Layer 2 · In depth

A compact across differences

The reasoning behind the seven ground rules - four starting principles and how each was refined, comparisons with other countries, overlapping consensus, the mechanisms that could make a compact real, and where it could fail.

Updated 21 Nov 2026 · 28 min read

On this page
  1. The four starting principles
  2. Five failure modes, and how each principle was refined
  3. Why the self-enforcement gap matters
  4. Where Bangladesh starts
  5. Where the rules collide
  6. Party, supporters and individuals
  7. Overlapping consensus in plain words
  8. What other countries did
  9. Lowering the stakes of disagreement
  10. Institutions for competing interests
  11. How the four became seven
  12. Mechanisms that could make a compact real
  13. Three first packages
  14. Sequencing, relative to adoption
  15. What to publish
  16. Where the compact could fail
  17. Where the rules go next

This page holds the reasoning behind the seven ground rules. It is a proposal for discussion and not an existing agreement. Its aim is to help people who differ, in party, faith, class or generation, to compete over policy while sharing rules they can all trust. The page starts from four principles proposed for discussion, tests each for weak points, compares other countries, and then sets out mechanisms and failure risks. How the rules change each reform is on strategy and deals.

How to read the evidence

Historical facts about other countries and the philosophical and religious references are unverified beyond the sources named in the footnotes. Where a source is named without a link, we say so. We have not independently re-checked every detail. Statements about what Bangladesh should do are our assessment and are labelled as such.

The four starting principles

Four principles were proposed for discussion as a possible common ground:

  1. We all want Bangladesh to make progress economically.
  2. No lawbreaking is tolerated. People may protest and argue against laws they dislike, but laws in force must be respected, prosecutions proceed independently, and law enforcers and the government are bound by the law too.
  3. Past grievances may remain personal or group grudges, but cannot be used to break laws or to hinder national development.
  4. Ideals and ideologies are sacred to individuals and groups and must be respected, but they cannot supersede the country’s laws or hinder economic development.

They identify real obstacles: doubt about the shared goal, selective enforcement, retaliation, and attempts to impose one group’s beliefs through public power. Testing them against hard cases shows that each needs sharpening, and that all four together leave a fifth problem.

Five failure modes, and how each principle was refined

Starting principle What it gets right Where it can break How it was refined
1. Shared economic progress A common aim can make political competition more productive. The distribution trap. Agreement on “growth” leaves open who pays, who gains and when. Growth can coexist with weak wages, insecure work or displacement, and a claim of unanimity can make criticism look disloyal. Define progress as widely shared gains in jobs, real incomes and services, measured honestly and by region and group. Publish who pays. (Rule 1, Rule 7)
2. Strict, symmetric law Even-handed enforcement protects people and investment. Binding the state is the most important part. The weaponised-law paradox. A published law can violate the Constitution. Automatic prosecution of every alleged breach treats allegation as proof and invites selective crackdowns, and “obey everything” can shield those in power while courts and police are not yet independent. Follow valid laws and court orders; keep the right to challenge them; require individual evidence, fair trial, reasons and remedies for official abuse. Allegation, arrest and guilt are different. (Rule 2)
3. Past grievances It rejects retaliation, intimidation and collective blame. The suppressed-wound trap. Calling an unresolved killing, disappearance or dispossession a “grudge” can deny a victim a remedy. A lawful investigation or compensation claim may delay a project and still be right. Protect memory, lawful advocacy and claims for remedy; pursue individual responsibility through fair, prompt process; ban reprisals and collective punishment. (Rule 4)
4. Respect ideals, subject to law and the economic goal People can keep deep commitments while working together; office is no licence to coerce. The sacred-values backlash. Placing growth “above” faith sounds like secularism by stealth to a believer. “Hindering development” is vague and can silence unpopular views. Deference to a leader can also silence dissenters inside a community. Protect conscience, worship, association, criticism and peaceful advocacy. Limit conduct only through lawful, necessary, proportionate measures. Protect individuals inside communities. (Rule 3)
(Missing) Self-enforcement The self-enforcement vacuum. Why obey first if a rival may cheat? Principles on paper are a prisoner’s dilemma. Add safe political loss, independent referees, visible payoffs and public scorecards. (Rules 5 and 6, Rule 7)

The fifth row is not a proposed principle. It is the structural gap that the first four leave open.

Three qualifications that matter most

“Grudges.” Treating past harm as a private feeling is wrong where there is no fair channel for a remedy. The refined rule keeps a public remedy open and separates it from reprisal. Unconditional obedience to law. Citizens owe respect to laws that are public, constitutional, equally enforced and open to challenge before independent courts. The Constitution itself provides that laws inconsistent with fundamental rights are void to that extent (Article 26), and sets out the remedies in Articles 44 and 102.1 Growth ahead of rights or faith. Growth is the aim of economic policy. It does not outrank rights or conscience; those govern how policy is pursued.

Why the self-enforcement gap matters

Three things make rules self-enforcing. First, a credible referee: independent courts, an independent election authority and police oversight, so that no side must trust the other’s goodwill. Second, a safe floor for losers: if losing an election means prison or ruin, no party will keep to the rules. Third, a public focal point: clear, widely known rules let millions of people, journalists, firms and foreign partners recognise at once when a line is crossed and push back together.2 The 2026 provision that let former bank owners buy back banks was repealed within months after public, opposition and industry objections; our reading is that this shows what a shared, public standard can do, but we cannot show the rules caused it.3

Why the focal point matters

Scattered frustration does not stop capture; a shared standard can turn it into a coordinated response. This is an assessment of how coordination works, not a measured effect of any rule.

From slogans to rules that hold

Paper principles

  • High-minded principles only

The trap

  • Distrust: if I keep the rules and my rival cheats, I lose
  • Pre-emptive defection: capture referees, hoard rents, mobilise the street

What changes the game

  • Independent referees
  • A safe way to lose
  • Visible payoffs and scorecards

The result

  • Cheating is costly and losing is survivable
  • Groups compete inside the rulesa hope to be tested, not a guarantee
Our reading of why rules need enforcement and incentives to hold.

Where Bangladesh starts

The World Justice Project (WJP) Rule of Law Index scores countries from 0 to 1 on eight dimensions, including limits on government power, absence of corruption, fundamental rights and civil and criminal justice. Bangladesh’s 2026 score is reported at about 0.39, rank 123 of 143, up from 127th in 2024 and 125th in 2025. The same sources report civil justice at 132nd and fundamental rights at 133rd.4 The comparison with others is on key statistics.

The WJP gives descriptive context only. Prosperous Asian peers such as Malaysia and Indonesia scored well below the Scandinavian top, yet grew and attracted investment, so a predictable floor may matter more than perfection. That is our reading of a pattern, not a tested causal claim.

Bangladesh’s own record also shows why the first principle needs a distributional test. The World Bank’s 2025 Poverty and Equity Assessment finds that poverty reduction slowed after 2016 despite continued growth, and names more and better jobs, connectivity and protection against shocks as priorities.5

Where the rules collide

Real life produces cases where two good rules pull apart. The compact does not remove these; it gives a way to handle them.

  • Law and constitutional challenge. The Constitution states its own supremacy (Article 7), provides that laws inconsistent with fundamental rights are void to that extent (Article 26) and offers enforcement routes (Articles 44 and 102). Articles 27 to 28 address equality, Articles 31 to 35 legal protection and trial, and Articles 37 to 39 assembly, association, thought, conscience and speech. These have distinct texts and limits. Binding court orders govern the parties while a case is pending. Enforcement should follow published priorities, keep review available and avoid targeting by viewpoint or affiliation.1
  • Belief and public obligations. Article 2A names Islam as the state religion and gives equal status and equal right in the practice of other religions. Article 12 addresses secularism as a principle of state policy, and Article 41 protects religious freedom subject to its stated conditions. The principles in Part II have a different legal status from the enforceable rights in Part III (Article 8(2)). The compact can draw support across these provisions by protecting everyone’s conscience and equal treatment by public bodies. A workplace can accommodate prayer or religious dress where practicable while applying the same safety and job requirements to everyone. People within a family or faith community keep their own rights.1
  • Protest and disruption. A march that delays traffic, a strike that suspends production and an attack on workers call for different responses. The UN Human Rights Committee’s reading of the ICCPR recognises that peaceful assemblies may disrupt ordinary activity, and nonviolent civil disobedience can fall within protected assembly depending on the circumstances. Officials should facilitate peaceful assembly, negotiate emergency access where needed, give reasons for restrictions and act against specific unlawful conduct on evidence. Where conduct is lawfully restricted, due process and proportionality still apply.6
  • Accountability and political peace. The UN Human Rights Office’s fact-finding on the July-August 2024 protests is summarised by the UN as finding systematic repression, with attacks on protesters that formed part of an official policy, and as raising concerns that crimes against humanity may have been committed and require investigation. These are the investigators’ findings and concerns; they are not findings of guilt against any named person, and nothing in them attributes responsibility to any party’s supporters as a group. The report recommends accountability and remedies. A credible settlement would preserve evidence, protect victims and witnesses, and fund fair casework and reparations under published criteria. Fact-finding can inform investigations, reforms and victim assistance. Courts, not reports, decide individual criminal responsibility.7

Party, supporters and individuals

When a barred or contested party is discussed, three questions must stay apart.

Question Who decides What the compact says
The party’s legal status (for example a bar from contesting an election) The law, as applied by the competent authorities and courts A matter of verified legal fact. It does not by itself answer the other two questions.
Ordinary supporters’ rights The Constitution and equal laws Civic rights and protection from arbitrary action extend to people regardless of party affiliation. Ordinary political participation is protected, subject to law.
Individual legal proceedings Independent prosecutors and courts, on individual evidence An allegation is not guilt. Proceedings go forward on individual evidence and fair trial, and the same fairness protects anyone leaving office.

Inclusion and accountability are compatible duties. Neither is a payment for the other. The compact does not trade justice for 2024 against political stability. Affiliation never implies individual guilt.

Overlapping consensus in plain words

The philosopher John Rawls asked how people with deep, incompatible beliefs can live under one set of public rules.8 His answer separates two things. People hold comprehensive doctrines (a religion, secular humanism, socialism, nationalism, and so on) that answer ultimate questions. The state cannot impose one of them, and it should not ask people to rank their own doctrine second. A truce (modus vivendi) holds only while power is balanced; whoever gains a big majority may break it. An overlapping consensus is different: each group endorses the same public rules, but for reasons from inside its own tradition.

These are illustrations only (unverified; no primary source located), and each tradition’s own scholars and members are the right people to say what their tradition teaches.

Tradition Why it might endorse fair rules, honest dealing and non-coercion
Faith traditions Justice (adl), trust (amanah), fulfilling contracts, and the prohibition of oppression (zulm), bribery (rishwah) and embezzlement of public funds. For non-coercion, the principle often rendered “no compulsion in religion” and the protection of private homes from intrusion.
Secular and constitutional views Fundamental rights, civic equality and due process.
Nationalist or traditionalist views National sovereignty and dignity, and the sacrifices of 1971 and 2024, which they may see as demanding a fair and stable state.
Business, labour and youth Predictable rules protect livelihoods, investment, merit-based opportunity and safety.

The practical meaning is that nobody is asked to choose economics over faith. The shared ground is that fair rules and widely shared prosperity can be reached from each starting point.9 How this bears on religion and development in Bangladesh, including the right of individuals to decline and dissent, is on culture and religion.

What we are unsure of

An overlapping consensus is easiest to describe and hardest to build. It needs trusted voices inside each tradition to explain the rules in their own language, and it cannot be declared from a conference room. We think it is worth attempting because the alternative, a truce held by power, breaks when power changes. We cannot show it will hold.

What other countries did

The cases illustrate different functions. None can be copied whole, and these comparisons cannot isolate the effect of any one institution on growth. Income, state capacity, history and access to regional and global markets also shape results. The WJP index separates order and security from constraints on government, fundamental rights and justice, so a high overall score can hide trade-offs.10

Lowering the stakes of disagreement

Case What happened What could transfer Where the comparison ends
Japan, 1960 After mass protests over the US-Japan Security Treaty in 1959-60, Prime Minister Kishi resigned. His successor Hayato Ikeda adopted a “low posture” of tolerance and dialogue and, in December 1960, an Income Doubling Plan. National income is reported to have more than doubled in under seven years (unverified here).11 Our interpretation (see the judgement note below): lower the temperature first and tie unity to visible household gains such as wages, prices and dignity, not just headline GDP. Japan later had decades of single-party dominance. Bangladesh has no such wage-driven boom guaranteed.
Spain, 1977 The Moncloa Pacts (25 October 1977) tied union wage restraint and fiscal tightening to wider political freedoms and the road to the 1978 Constitution. These pacts were separate from the Amnesty Law of 15 October 1977, which covered offences by officials and law-enforcement personnel (Article 2(e)-(f)). That law was an accountability cost for victims of Franco-era abuses, not a mere postponement: it limited prosecution of such offences.12 Rival parties can sign a binding economic pact in a crisis if restraint is paired with real, checkable political rights. Grievances are reported to have resurfaced later, including in Spain’s 2007 Historical Memory Law (unverified here). Our interpretation: grievances need a bounded, transparent channel, not silence or a blanket amnesty.
Indonesia, after 1998 Pancasila (1945) names belief in one God while rejecting both theocracy and anti-religious secularism. After 1998 came term limits, competitive elections, a constitutional court and decentralisation, so opposition groups could govern provinces. Large moderate Muslim organisations served as civic anchors for pluralism.13 Include faith respectfully inside the national frame; decentralise so that losing the capital is not a total loss; build on moderate civic institutions. Indonesia still faces religious majoritarianism, blasphemy prosecutions and local corruption. Its 2004 law to end military businesses was not fully carried out, a caution for any military reform.14
Malaysia, from 1969 After riots in May 1969 exposed ethnic economic inequality, Malaysia proclaimed the Rukun Negara (1970), whose principles include belief in God, supremacy of the Constitution and rule of law, and launched the New Economic Policy. It also built globally open industrial hubs.15 Recognise cultural dignity while keeping the economy open and predictable. The New Economic Policy’s long-running ethnic preferences are widely criticised as having encouraged rent-seeking (our interpretation, see the judgement note below). This is the origin of the lesson to sunset every concession (Rule 7).
South Africa, 1994-96 The 1996 Constitution, an independent Constitutional Court and a Bill of Rights were paired with a Truth and Reconciliation Commission. Amnesty required full disclosure of the relevant facts and a political objective. Applications involving no gross human-rights violation could be decided without a public hearing, so public hearings were not universal.16 Transitional justice can be bounded, truthful and non-partisan. The settlement left deep economic inequality largely untouched, which later fed disillusionment. Justice and visible opportunity need to travel together.
Singapore (religious harmony) After the 1964 riots and independence in 1965, the Maintenance of Religious Harmony Act (1990) bars religious platforms from being used to stir political disaffection or enmity, with restraining orders reviewed by a multi-faith Presidential Council.17 Borrow only consultation and review arrangements: a multi-stakeholder body that reviews and advises. Do not borrow restrictions on peaceful political criticism. Singapore’s official framework extends to political causes and disaffection against government, which would conflict with Rule 3. Any such council’s appointments and powers should pass the rival test: would its sponsors accept them under their strongest rival? Singapore’s high order-and-security scores coexist with strict limits on protest, assembly and speech; we do not claim the limits caused the scores. A city-state’s scale also differs from a mobilised society of about 170 million people.

How we read these cases

These are our interpretations, not findings. They are consistent with the cases but do not prove cause. Japan: a period of rising household incomes may have eased street confrontation, but other factors also changed. Malaysia: preferences that run without an end date are open to rent-seeking, which is why the compact asks for sunsets. Singapore: strong administration and tight limits on protest appear together; the evidence cannot separate them. Descriptions of Japan, Spain (outside the amnesty law), Indonesia and Malaysia rest on sources named without links and are unverified here.

Institutions for competing interests

Case How it handles competing interests What Bangladesh could adapt, and where it ends
Germany The Basic Law (1949) binds public power to constitutional rules, protects equality, religion, expression and assembly, and offers judicial recourse against public authority. Its Article 79(3) “eternity clause” puts human dignity and the democratic order beyond amendment. “Constitutional patriotism” directs civic loyalty to shared rules, not to a ruler, party or creed. A majority that removes the chancellor must elect a successor (a “constructive vote of no confidence”).18 Publish reviewable rules for appointments and transitions; keep executive continuity compatible with opposition rights. A constructive vote is an option to evaluate beside Bangladesh’s own confidence and floor-crossing rules. Germany’s federal and postwar history limits direct transfer.
Finland The Constitution protects equality, conscience, expression and assembly. A prime-minister-led Economic Council brings organised interests and government together as an advisory forum, and the Parliamentary Ombudsman supervises the legality of administration.19 Pair economic consultation with independent scrutiny, and give less organised groups access through open submissions and rotating seats. Finland’s established organisations and administrative capacity help; Bangladesh would need to test whether its own forum represents affected people.
Ireland A National Economic Dialogue consults unions, employers and community interests before budget decisions. Earlier national social partnership broke down in the financial crisis when public-sector talks collapsed.20 Hold recurring budget dialogue, publish costings and minority positions, and agree beforehand how elected institutions decide when talks stall. Fiscal shocks can break bargains that depend on an affordable package for all.
New Zealand The Public Finance Act requires a pre-election economic and fiscal update prepared by the Treasury and disclosure of material decisions and risks, subject to statutory exceptions.21 Give voters and incoming governments one consistent view of commitments such as bank support, energy arrears and guarantees. Disclosure needs reliable underlying accounts and independent scrutiny.
South Korea After the martial-law declaration, the National Assembly brought impeachment. In April 2025 the Constitutional Court reviewed its legality, upheld it and removed the president.22 Make review of extraordinary executive action credible in advance, with courts able to act and the legislature able to meet. It shows a safeguard working during a severe breakdown of trust, and offers no guarantee where judges or legislators lack practical independence.
Singapore (administration) A specialised anti-corruption bureau investigates under the Prevention of Corruption Act. Speech, assembly, association and religion are protected under provisions with different limits, and public assemblies generally need a permit unless exempt.23 Study investigative capability, clear administration and compliance. Judge assembly rules independently against Bangladesh’s constitutional needs; strong administration can coexist with tight protest limits.
Mauritius A small island economy diversified substantially. A World Bank country report from 2026 still finds a need for better institutions, infrastructure and incentives to support productivity and jobs.24 Diversification is a continuing process. Review sector incentives for competition, productivity and jobs, and end programmes that fail the test. Size and history limit inference about Bangladesh’s political settlement.

What seems most transferable

The most transferable combination in these cases is constitutional protection, representative dialogue, independent scrutiny and transparent fiscal commitments. Each does a separate job. Dialogue can show the costs of reform. A legislature can authorise and fund it. Review bodies can enforce rights and probity. Elections can replace decision-makers and redirect policy. Asking one institution to do all four invites capture or paralysis. This is our synthesis, not an empirical finding.

How the four became seven

The failure-mode table above is the canonical account of how each starting principle was refined, and the seven rules are stated on the overview page. The only link not shown there is that Rules 5, 6 and 7 come mainly from the self-enforcement gap rather than from a single starting principle. The diagram shows how the rules support each other.

How the seven rules reinforce each other

The shared goal

The shared goalRule 1: Shared prosperity, fairly shared costs
Fed by Rule 2: One law for everyone, including the state, Rule 5: Losing power is safe; government can act, Rule 6: Public office is a public trust, Rule 7: Open numbers, sunsets, correct mistakes

Guardrails that make rules self-enforcing

Guardrails that make rules self-enforcingRule 2: One law for everyone, including the state
Feeds into
  • Rule 1: Shared prosperity, fairly shared costs (predictable rules protect livelihoods)
Fed by Rule 5: Losing power is safe; government can act, Rule 6: Public office is a public trust, Rule 7: Open numbers, sunsets, correct mistakes
Guardrails that make rules self-enforcingRule 5: Losing power is safe; government can act
Feeds into
  • Rule 2: One law for everyone, including the state (referees need no fear of the loser)
  • Rule 1: Shared prosperity, fairly shared costs (longer horizons for investors)
Fed by Rule 3: Conscience and peaceful disagreement, Rule 4: Fair justice for past wrongs
Guardrails that make rules self-enforcingRule 6: Public office is a public trust
Feeds into
  • Rule 2: One law for everyone, including the state (equal enforcement needs clean offices)
  • Rule 1: Shared prosperity, fairly shared costs (legal path beats illegal path)
Fed by Rule 7: Open numbers, sunsets, correct mistakes

Social peace and pluralism

Social peace and pluralismRule 3: Conscience and peaceful disagreement
Feeds into
  • Rule 5: Losing power is safe; government can act (less culture-war stakes in elections)
Social peace and pluralismRule 4: Fair justice for past wrongs
Feeds into
  • Rule 5: Losing power is safe; government can act (fair process makes leaving office safer)

Public record against capture

Public record against captureRule 7: Open numbers, sunsets, correct mistakes
Feeds into
  • Rule 2: One law for everyone, including the state (public record exposes capture)
  • Rule 6: Public office is a public trust (costs, owners and review dates)
  • Rule 1: Shared prosperity, fairly shared costs (honest numbers keep support)
Our reading of the links between the rules. They are judgements about how the rules support each other, not measured effects.

Mechanisms that could make a compact real

A compact only changes behaviour when it is built into political and civic routines. The mechanisms below are proposals to test, each with its legal mandate, budget and appeals confirmed before launch.

Mechanism What it is Notes
Cross-party minimum compact Instead of agreeing on ideology or history, parties, business and labour bodies and respected civic and religious elders endorse a short compact on a few binding commitments. The research suggested three: macroeconomic and export stability (central-bank independence, bank resolution without insider return, a published tariff and export schedule); political safety and peaceful competition; and no street extortion and no mob coercion. This is the “Moncloa moment” idea. Signing can signal intent; carrying it out needs existing legal powers, appropriations, legislation or constitutional procedure. The research tied it to a constitutional amendment then on Parliament’s calendar for January 2027. That date is a planning assumption pending confirmation, so we use adoption-relative milestones below. If the amendment slips, the minimum compact can proceed through ordinary legislation, parliamentary resolution or an agreed code of conduct.
Multi-tradition endorsement Do not market the compact only through English-language Dhaka think tanks. Co-author plain-Bangla explanations in each community’s own terms: religious scholars, business and SME bodies, student unions, women’s groups and diaspora networks. The aim is that people hear the rules from voices they trust, in their own language. Each tradition states its own reasons; we do not do it for them.
Compact scorecard A quarterly scorecard, jointly published by independent economists, think tanks, business chambers and citizen journalists, scoring government and opposition on verifiable indicators. Indicators proposed include inflation, bad-loan recovery, extortion, port dwell time, court clearance, political and mob violence, women’s workplace safety and budget transparency. Higher complaint counts can mean greater confidence in reporting, so pair them with surveys and outcomes.
Pilot “islands” Visible pilots of fair rules in limited places before nationwide roll-out. See the three first packages below.
Consequential consultation A published-mandate forum with a small analytical secretariat, broad and transparently chosen membership, rotating seats, accessible Bangla materials and written submissions for people outside organised groups. The forum advises; Parliament, authorised agencies and courts keep their legal roles. It must publish draft options, distributional effects, conflicts of interest, disagreement and the authority’s response. For each decision, say who gains, who pays, what relief is justified and when it ends.
Deadlock procedures A pre-announced timetable for submissions, a reasoned decision by the competent body and normal channels of review. A genuine emergency may shorten consultation where the law allows, with written reasons and later scrutiny. Deadlines for advice stop extra veto points from delaying legitimate decisions.
Handover records At a change of government, hand over a public inventory of outstanding commitments, liabilities, contracts, pilot findings and pending reviews. Sensitive case and personal data stay lawfully protected. Review the compact after each general election in public, including commitments the incoming government intends to revise.
Privacy safeguards Protected submissions; confidentiality of complainants and witnesses; data collected for measurement kept apart from enforcement. Measure outcomes across regions and groups using privacy safeguards. Protect complaint data before using performance pay.
Realistic pilot evaluation Each pilot gets a review date, a cost ceiling, a primary outcome and safeguards against serious harm; baselines are set before enrolment. Scale only when benefits persist, costs are acceptable and rights and service standards hold at scale. Revise or close a pilot if harm or capture persists.

Three first packages

The research proposed three narrow first packages that can be run through existing institutions where powers and resources allow:

  1. Bank and budget credibility. Publish bank-resolution criteria and a fiscal-risk statement, commission independent verification of asset quality, identify who bears losses under the applicable framework and require Parliament to authorise material new liabilities as the law provides. See the banking crisis.
  2. One trade and logistics corridor. Pick a corridor by open criteria, publish the fee schedule and port baseline, pair protected complaints and traceable payments with a route to challenge decisions, and add worker safety and transparent contracting.
  3. Fair access to study and work. Pair transparent dorm allocation or an existing training provider with paid placements, independent appeals and measures for women’s access, including religious and general education providers where demand justifies it.

Accountability and political safeguards run in parallel. Publish the legal status of institutional reform proposals, define how oversight bodies get information, and fund victim and witness support alongside investigative capacity. Delaying them until output recovers would leave reforms open to retaliation or capture exactly when people are asked to accept losses.

Sequencing, relative to adoption

Because domestic dates are uncertain, milestones count from the day a government or parliamentary sponsor adopts the proposal.

  • First 90 days. Publish a common baseline for bank and fiscal risks, jobs, trade exposure, service reliability and case delays, with the series, reporting periods, limitations and who verifies updates. Require a legal-authority and rights assessment for each pilot.
  • Months 4 to 12. Run the budgeted pilots and hold a public review. Move promptly on bank cases where criteria and evidence are clear. Time larger fiscal or tariff adjustments alongside household protections and a verified revenue plan. Proceed with hearings and oversight reforms under existing authority.
  • From the first review to LDC graduation. Expand corridor, customs and training measures where pilots show benefits under ordinary staffing. The UN General Assembly approved extending Bangladesh’s graduation from the least developed country category to 24 November 2029; this is a planning milestone and market-by-market rules differ.25 Publish what exporters and workers must do under each plausible market-access scenario.
  • Beyond graduation and through electoral transitions. Keep baseline reporting, review subsidies, zone benefits and fiscal risks annually, and move successful programmes into regular budgets as evidence and fiscal capacity allow.

What to publish

Public question Suggested measures How to avoid a misleading result
Are households better off? Real earnings, poverty, employment and underemployment, output per worker, from national survey and accounts series. Publish breakdowns by sex, age, region and income. Show survey frequency and explain revisions.
Are commitments affordable? Inflation, bank losses and deposit access, credit to productive firms, tax expenditure, arrears and guarantees. Separate stocks from flows, approved spending from contingent liabilities, and reclassification from real recovery.
Is government fairer and more reliable? Service and court processing times, reasoned decisions and appeals, independent complaint and victimisation surveys, audit follow-up. Pair administrative counts with surveys, outcomes and retaliation checks. Protect identities.
Are opportunities expanding? Foundational learning, job retention after training, customs times, women’s access to work, service reliability. Set a baseline and compare with similar places or a phased roll-out where possible. Record spillovers before crediting a reform.

Where the compact could fail

  • A pact among established organisations can entrench them. Open meetings miss people who face retaliation or cannot travel or organise. Use protected submissions, local outreach and independent surveys, and respond specifically to objections from affected groups.
  • Oversight bodies can inherit the politics they are meant to check. Design appointment, removal, budget, audit and reporting rules together. Test whether each body can secure records and act on a complaint involving a politically connected person. More veto points can also delay legitimate decisions, so set deadlines for advice and keep prompt remedies for rights violations.
  • A short-term bargain can reward the conduct a reform seeks to end. A discretionary loan settlement or an apprenticeship reserved for a party network teaches others to hold out for the same. Use transparent, objective eligibility, audit exceptions, publish exit conditions, and protect depositors and people in need under general rules.
  • Economic hardship can split a stable-looking coalition. Stress-test reform costs against inflation, power disruption, trade shocks and low revenue. Agree adjustment triggers and which household protections continue when savings fall short. Let successors change policy while preserving lawful process and disclosure.
  • A pilot may succeed because it gets unusual resources. Count those costs, check whether problems move next door, and replicate under ordinary conditions. Some reforms need national institutions from the start, including legal protection and financial supervision.
  • The principles may be heard as a demand for silence or as secularism by stealth. The remedy is in the rules: a public channel for grievances (Rule 4) and the protection of conscience, including inside communities (Rule 3).
  • The rules could become a slogan. Without referees, a safe way to lose and visible results, the compact is a hope. The mechanisms above are the answer, and each is a proposal that still has to prove itself.

Evidence gaps

The research did not verify a legal inventory of July Charter measures and bank-resolution powers, credible local measures of extortion and service access, or market-by-market estimates of workers and products exposed to preference changes. Those three pieces of evidence would most change this assessment. The project’s own forecasts, political dates, incentive scores and proposed spending amounts remain inputs to test.

Where the rules go next

The deal-by-deal application is on strategy and deals; the group-by-group view is on group incentives; the loops they would cut are on circular dependencies; and the objections are on objections and critique. The overview is ground rules for working together.


  1. Constitution of Bangladesh, Articles 2A, 7, 8, 12, 26, 27-28, 31-35, 37-39, 41, 44 and 102, as cited (unverified; no primary source located). We did not independently re-read each article. ↩↩↩

  2. The “focal point” argument is the research’s own analysis of coordination; it is a judgement, not a sourced empirical finding. The input pointed to the World Bank’s governance framework for related problems of credible commitment, coordination and cooperation: World Development Report 2017. ↩

  3. See the repeal of Section 18(a) on the banking crisis and the sources there; BSS, repeal; Daily Star, bill to abolish the provision. ↩

  4. World Justice Project, Bangladesh results, 2026 (accessed 2026-11-21, in editorial review): score about 0.39, rank 123 of 143, civil justice 132, fundamental rights 133. The earlier ranks (127 of 142 in 2024; 125 of 143 in 2025) were reported in the research input and are not individually traced. The WJP Bangladesh country sheet is the 2025 edition and supports 2025 claims only. ↩

  5. World Bank, Bangladesh Poverty and Equity Assessment 2025. ↩

  6. ICCPR text; UN treaty status; UN Human Rights Committee, General Comment No. 37 on the right of peaceful assembly. ↩

  7. UN Human Rights Office, fact-finding report, February 2025; UN Bangladesh summary. ↩

  8. John Rawls, Political Liberalism (1993), as summarised in the Stanford Encyclopedia of Philosophy (plato.stanford.edu); the input gave no full URL. ↩

  9. The religious references are illustrations (unverified; no primary source located) and are not a theological ruling. They are for scholars and members of each tradition to confirm or correct. ↩

  10. World Bank, income groups; World Justice Project, Rule of Law Index. ↩

  11. Britannica, the Tokyo Foundation and the National Diet Library of Japan, named without links (unverified; no primary source located); details such as growth rates are not independently checked here. ↩

  12. For the amnesty: Spain, Ley de Amnistia, 15 October 1977, Article 2(e)-(f) (official gazette; link supplied in editorial review). For the Moncloa Pacts: Real Instituto Elcano, Cambridge University Press and Fundacion Juan March, named without links (unverified; no primary source located). ↩

  13. Middle East Institute, University of Melbourne and the Global Centre for Pluralism, named without links (unverified; no primary source located). See also Ash Center, From Reformasi to Institutional Transformation. ↩

  14. Human Rights Watch, Unkept promise: failure to end military business activity in Indonesia. ↩

  15. Ministry of Economy Malaysia, the London School of Economics and malaysia.gov.my, named without links (unverified; no primary source located). ↩

  16. South Africa Department of Justice, explanation of the amnesty process (link supplied in editorial review). For the Constitution and Constitutional Court: Constitute Project and Britannica, named without links (unverified; no primary source located). ↩

  17. Singapore Ministry of Home Affairs, Framework on race and religion (link supplied in editorial review; it confirms the broader scope covering political causes and disaffection against government). Dates and the council’s role: the National Library Board and the National University of Singapore, named without links (unverified; no primary source located). ↩

  18. German Basic Law, Articles 1, 3-5, 8, 19-20 and 67; the Article 79(3) and constitutional-patriotism points come from the Federal Constitutional Court and deutschland.de, named without links (unverified; no primary source located). ↩

  19. Finnish Constitution, sections 6, 11-13 and 109; Economic Council composition. ↩

  20. National Economic Dialogue 2026; Eurofound, end of social partnership as public-sector talks collapse. ↩

  21. New Zealand Public Finance Act 1989, sections 26S and 26U-26V. ↩

  22. Constitutional Court of Korea, case 2024Hun-Na8. ↩

  23. Singapore CPIB; Singapore Constitution; Singapore Police permit guidance. ↩

  24. World Bank, Mauritius Country Growth and Jobs Report. ↩

  25. UN General Assembly, resolution 81/8; proceedings, 29 October 2026; European Commission, EU preference scheme. ↩

Go deeper

Where this comes from

Synthesised from 6 research reports, checked against the sources cited above and reviewed before publication. How we work · All sources.