The cost of the middle

A Bangladeshi worker can owe more than a year of his earnings abroad before he has worked a single day of them. Between a European buyer and a Bangladeshi factory sit layers whose price is not published either. On what the middle costs — and why the figure is so hard to see.

Just over 1.1 million people left Bangladesh for work abroad in 2025. Before any of those jobs began, money moved in the opposite direction.

How much money is not a matter of opinion, and it is worth saying why. Recruitment cost is a United Nations development indicator — SDG 10.7.1 — and of all the countries that have run the survey, only four have published their results as official data. Bangladesh is one of them. What follows is not an outsider's estimate of Bangladesh. It is Bangladesh's own measurement.

There is a published price, and it is not the price

The indicator asks one question: how many months must a migrant work at his destination to earn back what he paid to get there? In the Labour Force Survey of 2022, published in October 2023, the answer was 14.6 months — 15.1 for men, 7.8 for women — on an average recruitment cost of 381,967 taka against average monthly earnings abroad of 26,172.

The same table cuts one more way. Migrants who had passed no class paid the smallest sum of anyone — 311,237 taka — and needed the longest to earn it back: 18.7 months. A graduate paid 492,799 and was clear in 13.5. The least educated pay the fewest taka and wait the most months for them.

None of this is unregulated on paper. Bangladesh's ministry fixed maximum migration costs by destination in 2017 — 165,000 taka for Saudi Arabia — and that figure has not been revised since. Against it, the same survey measured what people actually paid to go there: 388,210 taka. Two and a third times the legal maximum.

The sharpest version is the Bangladesh–Malaysia corridor, where a memorandum between the two governments had bound recruitment fees to US$420 and researchers measuring the corridor in 2021 found an average of US$3,900 — "roughly 9 times higher than the agreement amount", in the study's own words. Bangladeshi reporting in December 2025 still put the live market at three to four times the fixed rate.

So the number is published. It is simply not the number that is charged, and the person paying has no way of making the published one bind.

And it is borrowed

Almost nobody has the money. Bangladesh's 2023 socio-economic survey found 58 per cent of migrants borrowing to cover it, and the World Bank finds about a third of all migrants using neither a bank nor an NGO but unregulated moneylenders. In the Malaysia corridor, family loans, land sales and savings financed the trip in roughly equal measure — and among people going a second time, land sales were the largest single source. Measured against the household rather than the individual, the World Bank puts costs above US$4,000 at about thirty months of a migrant household's income before departure.

Two and a half years of everything a family earns, paid before the first wage.

So the debt exists before the work does. And under the sponsorship arrangements governing most of these destinations — which the World Bank describes as requiring a migrant to hold an employer's sponsorship, with the end of the contract meaning return — a worker who arrives to find the job is not the one that was described to him has very little room to argue. The ILO has a name for the common version of this: contract deception and substitution. The paperwork is elsewhere, the money is already spent, and the arithmetic of walking away is unforgiving.

That is not a story about a country. It is a story about when the money changes hands, and who could see the price at the time.

The half that is not a wound

It would be dishonest to stop there, because the same decision also works.

Low-skilled Bangladeshi workers can expect to earn roughly four times more overseas than at home, which is why a million people a year make a choice that looks reckless from outside and is not; the World Bank's return-migrant survey finds 88 per cent eventually earning back more than it cost them. And the money comes home. Remittances reached a record US$35.56 billion in the 2025-26 financial year, up about 17 per cent, against ready-made garment exports of US$38.7 billion in the same year — and those were falling. Bangladesh's migrant workers now send home very nearly what its entire garment industry exports, into households rather than onto anyone's balance sheet.

Migration is not a tragedy. It is one of the most effective things a poor household can do. Which is exactly why the cost of getting in matters: it is the toll on the best road available to them.

The same journey, at two prices

Here is the fact that decides the argument.

A small number of placements are arranged government to government, with the two states handling it directly and no private intermediary in the middle. Under the Korean Employment Permit System a migrant pays about US$900; to the same destination, privately intermediated, more than US$4,000. Under the Bangladesh–Malaysia agreement for palm-oil work, about US$500. The private route to that same work: about US$4,500.

Same country, same job, same worker — and, in the Malaysian case, nine times the price. The evaluation of that programme found it cut migration costs eightfold and raised net earnings from a three-year stay by 87 per cent.

These schemes stay small, because two governments can only administer so many places — which is what makes the last figure the important one. When 30,000 vacancies were opened on those terms, about 1.4 million eligible people applied.

So the cost is not a law of nature. Not distance, not skill, not the price of a ticket, not the state of the labour market. It is a property of how the middle is arranged, and whether its price is set anywhere a person can look at it.

It is not the middle. It is the opacity of the middle.

This has to be put carefully, because the easy version is wrong. The intermediaries are not superfluous. The World Bank's own account is that the informal subagents — the dalals — "serve the important function of linking migrants with recruitment agents", and that people go to them "because of trust, proximity, and access to support services". In a village with no other channel to a foreign labour market, that is a real service, performed by someone the family knows.

The problem is narrower and harder. The destination states legislate: Bahrain, Oman, Qatar, Saudi Arabia and the United Arab Emirates all have laws prohibiting workers being charged recruitment fees. But as the ILO's second global study on recruitment fees observes, those prohibitions "do not often apply to costs incurred in the country of origin". The rule stops at the border; the charge happens on the other side of it, to a person with no way to compare it against anything. On the World Bank's estimate, intermediary costs alone average US$2,000 a migrant — more than half of everything he pays.

That is what makes it collectable. Not the existence of the middle. The fact that nobody standing at either end can see what was taken inside it.

I ran garment factories in Bangladesh for seven years, and the trade I work in has the same shape — though not the same weight, and that should be said before the comparison rather than after. A commercial margin is not a family's land. But between a European brand and a Bangladeshi factory there are usually several parties, and the buyer cannot see what the worker cannot see. He knows the number he pays. The factory knows the number it receives. Neither knows the shape of the space between, and neither is placed to ask. Some of those parties earn every taka they take — a good agent knows which factory can hold a delivery date in October and which sample means something. But the information is not being withheld by anybody in particular. It simply is not anywhere.

Paid in full, and still without status

One more step closes the mechanism.

A visa can be sold by a sponsor who has no work to give. It has a name — the free visa, or visa trading — and the official descriptions are matter-of-fact: workers who "pay an employer a recurring fee to sponsor their work visa while performing work for other employers", and companies that take payment for visas but provide no "legitimate employment, rendering the worker without status or legal protections". The ILO's footnote is the tidiest definition: a visa that is "not free of cost, but free of an employer."

So the full price can be paid for something void. Nor is that the only route there. The ILO's regional white paper lists the ordinary ones: a permit the employer does not renew, a passport not returned, a transfer whose paperwork was never filed, a false report of absconding. Most workers who become irregular, it records, "do so beyond their control and through no fault of their own" — and are nonetheless "subject to arrest, detention and deportation."

Then the part that decides everything: such a person cannot complain, because the complaint is the exposure. Official reporting describes workers who left a sponsor and "chose not to report abuse to authorities out of fear of deportation". Of Bangladeshi return migrants surveyed in 2018-19, 19 per cent had been threatened with deportation or denouncement to the authorities, and for 12 per cent the reason for coming home was expulsion over a permit or a visa.

Ordinary fraud is checked by the buyer's ability to object. Here the defect in the goods silences the person who paid for them, and that is what makes them safe to sell. How much of the trade works this way, nobody has measured: the practice is documented, its scale is not.

This is not an abstract subject for me, and the reason I cannot tell you how I know it is the whole of what this piece is about.

What is actually being done

A piece that names a problem and no repair is a complaint. So:

The rule already exists. Behind the ILO's fair-recruitment guidelines sits Convention No. 181, whose Article 7(1) runs to a single sentence: "Private employment agencies shall not charge directly or indirectly, in whole or in part, any fees or costs to workers." The business version, from the Dhaka Principles for Migration with Dignity, is shorter still — no worker should pay for a job.

Things have moved. Since 2019 Qatar has opened fourteen visa centres in six origin countries, Bangladesh among them, where a worker has the terms explained to him and signs his actual contract before he leaves — precisely what contract substitution depends on nobody doing. The Saudi–Bangladesh recruitment agreement of October 2025 likewise requires a documented offer and contract before departure. And money has moved back the other way: one remediation programme reports facilitating repayment of over US$111 million to more than 86,000 workers.

The newest item is two days old. On 28 August 2026 Bangladesh's expatriates' welfare ministry announced that Malaysia has agreed to take 10,000 workers on a "zero-cost" basis, the first group flying through the state employment company "without airfare or other migration-related costs". The ministry's instruction to the public, in the same statement, gives away everything above: make no payments, undergo no medical tests and submit no passport until an official notice is issued.

The limits belong in the same breath. Qatar's visa centres have no mandate to monitor fees. The ILO records costs paid by Kenyan workers rising despite fair-recruitment campaigns and destination-country prohibitions. And the group of companies that set out in 2016 to eradicate worker-paid recruitment fees everywhere gave itself ten years to do it; that deadline is this year.

Legislation is the cheap half. The ILO's own recommendation names the target in the plainest possible words: information asymmetry. Not greed, not any nation. The gap between what one party knows and what the other can find out.

The best answer to a person leaving

One last thing, and it is a preference rather than a policy.

The World Bank's account of why so many Bangladeshis go is not mysterious: good formal jobs are not being created quickly enough to absorb the hundreds of thousands of young people entering the labour market each year. Migration fills the gap. So the number of people for whom leaving is the only arithmetic that works is not a fixed quantity. It is an outcome, and outcomes can be changed at the other end.

A sewing unit an hour from a village, doing ordinary work at an ordinary standard, does not sound like an answer to anything. It employs forty people who would otherwise be among those hundreds of thousands. Multiply it and you have not solved migration, and you should not want to — the four times still holds, and the remittances are real. What you have done is make the choice a choice again, taken by people who are not staking the family's land on a number nobody would show them.

One last piece of arithmetic.

No toll in this account is large enough, on its own, to be worth a quarrel. Bangladesh's own transparency researchers surveyed 15,715 households in the year to October 2025 and put the average unofficial payment a household made for public services at 5,124 taka across the whole year — a nuisance, not a catastrophe. Scaled up, they put the national total at 126.3 billion taka, counting only the sectors they looked at. Of the households that experienced it, 98.1 per cent were forced to pay — and of those, 81.5 per cent said the service simply could not be obtained without it. Intermediaries were the third most frequently named recipient; in some sectors, the researchers concluded, their influence "has become institutionalised." Where a channel only half works, someone who knows the way becomes the way — and charges for it.

What nobody has measured is what a toll of that shape, repeated at every counter, costs a country in what a surplus would otherwise turn into: a second line, a training programme, a wage. The transfer has been counted. The compounding has not, and I will not invent the number. Its absence is the point. The cost of the middle is the one cost nobody is ever required to publish.

Make the middle visible and the middle has to earn its keep. That is true of a recruitment fee in a Bangladeshi village and of a sourcing margin between Copenhagen and Gazipur, and it is the only thing I have ever seen work in either place.

— A.K. · TB Textile Sourcing · Slovenia, EU
(European who ran knit factories in Bangladesh for seven years. Every figure above is published; the central ones were measured by Bangladesh itself.)