In short

This open guide explains how Bangladesh’s garment export trade works, in five cited chapters: how orders move from buyer to factory, buyers’ purchasing practices, back-to-back credit and bonded warehouses, why export, factory and worker figures disagree, and why knitwear overtook woven clothing. Every figure is cited to its source, and every table is free to reuse with credit.

Want a costing for your own style? Send us your enquiry

This is an open guide to how Bangladesh’s garment export trade works: how an order travels from a foreign buyer to a sewing line, what buyers pay and how they buy, how factories finance imported fabric, how the trade is counted, and why knitwear became the larger half of it. It is written for doctoral and master’s students, academics in trade, supply chains and development, journalists and policy researchers. Each chapter sets out what the evidence shows, where it is thin, and what is still argued about, with every figure tied to the source it came from.

The scale is the reason it matters. Garments earned Bangladesh US$38.70 billion in FY2025-26 (July to June), 80.6% of the value of every good the country sold abroad, and 53.3% of that garment figure was knitwear (BGMEA n.d.-b). The same series begins in FY1983-84 at US$32 million. Behind those totals sit questions that the published research answers only in part.

What this guide covers, and what it leaves to others

The guide is about the trade: the transactions, contracts, finance and statistics that connect Bangladeshi factories to buyers abroad. It does not retell the industry’s history, which has its own cited page on this site, the Bangladesh garment industry from 1960 to 2026, with the Desh and Daewoo story, the quota years, Rana Plaza and the wage rounds. Labour, women’s work, factory safety and the environment are covered in depth by the open research guide of Collective Studio Ltd., our sister company, where I am Head of Growth. The two guides were written separately and cite their own sources; where this guide reaches a subject the other covers, it links there rather than repeating it.

Two choices follow from this. The chapters give the trade’s own documents (central bank circulars, customs rules, importers’ statistics, surveys of buyers and suppliers) the same weight as the academic literature, because much of the trade is governed by documents that, in my reading, scholars rarely cite. And they treat a buyer, a buying house and a factory as parts of one transaction, because many of the disagreements in the literature turn on which of the three is doing the deciding.

Five questions, one per chapter

1. Who stands between a foreign buyer and a sewing line?

The theory of buyer-driven chains explains how retailers and brands that own no factories can still set the terms (Gereffi 1994; Gereffi, Humphrey, and Sturgeon 2005). In Bangladesh an order may pass through the buyer’s own Dhaka office, an international agent, a local buying house or a factory that subcontracts part of it. The three published figures on how much of the trade goes through intermediaries count different things (buyers, orders and firms), and customs data may not settle it, because local intermediaries rarely take ownership of the goods (Cajal-Grossi, Macchiavello, and Noguera 2023, 2409). Published lead times for sea shipments mostly run from about 60 to 134 days, but sources start and stop the clock at different points. Chapter 1 lays out the actors, the steps and the evidence on each.

2. Do buyers’ purchasing practices squeeze suppliers?

Supplier surveys agree that buyers of Bangladeshi clothing often give short lead times, long payment terms and late changes, and that many suppliers have accepted some orders priced below their own cost (ILO 2017; Anner 2020b). On US import statistics, the average value of a square metre equivalent of Bangladeshi apparel rose in current dollars between 1997 and 2021 but fell by about a quarter once inflation is removed, a calculation the chapter shows in full. The links from purchasing terms to wages, hours and subcontracting are consistent across studies but rest on cross-sectional surveys. Chapter 2 also weighs the case against a simple buyer-power story and the evidence on who carried the cost of the orders cancelled in 2020.

3. How do factories pay for imported fabric?

Through two instruments that date from the industry’s first years: the back-to-back letter of credit, opened against the buyer’s export credit, and the special bonded warehouse, which lets inputs in free of duty (Rhee and Belot 1990). The sources do not agree on when either began, giving Desh’s first years, 1980 or 1984 (Yunus and Yamagata 2012; Fakhruzzaman 2025). Both are governed by rules that change often; today they are set out in a consolidated import circular of August 2026 and in customs rules of 2024 made under the Customs Act, 2023. Central bank dollar loans were cut from 2023, and export incentives were cut in 2024 and then, for one scheme tied to local yarn and fabric, raised in July 2026. Chapter 3 reads the rules from the circulars themselves and asks what changes when Bangladesh leaves the least developed country group.

4. How big is the industry, and how do we know?

Export figures were revised in 2024, after the central bank found that the official series had run ahead of export receipts; the revision cut FY2022-23 and FY2023-24 by about US$9 billion and US$10 billion, roughly a fifth of each (Ahmed 2024). In my comparison with what importers recorded, the revised figures pass the test for the United States and the European Union, while Canada, Japan and some smaller markets record far more than Bangladesh does. Published counts of factories run from under 1,800 to over 7,000, of workers from about 2.6 to 5.1 million, and of women’s share from 35% to 65%, depending on who counts and how. Chapter 4 explains each gap and says which series suits which question.

5. Why did knitwear overtake woven clothing?

On BGMEA’s series, knitwear first passed woven garments in FY2007-08, fell behind from FY2011-12 to FY2019-20, and has led every year since FY2020-21 (BGMEA n.d.-b). The usual explanation is fabric made at home, in vertical factories that knit, dye and sew under one roof, but the published estimates of how much fabric (or, in one source, yarn) is made locally, among those I could trace to their source, run from about 65% to 90% for knit and from about 15% to 40% for woven. In my reading, the EU’s change to its rules of origin in 2011, which let woven makers use imported fabric and raised their exports to the EU, helps explain the years when woven pulled ahead (Sytsma 2021; Sytsma 2022). Chapter 5 follows knitwear from cotton to export and asks whether vertical firms did better.

The literature, read by question

The chapters each review the work on their own subject. Read together, five lines of research run through all of them.

Who governs the chain

Gereffi’s distinction between producer-driven and buyer-driven chains (Gereffi 1994) and the later typology of value chain governance (Gereffi, Humphrey, and Sturgeon 2005) are a common starting point. Applied to Bangladesh, the Duke and World Bank studies of 2011 to 2016 reported, citing an earlier World Bank study, that two-thirds of apparel firms were in cut, make and trim work by 2000 (Fernandez-Stark, Frederick, and Gereffi 2011). They also found large buyers moving to direct sourcing through their own offices while smaller ones kept using intermediaries (Staritz 2011), and buyers ranking Bangladesh well on cost but in the weakest group on lead time and reliability (Lopez-Acevedo and Robertson 2016). In my reading the most important recent addition is the evidence that buyers are not all alike: some keep relational ties with their suppliers and others buy on the spot, and exporters earn higher markups on otherwise identical orders for the first kind (Cajal-Grossi, Macchiavello, and Noguera 2023, 2391, 2410). That finding matters for every chapter, because the same order process can carry different risks for the factory depending on which kind of buyer placed it.

Buyer power and its limits

Anner’s is the most cited argument that buyers squeeze suppliers: it links falling prices and shorter lead times to pressure on wages and workers’ rights (Anner 2020b). The ILO’s global survey of 1,454 suppliers in 87 countries gave a large cross-country picture (ILO 2017). The counter-arguments deserve a fair hearing. In the ILO’s own survey, more suppliers said they took orders below cost to beat other suppliers or secure future orders than because of buyer pressure (ILO 2017), so overcapacity among suppliers is part of the story. And the evidence on whether buyers reward good factories points both ways: a study of more than two thousand establishments in 36 countries links achieving compliance to more purchasing (Distelhorst and Locke 2018), while a study of Bangladeshi workers after Rana Plaza concludes that buyers insisting on compliance did not necessarily reward compliant factories with better terms (Kabeer, Huq, and Sulaiman 2020, 32). Chapter 2 sets the two sides out at length.

Instruments as industrial policy

The literature on how Bangladesh’s exports began treats the back-to-back letter of credit and the bonded warehouse as the tools that let firms export without the capital to buy their own fabric (Rhee and Belot 1990; Yunus and Yamagata 2012), although Mostafa and Klepper, from interviews with the first entrepreneurs, give most of the credit to the people and treat these policies as a short list of things the state did right (Mostafa and Klepper 2009). Much less has been written about how the instruments work today, about their costs, or about what the incentive schemes achieved. Recent work on graduation from the least developed country group (Razzaque 2021; Razzaque, Islam, and Rahman 2023) and on the textile and clothing chain (Kabir, Singh, and Ferrantino 2019) is beginning to fill that gap, but most of what is known still comes from the circulars and customs rules themselves, which is why chapter 3 leans on them.

Measurement as a research problem

The revision of Bangladesh’s export series in 2024 (Ahmed 2024; IMF 2025) and the long argument over how many factories and workers there are (Labowitz and Baumann-Pauly 2015; Anner and Bair 2016) make measurement a subject in its own right. With no official count of export factories that I could open, the Mapped in Bangladesh census (BRAC University CED 2024; BRAC University CED 2025) fills the gap, though its totals differ from employers’ figures and from household surveys, and even its own totals change in ways its publications do not explain. Anyone writing about growth rates across 2022 and 2023 should read chapter 4 first.

Backward linkage and upgrading

The case that local fabric explains knitwear’s rise is made mainly in World Bank and IMF studies and in industry statements (Mlachila and Yang 2004; Staritz and Frederick 2012), and tested at firm level only rarely. The largest survey of knitwear firms I found covered 232 BKMEA members in 2001 and found no significant difference in profit rates for firms that knitted their own fabric (Bakht et al. 2008, 2, 12), and Sytsma’s study of the EU’s 2011 change uses customs records of every Bangladeshi export shipment from 2008 to 2013 (Sytsma 2021; Sytsma 2022). The gap between how often the backward-linkage story is told and how rarely it is tested is, in my reading, the largest in this field.

A research agenda: what nobody has measured yet

Each chapter sets out the questions the published evidence cannot answer. These are the ones I would put first, because each could be answered with data that exists but is not yet public or not yet linked.

  • The share of export orders that pass through buying houses and agents, and what intermediaries add or take on price, lead time and compliance (chapter 1).
  • A price series for the same defined garment over time, from order-level customs data, to replace unit values that mix products (chapter 2).
  • Evidence that links buyers’ order records to wages and hours in the same factories, rather than surveys in which one manager reports both sides (chapter 2).
  • A monthly series of Export Development Fund lending to garment firms, and a study of what the cash incentives did, using the rate changes of 2024 as a natural experiment (chapter 3).
  • An estimate of how much bonded fabric leaks into the home market, from the utilisation declarations that were built to track it (chapter 3).
  • A revised export series for FY2018-19 to FY2021-22, and a published list of the customs headings counted as garments (chapter 4).
  • An explanation for why Canada and Japan record more imports from Bangladesh than Bangladesh records exports to them (chapter 4).
  • A dated survey of local fabric sourcing by firm, repeated over time, to test the backward-linkage story directly (chapter 5).
  • Firm-level evidence, after 2011, on whether vertically integrated knit firms grew, upgraded or survived shocks better (chapter 5).

When sources give different figures

Garment statistics disagree more often than they agree, so the chapters follow the same rules.

  • When two sources give different figures for the same thing, both are shown, with whose figure is whose. I do not pick one unless one source has corrected the other, and then I say so.
  • Each figure is cited to the source that first measured or reported it, where I could open that source. Where I could read a figure only as another work quotes it, or a document only as a newspaper reports it, the sentence says so.
  • Where a chapter does its own arithmetic (a share, an inflation adjustment, a sum of importers’ records), the text says the calculation is mine and gives the inputs, so a reader can repeat it.
  • Readings of my own, where the evidence is suggestive but not proof, are marked “in my reading”.
  • Facts I saw only in a search summary, and could not open at the source, are left out. Each chapter lists what it could not verify.

Data you can reuse

Each chapter’s tables come with a CSV link underneath, and all of them are gathered on the data page. They carry the CC BY 4.0 licence, which allows reuse, adaptation and republication for any purpose, commercial included, as long as the page they came from is credited. Where a table repeats figures from another publisher, cite that publisher as well; the source of each figure is in the table or its caption.

Who wrote this guide, and why that matters

I am AB Rahman. I co-founded CMB Sourcing Ltd., which makes knitwear and caps in its own factory, the factory of our sister company, Collective Studio Ltd., where I am Head of Growth, and works as a buying house for other garments, sourced from partner factories. I live in Western Australia and hold a master’s degree in marketing and consumer behaviour from Curtin University. I am not an economist or a lawyer by training, which is one reason every calculation in the guide shows its inputs and every rule is cited to the document that sets it.

A firm like mine has interests on both sides of most questions here: it sells to buyers and, for some products, buys from factories. I have tried to give the evidence against my own side’s view as fully as the evidence for it. Nothing in the guide draws on CMB’s own orders, prices or records.

The method is a desk review of published research, official documents and public data, carried out in November 2026 and current to 22 November 2026. I did not interview anyone, visit archives or run surveys for it. For journal articles I read the full text where I could open it and the abstract or the bibliographic record where I could not; each chapter’s method note says which. I did the review with the help of AI research tools, which searched for sources, read them, extracted the figures and drafted the text to my brief. Before publication a separate AI check went through each chapter independently, reopened its sources and tried to prove each figure, quotation and citation wrong, and the errors it found were corrected. I set the questions and the rules, and I am responsible for every sentence published here. Sources that could not be opened are named in each chapter’s method note.

If you find a mistake, please write to the address in the “Cite this page” box at the end of the page. I will correct it and note the change on the page.

Citing the guide

Each page has a “Cite this page” box at the end, with entries in Chicago, APA and Harvard styles. Cite the chapter you used rather than this page where you can. For a single figure, cite the original source given in the chapter as well, since that is where the figure was measured.

Bibliography

Every work cited in the five chapters, in Chicago author-date style. Web pages were read on 22 November 2026 unless the entry says otherwise.

Sources

Checked . Rules and figures change, so confirm anything that affects your pricing.

  1. BGMEA, “Export Performance”, fiscal-year tables, read 22 November 2026 (opens in a new tab)

All guides