Layer 1

Overview

The findings and guidance in brief. Start here for the clear picture in a few minutes.

Where Bangladesh stands

The state of the economy in late 2026 in plain terms - slow growth, high prices, a banking system that has stopped lending, record remittances and a three-year reprieve on trade preferences.

  • Growth has slowed for four years, and in FY26 it was 3.4% (World Bank) or 4.14% (the statistics bureau); inflation has stayed above 8%.
  • About one-third of outstanding bank lending, by value, is classified as non-performing, so banks lend very little, and private credit grew only 4.5%.
  • Remittances reached a record $35.6bn and have kept the currency steady, but poverty has risen to about 22.5%.
  • The UN postponed Bangladesh's graduation from least-developed-country status to November 2029, which starts the clock on losing trade preferences; the EU gives a further three-year transition, and other markets differ.

Why progress has stalled

Why Bangladesh's growth model is running out of road - in many places taking pays better than making - explained in four layers (institutions, politics, skills, norms), and why culture alone does not explain the outcome.

  • In many parts of the system, taking pays better than making, so capital, talent and effort flow to power more than to productivity.
  • The explanation comes in layers - history, institutions and politics, skills and organisations, norms - and none is enough alone.
  • Religion and culture matter mainly through specific channels such as women's work, and do not explain outcomes by themselves.
  • Because the layers feed each other, fixing one alone tends to fail; the aim is to change what each group is paid for.

Why fixing one thing is not enough - the loops

Bangladesh's problems feed each other in circles, so a reform that touches only one of them is pulled back by the rest. Where the circles can be cut, and why that is good news.

  • Weak banks, few jobs, idle young people, party money and thin tax revenue feed each other in loops.
  • A reform that touches only one link tends to be pulled back by the others.
  • Loops can be cut: choose a link that is cheap to change and held by someone who gains from changing it, and use an outside deadline when nobody inside will move first.
  • The best first cuts are the banks, the 2029 graduation deadline, party money and extortion, and safety for election losers.

Everyone has a stake - groups and incentives

Every group that matters for the economy acts sensibly on what it is paid for today. Nineteen groups, what each wants, and what could change the payoffs so that each is better off and the country prospers.

  • Every group is acting sensibly on its payoffs: leaders, cadres, officials, borrowers, clerics, workers and exporters alike.
  • The task is to change what each group is paid for, at a price the country can afford, not to appeal to goodwill.
  • Some of the groups pulling hardest the right way (women workers, migrants, small firms) have the least power; the most powerful actor can tip either way.
  • Status, belonging and belief move people as well as money, so some of the cheapest levers are not financial.

Ground rules for working together

Seven plain ground rules that people of every party, faith and outlook could accept for the same reasons they hold dear, and one test to keep them fair. A proposal for discussion, not an existing national agreement.

  • Unity does not need agreement on everything. It needs agreement on how power is used, how disputes are settled and how results are judged.
  • Seven rules: shared prosperity with fairly shared costs; one law for all, including the state; freedom of conscience and peaceful disagreement; fair justice for past wrongs; safe loss of power; public office as public trust; open numbers and sunset clauses.
  • The rival test: would the sponsors accept the same rule if their strongest rival held power?
  • This is a proposed compact for discussion. No such national agreement exists today.

The path forward - deals, order and a realistic payoff

A plan built as exchanges that make production pay better than taking. Design rules, the nine deals, the order to do them in, and what growth can realistically be expected.

  • The plan is nine deals: each gives a group something it values in exchange for a change that rewards making things over taking.
  • Order matters more than the list: tell the straight story, resolve the banks, make losing safe in January 2027, use the November 2029 graduation deadline, and pilot the untested deals.
  • Expect little payoff before 2029; our judgement of the realistic growth ranges is on the roadmap page.
  • Every deal has a price and a risk, stated openly, and should leave a rule behind, not a private favour.

What to watch - the scorecard for the next 12 months and beyond

The dates and indicators that will show whether Bangladesh is moving toward recovery, from the January 2027 amendment and the IMF talks to bank losses, extortion surveys and the 2029 graduation deadline.

  • The next twelve months are a test of credibility: an IMF programme, published bank losses, the January 2027 amendment and a settlement window that closes on its date.
  • Process indicators (was it published, was it passed, did it close) matter as much as numbers, because they show whether the rules are real.
  • Expect little growth improvement before 2029; judge progress by credit, trust and jobs first.
  • A public quarterly scorecard from bodies the government does not control is part of the plan.

Risks and open questions - what could go wrong, what we do not know

The strongest objections to the plan, what survives each, the weak point of political feasibility, and the gaps in the evidence that later iterations should close.

  • The weakest point is who carries the plan out: it asks the most powerful actor in the system to help change the system, and confidence in this is low.
  • What holds up best is the order of work (candour and banks first), the use of outside deadlines and the habit of piloting untested deals with published numbers.
  • Many things are unmeasured: the size of extortion, party finance, madrasa graduates' earnings, even the true size of GDP.
  • Readers who disagree with parts of this analysis are not opponents. Where evidence is disputed the site says so.