0

National Graduate Institute for Policy Studies Global Center of Excellence Economics Working Paper No. 21 August 2009

An Inquiry into the Rapid Growth of the Garment Industry in

Bangladesh

Khondoker Abdul Mottaleb National Graduate Institute for Policy Studies, Tokyo

and Tetsushi Sonobe

Foundation for Advanced Studies on International Development, and National Graduate Institute for Policy Studies, Tokyo

Abstract

The export-oriented garment industry in Bangladesh has grown rapidly for the last three

decades and now ranks among the largest garment exporters in the world. While its

early success is attributed to the initial technology transfer from South Korea, such a

one-time infusion of knowledge alone is insufficient to explain the sustained growth for

three decades. This paper uses primary data collected from knitwear manufacturers

and garment traders to explore the process of the continuous learning of advanced skills

and know-how. It finds, among other things, that the high profitability of garment

manufacturing due to the initial infusion of specific human capital attracted a number of

highly educated entrepreneurs to the industry, that the division of labor between

manufacturers and traders has facilitated the expansion of the industry, and that

enterprise growth has lasted long because of the continuous learning from abroad by the

highly educated entrepreneurs.

1

1. Introduction

The phenomenal growth of the garment industry in Bangladesh is widely known, due in

no small part to the popular book by Easterly (2002) and an earlier paper by Rhee

(1990). The industry now ranks among the largest garment exporters in the world. It

accounts for 75 percent of the country’s export earnings and 25 percent of GDP and

provides ample job opportunities for females. The garment industry started from

scratch three decades ago. Since its exports were negligible in those days, the country

was not subject to export restrictions under the Multi-Fiber Arrangement (MFA), which

attracted the attention of Daewoo Corporation of South Korea, one of the rising garment

manufacturers suffering severely from the quota system. The company planned to

develop a production base in Bangladesh and teamed up with an indigenous new

enterprise called Desh Ltd. in 1979. Desh sent 130 new employees to Daewoo’s

factory in South Korea, where they participated in an eight-month intensive training

course covering diverse topics from sewing skills to factory management and

international marketing. By two years after the training, however, almost all the

trainees had left Desh to start their own garment businesses (Rhee, 1990). Easterly

(2002, pp.147-148) comments: “This explosion of garment companies started by

ex-Desh workers brought Bangladesh its $2 billion in garment sales” in the late 1990s.

By 2007/08, the industry was earning more than $10 billion.

Like the Bangladesh of three decades ago, countries in Sub-Saharan Africa are

now given favorable treatments to their garment industries, such as duty-free, quota-free

access to the US market provided by the African Growth and Opportunity Act (AGOA).

What can be done for the garment producers in Africa to allow them to take full

advantage of such a helpful situation? Existing studies of the Bangladeshi garment

2

industry agree with Easterly (2002) and Rhee (1990) that a central role in the industry’s

development was played by technology transfers from abroad (e.g., Quddus and Rashid,

2000; Hoq, 2004; Khan, 2004; Mlachila and Yang, 2004; Rahman, 2004; Siddiqi, 2005).

This paper uses primary data of garment manufacturers and traders in Bangladesh

to investigate further the role and mechanism of technology transfers. While Rhee

(1990) describes vividly the technology transfers that occurred in the early stage of the

industry’s development, his study does not cover the recent period. From the

description provided by Easterly (2002), one may have an impression that the

proliferation of manufacturers was the major way in which the industry has grown. In

the proliferation process, the Desh-Daewoo trainees established their factories and

taught their workers, who later started their own factories and disseminated the

knowledge originally from Korea. While this is true, it is not the whole story. Many

of the original Desh-Daewoo trainees did not become garment manufacturers but

instead became garment traders, who act as intermediaries between manufacturers and

foreign buyers. The subsequent proliferation of local garment traders suggests that

their activities have been useful for the development of the industry. A question arises

as to what roles local traders play in industrial development in developing countries.

The economics literature, however, is silent about this question.1 In an attempt to

examine what role local traders play, we collected data of traders as well as

manufacturers.

The export-oriented garment industry in Bangladesh today is large not just in 1 The roles of agricultural traders in counteracting the imperfections of market institutions are extensively discussed by Fafchamps (2004) and Hayami and Kawagoe (1993). Gereffi (1995), Schmitz and Knorringa (2000), and Bazan and Navas-Aleman (2004) among others investigate how global buyers organize international trade networks and how learning takes place in the network. To our knowledge, however, there are few studies of local traders in industries in developing countries.

3

terms of the number of manufacturers (over 4000), but also because these enterprises

are large in size. During the last three decades, the export-oriented garment industry

in other countries, such as the Philippines, has experienced prosperity and then a sharp

decline. This fact suggests that there is an interesting story not just about the

beginning of the industry but also about the subsequent process of enterprise growth.

According to Gereffi (1999), Schmitz and Knorringa (2000), and Bazan and

Navas-Aleman (2004), opportunities for continuous learning are built in to operation

within the global commodity chain, and it is up to local suppliers, such as garment

manufacturers in Bangladesh and the Philippines, whether they learn advanced skills

and know-how. What are the determinants of enterprises’ continuous learning from

abroad and own experiences? One answer is the profitability from learning, which

will in turn depend on labor costs and other conditions. In search of the fundamental

determinants of the profitability, we attempt to identify the entrepreneur’s background

attributes that affect the enterprise performance.

Since information on entrepreneurs is unavailable from secondary data, we

conducted personal interviews with them. It was, however, difficult to meet with

entrepreneurs, especially with traders, because they always move from place to place.

It is no wonder that empirical studies of traders are scarce. Our data indicate that

traders provide manufacturers with services by using specific skills that newly

established manufacturers tend to lack, and that manufacturers reduce their dependence

on traders as they learn such skills. These findings suggest that the division of labor

between traders and manufacturers has facilitated the reproduction of manufacturers and

the formation of a cluster. Our data also indicate that not just enterprise size but

enterprise growth depends much on the general human capital of the entrepreneur. We

4

conjecture that highly educated entrepreneurs have been attracted to this industry by the

high profitability which was boosted initially by the Desh-Daewoo infusion of Korean

skills and know-how.

The rest of the paper is organized as follows. Section 2 reviews the historical

development of the garment industry in Bangladesh. Section 3 advances three testable

hypotheses. After the descriptive analysis is presented in Section 4, regressions

analyses are carried out in Section 5. Section 6 summarizes the findings and discusses

the implications.

2. Brief History

In the 1970s, Bangladesh was dominated by the socialistic ideology. The government

controlled foreign trade nearly completely and regulated private investments. When

Desh Garment made its debut on the world market in 1980, the government hardly

recognized the potential of the garment industry. In 1982, however, the government

began to provide various incentives to the garment industry, such as the duty free import

of machinery, bonded warehouse facilities, and cash incentives, and donated land to

garment producers in Narayanganj and Gazipur.2 Furthermore, the government sent

missions to the US government to lobby for duty free access to the US market. The

government also helped organize trade fairs at home and abroad. Thus, it is clear that

the government has played a role in the development of the garment industry in

Bangladesh (e.g., Siddiqi, 2005; Quddus and Rashid, 2000).

Textile and garment companies in South Korea and other newly industrialized 2 Large garment clusters have emerged in Gazipur and Narayanganj on the land donated by the government. Very recently, the government has donated another 300 acres of land exclusively to develop a garment village in Munshiganj, a suburb of Dhaka, where at least 390 production units will be accommodated.

5

countries (NIEs) in East Asia continued to invest in their factories in export processing

zones in Dhaka, the capital city, and Chittagong, the largest port city, in Bangladesh.

Like the Desh-Daewoo training, these foreign ventures sent Bangladeshi workers to

their home countries for intensive training, and some time after returning to Bangladesh,

many of these trainees moved to other factories or started their own factories or trading

houses.

As a result of its rapid growth, this industry was brought under the MFA quota

system in 1986. The system is said to have favored Bangladeshi producers because it

protected them from foreign competition and because a more generous export quota was

given to Bangladesh than to India and Sri Lanka (e.g., Mlachila and Yang, 2004; Siddiqi,

2005; Saxena and Wiebe, 2005). Another important event in 1986 was that the Export

Promotion Bureau of Bangladesh and the United Nations Development Programme

(UNDP) sent several Bangladeshi knitwear producers to Italy, Germany and Poland,

which helped the producers understand the latest production techniques and export

market conditions, according to Hoq (2004).

From the late 1980s through the 1990s, the center of gravity of the industry

shifted from foreign ventures to indigenous producers and traders, who increased

exports to large retailers and branded marketers in Europe and North America. These

customers make new designs at their headquarters and sell the products under their own

brands through their own distribution channels. However, production is subcontracted

to suppliers in developing countries. The case studies of global value chains covering

Asia and Latin America find that working to the specifications set by global buyers has

helped local suppliers improve production efficiency and product lines, and that some

local suppliers have been able to obtain more advanced know-how, such as designing,

6

marketing, and branding (e.g., Gereffi, 1999; Tewari, 1999; Schmitz and Knorringa,

2000; Bazan and Navas-Aleman, 2004). The same seems to hold true in the case of

the suppliers in Bangladesh.

In 1994, the international community decided to abolish MFA gradually. The

phased abolition of the quota system began in 1995 and was completed in 2005.

Because the quota elimination allows global buyers to import as many garments from

their favorite suppliers as they like, the phase-out of the quota has increased competitive

pressure on suppliers. The Bangladeshi garment industry has been a winner. From

1983 to 2008, as shown in Table 1, the number of garment producers, their employment

and export value, and the share of garment export in the country’s total export earnings

continued to increase. In 2007/08, the industry earned more than 10.7 billion US

dollars from exports, comprised over 4700 factories, employed two and a half million

workers, and accounted for more than 76 percent of the country’s total export earnings.

Although not shown in the table, garment traders who help to intermediate

between local producers and global buyers have also increased in number. Although

garment traders have no legal obligation to register with the Bangladesh Garment

Manufacturers and Exporters Association (BGMEA), 846 traders are registered with

BGMEA today. The location of the garment manufacturers and traders is concentrated

in the greater Dhaka area and the greater Chittagong area. The Dhaka area has

approximately three times as many manufacturers as Chittagong. While Dhaka-based

manufacturers are sprawling out to the suburbs, almost all the traders have their offices

in the urban center because foreign buyers hesitate to visit their suppliers located in the

suburbs. One can start a garment trading house with a much smaller initial investment

than for a garment factory. A number of garment businessmen started their careers as

7

traders and later branched out into garment manufacturing.

One of the impacts of the phase-out of the quota on the Bangladeshi garment

industry is the rise of the knitwear sector and the comparative decline in the woven

garment sector. The emphasis on quick delivery due to the intensified competition has

put the producers of woven garments, such as shirts and pajama sets, at a disadvantage

compared to producers of knitwear, such as polo shirts and sweaters. This is because

the woven garment producers depend heavily on imported woven textile, which is prone

to delays in delivery, whereas the knitwear producers have an abundant and quick

supply of domestically produced yarn and fabrics (Siddiqi, 2005). Moreover, knitwear

producers in Bangladesh have managed to find higher value-added markets for sweaters

and cardigans. As a result, the knitwear sector has grown much faster than the woven

garment sector since 1995 in terms of both the number of manufacturers and export

earnings per manufacturer, as shown in Table 2. In 2007/08, the knitwear sector

earned more export revenue than the woven garment sector for the first time.

Global value chains are influenced by not only market demands and trade

agreements. Global buyers fear that trade unions, NGOs, human rights groups and

consumers’ associations in developed countries may accuse them of encouraging their

suppliers in developing countries to run sweatshops and use child labor. To avoid such

accusations, they have set parameters on standard rules and regulations regarding

product quality, safety, labor standards, working environment, and child labor issues,

and push local suppliers to follow these standard parameters, which are known as the

codes of conduct (Humphrey and Schmitz, 2004). In Bangladesh, the garment

industry has undertaken large investments in equipment and working conditions in order

to comply with these codes of conduct (e.g., Rahman, 2005).

8

Still, wages are low in Bangladesh. As Figure 1 shows, the real wage index,

which was 100 in 1969/70, remained largely the same in the early 1980s. This

suggests that real wages remained more or less constant for more than a decade. It

rose slowly in the next 15 years to 140, and more rapidly in the early 2000s.

According to the Japan External Trade Organization’s (JETRO 2009) compilation of

investment cost information, an ordinary factory worker in Dhaka is paid less than half

the wage for her counterpart in the Philippines. The labor cost difference may partly

explain why the garment industry in Bangladesh has been rising in contrast to that in the

Philippines, which has been declining. However, garment exports have been growing

as rapidly in Turkey as in Bangladesh, and wages are much higher in Turkey than in the

Philippines. The garment industry in Turkey focuses on a market segment for more

fashionable garments than Bangladesh and is said to have an advantage of being

geographically closer to Europe. In any case, it is clear that labor cost is not the only

determinant of the development of the garment industry.

3. Hypotheses

We agree completely with many other authors that the Desh-Daewoo training

kick-started the garment industry in Bangladesh. The thorough training formed a mass

of specific human capital with high skills not only in production techniques but also in

factory management, international procurement, and international marketing (Rhee

1990). For those with high levels of general human capital and abundant financial

capital, the massive infusion of specific human capital meant that their production

functions shifted upward substantially. According to our interview with one of the 130

Desh-Daewoo trainees, highly educated people and wealthy families in Bangladesh

9

were not at all interested in the garment business and any kind of manufacturing.

However, looking at Desh’s good start in exporting, some of them changed their mind

and decided to supply their resources to the garment industry; that is, financial capital

and high levels of general human capital.

Among those who had received good training, some had sufficient financial

capital to establish their own garment factories, and others became garment traders or

worked for other manufacturers as managers.3 Probably, traders would work with those

manufacturers who started garment manufacturing without sufficient specific human

capital. Traders’ services would be primarily related to but not limited to marketing.

Traders who received better training would provide a variety of more valuable services

to client manufacturers. With the help of traders, a large number of manufacturers can

start production and export without wasting time on trial and error, which would

otherwise be needed to find their way into foreign markets. The collaboration with

traders will also reduce the risk of manufacturers’ bankruptcy. As manufacturers

acquire specific human capital, they would reduce their dependence on traders and

increase direct transactions with foreign buyers.4 Traders will divert their resources

away from such graduating manufacturers to newly established manufacturers. As

Sonobe and Otsuka (2006) argue based on case studies in East Asia, the division of

labor between manufacturers and traders facilitates the expansion of an industry.

Easterly (2002, Ch. 8) refers to Romer’s (1986) model of endogenous economic

growth to explain the two special attributes of knowledge as a good: one is that

3 Those who received formal training abroad were highly educated because only those who were fluent in English were given training opportunities. 4 Sonobe, Hu, and Otsuka (2002) report that garment manufacturers in Zhejiang province in China depended on local traders initially and then increased direct transactions with retailers in remote large cities.

10

knowledge leaks, and the other is that the same knowledge can be used repeatedly

without causing rivalry. He uses the anecdote of the Bangladesh garment industry to

illustrate how the non-rivalry and spillovers of knowledge as a production input lead to

external increasing returns. In our view, transactions between traders and

manufacturers are an important channel of knowledge spillovers, and external

increasing returns arising from knowledge spillovers are strong where traders and

manufacturers are agglomerated, because their proximity to each other in an

agglomeration reduces transaction costs, thereby encouraging transactions as the

theoretical model developed by Becker and Murphy (1992) predicts. Probably, the

traders’ provision of services to manufacturers eases the entry of new manufacturers,

leading to the expansion of agglomeration, which in turn facilitates the division of labor

between traders and manufacturers.

Our argument concerning enterprise growth begins with a general observation

that a growing enterprise has increasingly complex flows of cash, credit, and debit, an

increasing number of workers, and an increasing number of transacting partners. Thus,

growth requires good management, which in turn requires management skills.

Moreover, the global garment chains keep pressure on local suppliers to shorten

delivery time, improve quality, and reduce unit costs (Gereffi, 1999; Schmitz and

Knorringa, 2000). Thus, enterprise growth will depend on the pace at which the

enterprise learns advanced skills and know-how. When the enterprise is small, the

entrepreneur himself can devote much time to learning. As the enterprise grows

further, however, he needs to hire managers and experts and delegate authority to

managers because his time and capacity are limited (Aghion and Tirole, 1997; Hart and

Moore, 2005). As a result, the enterprise’s organizational learning no longer depends

11

on the entrepreneur’s learning ability but on his selection of managers and experts as

well as the quality of his guidance to them. We conjecture that the entrepreneur’s

ability to make good decisions related to organizational learning as well as his learning

ability depends on his general human capital.

To substantiate these arguments, we will analyze some portion of the arguments

with data. The first concerns the provision of services by traders to manufacturers as

follows:

Hypothesis 1: Traders with good training and rich experience in the business

provide manufacturers with higher-valued services and earn greater revenues,

whereas manufacturers with high education and rich experience depend less on

traders’ services.

We have argued that the growth of an enterprise depends on its organizational learning

because not just the scale but the complexity of activities increases as the enterprise

grows. We have also argued that organizational learning depends on the general human

capital of the entrepreneur. Since the activities of trading houses are much smaller in

scale and less diverse, it follows that the growth effect of the entrepreneur’s general

human capital is expected to be weaker for the trader than the manufacturer. If our

argument is correct, however, it also follows that the trader’s general human capital

assumes importance when his business branches out into manufacturing. We thus

formulate a hypothesis as follows:

Hypothesis 2: The trader’s education level has only a weak effect on the growth of

12

his trading business but a positive and significant effect on his entry into

manufacturing.

We have argued that the manufacturer’s education level is likely to have a positive

effect on enterprise growth. From formal education, one learns how to learn. Does

this apply to formal training? To the extent that formal training concentrates on the

acquisition of specific skills and know-how, there is no convincing reason to expect that

it has a lasting growth effect, and thus it seems reasonable to formulate the following

hypothesis:

Hypothesis 3: The manufacturer’s education level has positive effects on the size

and growth of his enterprise, but his prior formal training has only a level effect.

4. Data

In August and December 2005, we conducted preliminary surveys of producers, traders,

and consultants for four weeks in total. After visiting both knitwear factories and

woven garment factories of various sizes, we decided to study the development of the

knitwear sector and the trading sector, because the knitwear sector has been growing

faster than the woven garment sector in recent years and because the roles of traders in

industrial development remain an enigma. Our sample was selected randomly from

the member lists of three associations: BGMEA, the Bangladesh Knitwear

Manufacturers and Exporters Association (BKMEA), and the Bangladesh Garments

Buyers Association (BGBA). After establishing rapport with some leaders of these

associations, we were able to request introductions to the entrepreneurs whom we

13

selected randomly. For three months, from late December 2005 to early March 2006,

we visited 40 traders and 100 knitwear producers in the greater Dhaka area,5 and we

were able to interview the entrepreneur at each enterprise except for eight factories.

Our sample consists of 40 traders and 92 knitwear manufacturers. They

provided us with recall information on the number of workers, production and costs,

export revenues in 1998, 2000, 2002, 2004, and 2005, as well as information on their

educational and occupational backgrounds. For simplicity, we refer to the knitwear

manufacturers in the sample as manufacturers. Note, however, that they do not

represent woven garment manufacturers. By contrast, the traders in the sample

represent the whole population of traders, because they were selected randomly

regardless of their specialization. The sample traders deal in both woven garments and

knitwear. Of the 92 sample manufacturers, 37 produce sweaters and cardigans, while

the remaining 55 do not produce these items but polo shirts and T-shirts instead.

The sample entrepreneurs’ background attributes are summarized in Table 3.

Both the traders and manufacturers went to school for 15 years on average, and about 60

percent of them went to school for 16 years or more.6 These numbers are very high,

compared with the education levels of entrepreneurs in garment and other light

manufacturing industries in East Asia reported by Sonobe and Otsuka (2006), Vietnam

by Nam et al. (2009a, b), and Africa by Sonobe et al. (2009a, b). The high education

level of entrepreneurs is characteristic of the Bangladeshi garment sector. Note,

however, that highly educated people became interested in garment business only after

5 The Greater Dhaka area consists of Dhaka and the neighboring two districts, i.e., Narayanganj and Gazipur. Narayanganj used to be the center of the production of hosiery, but it has many T-shirt factories now. Gazipur is a newly developed cluster of garment factories, especially those producing sweaters. 6 In Bangladesh, it takes 16 years for a student to receive a specialized degree.

14

the Desh-Daewoo technology transfer, as mentioned in the previous section. In our

view, high education is characteristic of this industry, simply because comparably

sizable technology transfers, which shift production functions substantially upward and

attract the highly educated to manufacturing, rarely take place.

By “years in prior experience in garment trading,” which appears in the middle of

Table 3, we mean the number of years the entrepreneur has engaged in the garment

trading business, as a worker, manager, or unpaid family worker, prior to the

establishment of his own business. Similarly, “prior experience in garment

production” is the number of years the entrepreneur has worked at a garment factory

before he started his own trading house or factory, and “prior experience in other

sectors” is the number of years the entrepreneur has worked in a non-garment sector.

The second row from the bottom of Table 3 shows the percentage of

entrepreneurs who received formal training in garment businesses in Korea and other

foreign countries. Three traders and two manufacturers in the sample received

intensive training in South Korea and Singapore. Among them is one of the 130

Desh-Daewoo trainees. UNDP provided support for the training of three sample

manufacturers in Europe. Similarly, the Association for Overseas Technical

Scholarship (AOTS) of Japan supported the training of a sample trader in Japan.

Three other sample traders and one sample manufacturer received intensive training in

China, Taiwan, and Sri Lanka at their own expense. We regard these entrepreneurs as

the recipients of good prior training and distinguish them from the other sample

entrepreneurs in the analysis below. The education levels of these entrepreneurs with

good prior training tend to be higher than the sample average. Among the six such

manufacturers, the lowest education level is 16 years of schooling, and the mean is 17.3

15

years.

Tables 4 and 5 present data on the size and activities of the trading and

manufacturing enterprises, respectively. Since new entrants tend to be smaller in size

than incumbent enterprises, it is difficult to see how rapidly enterprises grow if

enterprises of very different ages are mixed. This is why these tables divide the

sample enterprises into three cohorts: those established in 1994 or earlier, between 1995

and 1999, and between 2000 and 2005. These tables clearly indicate that the growth of

trading enterprises in employment and sales revenues (or export earnings) is slower than

the growth of manufacturing enterprises in employment and value added, and that the

former becomes even slower as trading enterprises mature whereas fast growth is

sustained in the manufacturing case.

As indicated in the second row of Table 4, there are five foreign ventures in our

trader sample. They are foreign ventures in the sense that their headquarter offices are

located in foreign countries, even though the heads of their Bangladesh offices are

Bangladeshi except for one, who is Italian. None of the sample manufacturing

enterprises are foreign ventures. With the small number of foreign ventures in our

sample, the impacts of foreign direct investment, if any, are impossible to analyze with

our data. In the garment industry in Bangladesh, a long-run tendency is that foreign

enterprises have been replaced by indigenous enterprises as service providers,

manufacturers, and material suppliers (e.g., Siddiqi, 2005; Quddus and Rashid, 2000;

Yamagata, 2006). The proportion of trading houses that hire foreign experts in

marketing is high among relatively long-established trading houses and low among

newly established trading houses, which may be another indication of the long-run

tendency of decreasing dependence on foreign human capital. We will return to this

16

point in the next section.

To test Hypothesis 1, we need a measure of high-valued services provided by

traders. All traders provide services related to the procurement of materials from

abroad and the marketing and shipment of product to abroad. It is practically

impossible, however, to measure the quality or the value of such service. Instead we

will take advantage of the fact that a limited number of traders provide services related

to manufacturing. An example is to make counter samples on behalf of a manufacturer,

following the sample or specifications set by its buyer abroad. By inspecting counter

samples submitted by local suppliers, the global buyer decides with whom to place the

order. Traders provide sample making services because some manufacturers are too

busy and some lack the skills necessary to make a satisfactory counter sample. As

shown in Table 4, the majority of traders have equipment and staff for sample making.

Another example is design re-engineering. While designs are sent by global

buyers to local suppliers, some designs require modifications in order to become more

attractive to consumers or easier to produce in large quantities. Some global buyers

allow their local suppliers to re-engineer designs. When a manufacturer is allowed to

re-engineer a design but cannot do so satisfactorily, he may ask a capable trader for help.

Table 4 shows the mean of the percentage of designs that a trader re-engineered in the

specified year. Design re-engineering is regarded as a higher-valued service than

sample making, according to the result of our interviews with traders and manufacturers.

In the next section, we will use this percentage or fraction of designs re-engineered and

a dummy variable indicating whether the trader has the capacity to make counter

samples as proxies for the provision of high-value services.

Table 4 ends with the data on the entry of the traders into manufacturing, which

17

we use to test Hypothesis 2. If the three cohorts are mixed, slightly more than half of the

traders were engaged in garment manufacturing in 2005. According to our interviews

with traders who owned factories, their products are exported through their own trading

houses. Thus, if a trader starts operating a garment factory, his export sales revenue

increases. The trading houses which were established between 1995 and 1999 had a

high average of sales revenues in 2005 because as many as 68.7 percent of them had

started manufacturing.

Turning to the manufacturing enterprises, Table 5 shows the percentage of their

direct export sales revenues. This percentage is equal to 100 minus the percentage of

their sales revenues coming from or through traders. The percentage of direct export

by the relatively long-established enterprises increased over time and is higher than that

of the direct export by the newly established enterprises. These observations lend

strong support to our argument that the manufacturers reduce their dependence on

traders as they learn skills and know-how that they did not have when they started

production.

Table 5 also shows the percentage of manufacturers with an international

certificate. Global buyers that specialize in high-quality segments of the world

garment market tend to prefer dealing directly with local manufacturers, rather than

through traders. Probably, this is because direct transactions allow tight control of

local manufacturers in terms of product quality, delivery time, and the codes of conduct.

For local manufacturers, direct transactions with these demanding buyers involve large

investments in equipment, buildings, and training workers but are profitable since only

a limited number of local manufacturers in the world can satisfy these buyers’ needs.

To attract the attention of these buyers, a number of manufacturers in Bangladesh have

18

obtained certificates from international auditing bodies, such as International

Organization for Standardization (ISO). This is why the data on the percentage of

manufacturers with such certificates show a similar tendency as the data on the

percentage of direct exporting sales revenues.

Direct transactions with global buyers, especially those with quality-conscious

buyers, require the use of expensive machines and tight quality control, which in turn

require the input of engineering expertise. Our respondents informed us that human

resources with such expertise have become scarce in the industry and that garment

manufacturers have increased the employment of foreign experts mainly from Sri Lanka

and India. The data shown in Table 5 are consistent with this information.

5. Regression Analyses

Tables 6 to 8 present the results of the reduced-form regressions. Table 6 and the first

column of Table 7 concern Hypothesis 1 on the symbiotic relationship between traders

and manufacturers. The rest of Table 7 is about Hypothesis 2 on the effects of traders’

education levels on the growth of their businesses and their entry into manufacturing.

Table 8 concerns Hypothesis 3 on the determinants of the size, productivity, and growth

of manufacturing enterprises.

Column (1) of Table 6 presents an estimate of the function explaining the fraction

of designs re-engineered as a proxy for the provision of high-valued services by the

trader. The major determinants of this variable are the trader’s formal training in a

foreign country, the number of years of operation as an indicator of organizational

experience as a trading house, the dummy variable indicating whether the trading house

is a foreign venture, and the entrepreneur’s experience of working at a foreign venture

19

prior to the establishment of his trading house. Column (2) reports the estimated

probit model for the determination of whether to provide sample making services. The

estimates in columns (1) and (2) are qualitatively similar, with two exceptions. One is

that the foreign venture dummy has a much stronger effect on design re-engineering

than on sample making. Another is that the entrepreneur’s experience of working at a

foreign venture has a positive effect on design re-engineering and a negative effect on

sample making. Column (3) presents the OLS estimates of the function explaining the

sales revenue of the trading house. Note that the effects of the regressors on sales

revenues are qualitatively similar to those on design re-engineering and sample making.

These results indicate that the trading houses with the skills of design re-engineering

and sample making earn high sales revenues, lending support to the first half of

Hypothesis 1.

The regression results shown in column (4) indicate that the trading house is more

likely to employ foreign experts if the entrepreneur is not highly educated, did not

receive prior training in the garment business, had prior experience of working at a

garment factory or a garment trading house, and the trading house is long established

and foreign owned. The negative and significant effects of the education variable and

the prior formal training variable suggest that foreign experts are not useful for the

traders with high education, high skills and know-how specific to the garment industry.

The positive and significant effects of the prior experience in garment production and

marketing on the employment of foreign experts, together with the insignificant effects

of these variables on sales revenues, suggest that such experience as a worker are poor

substitutes for formal training in a foreign country. The coefficients on the four

year-dummy variables are negative, and those for the most recent two years are

20

significant in column (4). These results indicate that the employment of foreign

experts has been decreasing over time, probably reflecting the increasing presence of

the Bangladeshi garment industry in the world garment market as well as the increasing

skills of the indigenous garment traders. According to column (4), however, the

long-established trading houses tend to employ foreign experts, and according to

columns (1) to (3), they tend to perform well, relative to the newly established trading

houses. These results suggest that the employment of foreign experts has been an

important channel of technology transfer.

Table 7 reports the regression results concerning manufacturers’ direct exporting,

traders’ growth in revenues, and traders’ entry into manufacturing. Column (1)

presents the estimated tobit model in which the dependent variable is the fraction of the

manufacturer’s sales revenue from direct exporting. This direct export ratio is one

minus the rate of dependence on traders. In this column, the years of schooling of the

entrepreneur, the years of operation, and the sweater production dummy have positive

and significant effects. These results support the second half of Hypothesis 1. A

question arises as to what the number of schooling years represents. It may represent

the general human capital of the entrepreneur, or the wealth of his family or parents, or

both. Manufacturers need much greater investments to start businesses than traders.

Thus, we should be careful when we discuss the effects of the education and prior

experience of manufacturers on their business results. We will return to this point

later.

Columns (2) and (3) of Table 7 present the results of the growth regressions for

traders, which include a lagged dependent variable on the right-hand side. In column

(2), the dependent variable is the logarithm of the sales revenue in 2005, and the

21

corresponding variable on the right-hand side is the logarithm of the sales revenue in

2000. This is a cross section regression with a small sample size of 33 trading houses.7

In column (3), the four periods (i.e., 1998-2000, 2000-2002, 2002-2004, and

2004-2005) are pooled to increase the sample size. In neither column does the

education variable have a significant growth effect, which is consistent with the first

half of Hypothesis 2. The results of these regressions indicate that the trading houses

with older entrepreneurs have grown in size more slowly than those with younger

entrepreneurs, and that the foreign trading houses have grown in size more rapidly than

the indigenous trading houses.

Column (4) presents the estimated probit model for the trader’s entry into garment

manufacturing. The foreign trading house dummy has a negative and significant effect

on the entry into manufacturing, suggesting that the foreign trading houses are not

interested in manufacturing. An indigenous trader is more likely to own a factory if he

is highly educated, received formal training abroad, has prior experience in garment

production or trading, and has no experience of working at foreign ventures and if his

trading house was established long ago. To finance entry into manufacturing, the

trader has to earn large profits every year and accumulate sufficient financial assets,

which may be reflected in the positive and significant effects of the prior formal training

dummy and the years of operation. The positive and significant effect of the education

variable is consistent with the second half of Hypothesis 2. This effect may be a

reflection of the financial requirement for the entry into manufacturing. It may also be

a reflection of the requirement of the sufficient learning ability of the entrepreneur for

7 While there are 40 traders in our sample, the data on the sales revenue in 2000 are available from only 33 traders. An even smaller sample offers data on sales revenue in 1998. The period from 2000 to 2005 is the longest period for which we can run a growth regression.

22

the operation of a manufacturing enterprise.

Table 8 presents the regression results for manufacturers’ enterprise size and

growth as well as attempts to strengthen their competitiveness. The entrepreneur’s

education level, his prior experience in garment production, garment marketing, and

other sectors, the enterprise’s years of operation, and the sweater production dummy

have positive and significant effects commonly on the three dependent variables. The

entrepreneur’s age has negative effects on the three dependent variables. There are

also dissimilarities, however. For example, the prior formal training dummy has no

significant effect on value added, a negative and significant effect on employment, and a

positive and significant effect on the likelihood that the enterprise has received a

certificate from an international standards organization. These results suggest that the

enterprise tends to have higher labor productivity and transact with more quality

conscious buyers if its entrepreneur had good prior training.

The coefficients of the year dummies in columns (2) and (3) indicate that the

employment and the proportion of manufacturing enterprises with international

certificates have increased in recent years, even after controlling for the strong effects of

the years of operation. In column (1), the effects of the first three year dummy

variables on value added are negative, but the effect of the year 2005 dummy and that of

the year 2000 dummy are positive and marginally significant. Thus, there was a

tendency among the manufacturing enterprises for both employment and value added to

increase. Note that the expansion of enterprise size was not accompanied by

improvement in labor productivity. Presumably, the size expansion was a response to

the recent tendency among the global buyers to focus on local suppliers with large

production capacity.

23

Column (4) shows the estimated probit model for the employment of foreign

experts. According to our interviews with manufacturers, they employ Indian and Sri

Lankan experts because these experts have skills useful for meeting the needs of quality

conscious global buyers. Thus, there is small wonder that columns (3) and (4) share

similar patterns of signs of coefficients. Column (5) presents the result of a growth

regression in which the dependent variable is the logarithm of value added in 2005 and

one of the regressors is that in 2000. It shows that the entrepreneur’s education level

has a highly significant effect on growth in value added, whereas his prior formal

training does not, which lends support to Hypothesis 3.

In the sample of manufacturers, there is no entrepreneur who used to work as a

garment factory worker, and those entrepreneurs with prior experience in garment

production used to work as managers or work at factories owned by their family

members or relatives. Thus, if the years of prior experience in garment production

takes a positive value, it is likely that the entrepreneur is from a wealthy family.8

Similarly, if the years of prior experience in other sectors take a positive value, it is

likely that the entrepreneur or his family own a business other than garment

manufacturing. Thus, the positive effects of these variables on the growth in value

added are attributed largely to the ability of the entrepreneur to finance enterprise

growth.

These considerations may cast doubt on our interpretation of the positive effect of

the entrepreneur’s education level on enterprise growth. Does it reflect the effect of

the general human capital useful for enterprise growth? Or should it be solely 8 In the case of traders, this variable and the variable called the years of prior experience in garment marketing are unlikely to represent the wealth of the entrepreneur’s family since the traders with prior experience in garment production or marketing are mostly ex-workers of garment factories or garment trading houses.

24

attributed to the association between the education level of the entrepreneur and his

family’s wealth? To cast a light on this issue, we estimated a function explaining the

employment size in the first year of the enterprise’s operation. If entrepreneurs from

wealthy families tended to build large factories, and if the education variable represents

the wealth of the entrepreneur’s family rather than his general human capital, this

variable is expected to have a positive effect on the initial employment size. The

estimation result is reported in column (6) of Table 7. The effect of the education

variable is positive but insignificant. This is a result favorable to our argument in

Section 3 that the entrepreneur’s general human capital has a growth effect. The years

of prior experience in garment production and marketing have positive and significant

effects on the initial employment size, which is consistent with the assumption that

entrepreneurs from wealthier families tended to build large factories. The negative and

significant effect of the years of operation indicates that the initial employment size was

smaller in the remote past than in recent years. This result is consistent with the

tendency for enterprise size to increase.

6. Conclusions

This paper has examined the successful development process of the Bangladesh

garment industry and explored the key to its success by using primary data on garment

manufacturers, trading houses, and the entrepreneurs operating these enterprises. The

results of our regression analyses indicate that the high education of the garment

manufacturer has had a positive effect on enterprise growth. Presumably, this is

because manufacturers have to upgrade their skills and know-how continuously in order

to survive the intense competition in the world garment market and because the high

25

levels of the general human capital of the entrepreneur are needed to manage an

increasing number of managers and experts. The high education level of the

entrepreneurs is likely to be a major reason why the garment industry in Bangladesh has

continued to grow for the last three decades.

The regression results also indicate that the formal training that the garment

entrepreneur has received in a foreign country has not had a significant effect on

enterprise growth. The entrepreneur’s experience of working at a garment enterprise

also has not had a growth effect. However, the formal training and on-the-job training

have had significant effects on enterprise size or productivity or both even many years

after. Those garment workers who had acquired skills and know-how but could not

afford to start factories started trading houses and helped those new manufacturers

without marketing skills. It has also turned out that foreign owned trading houses

perform better than indigenous trading houses, which suggests that there still exist skills

and know-how to be learned from foreign countries. Thus, technology transfer seems

to be a long-lasting process, and its effect also seems long-lasting. These findings of

this paper strongly support the proposition that sizable technology transfers, which

include marketing, procurement, and management as well as production technologies,

are critically important for the promotion of industrial development in developing

countries.

26

References

Aghion, Philippe, and Tirole, Jean. 1997. “Formal and Real Authority in Organization.”

Journal of Political Economy 105, 1-29.

Bangladesh Bureau of Statistics (BBS). Statistical Yearbook of Bangladesh. Various

Issues. Dhaka: Bangladesh Bureau of Statistics.

Bangladesh Export Promotion Bureau. 2005. Export from Bangladesh 1972-73 to

2004-2005. Dhaka: Export Promotion Bureau.

Bangladesh Garment Manufacturers and Exporters Association (BGMEA). BGMEA

Members Directory. Various years. Dhaka.

Bangladesh Garment Manufacturers and Exporters Association (BGMEA). 2008.

Online: http://bgmea.com.bd/images/member (accessed July 15, 2009).

Bangladesh Knitwear Manufacturers & Exporters Association (BKMEA). 2005.

BKMEA Members Directory 2005. Dhaka.

Bangladesh Knitwear Manufacturers & Exporters Association (BKMEA). Online:

http://www.bkmea.com/facts_figures.php (accessed July 09, 2009).

Bazan, L. and Navas-Aleman, Lizbeth. 2004. “The Underground Revolution in the

Sinos Valley: A Comparison of Upgrading in Global and National Value Chains.”

In Local Enterprises in the Global Economy: Issues of Governance and

Upgrading, ed. Hubert Schmitz. Cheltenham: Edward Elgar..

Becker, Gary S. and Murphy, Kevin M. 1992. “The Division of Labor, Coordinating

Costs, and Knowledge.” Quarterly Journal of Economics 107 (4), 1137-60.

Easterly, William. 2002. The Elusive Quest for Growth: Economists' Adventures and

Misadventures in the Tropics. Massachusetts: The MIT Press.

Fafchamps, Marcel. 2004. Market Institutions in Sub-Saharan Africa: Theory and

27

Evidence. Cambridge MA: The MIT Press.

Gereffi, Gary. 1999. “International Trade and Industrial Upgrading in the Apparel

Commodity Chain.” Journal of International Economics 48 (1), 37-70.

Hart, Oliver, and Moore, John. 2005. “On the Design of Hierarchies: Coordination

versus Specialization.” Journal of Political Economy 113, 675-702.

Hayami, Yujiro, and Kawagoe, Toshihiko. 1993. The Agrarian Origins of Commerce

and Industry: A Study of Peasant Marketing in Indonesia, London: Macmillan.

Hoq, Monzurul. 2004. “Knitwear Industry in Bangladesh: An Untold Story.” Explore

the Galore of Bangladesh Knitwear 1st Bangladesh Knitwear Exhibition. Dhaka:

BKMEA.

Humphrey, John, and Schmitz, Hubert. 2005. “Governance in Global Value Chains.” In

Local Enterprises in the Global Economy: Issues of Governance and Upgrading,

ed. Hubert Schmitz. Cheltenham: Edward Elgar.

Japan External Trade Organization (JETRO). 2009. “Kakkoku Chiiki Betsu Joho

(Information by Country/Region).” In Japanese. Online:

http://www.jetro.go.jp/world/search/compare/ (accessed August 10, 2009)

Khan, Sharifa. 2004. “Textile and Clothing Sector in Bangladesh: Post MFA Challenges

and Action Plan.” Dhaka: Ministry of Commerce, WTO Cell.

Mlachila, Montfort, and Yang, Yongzheng. 2004. The End of Textiles Quotas: A Case

Study of the Impact on Bangladesh. IMF Working Paper, No. WP/04/108.

Nam, Vu Hoang, Sonobe, Tetsushi, and Otsuka, Keijiro. 2009a. “An Inquiry into the

Development Process of Village Industry: The Case of a Knitwear Cluster in

Northern Vietnam.” Journal of Development Studies, forthcoming.

Nam, Vu Hoang, Sonobe, Tetsushi, and Otsuka, Keijiro. 2009b. “An Inquiry into the

28

Transformation Process of Village-based Industrial Clusters: The Case of an Iron

and Steel Cluster in Northern Vietnam.” Journal of Comparative Economics,

forthcoming.

Quddus, Munir and Rashid, Salim. 2000. Entrepreneurs and Economic Development:

The Remarkable Story of Garment Exports from Bangladesh. Dhaka: The

University Press Limited.

Rahman, Mustafizur. 2004. Surviving in a Quota Free World, Will Bangladesh make it?

Report No. 72. Centre for Policy Dialogue, Dhaka.

Rahman, Mustafizur. 2005. “Bangladesh After MFA Phase Out.” South Asian Journal 8

(April-June) http://www.southasianmedia.net/Magazine/journal/8_phases_out.htm

(accessed on August 3, 2007).

Rhee, Yung Whee. 1990. “The Catalyst Model of Development: Lessons form

Bangladesh’s Success with Garment Exports.” World Development 18 (2),

333-316.

Romer, Paul M. 1986. “Are Nonconvexities Important for Understanding Growth?”

American Economic Review Papers and Proceedings 80 (2), 97-103.

Rushidan Islam Rahman. 2009. “An Analysis of Real Wage in Bangladesh and Its

Implications for Underemployment and Poverty.” Bangladesh Institute of

Development Studies. Online: http://www.peri.umass.edu/fileadmin/pdf/

conference_papers/khan/Rahman_Bangladesh.doc (accessed on July 25, 2009).

Saxena, Sanchita B. and Wiebe, Frank. 2005. The Phase-out of the Multi-fiber

Agreement: Policy Options and Opportunities for Asia. San Francisco: The Asia

Foundation.

Schmitz, Hubert and Knorringa, Peter. 2000. “Learning from Global Buyers”. The

29

Journal of Development Studies 37 (2), 177-205.

Siddiqi, Hafiz G. A. 2005. The Readymade Garment Industry of Bangladesh. Dhaka:

The University Press.

Sonobe, Tetsushi, Akoten, John, and Otsuka, Keijiro. 2009a. “An Exploration into the

Successful Development of the Leather-Shoe Industry in Ethiopia.” Review of

Development Economics, forthcoming .

Sonobe, Tetsushi, Akoten, John, and Otsuka, Keijiro. 2009b. “Growth Process of

Informal Enterprises in Sub-Saharan Africa: A Case Study of a Jua Kali Cluster in

Nairobi.” Small Business Economics, forthcoming.

Sonobe, Tetsushi, Hu, Dinghuan., and Otsuka, Keijiro. 2002. “Process of Cluster

Formation in China: A Case Study of a Garment Town.” Journal of Development

Studies 39 (1), 118-139.

Sonobe, Tetsushi, and Otsuka, Keijiro. 2006. Cluster Based Industrial Development: An

East Asian Model. Basingstoke: Palgrave Macmillan.

Tewari, M. 1999. “Successful Adjustment in Indian Industry: The Case of Ludhiana’s

Woolen Knitwear Cluster.” World Development 27 (9), 1651-72.

World Trade Organization. 2008. International Trade Statistics 2008.

http://www.wto.org/english/res_e/statis_e/its2008_e/section2_e/ii69.xls (accessed

July 15, 2009).

Yamagata, Tatsufumi (2006). “Two Dynamic LDCs: Cambodia and Bangladesh as

Garment Exporters.” EIC Economic Review (Economic Institute of Cambodia) 3

(3), 81-136..

Table 1: Growth of the Garment Industry in Bangladesh

30

Fiscal year a Number of Garment Factories

Employment (million workers)

Export value (billion USD)

% of garments in the country’s export earnings

1983/84 134 0.04 0.03 3.9

1987/88 685 0.28 0.43 35.2

1991/92 1163 0.58 1.18 59.3

1995/96 2353 1.29 2.55 65.6

1999/00 3200 1.6 4.35 75.6

2004/05 4107 2.1 5.17 74.2

2007/08 4740 2.5 10.7 75.8 a. Note: In Bangladesh, a fiscal year starts in July and ends in June. Sources: Bangladesh Export Promotion Bureau (2005) and Bangladesh Garment Manufacturers and Exporters Association (BGMEA), RMG Export Statistics, http://bgmea.com.bd/index.php?option=com_ content&task=view&id=56&Itemid=175 (Accessed July 24, 2009).

31

Table 2: The Number of Manufacturers and Export Value by Sector, 1995-2008

Woven garment sector Knitwear sector Export value

(million USD) Export value

(million USD)

Year

No. of manufacturers

Total Per manufacturer

No. of manufacturers

Total Per manufacturer

1995/96 1646 1835 1.11 350 393 1.12

2000/01 2318 3364 1.45 1044 1496 1.43

2002/03 1997 3258 1.63 1229 1654 1.35

2004/05 1928 3220 1.67 1523 2536 1.66

2007/08 4076 5167 1.27 1500 5533 3.69

Sources: BGMEA (various years), Export Promotion Bureau (2005), and BKMEA (2005): Online: http://www.bkmea.com/facts_figures.php (accessed July 09, 2009).

32

Table 3: Means of Entrepreneurs’ Background Attributes Data

Traders Manufacturers

Number of sample entrepreneurs 40 92

Age (in 2005) 43.0 44.7

Years of schooling 15.2 15.0

Percentage of entrepreneurs with 16 years or more education 60.0 58.7

Years of prior experience in garment trading before starting the current business 4.0 3.8

Years of prior experience in garment production before starting the current business

2.1 3.6

Years of prior experience in working at a foreign venture before starting the current business

8.9 n.a.

Years of prior experience in other sectors before starting the current business n.a. 2.0

Percentage of entrepreneurs who received formal training in garment business abroad 17.5 6.5

Years of operation of the current business 9.8 6.8

“n.a.” indicates that data are not available. Few sample traders had worked in other sectors and few sample manufacturers had worked at foreign ventures.

33

Table 4: Means of the Trading Enterprise Data by Period of Entry

Period of Entry By 1994 1995-99 2000 and

onwards No. of Traders 16 16 8

No. foreign trading houses 1 3 1 Number of workers

1998 16.0 2002 22.6 17.0 2005 27.6 20.2 20.2

Export earnings (million USD) 1998 4.3 2002 4.4 3.4 2005 4.6 6.0 3.4

% traders hiring foreign experts 1998 31.3 2002 37.5 18.8 2005 43.8 18.8 12.5

Percentage of traders making samples

1998 68.7 2002 81.3 62.5 2005 81.3 87.5 37.5

Percentage of designs re-engineered

1998 8.8 2002 11.3 33.8 2005 24.4 41.3 12.5

Percentage of traders who own a garment factory

1998 31.3 2002 50.0 37.5 2005 50.0 68.7 37.5

34

Table 5: Means of the Manufacturing Enterprise Data by Period of Entry

Period of Entry By 1994 1995-99 2000 and

onwards No. of Producers 25 34 43

% producing sweater 32.0 20.8 55.8 Number of workers

1998 734.0 2002 1287.0 677.6 2005 1662.4 1306.3 939.7

Value added (million USD) 1998 3.1 2002 4.6 3.0 2005 6.3 3.6 3.1

Percentage of sales revenue from direct exporting

1998 48.1 2002 58.0 32.9 2005 65.8 45.6 40.4

Percentage of manufacturers with an international certificate

1998 12.5 2002 64.0 45.8 2005 76.0 66.7 34.9

Percentage of manufacturers hiring foreign experts

1998 4.2 2002 20.0 8.3 2005 32.0 12.5 18.6

Percentage of manufacturers with a composite factory

1998 25.0 2002 28.0 4.2 2005 40.0 16.7 13.9

35

Table 6: Estimated Functions Explaining Behavior and Performance of Traders

(1) (2) (3) (4) Fraction of

Designs re- engineered

Sample making dummy

ln(Export revenue)

Foreign expert

employment Tobit Probit OLS Probit Years of schooling -0.02

(-0.68) 0.05

(0.59) -0.001 (-0.01)

-0.37*** (-3.17)

Prior formal training dummy 0.59*** (4.60)

1.58*** (3.68)

0.83*** (4.60)

-0.87** (-2.22)

Age -0.02 (-1.49)

-0.043* (-1.71)

-0.04** (-2.11)

0.04 (1.30)

Years of prior experience in garment marketing

0.015 (0.83)

0.03 (0.78)

0.03 (1.32)

0.12*** (3.28)

Years of prior experience in garment production

-0.01 (-0.57)

-0.07* (-1.91)

0.02 (1.07)

0.083* (1.85)

Years of operation 0.04*** (3.11)

0.13*** (4.76)

0.15*** (6.28)

0.09*** (4.04)

Foreign trading house dummy

1.67*** (8.58)

0.60 (1.49)

1.01*** (5.19)

2.58*** (5.21)

Foreign venture experience dummy

0.41*** (3.89)

-0.65** (-2.48)

0.19 (1.01)

0.009 (0.04)

Year 2000 dummy -0.03 (-0.22)

-0.013 (-0.04)

0.05 (0.18)

-0.35 (-0.81)

Year 2002 dummy -0.02 (-0.14)

-0.18 (-0.49)

-0.01 (-0.03)

-0.51 (-1.22)

Year 2004 dummy 0.04 (0.28)

-0.20 (-0.50)

0.08 (0.26)

-0.82* (-1.92)

Year 2005 dummy 0.14 (0.95)

0.07 (0.18)

0.15 (0.45)

-0.84** (-1.97)

Constant 0.19 (0.29)

0.52 (0.32)

14.6*** (12.29)

2.29 (1.60)

N 176 176 176 176 Left censored Right censored

107 19

R2 0.38 Numbers in the parentheses are the t- or z- statistics based on heteroskedasticity-robust standard errors. ***, **, * indicate the 1, 5, and 10 percent significance levels, respectively.

36

Table 7: Estimated Functions Explaining Manufacturers’ Direct Exporting and Traders’ Sales Revenue, Growth, and Entry into Manufacturing

(1) (2) (3) (4) Manufacturer’s

direct exporting

Trader’s ln(revenue)

Trader’s ln(revenue)

Trader’s own factory dummy

Tobit OLS OLS Probit Years of schooling 0.16***

(5.90) 0.043 (0.34)

-0.006 (-0.22)

0.21** (2.15)

Prior formal training dummy

0.17 (0.92)

0.39 (0.95)

0.05 (0.52)

0.68** (2.06)

Age

0.01 (0.86)

-0.04 (-1.34)

-0.02*** (-2.92)

-0.02 (-0.86)

Years of prior experience in garment production

-0.001 (-0.13)

-0.006 (-0.10)

0.008 (0.54)

0.07** (2.03)

Years of prior experience in garment marketing

0.0097 (1.31)

-0.01 (-0.15)

0.001 (0.12)

0.09*** (2.63)

Years of prior experience in other sectors

-0.009 (-1.09)

Foreign trading house dummy

0.22 (0.54)

0.053 (0.54)

-0.61** (-2.30)

Foreign venture experience dummy

0.67* (1.89)

0.33*** (3.16)

-0.61** (-2.30)

Years of Operation 0.06*** (6.50)

0.002 (0.05)

0.004 (0.36)

0.07*** (2.58)

Sweater dummy 0.67*** (6.81)

Year 2000 dummy -0.15 (-0.94) 0.04

(0.12) Year 2002 dummy -0.14

(-0.91) 0.012 (1.20)

0.19 (0.54)

Year 2004 dummy -0.13 (-0.88) 0.143

(1.19) 0.28

(0.78) Year 2005 dummy -0.12

(-0.81) 0.065 (0.58)

0.67* (1.78)

Lagged dependent variable

0.48*** (3.64)

0.76*** (14.04)

Constant -2.45*** (-5.32)

8.91*** (3.82)

4.60*** (4.44)

-3.73** (-2.42)

N 341 33 136 176 Left censored Right censored

83 114

R2 0.56 0.87 Numbers in the parentheses are the t- or z- statistics based on heteroskedasticity-robust standard errors. ***, **, * indicate the 1, 5, and 10 percent significance levels, respectively.

37

Table 8. Estimated Functions Explaining the Size and Growth of Manufacturing Enterprises.

(1) (2) (3) (4) (5) (6) ln(value

added) ln(current number of workers)

Int’l certificate

Foreign expert em-ployment

ln(value added in

2005)

ln(initial number of workers)

OLS OLS Probit Probit OLS OLS Years of schooling 0.15***

(3.34) 0.08*** (4.41)

0.35*** (6.40)

0.09 (1.25)

0.19*** (3.75)

0.02 (0.38)

Prior formal training dummy

0.01 (0.04)

-0.31** (-2.55)

1.25*** (3.32)

1.01*** (2.85)

-0.53 (-1.15)

0.03 (0.09)

Age

-0.01 (-1.05)

-0.02***(-2.87)

-0.03***(-2.62)

0.002 (0.14)

-0.01 (-1.04)

0.01 (0.88)

Years of prior experience in garment production

0.07*** (5.77)

0.05*** (6.60)

0.03* (1.88)

0.03* (1.85)

0.04** (2.04)

0.04** (2.08)

Years of prior experience in garment marketing

0.03*** (3.80)

0.02*** (3.58)

0.03** (2.55)

0.023 (1.50)

0.01 (0.64)

0.034** (2.00)

Years of prior experience in other sectors

0.034*** (3.48)

0.04*** (6.29)

0.04** (2.42)

-0.013 (-0.50)

0.03* (1.84)

0.01 (1.07)

Years of Operation 0.10*** (8.27)

0.09*** (11.38)

0.09*** (3.58)

0.052** (2.30)

0.04 (1.36)

-0.06** (-2.43)

Sweater dummy 0.76*** (5.44)

1.21*** (15.80)

0.54*** (2.96)

0.95*** (3.74)

0.57** (2.59)

0.79*** (4.24)

Year 2000 dummy -0.22 (-0.84)

0.10 (0.69)

0.79* (1.81)

0.08 (0.19)

Year 2002 dummy -0.17 (-0.70)

0.24* (1.78)

1.80*** (4.25)

0.38 (0.91)

Year 2004 dummy -0.03 (-0.16)

0.36*** (2.71)

2.01*** (4.86)

0.41 (1.04)

Year 2005 dummy 0.25 (1.20)

0.51*** (3.80)

2.01*** (4.92)

0.70* (1.82)

ln (value added in 2000) 0.49*** (5.63)

Constant 11.28*** (14.61)

4.19*** (12.79)

-6.95***(-6.92)

-4.12***(-3.36)

11.82*** (13.72)

4.47*** (6.78)

N 341 341 341 341 53 92 R2 0.31 0.59 0.69 0.37

Numbers in the parentheses are the t- or z-statistics based on heteroskedasticity-robust standard errors. ***, **, * indicate the 1, 5, and 10 percent significance levels, respectively.

38

Figure 1: Real Manufacturing Wage in Bangladesh, 1983/84 to 2005/06 (1969/70 = 100)

Sources: Statistical Yearbook of Bangladesh (various issues) and Rushidan Islam Rahman (2009) “An Analysis of Real Wage in Bangladesh and Its Implications for Underemployment and Poverty” Online: http://www.peri.umass.edu/fileadmin/pdf/ conference_papers/khan/Rahman_Bangladesh.doc (accessed on July 25, 2009).

0

20

40

60

80

100

120

140

160

180

200

1983/84 1987/88 1991/92 1995/96 1999/00 2005/06