33 Higher education, research and brain drain

Chapter 33 of 60 in the Bangladesh 2036 research base. Contents of the series.

A credential machine with no research engine meets the decade that needs both

Chapter 09 measured the school system and stopped where the tiers end: tertiary gross enrolment of 23.73 percent in 2024 [WB WDI 2026], 171 university institutions holding 1,069,579 students [BANBEIS 2023, tier table], and the tertiary decision named but not taken. This chapter goes into that tier, its financing and its export market, and the record it exposes is a credential machine with no research engine. The system certifies at scale: 3,946,284 people aged 20 to 24 were attending tertiary education on census day 2022 [BBS Census 2022]. It does not measurably research: no series in the sources this chapter draws on carries a national research and development expenditure figure, which is itself the finding, resident patent applications ran between 40 and 74 a year across 2019 to 2021 [WB WDI 2026], and receipts for the use of intellectual property were 1.3 million USD in 2024 [WB WDI 2026]. Its best people exit: the Bangladesh born stock of the four settlement destinations, the United States, the United Kingdom, Canada and Australia, rose from 130,312 in 1990 to 619,452 in 2020, 4.75 times, by this chapter's arithmetic on the destination tables [UN DESA 2020]. The thesis is that the decade to FY36, the horizon the chapter 15 scenarios assume, splits this tier into two regimes. In the credential regime the state keeps founding campuses and certifying graduates, the research deficit compounds, and the settlement channel drains the top of the ability distribution. In the research regime the money already committed to the tier, 26,347.13 crore BDT of committed project cost across the 44 University Grants Commission development lines of the FY26 revised programme [Planning Commission ADP 2025], converts into laboratories, doctoral output and accreditation, and the diaspora becomes an input rather than a loss. At FY26 the largest research oriented line in that programme has executed 0.40 percent of its committed cost [Planning Commission ADP 2025], which is why the credential regime is the default path and the research regime is a decision that has not yet been made.

The record: mass enrolment, thin research, an exodus that changed scale

The enrolment base grew 78.7 percent in five years. Enrolment across all tertiary programmes, universities and affiliated colleges together, rose from 2,068,355 in 2014 to 3,695,233 in 2019 [WB WDI 2026]. The census counts 3,946,284 people aged 20 to 24 attending tertiary in 2022 [BBS Census 2022]. The gross enrolment rate sat at 23.73 percent in 2024, men at 26.95 percent and women at 20.79 percent, a gender parity index of 0.771 [WB WDI 2026]. Chapter 09 carries the pipeline that feeds this tier: secondary completion of 94.30 percent at primary exit and 73.81 percent at lower secondary [WB WDI 2026], so the demand pressure on tertiary seats is demographic, not discretionary, and the parity gap at the top of the system, 0.771, is the residue of the parity advantage the same pipeline holds at the bottom.

The university tier is majority public, privately supplemented and thin at the faculty. The 2023 tier table counts 171 university institutions, 1,069,579 students of whom 37.47 percent are women, 32,917 teachers of whom 29.11 percent are women, and a student teacher ratio of 32.5 [BANBEIS 2023, tier table]. The 2022 census management table splits the tier: 53 public universities, 109 private and 2 international; public institutions held 720,379 students, 69.65 percent of tier enrolment by this chapter's arithmetic, with 40.51 percent women against 30.64 percent in private institutions [BANBEIS 2022, education census management table]. The public institutions run the fuller classrooms, a student teacher ratio of 47 against 20 in the private universities [BANBEIS 2022, education census management table]. The public network is still expanding from a small base: the University Grants Commission panel counts 6 public universities in 1985, 31 in 2010 and 49 in December 2022, with 15,658 teachers of whom 4,215, 26.92 percent by this chapter's arithmetic, were women [UGC 2022]. The panel's enrolment and ratio columns switch definition between years, printing 820,430 students for 2020 against 289,645 for 2021, so this chapter uses it only for the institution count and the teacher corps, which move on one definition, and cites the census for levels.

The largest segment of higher education is not in universities at all. The census counts 3,301 colleges with 4,388,005 students, 49.97 percent of them women, a tier more than four times university enrolment by this chapter's arithmetic on the same table [BANBEIS 2022, education census management table]. These colleges teach the pass degree programmes of one affiliating university, and their public segment runs 86 students per teacher, twice the public university ratio [BANBEIS 2022, education census management table]. The private share of all tertiary enrolment was 36.49 percent in 2019 [WB WDI 2026], the balance the credential machine needs from a private university sector created in the 1990s and unranked in any assessment this machine holds.

What the machine produces is measured in the graduate mix, and the mix is not a research economy's. Social sciences supplied 25.78 percent of 2019 graduates, science 7.97 percent, health 1.97 percent and engineering 1.34 percent [WB WDI 2026]. Doctoral attainment, the stock a research system draws on, was 0.04 percent of the population aged 25 and over in the 2023 print of a series too volatile across prints to trend [WB WDI 2026]. The research output record has one bright line and nothing behind it: scientific and technical journal articles rose from 1,426.96 in 2010 to a peak of 7,551.78 in 2021 and 7,056.17 in 2022, roughly 42 articles per million people by this chapter's arithmetic on the 2022 population of 169.38 million [WB WDI 2026]. Behind the article count sits almost no appropriable research: resident patent applications of 68 in 2019 and 74 in 2021, against 373 applications from nonresidents in 2021 [WB WDI 2026], intellectual property receipts of 1.3 million USD in 2024 [WB WDI 2026], and a Global Innovation Index score of 21.4, rank 106 of 133 in 2024 [WIPO GII 2024]. The national research and development expenditure share, the headline input every comparator reports, exists in no series in the sources this chapter draws on, so any figure for it is not established, with the Ministry of Science and Technology and the UNESCO institute the resolving sources.

The exodus record changed scale, not just direction. The Bangladesh born stock of the United States rose from 21,414 in 1990 to 261,496 in 2020, the United Kingdom stock from 102,380 to 241,799, Canada from 4,203 to 63,250 and Australia from 2,315 to 52,907, a four destination stock of 619,452 against 130,312 in 1990, by this chapter's arithmetic on the destination tables [UN DESA 2020]. These are the settlement destinations of the skilled channel, the counterpart of the Gulf employment corridor chapters 08 and 21 carry. The student flow that feeds the stock, the outbound internationally mobile count the UNESCO institute compiles for every comparator, is not held in the sources this chapter draws on, so any figure for students abroad is not established, with the UNESCO institute the resolving source. The best local evidence on the intention behind the flow is a July 2024 survey of 189 technology and STEM professionals: 86.24 percent agreed they prefer working abroad over domestic employment, 76.88 percent said they would return if domestic opportunities and compensation improved, and 97.87 percent rated government retention policy insufficient; the dataset is held here with its upstream publisher unidentified, so each figure could not be confirmed from the sources this chapter draws on.

Mechanism: an affiliation machine, a campus budget and an open exit door

The credential machine runs on affiliation and examination, and its economics explain the research deficit. The 109 private universities and the 3,301 affiliated colleges were the state's answer to demand the public sector could not seat, and both operate on a model that prices teaching and does not price research: colleges collect examination linked fees under an affiliating university that standardises syllabus and assessment, private universities collect tuition, and public universities ration seats through admission tests rather than expand faculty [BANBEIS 2022, education census management table]. Nothing in that revenue structure funds laboratories, doctoral stipends or the sabbatical and promotion incentives that produce the 42 articles per million people the record shows, and the graduate mix the structure produces, a quarter social scientists and 1.34 percent engineers, is the mix an examination system optimises for [WB WDI 2026]. The surplus it certifies into a thin labour market is the NEET stock chapter 08 measures at 15.43 percent of youth in 2024 [WB WDI 2026]: the credential machine converts demand for education into credentials faster than the economy converts credentials into wages.

The budget machine funds buildings, and its development programme reads as a found campus habit alongside an unstarted research agenda. The University Grants Commission's 44 development lines in the FY26 revised programme carry 3,285.75 crore BDT of allocation against 26,347.13 crore BDT of committed cost and 10,016.20 crore BDT of cumulative expenditure, 38.02 percent executed by this chapter's arithmetic [Planning Commission ADP 2025]. The committed cost sits overwhelmingly in campuses: Jagannath University's new campus at 1,746.06 crore BDT, 63.8 percent executed, Comilla University's development at 1,655.50 crore BDT, 87.0 percent executed, Netrokona University at 645.61 crore BDT, 96.5 percent executed, the Khulna University of Engineering and Technology expansion at 1,225.70 crore BDT, 53.6 percent executed, and Dhaka University's development line at 2,840.39 crore BDT of committed cost with zero cumulative expenditure [Planning Commission ADP 2025]. The two research oriented lines are the ones that have not moved: the Higher Education Acceleration and Transformation project, the tier's flagship quality and research instrument, carries 4,016.57 crore BDT of committed cost with 16.03 crore BDT of cumulative expenditure, 0.40 percent, and the computer and software engineering tertiary education project carries 1,219.80 crore BDT committed with 34.14 crore BDT cumulative, 2.80 percent [Planning Commission ADP 2025]. Meanwhile the found campus habit continues at the margin: the FY26 programme carries feasibility studies for a Kurigram Agricultural University at 4.87 crore BDT and a Maulana Bhashani Science and Technology University infrastructure survey at 2.85 crore BDT, both with zero cumulative expenditure [Planning Commission ADP 2025]. The recurrent side of the tier, the block grants the commission distributes to public universities, exists in no series in the sources this chapter draws on, so the public university financing total is not established, with the commission's yearbook the resolving source. The execution ratios are cumulative expenditure against committed cost at the FY26 programme cut, the caveat chapter 04 records that the fields do not reconcile across project life.

The governance machine rules by appointment, and the public university is the most consequential political infrastructure the state operates. The President is Chancellor of every public university, vice chancellor appointments route through the government of the day, and the quota dispute that fed the July 2024 movement began as a public university admissions question; the sequence and the current appointment rules after the 2024 transition are not documented in any instrument available to this chapter, and could not be confirmed, with the commission and the Ministry of Law the resolving sources. The state's quality instrument, accreditation, is in the same condition: no assessed result for any of the 109 private universities is verifiable from the sources this chapter draws on, and the accreditation council's operational status could not be confirmed, with the council the resolving source. A system that appointments, examines and builds, but neither funds research nor assesses quality, produces the governance cycle the public campuses replay roughly every decade.

The exit machine is the wage and institutional arbitrage the three other machines create. The settlement channel moves the top of the distribution permanently, 619,452 people across the four destinations in 2020 [UN DESA 2020]; the employment channel prices the mid distribution at the Gulf rate, with 24.82 percent of 2023 departures recorded as skilled, the series chapter 21 carries [BMET 2023]; and the student channel, unmeasured here, moves the undergraduates who become the settlement stock of the following decade. The state's own instrument for faculty development confirms the direction of flow: the commission's overseas Masters, M.Phil and PhD scholarship for university teachers, circularised under its 2025 policy with applications due 5 April 2026, finances public university teachers to take their doctorates at foreign universities [UGC 2026]. The doctorate the system needs is one the system itself does not yet produce at scale, and the state's answer is to send its faculty to the systems that do. The remittance counterweight is real, chapters 03 and 21 document it, but it prices labour migration, not the settlement channel that removes researchers permanently.

The decade ahead: five decisions decide which regime the tier lands in

The demand arithmetic eases before the decade ends. The cohort projections chapter 08 and chapter 09 carry on the UN medium variant [UN WPP 2024] imply that the tertiary age population stops growing and starts shrinking within the decade, so the enrolment rate the 2024 print holds at 23.73 percent [WB WDI 2026] will face a flat denominator on the UN's projected fertility path. The seat constraint that justified the found campus habit loosens while the quality deficit does not. A tier whose intake cohort shrinks can hold its budget flat and let per student resources rise, the dividend chapter 32 prices for the school tiers, or keep building; the research regime requires the first.

Five decisions carry the regime choice, each with a named author. The research funding decision, authored by the Finance Division with the Ministry of Science and Technology and the commission, is whether a measured national research expenditure line, funded recurrently rather than through development projects, appears in the budget; the revealing instrument is the Higher Education Acceleration and Transformation execution, 0.40 percent at FY26 [Planning Commission ADP 2025], because the project carries the laboratories and the doctoral financing the tier has never had. The accreditation decision, authored by the commission with the accreditation council, is whether the 109 private universities face assessed quality results a student or employer can read; the current status could not be confirmed, with the council the resolving source. The governance decision, authored by the Ministry of Education with the Chancellor's office, is whether vice chancellor appointment rules are settled on merit criteria after the 2024 transition; the reform proposals' status could not be confirmed, and the ministry would settle it. The faculty pipeline decision, authored by the commission with the universities, is whether doctoral production is domesticated before the current teacher corps of 32,917 retires, from a base of 0.04 percent doctoral attainment among adults [WB WDI 2026] and an overseas scholarship scheme that is the largest financed pipeline [UGC 2026]; the scheme's own evaluation of how many scholars return is not available to this chapter, and could not be confirmed, with the commission the resolving source. The talent conversion decision, authored jointly with the innovation system chapter 55 prices and the diaspora instruments chapter 60 carries, is whether the 619,452 person settlement stock [UN DESA 2020] and the 76.88 percent return willingness of the local survey print convert into faculty hires, adjunct appointments and funded collaborations, or stay remittances and conferences. No independent forecaster projects this tier's trajectory, so the chapter 15 scenarios carry the assumption: the reform scenario assumes the research funding and accreditation decisions are taken, the stall scenario assumes the campus budget continues.

Three risks print in project tables and index ranks, and the upside is already committed

Risks. First, a research regime lock out: if the Higher Education Acceleration and Transformation project stays below a quarter of committed cost through the decade and no recurrent research line appears in the budget, the article count will plateau the way the patent count did, and the tier will keep certifying 4 million students a system no innovation system employs; the revealing indicators are the project's cumulative expenditure line and the first official research expenditure measurement. Second, a governance repetition on the public campuses: the appointment cycle that produced the quota confrontation can re produce itself, and each closure costs semesters in a system where the public segment already runs 47 students per teacher [BANBEIS 2022, education census management table]; the revealing indicators are the Dhaka University development line, zero executed at FY26 [Planning Commission ADP 2025], and the session length statistics the commission's yearbook publishes, currently not established in the sources this chapter draws on. Third, an accelerated drain: if the settlement stock, 619,452 in 2020 [UN DESA 2020], keeps compounding through the next tabulation while the faculty pipeline stays outbound, the system loses its replacement candidates at both ends, graduates to the settlement channel and teachers to the overseas scholarship scheme; the revealing indicators are the next UN tabulation and the skilled share of departures in the manpower bureau record.

Upsides. First, the publication base is real and young: 7,056.17 articles in 2022 is 4.9 times the 2010 level of 1,426.96 [WB WDI 2026], a research community exists and demonstrably publishes, so a funded line converts an existing habit rather than creating one, and the patent and licensing counters the habit has never fed are the indicators to watch. Second, the money is already committed: 26,347.13 crore BDT of development cost across the commission's portfolio [Planning Commission ADP 2025] means the research regime is a reallocation and an execution question, not a new fiscal claim, and the campus lines that finish, Comilla at 87.0 percent and Netrokona at 96.5 percent, free allocation for the lines that have not started. Third, the diaspora is at a scale where conversion is measurable: a 619,452 person stock in the four settlement economies [UN DESA 2020], a survey print of 76.88 percent return willingness among technology professionals, and the commission's own outbound scholars form the pipeline the faculty decision can hire from; the revealing indicator is the foreign trained and returnee share of new public university recruitment, currently unmeasured and unconfirmed, with the commission the resolving source.

What to watch: five indicators whose thresholds mark the regime

1. Higher Education Acceleration and Transformation execution. Current value 0.40 percent of committed cost, 16.03 of 4,016.57 crore BDT, at the FY26 programme cut [Planning Commission ADP 2025]. Threshold: cumulative expenditure above half of committed cost by FY30 is the pace the reform scenario assumes; stagnation below a quarter marks the research lock out this chapter's stall path assumes.
2. Scientific journal articles. Current value 7,056.17 in 2022, against a 7,551.78 peak in 2021 [WB WDI 2026]. Threshold: sustained prints above 12,000 a year alongside a first official research expenditure measurement signal the research regime; a fall below the 2021 peak confirms the plateau.
3. Resident patent applications. Current value 74 in 2021, against 68 in 2019 [WB WDI 2026]. Threshold: prints above 200 a year signal that research is reaching appropriation; a return below 50 confirms the system produces papers it never files.
4. Settlement diaspora stock in the United States, the United Kingdom, Canada and Australia. Current value 619,452 in 2020 [UN DESA 2020]. Threshold: the next tabulation above 900,000 confirms the drain regime the credential path implies; a flat stock alongside measured returnee recruitment marks the conversion regime.
5. Tertiary gross enrolment rate and its parity gap. Current values 23.73 percent in 2024 and a gender parity index of 0.771 [WB WDI 2026]. Threshold: a rate above 30 percent with research spending still unmeasured confirms the credential regime the stall scenario assumes; a rate rising while the parity index passes 0.85 and the research line appears signals the quality turn.

Sources used

[WB WDI 2026] World Bank World Development Indicators via bdpolicy parquets wb_full_bd and wb_edstats_bd: tertiary enrolment rates by sex and gender parity index SE.TER.ENRR, SE.TER.ENRR.MA, SE.TER.ENRR.FE, SE.ENR.TERT.FM.ZS, tertiary enrolment counts SE.TER.ENRL, private tertiary share SE.TER.PRIV.ZS, graduate field shares SE.TER.GRAD.SS.ZS, SE.TER.GRAD.SC.ZS, SE.TER.GRAD.EN.ZS, SE.TER.GRAD.HL.ZS, doctoral attainment SE.TER.CUAT.DO.ZS, primary and lower secondary completion SE.PRM.CMPT.ZS, SE.SEC.CMPT.LO.ZS, NEET SL.UEM.NEET.ZS, journal articles IP.JRN.ARTC.SC, patents IP.PAT.RESD and IP.PAT.NRES, intellectual property receipts BX.GSR.ROYL.CD, population SP.POP.TOTL, values in data/33-higher-education-research-brain-drain.csv. [BANBEIS 2023] Bangladesh Bureau of Educational Information and Statistics via bdpolicy parquet banbeis_institutions_students_by_tier, university tier institutions, students, teachers by sex and student teacher ratio, 2023. [BANBEIS 2022] Bangladesh Bureau of Educational Information and Statistics via bdpolicy parquet banbeis_education_census_2022, university, college, public, private and international management rows of the education census management table, 2022. [BBS Census 2022] Bangladesh Bureau of Statistics, Population and Housing Census 2022, persons aged 20 to 24 attending tertiary education, via bdpolicy.db series census2022_tertiary_students_20to24_nos. [UGC 2022] University Grants Commission public university panel via bdpolicy lake parquet ugc_public_universities_panel, institution count trajectory 1985 to 2022 and teacher corps; the panel's enrolment and ratio columns switch definition between years and are not used for level claims. [UGC 2026] University Grants Commission circular for the UGC Overseas Masters/M.Phil/PhD Scholarship for University Teachers under the 2025 policy, application deadline 5 April 2026, via ocr_text/ugc_extra. [WIPO GII 2024] WIPO Global Innovation Index via bdpolicy parquet wipo_gii_bd, Bangladesh score 21.4 and rank 106 of 133, 2024, as cited in chapter 13. [UN DESA 2020] UN Department of Economic and Social Affairs, International Migrant Stock 2020 origin destination tables via bdpolicy parquet humanitarian/undesa_migrant_stock_bd, Bangladesh born stock by destination, 1990 to 2020. [Planning Commission ADP 2025] Programming Division, Planning Commission, RADP FY2025-26 project table via bdpolicy.db adp_projects, University Grants Commission agency portfolio of 44 projects and named project lines, allocations, committed costs and cumulative expenditure in lakh taka converted to crore BDT, the caveat chapter 04 records on the RADP fields applies. [UN WPP 2024] UN World Population Prospects 2024 via bdpolicy parquet, series un_wpp2024_bd medium variant, the cohort projection chapters 08 and 09 carry. [BMET 2023] Bureau of Manpower, Employment and Training, overseas employment statistics via bdpolicy.db, series bmet_overseas_employment_skilled and bmet_overseas_employment_total, calendar year 2023, the series chapters 21 and 32 carry.

---

Verified line by line against primary sources: 89 claims checked, 3 corrected.

Previous: 32 Education, TVET and skills

Next: 34 Poverty, inequality and social protection

All sixty chapters and the ten narrative chapters.