09 Human capital

Chapter 09 of 60 in the Bangladesh 2036 research base. Contents of the series.

Bangladesh built the school seats and exited the clinic, and the decade turns on whether the state pays for learning and risk pooling before the demographic and epidemiological clocks close

The human capital record of FY10 to FY25 is two achievements and two absences. Stunting among children under five fell from 41.0 percent in the 2011 survey to 24.0 percent in the 2022 survey [DHS Program 2022], under five mortality fell from 48.7 to 30.5 deaths per 1,000 live births between 2010 and 2024 [WB WDI 2026], and the primary net enrolment rate reached 97.56 percent in 2022 [BANBEIS 2022]. The absences are measurable: 51.22 percent of children at the end of primary age cannot read and understand a simple text [WB WDI 2026], households pay 79.31 percent of current health spending out of pocket, and the government's own health expenditure fell to 0.31 percent of GDP in 2023, the lowest reading in the WHO series since 2000 [WHO GHO 2023, OOP share of CHE, GGHE-D share of GDP]. The thesis of this chapter is that Bangladesh assembled a middle income human capital stock on a low income financing base, and that the decade to FY36, the horizon the chapter 15 scenarios assume, turns on whether the state moves from financing school places and vertical health programmes to financing quality and risk pooling before the ageing that chapter 08 dates and the non communicable disease transition make the current model unaffordable. The World Bank's Human Capital Index Plus places the composite at 146.7 on a 325 point scale in 2025, up from 128.5 in 2015, with the female score 41 points below the male [WB HCI 2025, HD_HCIP_OVRL_TO, HD_HCIP_OVRL_FE, HD_HCIP_OVRL_MA].

Near universal seats and falling mortality stand beside a learning deficit and household financed care

The education expansion is real and nearly complete at the bottom. The primary net enrolment rate stood at 97.56 percent in 2022 and the gross enrolment rate at 106.49 percent in 2024, above 100 for over age pupils [BANBEIS 2022] [WB WDI 2026]. The primary completion rate reached 94.30 percent of the relevant age group in 2024, girls at 102.83 percent against 86.45 percent for boys [WB WDI 2026]. Retention improved fastest of all: entrants surviving the cycle rose from 67.25 percent in 2010 to 86.85 percent in 2022, and the dropout rate fell from 39.80 to 13.95 percent [BANBEIS 2022]. Literacy followed the cohort: adult literacy reached 79.0 percent in 2022 from 65.1 in 2015, youth literacy 93.45 percent with a gender parity index of 1.02 [WB WDI 2026]. Mean years of schooling among adults was 6.79 years in 2023 [UNDP HDR 2023]. The soft spot is at the margin: primary age children out of school rose to 5.86 percent in 2024 from 3.28 percent in 2017 [WB WDI 2026], and the learning poverty measure below counts them.

What the expansion did not produce is learning. The learning poverty measure was 51.22 percent in the 2022 measurement, 54.17 percent for boys and 49.14 percent for girls [WB WDI 2026]. The lower secondary completion rate was 73.81 percent in 2024 [WB WDI 2026], the gross secondary enrolment rate 64.30 percent, with girls at 70.47 percent against a system total pulled down by boys [WB WDI 2026], and the gender parity index across primary and secondary enrolment stood at 1.146 in 2020 [WB WDI 2026].

Above secondary the system narrows: tertiary gross enrolment of 23.73 percent in 2024, women at 20.79 percent [WB WDI 2026]; the BANBEIS tier table counts 1,069,579 university students in 171 institutions in 2023, 37.47 percent women [BANBEIS 2023]. Technical and vocational education is small and male: 689,745 students in 2,547 institutions in 2023, 27.50 percent female, against 8.17 million secondary and 2.76 million madrasah students in the same table [BANBEIS 2023, institutions and students by tier]. Delivery is private below the tertiary tier, 93.90 percent of secondary enrolment in the last published share against 23.90 percent at primary [WB WDI 2026], a structure in which the state sets curricula while families carry the rest. Classrooms run full relative to teachers at both tiers: pupil teacher ratios of 30.1 in primary at the last measurement, 33.1 in secondary in 2023 [WB WDI 2026] [BANBEIS 2023, tier table].

Under five mortality fell from 48.7 deaths per 1,000 live births in 2010 to 30.5 in 2024, neonatal mortality from 28.8 to 17.9 over the same window, and the maternal mortality ratio from 275 per 100,000 live births in 2010 to 152 in 2020, with a pandemic jump to 194 in 2021 in the same World Bank series before it fell to 131 in 2022 and 115 in 2023 [WB WDI 2026]. Immunisation coverage is near universal, 97 percent of one year olds for DPT and 96 percent for measles in 2024 [WB WDI 2026]. Life expectancy at birth reached 74.9 years in the 2024 series [WB WDI 2026], against healthy life expectancy of 63.1 years in the WHO series for 2021 [WHO GHO 2021, HALE], a gap of more than a decade lived with illness or disability. Nutrition improved but has not closed: the survey record shows stunting falling from 41.0 percent in 2011 to 36.0 in 2014, 31.0 in the 2017 to 18 survey and 24.0 in 2022 [DHS Program 2022], the modelled series agrees at 25.1 percent in 2024 from 54.9 percent in 2000 [WB WDI 2026], underweight prevalence was 21.7 percent in 2022, and anaemia among children ages 6 to 59 months still ran at 43.1 percent at the last reading in 2019 [WB WDI 2026]. The mortality mix has moved with it: non communicable diseases were 63.1 percent of all deaths in 2021, up from 42.5 percent in 2000 [WB WDI 2026].

The health financing record is the mirror opposite of the outcome record. Total current health expenditure was 2.17 percent of GDP in 2023. Of that, the government paid 0.31 percent of GDP, 14.48 percent of current health expenditure, its lowest reading in the series since 2000 and down from 0.51 percent of GDP that year; donors paid 3.50 percent of current health expenditure; households paid 79.31 percent out of pocket, up from 61.82 percent in 2000 [WHO GHO 2023, CHE as share of GDP, GGHE-D as share of GDP, GGHE-D as share of CHE, EXT as share of CHE, OOP as share of CHE]. Out of pocket spending reached 42.40 USD per person per year in 2023 [WHO GHO 2023, OOP per capita]. The domestic accounts agree: the latest Bangladesh National Health Accounts, 2020, puts total health expenditure at 777,347 million BDT, 77,735 crore, with official shares of 68.53 percent out of pocket, 23.12 percent government and 5.01 percent development partners [MOHFW BNHA 2020]. Capacity sits where that financing predicts: a universal health coverage service coverage index of 54 of 100 in 2023 [WHO GHO 2023, UHC Service Coverage Index], 0.72 physicians and 0.66 nurses and midwives and 0.92 hospital beds per 1,000 people at the latest readings [WB WDI 2026], and a Global Health Security Index score of 35.5 of 100 in both the 2019 and 2021 editions with the health system category falling from 35.6 to 31.3 between them [GHS Index 2021]. The household cost is measured: in 2016, 24.42 percent of the population lived in households that spent more than a tenth of the household budget on health, up from 14.22 percent in 2010, and 8.45 percent in households that spent more than a quarter [WHO GHO 2016, catastrophic health expenditure at 10 and 25 percent of household budget], while health payments pushed 6.99 percent of the population below the 3.65 USD a day line, up from 4.34 percent in 2010 [WHO GHO 2016, population pushed below 3.65 USD a day].

The spending envelope explains both absences at once. Government expenditure on education was 2.03 percent of GDP in 2024 and 11.88 percent of total government expenditure in 2025 [WB WDI 2026], against the 4 to 6 percent of GDP benchmark the UNESCO Incheon framework sets, a figure that could not be confirmed against the framework text, which is not available in any source this chapter draws on [UNESCO UIS 2015]. The development budget repeats the pattern at the margin: the FY26 annual development programme allocates 28,557.43 crore BDT to education, 12.42 percent of the 230,000.00 crore BDT programme the National Economic Council approved on 18 May 2025, and 18,148.14 crore BDT to health, 7.89 percent on the same denominator, both shares as the programme itself prints them [Planning Commission ADP 2025, ADP at a glance table]. The combined operating claim of the two sectors on the state, 2.03 plus 0.31, is 2.34 percent of GDP by this chapter's arithmetic on the two series [WB WDI 2026] [WHO GHO 2023, GGHE-D as share of GDP], the fiscal expression of the model: Bangladesh mass produced human capital by keeping the unit cost of the public contribution low and shifting the marginal cost to households.

Access first financing built the enrolment record and routed the chronic disease caseload to households

The education mechanism was quantity first by design. Three instruments drove the enrolment and retention revolution: the primary education development programmes that nationalised teacher salaries and expanded the school network, the female secondary stipend programme that paid families to keep daughters enrolled, and NGO and private delivery that absorbed demand at unit costs the state could afford. The instruments worked on the margins they targeted, enrolment and completion, and not on the margin nobody measured, learning. A state spending 2.03 percent of GDP [WB WDI 2026] cannot fund access and quality at once, and it chose access first; the mechanism that delivered the enrolment miracle, low unit cost, is the mechanism that delivered the learning deficit, because teacher time, materials and assessment are the inputs cheap provision skimps on. The 2022 to 2024 curriculum reform, the first move from memorisation toward competence, was rolled back in part after the 2024 transition; the sequence and current status could not be confirmed; the Ministry of Education would settle it.

The health mechanism is a state that financed inputs and let households finance care. The achievements, immunisation, family planning, oral rehydration, maternal survival, were delivered through vertical programmes with dedicated funding and NGO delivery, which is why they reached near universal coverage on a public budget of 0.31 percent of GDP [WHO GHO 2023, GGHE-D as share of GDP]. The 46 against 66 split in the UHC service coverage sub indices is that inheritance in one number, 66 for the reproductive, maternal and child track and 66 for infectious disease against 46 for non communicable disease [WHO GHO 2023, UHC SCI by component]. The diseases the population now dies of, led by ischaemic heart disease, stroke, diabetes and cancers, are episodic, facility based and catastrophic in cost, and the financing model routes them to the household: an out of pocket share of 79.31 percent [WHO GHO 2023, OOP as share of CHE] is what a vertical programme system looks like from the paying side after the epidemiological transition. The catastrophic and impoverishment series in the record is the measured cost of that routing [WHO GHO 2016].

The skills mechanism is a production model that stopped needing training precisely when the labour market needed more of it. The export engine documented in chapters 02 and 08 hires at the bottom of the skill distribution and trains on the job at low cost, which made formal training, TVET and employer provided alike, optional for firms. The enterprise survey record shows the option being exercised less and less: firms offering formal training fell from 21.90 percent in 2013 to 6.42 percent in 2022 [WB ES 2022]. The World Bank's human capital accounts measure the same collapse from the worker side: the on the job learning pillar of the HCI Plus framework scores Bangladesh at 27.4 of a possible 87 in 2025, with the male score 50.3 and the female score 5.7 [WB HCI 2025, HD_HCIP_OTJL_TO, HD_HCIP_OTJL_MA, HD_HCIP_OTJL_FE]. The TVET tier that should backfill the gap is one vocational student per twelve secondary students by this chapter's arithmetic on the tier table, and the skill mix of the 2023 migration flow, 50.2 percent less skilled in the BMET record chapter 08 documents, prices the output of this system in the Gulf labour market. The mismatch is a post schooling investment problem with three parties who each expect the other to pay, firms, the state and households.

Smaller cohorts make quality affordable while ageing prices the household model out of reach

The demographic arithmetic changes the demand side before any policy does. The UN WPP 2024 medium variant projects fertility falling from 2.14 births per woman in 2024 to 1.92 by 2036 and the median age rising from 25.65 to 30.0 years [UN WPP 2024, medium variant], and annual births from 3.47 million in 2024 to 3.25 million in 2030 and 2.98 million in 2036 [UN WPP 2024, medium variant, births], so the 2030 birth cohort is 6.3 percent smaller than the 2024 cohort and the 2036 cohort 14.2 percent smaller, by this chapter's arithmetic on the medium variant. Smaller cohorts are the education opportunity of the decade: the system can hold its enrolment budget flat and watch unit costs rise, or hold unit costs and let quality spending absorb the difference, and the reform scenario in chapter 15 assumes the second. In health the arithmetic runs the other way: the old age dependency ratio rising 0.22 points a year on the trend chapter 08 records means the chronic disease caseload grows faster than the population, and the financing model that delivered the twentieth century outcomes, vertical programmes plus household payment, prices the twenty first century caseload out of reach.

The HCI Plus trajectory gives the benchmark: the overall score rose from 128.5 in 2015 to 137.7 in 2020 and 146.7 in 2025 [WB HCI 2025, HD_HCIP_OVRL_TO], about 18 points a decade, driven by the education pillar's rise from 63.9 to 77.5 of a possible 188 [WB HCI 2025, HD_HCIP_EDUC_TO] and the health pillar to 41.8 of 50 [WB HCI 2025, HD_HCIP_HLTH_TO]; this chapter's arithmetic assumes the 2015 to 2025 pace continues, which puts the score near 165 by FY36 in the reform scenario and materially lower in the stall scenario if the learning and financing deficits persist.

Five decision points define the window, each with a named author. The education financing decision, authored by the Finance Division with the Ministry of Education, is whether the education budget moves off 2 percent of GDP toward the plan target; the target could not be confirmed; the Eighth Five Year Plan document would settle it [GED Plans 2020]. The assessment decision, authored by the education ministries and the examination boards, is whether a national learning measurement at primary level becomes routine public data, because the learning poverty number is an international estimate and no domestic series in the data lake measures learning. The TVET decision, authored by the technical education directorate with industry, is whether the vocational tier scales from its 689,745 students toward the systems of Viet Nam and Malaysia; the sector's monitoring framework lists the youth enrolment proportion among its indicators and states that not all carry targets [DTE 2023]. The health financing decision, authored by the Health Services Division and the Health Economics Unit, is whether the prepayment and social health protection schemes in the sector strategy are funded at a scale that bends the out of pocket share; the strategy's coverage targets could not be confirmed; the ministry documents would settle it. The tertiary decision, authored by the University Grants Commission and the education ministry, is whether accreditation and research funding make the 171 institutions a quality system, because the graduate surplus that cannot find wage work is chapter 08's NEET stock in university clothes, 15.43 percent of youth ages 15 to 24 in 2024 and 19.73 percent among young women [WB WDI 2026]. The labour conventions attached to the GSP+ application that chapter 02 targets in its 2028 window lean on two of these five decisions, on workplace standards that assume trained workers and on freedom of association that assumes functioning employer institutions [EU GSP 2023].

Three risks print in published series and three dividends need no new money

Risks. First, a learning regime lock in: if the learning poverty rate stays near 51 percent, the cohorts entering the labour force in the 2030s carry reading deficits into an economy whose export regime after graduation, chapter 02's argument, depends on moving up from cut make trim toward design and compliance heavy segments; the revealing indicator is the next learning poverty measurement and the national assessment series that does not yet exist. Second, a medical impoverishment spiral: the non communicable disease share of deaths, 63.1 percent in 2021 [WB WDI 2026], grows with ageing while public financing stays at 0.31 percent of GDP [WHO GHO 2023, GGHE-D as share of GDP], and the catastrophic spending series [WHO GHO 2016] is already its ledger; the revealing indicator is the out of pocket share crossing 80 percent, bordered at 79.31 percent in 2023 [WHO GHO 2023, OOP as share of CHE]. Third, a skills famine at the window's close: if firm training stays below 10 percent [WB ES 2022] and the female on the job learning score stays near 5.7 of 87 [WB HCI 2025, HD_HCIP_OTJL_FE], the demographic window that chapter 08 dates to the 2030s closes on a workforce whose skill acquisition stopped at the school gate; the revealing indicators are the next enterprise survey wave and the BMET skill mix table.

Upside. First, the cohort dividend in education: the UN projects fertility at 1.92 births per woman by 2036 on the medium variant [UN WPP 2024, medium variant], which shrinks school age cohorts, so a flat education budget is a rising per student budget, the first decade in which quality spending does not compete with seats for the same taka. Second, the educated woman dividend: the schooling pipeline that favours girls at every stage, 1.02 youth literacy parity and 70.47 percent female secondary gross enrolment [WB WDI 2026], is the supply side precondition for the participation recovery chapter 08 prices and the path by which the HCI Plus gender gap of 41 points closes [WB HCI 2025, HD_HCIP_OVRL_FE, HD_HCIP_OVRL_MA]. Third, the rails dividend: the mobile money and digital identity rails of chapters 07 and 13 are the distribution system a prepayment or micro insurance scheme needs to pool the risk households currently finance at 68.53 percent of health spending [MOHFW BNHA 2020].

What to watch: five indicators whose thresholds mark the human capital regime

1. Learning poverty rate. Current value 51.22 percent in the 2022 measurement [WB WDI 2026]. Threshold: a print below 40 percent in the next measurement round signals the quality turn; a print above 55 percent confirms the lock in the stall scenario assumes.
2. Combined government education and health expenditure. Current value 2.03 percent of GDP for education in 2024 [WB WDI 2026] plus 0.31 percent of GDP for health in 2023 [WHO GHO 2023, GGHE-D as share of GDP], 2.34 percent combined by this chapter's arithmetic. Threshold: a combined print above 3.5 percent of GDP by FY30, education above 3 alone, is the funding turn the reform scenario assumes; a combined print below 2.0 confirms the fiscal choice is permanent.
3. Out of pocket share of health spending. Current value 79.31 percent in 2023 [WHO GHO 2023, OOP as share of CHE]. Threshold: a fall below 65 percent signals risk pooling at scale; a rise past 80 percent is the medical impoverishment regime, and the catastrophic spending series [WHO GHO 2016] is its welfare ledger.
4. Child stunting. Current value 24.0 percent in the 2022 survey [DHS Program 2022], 25.1 percent in the 2024 modelled series [WB WDI 2026]. Threshold: a fall below 15 percent, nine points under the 2022 base, six years at the survey trend of 1.5 points a year, marks the nutrition regime closing; stagnation above 20 percent confirms the stall.
5. HCI Plus overall score and gender gap. Current value 146.7 of 325 in 2025, with the female score 126.9 against the male 168.2 [WB HCI 2025, HD_HCIP_OVRL_TO, HD_HCIP_OVRL_FE, HD_HCIP_OVRL_MA]. Threshold: a female to male gap below 30 points alongside an overall score above 160 is the human capital regime that the reform scenario in chapter 15 assumes; an overall score stalling below 150 with the gap above 45 points is the stall.

Sources used

[WB WDI 2026] World Bank World Development Indicators pull via bdpolicy parquet wb_full_bd: education access, literacy, completion, parity, private share, staffing, spending, out of school children, NEET, mortality, immunisation, life expectancy, nutrition, causes of death and workforce density series, values in data/09-human-capital.csv. [WB HCI 2025] World Bank Human Capital Index Plus series via bdpolicy parquet wb_full_bd, overall, gender, education, health and on the job learning pillars, HD_HCIP codes. [WHO GHO 2023] World Health Organization Global Health Observatory via bdpolicy parquet who_gho_bd: health expenditure shares by financing source, out of pocket per capita, UHC service coverage index and disease component sub indices, 2023 vintage. [WHO GHO 2021] World Health Organization Global Health Observatory via bdpolicy parquet who_gho_bd, healthy life expectancy (HALE) at birth, series WHOSIS_000002, 2021 vintage. [WHO GHO 2016] World Health Organization Global Health Observatory via bdpolicy parquet who_gho_bd: population in households spending above 10 and 25 percent of the household budget on health (SDG 3.8.2, FINPROTECTION_CATA_TOT_10_POP and _25_POP) and population pushed below the 3.65 USD a day line (FINPROTECTION_IMP_NP365_POP), 2010 and 2016 tabulations. [GHS Index 2021] Global Health Security Index, Nuclear Threat Initiative and Johns Hopkins Center for Health Security, via bdpolicy parquet health/ghs_index_bd, overall and category scores, 2019 and 2021 editions. [DHS Program 2022] The DHS Program Bangladesh Demographic and Health Survey via bdpolicy series bdhs_nutrition_stunting, stunting among children under five, 2011 to 2022 waves. [BANBEIS 2022] Bangladesh Bureau of Educational Information and Statistics headline indicators via bdpolicy parquet banbeis_headline_indicators_bd, primary net enrolment, dropout and survival series. [BANBEIS 2023] Bangladesh Bureau of Educational Information and Statistics institutions and students by tier via bdpolicy parquet banbeis_institutions_students_by_tier, all tiers 2023 with student teacher ratios. [UNDP HDR 2023] UN Development Programme Human Development Report indicators via bdpolicy.db series undp_hdro_mean_schooling_years (same value in parquet owid_average_years_of_schooling_bgd), mean years of schooling, 2023. [UNESCO UIS 2015] UNESCO Incheon Declaration and its Framework for Action (SDG 4 targets), 4 to 6 percent of GDP education financing benchmark, text not held in the data lake, benchmark not established. [Planning Commission ADP 2025] Annual Development Programme 2025-2026, Programming Division, Bangladesh Planning Commission, June 2025, ADP at a glance sector table: education 28,557.43 crore BDT (12.42 percent) and health 18,148.14 crore BDT (7.89 percent), both on the 230,000.00 crore programme approved by the National Economic Council on 18 May 2025, https://adp.plancomm.gov.bd/book/2025-2026-ADP-BOOK.pdf, accessed 6 September 2026. The lake series adp_education_cr_bdt (27,641.03) and adp_health_cr_bdt (15,300.75) belong to the FY2023-24 programme book and are not used here. [GED Plans 2020] General Economics Division, Eighth Five Year Plan, spending targets resolving source, figures not held in the data lake. [DTE 2023] Directorate of Technical Education, TVET Sector Performance Monitoring Framework 2021-2025, February 2023, via the ocr_text/dte_bd lake, KPI list and its statement that not all KPIs carry targets. [WB ES 2022] World Bank Enterprise Surveys, Bangladesh 2022 wave, series IC.FRM.TRNG.ZS, firms offering formal training, 2013 and 2022 waves. [UN WPP 2024] UN World Population Prospects 2024 via bdpolicy parquet demography/un_wpp2024_bd, medium variant, fertility, median age and births projected to 2036. [EU GSP 2023] GSP regulation (EU) No 978/2012 as applied, GSP+ labour conventions, trade leverage on skills decisions. [MOHFW BNHA 2020] Health Economics Unit, Bangladesh National Health Accounts 2020 via bdpolicy.db bnha_ series: the_current, gov_health_expenditure, oop_health_expenditure, oop_share_pct, gov_share_pct, dev_partners_share_pct.

---

Verified line by line against primary sources: 134 claims checked, 6 corrected.

Previous: 08 Employment and demographics

Next: 10 Energy and infrastructure

All sixty chapters and the ten narrative chapters.