Executive finding
The Closing Ladder: measured 5-year entry probabilities into new export industries from 30 years of BACI HS92 data, Bangladesh's 627 entries since 1996, and the open paths ranked by conversion odds.
Measured entry probabilities from 30 years of trade data, and what they say about the LDC deferral years
BDPolicyLab · 2026-07-03
From 30 years of harmonized trade data (BACI HS92, 1995-2024, 134 countries), we measure how often a country not yet exporting a product competitively enters it within five years, given how close the product sits to its existing export basket. The answer has changed. Across all countries, the conversion rate from a well-positioned basket fell from 5.3% in the 1996-2003 window cohorts to 3.7% in 2011-2018, a 31% decline. For countries in Bangladesh's income group the best rungs roughly halved: the ninth density decile converted at 14.0% in 1996-2003 and 7.1% in 2011-2018. Bangladesh's own record shows it: 627 competitive entries since 1996, but the flow has collapsed from 105 in the windows closing in 2001 to 9 in 2023, and what does enter is still overwhelmingly textile-adjacent. The classic first rung has closed behind it: conversion into cotton T-shirts (HS 610910), the archetypal entry product, fell from 1.5% to 0.4% between the cohorts. The open paths that remain are mapped below, ranked by measured entry probability. With the CDP recommending deferral of LDC graduation to November 2029, the extra three years are exactly the window this data says is narrowing.
Key findings
- The global development ladder is closing: same basket position, 31% lower conversion than in the late 1990s. Across 134 countries, the probability of competitive entry (revealed comparative advantage crossing from below 0.5 to at least 1) within five years from a positioned basket (top three proximity-density deciles) fell from 5.33% (65,998 entries / 1,238,412 at-risk observations, 1996-2003 window starts) to 3.69% (46,538 / 1,260,737, 2011-2018). The decline holds at every density level except products unrelated to the existing basket, which never converted in any era. Over the same cohorts, China's average share of world exports per product rose from 8.6% to 18.4%.
- For Bangladesh's income group, the best rungs halved after 2004 (14.0% to 7.1%). Among the lowest income quartile of the panel (GDP per capita quartiles among panel countries; Bangladesh at $2,593 in 2024 sits in quartile 1), the ninth density decile, the sweet spot where products are close to the basket but not yet exported, converted at 13.99% in the 1996-2003 cohorts, 6.63% in 2004-2010, and 7.10% in 2011-2018. The step down coincides with China's post-WTO-accession expansion; the rate has not recovered since.
- Bangladesh made 627 competitive entries since 1996, but the flow fell from 105 a year to single digits. Counting every product where Bangladesh's revealed comparative advantage crossed from clearly absent to competitive within a five-year window: 105 entries in windows closing in 2001, 9 in 2023. Recent entries remain concentrated in textile-adjacent headings (woven and knitted fabrics, shawls and scarves, packing sacks) plus scattered items such as float glass, a sign that diversification is deepening within the cluster rather than branching out of it.
- Apparel entry is a mixed door: the knit T-shirt rung has closed, while cotton woven trousers still convert. The closing is real but heterogeneous inside apparel itself. Cotton T-shirts (HS 610910), the archetypal first export of a low-income entrant, converted at 1.5% of positioned countries in 1996-2003 and only 0.4% in 2011-2018. Women's cotton woven trousers (HS 620462), by contrast, converted at 1.9% then and 3.8% now: woven rungs remain climbable, which is exactly where new competitors keep arriving. The steepest textile closures are in specialty headings such as women's wool blouses (HS 620620, 20.2% to 3.7%). For Bangladesh this cuts both ways: its knit incumbency is protected by a door that has shut, while its woven position faces entrants through the door still open.
- The measured frontier: 15 open paths, led by rubber with a $11.5 billion world market. Ranking every product Bangladesh does not yet export competitively by its estimated five-year entry probability (its current basket position read against the 2011-2018 conversion rates for its income group), the top open paths cluster in agro-processing and natural-rubber value chains: technically specified natural rubber ($11.5 billion world market), cashew nuts ($8.3 billion), plants used in perfumery and pharmacy ($4.0 billion), plus further textile-adjacent headings. Each carries roughly a 7% measured five-year entry probability, several times the all-product base rate. These are probabilities, not recommendations: they say where the capability distance is short and history says entry happens, not where margins are best.
Bangladesh asked the UN to defer LDC graduation from November 2026 to November 2029, and on 1 June 2026 the Committee for Development Policy recommended granting it, with the caveat that the extra time must not delay reforms. This brief quantifies what the extra time is for. The ladder that carried Bangladesh from jute to apparel converts new positions into new industries at about 31% below the rate of the cohorts in which the apparel boom began, and at roughly half that rate for countries at Bangladesh's income level. Waiting costs measurably more than it did a decade ago.
What we measured
A country "enters" a product when its revealed comparative advantage crosses from clearly absent (below 0.5) to competitive (at least 1.0) within a five-year window, on three-year-averaged, classification-consistent trade data (BACI HS92, 1995-2024). Basket position is Hidalgo-style proximity density. Grouping the window starts into three cohorts (1996-2003, 2004-2010, 2011-2018) makes the eras directly comparable: identical density deciles mean identical absolute positions. Estimates are descriptive conversion rates with tight confidence intervals (roughly 400,000 observations per decile-era cell), not causal effects. Full method and interactive atlas: tradeweave.org/ladder; the Bangladesh page there lists every entry and open path.
What it says for policy
The window logic is real and it is measured. The probability that a well-positioned product converts into a competitive export industry has fallen about 31% since the cohorts in which Bangladesh's apparel complex was built, and roughly 50% for countries at Bangladesh's income level. Every year of delayed diversification is a draw from a shrinking distribution.
Deepening is happening; branching is not. Six of Bangladesh's last twelve measured entries are woven or knitted textile headings. That is the product space working as designed: nearby products convert. But the frontier list shows the nearest non-apparel rungs, agro-processing, natural rubber, botanical inputs to pharma and cosmetics, and these need different public inputs: phytosanitary certification capacity, cold chains, land-use decisions, and testing labs rather than bonded warehouses and back-to-back LCs.
Incumbency is now an asset to defend. The same closing that blocks new entrants protects Bangladesh's apparel position. Compliance, energy cost, and logistics failures that push buyers elsewhere would hand over a position that, at today's conversion rates, could not be rebuilt.
What this data cannot say. Entry probabilities are based on revealed comparative advantage in goods trade; they do not capture services exports, margins, or job quality, and a measured probability is not a guarantee. The frontier list is a screening tool for where capability distance is short, to be combined with cost, demand, and feasibility work, not a substitute for it.
Data and methodology
Directed, era-resolved product space estimated from CEPII BACI HS92 V202601 (time-consistent HS6 classification, 1995-2024). Exports are 3-year centered means; presence = Balassa RCA >= 1, absence (at risk of entry) = RCA < 0.5; windows run t to t+5 with starts 1996-2018 grouped into cohorts 1996-2003, 2004-2010, 2011-2018. Panel: 134 countries with population >= 1 million and exports >= $1 billion reporting in all 30 years. Density is the Hidalgo et al. (2007) proximity-weighted share of related products already exported, with proximity estimated within each era and decile cutoffs pooled across eras so identical deciles are identical absolute positions. Income quartiles use WDI GDP per capita (current US$) among panel countries, not World Bank income groups. Entry probabilities for the frontier are the 2011-2018 conversion rates for Bangladesh's income quartile at each product's current density decile; world market sizes are 2022-2024 mean world exports. Estimates are descriptive conversion rates, not causal effects. Pipeline and interactive atlas: tradeweave.org/ladder. Extract regenerated via scripts/extract_ladder_bgd.py; this brief interpolates every figure from that file.
Cite this
BDPolicyLab Research. (2026). The Closing Ladder: Bangladesh's Diversification Window. BDPolicyLab. https://bdpolicylab.com/publications/the-closing-ladder-bangladesh-s-diversification-window
Method and source
Source: Primary sources cited at point of use in the publicationAs of 3 Jul 2026