0610(0.0400) (0.0561)

Garment × Post × 2015 0.0493(0.0596)

Garment × Post × 2016 0.157**(0.0629)

Marginal Effects on the wage of a:Male in 2015 -0.0120P-value 0.802Male in 2016 0.0960P-value 0.0600Female in 2015 -0.153P-value 0Female in 2016 -0.244P-value 0

R2 0.407 0.410 0.345 0.346N 33500 33500 25614 25614

Notes: All regressions include controls for worker’s age, gender, level of schooling, region dummies, and dummy for

urban location, interacted with year dummies and gender dummies, and a triple interaction with year and gender.

Sampling weights included. Standard errors clustered at the primary sampling unit: *** p<0.01, ** p<0.05, * p<0.1

37

Table 5: Characteristics of workers in treatment and control groups, before and after Rana-Plazareforms

Females

Diff-in-diff S.e NCoeff

Age -0.288 [0.577] 17,440Education (years) 0.103 [0.233] 17,440Urban -0.032 [0.046] 17,440

Males

Diff-in-diff S.e NCoeff

Age -1.455*** [0.421] 35,548Education (years) 0.189 [0.212] 35,548Urban 0.040 [0.034] 35,548

All

Diff-in-diff S.e NCoeff

Age -0.640* [0.351] 52,988Education (years) 0.130 [0.179] 52,988Urban -0.017 [0.032] 52,988Share of males -0.033 [0.022] 52,988

Notes: Sample includes individuals in garments/textiles or other manufacturing (ISIC-3 codes 1500-3720). All

regressions include controls for region dummies, year dummies, and interactions between year dummies and gender

dummies. Sampling weights included. Standard errors clustered at the primary sampling unit: *** p<0.01, **

p<0.05, * p<0.1

38

Table 6: Heterogeneous effects of post Rana Plaza reforms on garment workers

Dependent Variable: log of hourly wage(1) (2) (3) (4) (5) (6)

Garment 0.0525* 0.0567* 0.0228 0.0371 0.0160 0.0172(0.0316) (0.0344) (0.0258) (0.0268) (0.0260) (0.0272)

Garment × Post -0.0754** -0.0715* -0.0840*** -0.0896*** -0.0456 -0.0513*(0.0338) (0.0367) (0.0281) (0.0293) (0.0284) (0.0296)

Garment× Female 0.224*** 0.213*** 0.241*** 0.197*** 0.240*** 0.237***(0.0527) (0.0566) (0.0519) (0.0531) (0.0521) (0.0546)

Garment× Female × Post -0.149*** -0.158** -0.171*** -0.153*** -0.157*** -0.144**(0.0577) (0.0616) (0.0570) (0.0585) (0.0573) (0.0597)

educ 7+ × Garment -0.0536* -0.0618*(0.0310) (0.0365)

educ 7+ × Garment × post 0.0612* 0.0566(0.0324) (0.0387)

educ 7+ × Female × Garment 0.0328(0.0592)

educ 7+ × Female × Garment × post 0.0070(0.0627)

Age 25 or under × Garment 0.0047 -0.0297(0.0233) (0.0261)

Age 25 or under × Garment × post 0.0931*** 0.104***(0.0230) (0.0273)

Age 25 or under × Female × Garment 0.105**(0.0432)

Age 25 or under × Female × Garment × post -0.0382(0.0470)

Children in household × Garment 0.0091 0.0074(0.0010) (0.0122)

Children in household × Garment × post -0.0033 0.0039(0.0092) (0.0119)

Children in household × Female × Garment 0.0039(0.0158)

Children in household × Female × Garment × post -0.0170(0.0173)

R2 0.436 0.436 0.432 0.433 0.430 0.430N 47637 47637 47637 47637 47637 47637

Notes: Sample includes individuals in garments/textiles or other manufacturing (ISIC-3 codes 1500-3720). Allregressions include controls for worker’s age, gender, level of schooling, region dummies, and dummy for urban

location, interacted with year dummies and gender dummies, and a triple interaction with year and gender.Sampling weights included. Standard errors clustered at the primary sampling unit: *** p<0.01, ** p<0.05, * p<0.1

39

Table 6 Panel B: Heterogeneous effects of post Rana Plaza reforms on garment workers

(1) (2) (3) (4) (5) (6)Dependent Variable: log of hourly wage

Garment 0.00596 0.0252 0.0401 0.0378 0.135*** 0.0875**(0.0351) (0.0365) (0.0303) (0.0307) (0.0396) (0.0428)

Garment × Post -0.0231 -0.0393 -0.0368 -0.0457 -0.0613 -0.0574(0.0378) (0.0392) (0.0332) (0.0334) (0.0420) (0.0451)

Garment× Female 0.239*** 0.147** 0.244*** 0.256*** 0.128** 0.245***(0.0515) (0.0703) (0.0525) (0.0855) (0.0600) (0.0673)

Garment× Female × Post -0.154*** -0.0790 -0.155*** -0.106 -0.0498 -0.0793(0.0568) (0.0832) (0.0578) (0.0941) (0.0643) (0.0729)

Dhaka district × Garment 0.0317 -0.00464(0.0465) (0.0474)

Dhaka district × Garment × post -0.0449 -0.0137(0.0510) (0.0516)

Dhaka district × Female × Garment 0.161*(0.0964)

Dhaka district × Female × Garment × post -0.137(0.110)

Urban × Garment -0.0390 -0.0339(0.0470) (0.0468)

Urban × Garment × post -0.0196 -0.00140(0.0515) (0.0511)

Urban × Female × Garment -0.0226(0.100)

Urban × Female × Garment × post -0.0831(0.111)

Firm size 10+ × Garment -0.175*** -0.0917**(0.0407) (0.0447)

Firm size 10+ × Garment × post 0.0462 0.0161(0.0417) (0.0456)

Firm size 10+ × Female × Garment -0.199***(0.0473)

Firm size 10+ × Female × Garment × post 0.0875*(0.0517)

R2 0.430 0.430 0.430 0.430 0.297 0.299N 47637 47637 47637 47637 39352 39352

Notes: Sample includes individuals in garments/textiles or other manufacturing (ISIC-3 codes 1500-3720). Allregressions include controls for worker’s age, gender, level of schooling, region dummies, and dummy for urban

location, interacted with year dummies and gender dummies, and a triple interaction with year and gender.Sampling weights included. Standard errors clustered at the primary sampling unit: *** p<0.01, ** p<0.05, * p<0.1

40

Appendix

41

Tab

leA

1:S

um

mar

yst

atis

tics

inal

tern

ate

contr

olgr

oup

s

All

All

Syn

Con

trol

P-v

alu

esfo

rte

stof

equal

ity

Min

us

Ag

Gar

men

tvs

Oth

er

Fem

ale

Mal

eF

emal

eM

ale

Fem

ale

Mal

eF

emal

eM

ale

Fem

ale

Mal

eF

emal

eM

ale

Age

34.5

99

35.6

89

34.0

7934

.971

35.2

9732

.702

0.00

00.

000

0.00

00.

000

0.00

00.

000

Yea

rsof

edu

cati

on

5.6

73

5.2

17

6.45

96.

731

4.87

66.

208

0.00

00.

000

0.00

00.

139

0.00

00.

000

Iscu

rren

tly

mar

ried

0.65

00.7

68

0.64

80.

750

0.61

40.

626

0.00

00.

000

0.00

00.

000

0.00

00.

002

Urb

an0.4

310.

310

0.50

30.

421

0.33

50.

451

0.00

00.

000

0.00

00.

000

0.00

00.

000

Age

25or

un

der

0.26

10.

258

0.26

60.

266

0.26

30.

390

0.00

00.

000

0.00

00.

000

0.00

00.

037

Dh

aka

dis

tric

t0.3

55

0.3

030.

405

0.35

10.

254

0.32

40.

000

0.00

00.

000

0.00

00.

000

0.00

0

Hou

rly

wag

e)41.1

4036.3

1444

.530

43.8

2931

.366

32.4

920.

000

0.00

00.

000

0.04

00.

000

0.00

0H

ours

last

wee

k44.2

6350.3

3344

.170

51.2

4844

.374

50.6

610.

000

0.00

00.

000

0.00

00.

000

0.00

01(S

ick

leav

e)0.3

050.

226

0.34

50.

305

0.26

50.

253

0.00

00.

000

0.00

00.

000

0.00

00.

000

1(M

ate

rnit

yle

ave)

0.2

38

0.1

13

0.26

90.

143

0.21

70.

144

0.00

00.

000

0.00

00.

000

0.00

00.

000

1(C

ontr

act

)0.

398

0.28

20.

458

0.38

20.

270

0.33

90.

000

0.00

00.

000

0.00

00.

000

0.00

01(I

nju

red

)0.

017

0.04

00.

020

0.04

70.

008

0.02

20.

128

0.00

00.

004

0.00

00.

002

0.21

61(A

bu

sed

)0.1

25

0.0

83

0.13

00.

083

0.13

10.

188

0.00

00.

000

0.00

00.

000

0.00

00.

000

1(D

anger

ous)

0.14

90.

200

0.16

70.

238

0.11

80.

399

0.00

00.

000

0.00

00.

000

0.00

00.

000

Fir

msi

ze10

+0.2

77

0.2

09

0.31

30.

276

0.14

70.

217

0.00

00.

000

0.00

00.

000

0.00

00.

000

Yea

r=

2003

0.1

150.

138

0.11

50.

116

0.10

80.

118

0.00

00.

000

0.00

00.

000

0.12

20.

000

Yea

r=

2005

0.1

250.

140

0.12

70.

124

0.10

90.

132

0.00

00.

012

0.00

00.

000

0.00

00.

000

Yea

r=

2010

0.1

450.

185

0.13

80.

172

0.12

30.

143

0.00

00.

000

0.00

00.

000

0.00

00.

000

Yea

r=

2013

0.2

330.

180

0.23

10.

186

0.56

30.

512

0.00

00.

034

0.00

00.

819

0.00

00.

000

Yea

r=

2015

0.1

830.

182

0.18

10.

203

0.04

70.

049

0.00

60.

000

0.00

10.

238

0.00

00.

000

Yea

r=

2016

0.1

980.

175

0.20

80.

199

0.04

90.

046

0.00

00.

000

0.00

00.

000

0.00

00.

000

N33

,311

173,5

29

28,4

8213

,007

117

,689

7350

No

tes:

Fir

msi

zea

vail

abl

eo

nly

20

10

on

wa

rds.

Co

ntr

act

ava

ila

ble

20

05

on

wa

rds.

Inju

red

,D

an

gero

us,

an

dA

buse

da

vail

abl

eo

nly

20

13

on

wa

rds.

42

Table A2: Distribution of industries in the control group

ISCI 3 code

number

of

workers

pct of

control

group ISCI 3 code

number

of

workers

pct of

control

group

1010 1 0.01 Manu. of veneer sheets, plywood, boards 1 0.01

1030 1 0.01 Manu. of builders' carpentry & joinery 2 0.02

Crude petroleum production 18 0.2 Manufacture of wooden containers 13 0.14

Extraction of natural gas 3 0.03 Basic chemicals, ex. fert. & nitro. com 3 0.03

Incidental to oil & gas ext. excluding 13 0.14 Manu.of other chemical products n.e.c 387 4.28

1156 1 0.01 3610 446 4.93

Prep. & pre. of meat & meat products 119 1.32 Wooden furniture and fixtures 1,073 11.86

Pro. & pre. of fish & fish products 132 1.46 Manu. of cane & bamboo furniture 99 1.09

Pro. & pre. of fruits & vegetables 12 0.13 Manufacture of chicks 17 0.19

Manu. inedible veg. oil, animal oil & f 72 0.8 3614 2 0.02

Edible veg. oil except hydro. veg. oil 46 0.51 Manu. of metal furniture & fixtures 96 1.06

Manufacturing of dairy products 328 3.63 Signs &advertising displays (non-electr 22 0.24

Grain milling except rice milling 277 3.06 3619 5 0.06

Manu. of starches & starch products 64 0.71 Wood, cane & bamboo handicrafts 448 4.95

Manufacture of prepared animal feeds 73 0.81 Paper and paper products handicrafts 11 0.12

Bone crushing 4 0.04 3623 1 0.01

Rice milling 637 7.04 China, ceramic & glass handicrafts 26 0.29

Grain Mill Products n.e.c 25 0.28 Metal decorative handicrafts 6 0.07

Bakery Products 236 2.61 Textile & sewing decorative handicrafts 258 2.85

Sugar 101 1.12 Bangles (except of proecious metal) 4 0.04

Cocoa, chocolate, sugar confectionery 528 5.84 Brooms and brushes 1 0.01

Macaroni, noodles, couscous & related 62 0.69 Decorative handicrafts n.e.c 44 0.49

1545 3 0.03 3655 1 0.01

Processing of tea and coffee 31 0.34 Manu. of jewellery & related articles 411 4.54

Blending of tea and coffee 115 1.27 Manufacture of musical instruments 2 0.02

Edible salt refining 14 0.15 Manufacture of games and toys 6 0.07

Manu. of other food products n.e.c 760 8.4 Pens, pencils & other office articles 3 0.03

Distilling, rectifying & blending of sp 4 0.04 Manu. of umbrellas & walking sticks 3 0.03

Manufacture of wines 48 0.53 Other manufacturing, n.e.c. 120 1.33

1553 8 0.09 Wholesale & retail sale of motor vehicl 1 0.01

Manu. of soft drinks & mineral waters 161 1.78

1600 272 3.01

Manufacture of cigarettes 28 0.31

Manufacture of cigars and cheroots 10 0.11

Manufacture of bidies 533 5.89

1604 4 0.04

Manufacture of zarda and quivam 17 0.19

Tobacco manufacture n.e.c 1 0.01

T. D of leather 71 0.78

luggage, H.bag & like saddlery & harnes 8 0.09

1920 435 4.81

Manufacture of leather footwear 137 1.51

Manufacture of other footwear n.e.c 2 0.02

Sawmilling and planing of wood 119 1.3243

Table A3: Summary statistics on specific abusive behaviors and dangerous conditions

Garments Other P-values for test of equalityManufacturing F vs M Gar vs Other

Female Male Female Male Gar Other F M

Abusive behaviorconstantly shouted/insulted 0.126 0.118 0.158 0.100 0.088 0.000 0.000 0.000beaten /physically hurt 0.006 0.003 0.008 0.006 0.002 0.213 0.327 0.004sexually abused 0.047 0.001 0.011 0.001 0.000 0.000 0.000 0.758others 0.003 0.002 0.000 0.004 0.020 0.001 0.008 0.001

Dangerous conditionsdust, fumes,noise or vibration 0.185 0.172 0.234 0.185 0.011 0.000 0.000 0.008fire, gas, flames 0.007 0.019 0.049 0.118 0.000 0.000 0.000 0.000extreme cold or heat 0.029 0.040 0.058 0.079 0.000 0.000 0.000 0.000dangerous tools 0.153 0.137 0.178 0.187 0.001 0.266 0.004 0.000work underground or at heights 0.002 0.004 0.011 0.028 0.083 0.000 0.000 0.000work in water/pond/river 0.003 0.003 0.001 0.009 0.827 0.000 0.172 0.000workplace too dark or confined 0.020 0.021 0.031 0.028 0.566 0.409 0.002 0.001chemicals/explosives 0.005 0.013 0.036 0.057 0.000 0.000 0.000 0.000other things (specify) 0.001 0.005 0.006 0.013 0.000 0.001 0.000 0.000N 10,386 11,363 2229 11,527

All variables available only 2013 onwards.

44

Table A4: Synthetic control industries

Two-Digit ISIC 3.1 Industry Weight: Females Weight: Males

1 Agriculture, hunting and related service activities 0.42915 Manufacture of food products and beverages 0.01636 Manufacture of furniture; manufacturing n.e.c. 0.34845 Construction 0.08180 Education 0.25693 Other service activities 0.63595 Activities of private households 0.235

as employers of domestic staff

45