Layer 2 · In depth
Root causes
A first-principles diagnosis of why Bangladesh's progress has stalled - making versus taking, four layers of explanation, corruption as a pattern of incentives, norms and hierarchy, human capital and organisational quality, and the thinkers and comparisons behind each claim.
On this page
- First principles: making versus taking
- Four layers of explanation
- Four mechanisms that rarely enter the standard debate
- Corruption and rent extraction
- Five channels of damage
- Why it persists
- An institutional reading
- Winner-take-all politics
- Norms, hierarchy and political culture
- Human capital and organisational quality
- How the causes interact
- Six tests for any explanation
- Thinkers and what they contribute
- Gaps
This page gives the reasoning behind Why progress has stalled. The facts it rests on are on The economy in late 2026, The state and its institutions, Education and skills and History and comparisons. The question of religion and culture has its own page, Culture and religion. How the causes feed one another is on The loops.
In one paragraph
Bangladesh does not lack hardworking people: the same workers build Dubai’s skyscrapers, drive the garment industry and send home $35.6bn a year. The difficulty is that inside the country many of the rewards go to taking, not making. A loan that is never repaid is cheaper than one that is. A party post can earn more than a factory job. A posting bought with a payment can earn more than one won on merit. Where these payoffs hold, more money, more plans and more targets leak away. The most useful question is not “what is wrong with the people or the culture?” but “which incentives, institutions and enforced norms keep low-performing practices in place, and what would make better ones self-sustaining?”
First principles: making versus taking
Income per person equals the share of people who work, times what each worker produces. What each worker produces depends on three things: tools and capital, skills, and whether capital and skilled people end up in the firms that use them best.
Bangladesh loses on each term, and for the same underlying reason:
- Share working. Most women and millions of young people are outside paid work (female participation 38.4%; the labour force fell by 1.7 million in 2024).
- Capital. Savings exist, but banks hand them to connected borrowers who do not repay, and to a government that borrows because it cannot tax (The banking crisis).
- Skills. Schools reward certificates and exam grades (Education and skills); in a 2018 survey, 47% of young people with higher-secondary education or above named a government job as their main goal.
- Allocation. Tariffs, licences, extortion and courts decide who wins, more than productivity does.
In each case resources are allocated by power, not by productivity. This is an old idea: talented people go where the returns are highest, and societies grow when making things pays more than taking things (Baumol 1990; Murphy, Shleifer and Vishny 1991; cited from memory in the research, not re-checked). It gives one test for every proposal: does it raise the return to making, or lower the return to taking, for a group strong enough to matter?
The core claim
That taking pays better than making for many powerful groups is our synthesis of the evidence, not a measured fact. It is consistent with the data on concentrated defaults, extortion, party finance and bought postings (see What the numbers miss). The groups act sensibly given what they are paid for; this is a description of a system, not a charge against anyone.
Four layers of explanation
No single cause is enough. The evidence supports a layered account.
- A low starting point built by history. About 62% of Pakistan’s income per head in 1972, after colonial land institutions, the partition of the jute economy and a war, with a famine to follow in 1974 (History and comparisons).
- Fast catch-up on a narrow model. Garments and remittances, run on reliable informal deals (“competitive clientelism”), which are now running out of road (History and comparisons).
- Binding constraints now. Trade protection that taxes exports, a financial sector that does not lend, an energy sector that loses money, a weak tax state and poor learning (The economy in late 2026).
- Mechanisms beyond the standard list. See below.
Four mechanisms that rarely enter the standard debate
(Confidence moderate; the ranking is a judgement.)
- The state taxes trade instead of incomes. With tax under 7% of GDP, tariffs serve as a revenue tool.1 That protects a domestic-market business class, so a fiscal weakness becomes an export weakness, and the lobby for protection is also the lobby against direct taxation.
- The banking system has worked as the financing arm of the political settlement. The Task Force recommended that banks no longer be licensed on political grounds, which shows that insiders see it the same way.1 Private credit at its lowest growth in 33 years is the cost.
- Half the potential workforce is under-used. About 85% of the jobs lost in the first half of FY2025 were women’s.2 The IMF models closing the gender gap as worth nearly 40% more output, an upper-bound scenario,3 and the World Bank estimated full convergence would add 1.8 points of annual potential growth.4
- Physical drags: heat, air pollution, nutrition and health costs (Heat, health and environment). They use different methods and cannot be added together.
Corruption and rent extraction
“Corruption” here means more than bribes. It includes patronage hiring, politically connected procurement, selective law enforcement, regulatory discretion, state-business collusion, preferential lending and unequal access to public resources. Transparency International’s 2025 score is 24 out of 100; it measures perceived public-sector corruption, not the amount stolen.5
Five channels of damage
- Resource diversion and a hidden tax on enterprise. Capital and management attention go to permits, access and protection instead of investment. Even a firm that never pays may lose to a favoured competitor.
- Adverse selection of firms and professionals. Connections replace competence, so there is less reason to build real capability.
- Erosion of state capacity. Tax administration, procurement, courts, customs, regulators and police are weaker when rules are selectively applied, and weak capacity makes corruption easier.
- Financial misallocation. Politically influenced lending and weak supervision send scarce capital to access, not productivity, and the losses become fiscal liabilities.
- Collapse of trust and compliance. When rule-following looks unrewarded, people rationally seek protection and access. It becomes a coordination failure, not only a collection of unethical individuals.
Why it persists
Corruption can be part of a political equilibrium: powerful actors benefit from discretion, while citizens and businesses cannot easily coordinate to change it. Elections alone may not fix this when parties depend on patronage finance or when enforcement bodies lack independence. This does not make attitudes to honesty irrelevant. It means that changing the payoff structure (transparent procurement, competitive recruitment, audited finance, credible sanctions, a less discretionary state) can change behaviour without first changing a national culture.
Jyoti Rahman argues that those who do not use public schools, hospitals or transport have less reason to support paying tax for them; this is his interpretation, and it underlies the tax loop on The loops.6
An institutional reading
One framing (unverified; no primary source located) casts the problem in the language of inclusive and extractive institutions (Acemoglu and Robinson). Inclusive institutions disperse power and protect property and contract rights broadly. Extractive ones concentrate power in a narrow elite. Applied to Bangladesh, it restates four familiar problems as outputs of one equilibrium. These are scholarly readings and judgements, not findings:
| Problem as usually stated | The same problem read as a system |
|---|---|
| The bureaucracy is inefficient | Its design inherited a revenue-collecting purpose (a Mughal-British hybrid, in the language of Bangladeshi scholars of the civil service) and political patronage was added on top; reforms stall because a well-functioning bureaucracy threatens the incumbent coalition |
| Education is low quality | Education is also a patronage asset: appointments, admissions, hall seats and tenders flow through it, so low quality is a cost the system tolerates |
| Religion resists change | Religious framing is genuine at the grassroots; at the political level it can also serve to defend the existing distribution of power |
| Politics is winner-take-all | The state is the largest single source of rent, so losing can mean prosecution and loss of assets (as events since 2024 show), which raises the stakes of every election |
Student politics are described in that account as the apprentice system of a patronage-based political order. Women’s exclusion from paid work is described as a material allocation of income and status that is defended with religious framing but explicable without it. The practical conclusion drawn is that a reform hitting only one of the four leaves the other three intact and they restore the equilibrium.
Handle with care
This reading is sharper than the other evidence allows on two points: it does not show that the bureaucracy’s design explains today’s outcomes, and it treats religious politics partly as a “shield”. We report it as one coherent view and test the religious claim separately on Culture and religion.
Winner-take-all politics
Where the ruling party controls the police, courts, tax authority, banks and procurement, losing power has meant false cases, ruin or exile. Politics then becomes a fight for survival rather than a debate over policy. First-past-the-post turned 49.97% of the 2026 vote into about 70% of seats; the facts are on The state and its institutions. Short time horizons follow: a ruler who expects to stay, or to leave safely, protects the economy he taxes; one who expects to be destroyed on losing strips it (Olson 1993, cited from memory). A 2026 refinement: much recent violence is mob killing and clashes within and among the winning camps, so the problem is not only an excluded opposition using the street.
Norms, hierarchy and political culture
A society’s informal rules affect who may speak, learn, work, travel and disagree. Some facilitate cooperation. Others restrict adaptation.
- Deference is useful until it becomes immunity from criticism. Hierarchy becomes dangerous when it disables feedback: weak upward reporting gives misleading information at the top, so decisions are poor and nobody corrects them. This is not specifically Islamic; authority-centred norms occur in religious, secular, corporate, military and family systems everywhere. What matters is the cost of disagreement and the quality of correction.
- Cultural measures. On Hofstede’s framework Bangladesh is cited at 80 for power distance, 20 for individualism, 60 for uncertainty avoidance and 20 for indulgence.7 These are old, survey-based national averages, and the inputs could not retrieve World Values Survey trust data. Japan and Korea combine high power distance with wealth, which weakens any inference from a single score.
- Campus and professional politics. The distinction is between political participation and politicised institutional capture. Student movements have played both constructive roles (the July 2024 uprising) and destructive ones. The facts are on The state and its institutions.
- Women’s exclusion costs households and the economy (below and Culture and religion).
- Time, holidays and work patterns. Workers can labour very long hours and produce little per hour if they lack capital, power, transport, skills and management. The target is output per hour, not hours; the evidence on hours worked is on Culture and religion.
Human capital and organisational quality
“Incompetence of the general public” is a poor causal category. It mixes innate ability, learning opportunity, job design, incentives, supervision and selection. A better decomposition:
- Foundational human capital: literacy, numeracy, reasoning, health, early childhood.
- Specialised skill: technical, scientific, managerial and professional expertise.
- Organisational capability: workflows, standard procedures, measurement, supervision, quality assurance and knowledge transfer.
- Incentive alignment: whether hiring, pay, promotion and procurement reward results.
- Learning capacity: whether errors are detected, discussed and turned into better processes.
Capability is partly a property of systems. A talented engineer cannot offset a procurement process that buys inferior equipment or an organisation that punishes bad news; a poorly prepared worker can improve quickly in an organisation with clear standards and feedback. Competence both affects institutions and is produced by them. The move from labour abundance to productivity growth needs more demanding capabilities: quality management, engineering, finance, logistics and institutions that correct errors quickly. In Zia Hassan’s view the country also has a “knowledge deficit”, which is why he sees foreign investment as the main exit, for the management systems and technology it brings as well as the dollars (Voices).
Evidence that skills follow rewards, not the other way round is summarised on Culture and religion. A 2018 survey found that 47% of young people with higher-secondary education or above named a government job as their main goal, which shows where many see the rewards; it does not measure motivation or party or visa preferences.8
How the causes interact
Political and institutional
- Weaker state capacity and enforcement
- Appointments and credit driven by connections
- Low-quality services and infrastructure
- Lower human capital and organisational capability
- Lower human capital and organisational capability
- Lower productivity, fewer jobs
Economic
- Poverty, insecurity, dependence on patrons
- Patronage, corruption, rent extraction
- Poverty, insecurity, dependence on patrons
- Low-quality services and infrastructure
Social and human
- Lower productivity, fewer jobs
- Reduced women's autonomy and participation
- High cost of dissent and criticism
- Lower productivity, fewer jobs
- Weaker state capacity and enforcement
- Lower human capital and organisational capability
Three factors
- Corruptionrent extraction, regulatory capture, politically directed finance
- Exclusionary normshierarchical deference, restricted voice, gender and religious coercion
- Capability deficitsweak learning, poor management, limited technical execution
Overlaps
- Corruption and normsclientelism legitimised by status; selective enforcement
- Corruption and capabilitypatronage hiring; low-quality procurement; unqualified supervision
- Norms and capabilityrote learning; suppressed feedback; talent lost through exclusion
Result where all three meet
- Weak accountability, low trust, poor adaptation, lower productivity
The three factors are mutually reinforcing mechanisms, not three independent root causes, and not fixed attributes of Bangladeshis. Their deeper determinants include political incentives, state capacity, historical institutional choices, elite bargains, economic structure, global conditions, geography and the interpretation and enforcement of social norms.
Six tests for any explanation
A theory of Bangladesh’s problems should pass these tests. Claims about corruption, gender exclusion, threats against dissent, patronage recruitment and organisational dysfunction pass them. Claims about lost hours from prayer or a universal psychological damage do not yet (see Culture and religion).
- Cause or outcome? Corruption can cause low capacity and low capacity can permit corruption; model both directions.
- Avoid ecological inference. Aggregate outcomes do not reveal the character, intelligence or diligence of individuals.
- Counterfactuals. If a factor is large, comparable groups with different values of it should show large, robust differences.
- Omitted variables. Geography, colonial inheritance, trade policy, conflict, energy, infrastructure, capital access, health and external shocks.
- Variation within a religion or culture. Differences among Muslim-majority countries, across regions and over time are evidence, not noise.
- Private belief versus public coercion. Voluntary observance does not limit anyone else; concern grows where rules are imposed on those who do not choose them.
Thinkers and what they contribute
| Lens | What it says | Use here |
|---|---|---|
| Baumol; Murphy, Shleifer and Vishny | Talent goes where returns are highest; growth needs making to pay more than taking | The making-versus-taking test |
| Acemoglu and Robinson | Inclusive versus extractive institutions; critical junctures | The institutional reading |
| Douglass North | Formal rules, informal norms and enforcement together determine incentives | Why laws on paper are not enough |
| Mushtaq Khan | Growth can come from managed rents; good-governance agendas give elites no workable reform path; target sectors where powerful actors gain from rules being followed | Competitive clientelism; islands of effectiveness |
| Persson, Rothstein and Teorell | In thoroughly corrupt systems, principal-agent fixes fail without trustworthy principals | Why reforms need a sponsor |
| Amartya Sen | Development is substantive freedom, not only GDP | Why girls’ schooling and women’s autonomy count in their own right |
| Max Weber | Beliefs can shape economic conduct through particular historical institutions | Not “one religion makes wealth” |
| Timur Kuran | Legal forms (inheritance, partnerships, endowments) shaped the rise of large commercial organisations | Path-dependent institutional history |
| Inglehart and Welzel | Values vary along traditional-secular and survival-self-expression axes | Useful, but averages hide internal diversity |
| Collective-action models | People comply with norms they privately oppose when challenging is risky | Why change can look impossible when seen statically |
Sources for these are on Sources.
What the best-supported claims are
The evidence best supports: a thin tax state, a captured financial system, trade protection, weak learning and women’s exclusion from work as the main economic constraints; patronage and winner-take-all politics as the main political constraints. It supports religious conservatism as a real, second-tier channel (Culture and religion). It does not support a single-variable story. The confidence is moderate.
Gaps
- No Bangladesh-specific causal estimates for several of the mechanisms; most cross-country work does not study Bangladesh directly.
- The relative weight of norms versus lack of demand for women’s work is unknown.
- The works of Baumol, Murphy-Shleifer-Vishny, Olson and Tay-Diener were cited from memory and not re-checked.
-
UN Women, State of the Bangladesh economy from a gender lens. ↩
-
IMF, Bangladesh can boost growth and climate resilience by investing in women. ↩
-
Daily Star, Female labour force could drive economy further: WB. ↩
-
Transparency International, Bangladesh country profile. ↩
-
Prothom Alo, interview with Jyoti Rahman, 2 August 2026. ↩
-
Prothom Alo, 6-D model of culture and Bangladesh (its ref 14) and a Scribd document (ref 15). ↩
-
Financial Express on the SANEM survey, youth support; BRAC 2018 survey, blog (47% of youth with higher-secondary education or above named a government job as their main goal). ↩
Go deeper
- Key statisticsData & sourcesThe important figures from the research, grouped by theme in tables, each with its value, date, source and link. Where sources disagree, see the disputed-figures ledger.
- Disputed figuresData & sourcesThe ledger of figures where our sources disagree, with both values, the likely reason and how this site handles each, plus a list of known gaps.
- SourcesData & sourcesThe consolidated bibliography of every external source cited in our research, grouped by type and deduplicated, with the links as given.
Summarised in
- Why progress has stalled Overview
Where this comes from
Synthesised from 7 research reports, checked against the sources cited above and reviewed before publication. How we work · All sources.