Bangladesh spends 2% of GDP on education — and half its secondary teachers are untrained

Introduction

Bangladesh has put almost every child in a classroom, but the resources behind that classroom have barely moved. General government spending on education was 2.03% of GDP in 2024, almost exactly where it stood in 1999 (2.13%).¹ In 2022, 51.2% of children of end-of-primary age were “learning-poor”: out of school, or unable to read and understand a short, age-appropriate passage.² And in 2024 only 55.0% of secondary teachers had the minimum required training, down from 65.0% a year earlier.³

2.03% of GDP

Government expenditure on education, Bangladesh, 2024 (UNESCO Institute for Statistics).

This article sets out what the official data show about how much Bangladesh invests in schooling, who is teaching, and how both compare with neighbouring countries. All figures are UNESCO Institute for Statistics (UIS) series published through the World Bank; they are administrative returns and modelled estimates, and definitions differ somewhat between countries, so comparisons indicate scale rather than a precise ranking.

A small share of income, a middling share of the budget

Among eleven Asian countries with recent data, Bangladesh’s 2.03% of GDP ranks eighth. Bhutan spends 5.85% (2023), India 4.10% (2022), Nepal 3.69% (2024) and Viet Nam 2.89% (2022). Only Pakistan (1.95%), Sri Lanka (1.83%) and Indonesia (1.28%) spend less.¹ Over the last 25 years Bangladesh’s share has stayed between 1.87% and 2.20% in every year with data except 2016 (1.28%) and 2019 (1.13%), two isolated values that may reflect reporting differences.¹

The picture is different when spending is measured against the government’s own budget. Education took 11.88% of government expenditure in 2025, close to Nepal (10.75%), Indonesia (10.56%) and Cambodia (11.55%), and above Pakistan (9.76%, 2023) and Sri Lanka (7.22%, 2024). It was lower than India (14.16%, 2022) and Viet Nam (15.45%, 2022).⁴ But the share has fallen from 20.49% in 2000.⁴

Taken together, the two measures say that schools do not get an unusually small slice of the budget by regional standards. The constraint is the size of the budget relative to the economy, which these series do not measure. A level share of GDP also means education spending has grown only as fast as the economy, not faster.

Primary teachers are catching up; secondary teachers are not

The share of primary teachers with the minimum required training was 77.9% in 2024, up from 48.5% in 2018.⁵ Secondary teachers tell a different story: the trained share rose from 53.6% in 2011 to 66.5% in 2017, held between 61.7% and 65.0% from 2020 to 2023, and then fell to 55.0% in 2024.³ The series has gaps, and the source does not say whether the 2024 drop is a real change in the workforce or a change in how teachers are counted.

Against neighbours the secondary gap is wide. Nepal reports 97.4% of secondary teachers trained (2025), India 92.2% (2025), Viet Nam 91.4% (2022), Sri Lanka 80.4% (2023) and Pakistan 68.9% (2024). Bangladesh, at 55.0%, is below all of them; only Indonesia reports less, at 36.6% (2025).³ At primary level Bangladesh (77.9%) is ahead of Pakistan (75.0%) and Indonesia (34.7%) but behind India (91.4%), Nepal (95.8%), Viet Nam (89.3%) and Sri Lanka (86.4%).⁵

“Trained” means the minimum organised pre- or in-service training a country itself requires, so a high share in one country does not guarantee quality, and Indonesia’s low figure shows how much definitions can matter.

Class sizes: old data, mixed signals

The latest pupil-teacher ratios available are for 2017–2018, so they describe the system before the recent changes in teacher training. The primary ratio fell from 43.0 pupils per teacher in 2010 to 30.1 in 2018, while the secondary ratio rose from 23.9 in 2005 to 35.1 in 2018, the highest among the thirteen Asian economies checked (data from 2017–2019).⁶⁷ Because the data are seven years old, they are best read as a prompt to find newer administrative figures, not as a description of today’s classrooms.

What the data cannot say

None of these series shows that more spending or more training would raise learning. The learning-poverty figure is a single 2022 data point, so it gives a level, not a trend.² Bangladesh also has a large non-government school sector and madrasa system whose resources these national aggregates may capture only partly. Causes of weak learning, such as curriculum, teaching practice, language, household circumstances and school management, are questions for assessment and evaluation studies, not for these indicators.

What the numbers do establish is an input gap that is concentrated and measurable: a low share of GDP, and a secondary teaching workforce that is less trained than in any of the comparable economies shown.

What it will take

• Set and publish a path for education spending. A stated target for the share of GDP, reported against UIS data, would turn a flat line into a policy choice. Spending has stayed near 2% for a quarter of a century.
• Repair the secondary teacher pipeline. Primary training coverage rose by about 29 percentage points between 2018 and 2024; secondary coverage needs a comparable push, starting with checking whether the 2024 fall is real.
• Measure learning every few years. One reading estimate for 2022 cannot show whether schools are improving. Regular, comparable assessments would let policymakers link training and spending to results.
• Publish current class-size data. Pupil-teacher ratios in the international series stop at 2018; up-to-date administrative figures would show whether secondary classes are still growing.
• Report on the whole system. Include non-government schools and madrasas in the data used to judge spending and teacher quality.