Convergence Without Absorption: Can emerging economies catch up when the escalator that carried workers has narrowed?

Hunter Hughes

Abstract

The question of whether emerging economies can still converge is usually asked as though convergence were one thing. It is at least three: output per capita relative to advanced economies, output per worker, and the number of people employed at rising productivity. For the four decades in which export manufacturing was the standard development pathway, those three moved together, because factory work was simultaneously more productive than what it replaced, tradable, and capable of absorbing large numbers of workers with limited formal education. Nothing that is now proposed in manufacturing's place has all four of those properties at once.

The evidence that the pathway has narrowed is established. Dani Rodrik's work on premature deindustrialization found that early industrialisers saw manufacturing employment peak at 30 to 37 per cent of the workforce at incomes of roughly US$11,000 to US$14,000 per head, while manufacturing began to shrink in Brazil at about US$5,000, in China at about US$3,000 and in India at about US$2,000. The World Bank's Global Economic Prospects for January 2026 reports that virtually half of all developing economies have failed since 2019 to narrow the income gap with advanced economies, with developing per capita income at about 12 per cent of the advanced level and per capita growth running roughly a percentage point below its 2000–2019 average.

The replacements are real and growing, and they are narrow. Digitally deliverable services now account for 56 per cent of global services exports; developing economies exported an estimated US$1.3 trillion of them and grew at 12 per cent against 9 per cent in developed economies. But the gains concentrate: in least-developed countries digitally deliverable services are only 16 to 20 per cent of services exports, about a third of the global average. Green industry, mineral processing and modern agro-processing share the same profile — tradable, productive, capital- or skill-intensive, and thin on employment relative to the factory floor they are replacing. Against that, the African Development Bank estimates that 10 to 12 million young Africans enter the labour market each year into an economy generating roughly three million formal jobs, and informal employment in Africa stood at about 85 per cent in 2024 against a world average of 58 per cent, barely changed in two decades.

The report's conclusion is that convergence is still available on output and largely unavailable on absorption, and that treating these as one objective produces policy that succeeds on the measure that is easiest to report. It proposes a two-track framing: an export track, which earns foreign exchange and builds capability through the narrow escalators and should be pursued without pretending it will employ the labour force; and an absorption track, aimed at productivity in the non-tradable sectors where most people already work — construction, retail and distribution, care, transport and food systems — where a one-off productivity gain reaches more workers than any plausible expansion of digital services exports. Both tracks depend on the same enabling layer of reliable power, connectivity and skills, which is where the polycrisis bites and where the case for investment is strongest.

Executive Summary

The honest answer to the title's question is: on one measure yes, on another no — and the two used to be the same measure.

FINDING 01

Convergence has split in three

Output per capita, output per worker, and employment at rising productivity moved together under the manufacturing escalator. They no longer do, and most discussion reports only the first.

FINDING 02

The factory was uniquely absorptive

Export manufacturing was productive, tradable, and able to employ millions of people without advanced schooling. Every proposed replacement has the first two properties and not the third.

FINDING 03

Two tracks, not one strategy

Pursue the narrow escalators for foreign exchange and capability — and pursue productivity where people already work, because that is where absorption has to come from. Conflating the two hides the failure.

Convergence is not going well. The World Bank's Global Economic Prospects for January 2026 reports that virtually half of all developing economies have failed since 2019 to narrow the income gap with advanced economies; per capita income in developing economies sits at roughly 12 per cent of the advanced level, and per capita growth is running about a percentage point below its 2000–2019 average.

The standard diagnosis is that the pathway has closed: manufacturing no longer absorbs labour at the incomes it once did. That diagnosis is right, and it is incomplete. What matters for policy is not that one escalator narrowed but that the replacements have a different shape — and the shape determines which kind of convergence is still on offer.

12%

Developing-economy per capita income as a share of the advanced-economy level

World Bank, GEP January 2026

$2–5k

Incomes at which manufacturing began to shrink in India, China and Brazil — against $11–14k for early industrialisers

Rodrik, premature deindustrialization

10–12m

Young Africans entering the labour market each year, against roughly 3 million formal jobs

African Development Bank

85%

Informal share of employment in Africa in 2024, against a world average of 58%

ILO modelled estimates

Section 1 separates the three convergences. Section 2 identifies the property that made manufacturing unique. Section 3 examines the replacements and what they absorb. Section 4 does the arithmetic against labour-force growth, Section 5 proposes a two-track response, and Section 6 asks what the polycrisis specifically changes. Sections 7 and 8 give recommendations and conclusions.

A country can raise its output per capita and its productivity while the number of people in decent work barely moves. That used to be impossible. It is now the central risk. The absorption problem

1. Which Convergence?

Three measures, one word. Under the old pathway they were interchangeable; today the choice of measure determines the answer.

Output convergence. GDP per capita approaching the advanced-economy level. This is the measure in almost every headline, and the one most responsive to a commodity boom, a capital inflow or a single successful export sector.
Productivity convergence. Output per worker approaching the frontier. A country can achieve this in an enclave — a port, a processing plant, a services campus — without the rest of the economy moving at all.
Absorption convergence. The share of the workforce employed at frontier-adjacent productivity. This is the one that changes household incomes at scale, and the only one that produces the social transformation people mean when they say a country developed.

For the four decades in which export manufacturing was the standard route, these were not really three measures. A factory that raised output per worker did so by hiring workers, so productivity and absorption moved together, and the wage bill carried output per capita with them. The escalator's defining feature was that it delivered all three from one investment.

Figure 1 — Where developing economies actually stand

Per capita income in developing economies as a share of the advanced-economy level, and per capita growth in 2026 against the 2000–2019 average. From the World Bank's Global Economic Prospects, January 2026. The two bars are different units and are shown together only to date the slowdown.

2. What the Escalator Actually Did

Export manufacturing was not merely productive. It was productive, tradable, and open to workers with limited schooling — and that combination is rare.

Four properties made the factory the engine of twentieth-century convergence. It was more productive than the smallholder agriculture or petty trade it drew people from. It was tradable, so demand was not limited by domestic purchasing power. It had a low entry requirement, absorbing workers whose formal education had ended early. And it was scalable in employment: expanding output meant hiring, not just installing.

Rodrik's work on premature deindustrialization documents how far that has narrowed. Early industrialisers — the United Kingdom, Germany, the United States — saw manufacturing employment peak at 30 to 37 per cent of the workforce at incomes of roughly US$11,000 to US$14,000 per head. In later developers, manufacturing began to shrink at a fraction of that income: about US$5,000 in Brazil, US$3,000 in China and US$2,000 in India. The hump-shaped relationship between industrialisation and income has moved down and towards the origin — countries are running out of industrialisation before they run out of poverty.

Figure 2 — The income at which manufacturing stopped growing

Per capita income (constant 1990 dollars) at which manufacturing employment peaked or began to shrink. Early industrialisers are shown as the US$11,000–14,000 range at which their manufacturing employment peaked, at 30–37 per cent of the workforce. From Rodrik.

The mechanism behind the narrowing is not mysterious: automation raised the capital intensity of manufacturing, and concentration in a few very large producers raised the scale needed to compete. An earlier essay from this house set out that argument and the case for alternative pathways — modern agro-processing, logistics, digital services — in detail. This report takes that as its starting point and asks the question it leaves open: what those pathways absorb.

3. The Replacements, and What They Employ

Every proposed successor to the factory is tradable and productive. None of them is tradable, productive, low-entry and employment-scalable at once.

3.1 Digitally deliverable services

The fastest-growing genuine escalator is services delivered across borders. UNCTAD reports that digitally deliverable services now make up 56 per cent of global services exports; developing economies exported an estimated US$1.3 trillion of them, growing at 12 per cent against 9 per cent for developed economies, whose exports came to about US$4.1 trillion.

Two qualifications matter more than the growth rate. The first is concentration: in least-developed countries, digitally deliverable services are only 16 to 20 per cent of services exports — around a third of the global average — and the gap between a few successful developing exporters and the rest has widened. The second is the entry requirement. The sector recruits from the tail of the education distribution, not the middle of it, which is the opposite of the factory's profile.

Figure 3 — Digitally deliverable services: growing, and concentrated

Digitally deliverable services exports by group, US$ trillions, with annual growth rates noted in the text. From UNCTAD. The developing-economy total is dominated by a small number of exporters.

3.2 The others

Green industry and mineral processing are capital-intensive by construction: a smelter or an electrolyser is a large investment operated by a modest workforce, and the strategy's logic — examined in an earlier report — is to capture value rather than employment. Modern agro-processing and horticulture absorb more labour and are constrained by logistics, standards compliance and perishability. Tourism absorbs a great deal of labour at low productivity and is highly exposed to exactly the shocks a polycrisis produces.

3.3 Where the old escalator still works

The argument of this report would be too neat if manufacturing had simply stopped absorbing labour anywhere, and it has not. Bangladesh's ready-made garment sector employs on the order of four million people, a majority of them women, and earned about US$38.5 billion in 2024 — roughly four-fifths of the country's export earnings. Vietnam's manufacturing workforce reached about 12 million people, some 23 per cent of total employment, in 2023, with more than 4.1 million employed in FDI projects across its industrial zones.

Two things follow, and they cut in opposite directions. The encouraging one is that the escalator is narrowed rather than closed: where wages, logistics, power and trade access line up, mass manufacturing employment is still achievable this decade, and it remains the fastest route to absorption anyone has found. The sobering one is what those two cases required — a currency and wage position competitive against incumbents, reliable power and ports, duty-free access to large markets, and a buyer base willing to relocate orders. That is a demanding list, and it is getting longer as compliance requirements are added to it.

The honest reading is therefore conditional rather than fatalistic. Countries that can assemble those conditions should pursue manufacturing employment aggressively, because nothing else absorbs at that rate. Countries that cannot — because power is unreliable, the coast is distant, the market access is absent or the wage gap has closed — should stop treating it as the plan and start building the two tracks in Section 5.

Table 1 — The escalators compared

Escalator	Productivity	Employment absorption	Entry requirement	Tradable
Export manufacturing (the historical case)	High	High	Low — basic literacy and training	Yes
Digitally deliverable services	High	Low	High — tertiary or specialised skills, English	Yes
Green industry and mineral processing	High	Low	Mixed — technical trades, few of them	Yes
Modern agro-processing and horticulture	Medium	Medium	Low	Yes, subject to standards and logistics
Tourism	Low	High	Low	Yes, and shock-exposed
Non-tradable services (where most people work)	Low	High	Low	No

Assessments are the author's, applied to the evidence cited in this section, and are directional rather than measured; cross-country employment data for tradable services are not comparable enough for a like-for-like ranking. The point of the table is the empty cell: no row after the first combines high productivity, high absorption and a low entry requirement.

4. The Arithmetic

The gap between what the new escalators employ and what the labour force supplies is not a rounding error. It is the largest number in development.

The African Development Bank estimates that 10 to 12 million young Africans enter the labour market every year, against an economy generating roughly three million formal jobs. The residual does not disappear; it goes into informality. ILO modelled estimates put informal employment in Africa at about 85 per cent of total employment in 2024, against a world average of 58 per cent — and the African share has barely moved in two decades, with published estimates sitting in the 83 to 85 per cent range since 2005 depending on vintage and coverage.

Figure 4 — Entrants and formal jobs

Millions of young people entering Africa's labour market each year against formal jobs created annually, per African Development Bank estimates. The entrant figure is published as a range and plotted as one.

Figure 5 — Informality, Africa against the world

Informal employment as a share of total employment. From ILO modelled estimates for 2024. The 2005 comparison is omitted from the chart because published figures for that year differ by vintage; all of them place the share within a couple of points of today’s, which is the relevant fact.

Set that against the escalators in Section 3. Suppose digital services exports from a mid-sized African economy tripled over a decade — an outstanding result by any standard. The employment created would be counted in tens of thousands. The annual entrant flow for the continent is counted in millions. No plausible expansion of the tradable escalators closes this gap, and a strategy that implies otherwise is not a strategy but a hope.

Why this is not an argument against the export track

Tradable sectors earn the foreign exchange that pays for imported capital goods, medicines and food; they build the technical capability that upgrades everything else; and they discipline quality in a way domestic markets rarely do. The export track is necessary. The error is to expect it to solve employment, and then to treat its success as evidence that the employment problem is being addressed.

5. Two Tracks

Separate the objectives and each becomes tractable. Combine them and the easier one is reported as progress on both.

Figure 6 — Three convergences, two tracks

A conceptual schematic of the argument in Sections 1 to 5, not a quantitative model. The dashed arrows indicate that both tracks depend on the same enabling investments.

5.1 The export track

Pursue the narrow escalators on their merits: foreign exchange, technical capability, quality discipline and tax base. Concentrate on the ones where the country has a genuine endowment — a clean power resource, a mineral position, a time zone and language advantage, a growing season — rather than on whichever sector is fashionable. And measure them by earnings and capability, not by employment, so that success is not mistaken for something it is not.

5.2 The absorption track

Most people in low-income economies already work, usually informally and at low productivity, in construction, retail and distribution, transport, care and food systems. These are non-tradable, which is why they are ignored by export-led strategy — and it is also why a productivity gain in them reaches an enormous number of workers at once. A distribution system that halves spoilage, a construction sector that adopts better methods, a transport network that cuts journey times: each raises real incomes for millions of people who will never be employed in a services export firm.

There is a reason this track is neglected, and it is not that anyone disputes the arithmetic. Non-tradable productivity gains are diffuse: they show up as slightly cheaper food, slightly shorter journeys and slightly higher margins for hundreds of thousands of small firms, none of which is attributable to a project or photographable at a ribbon-cutting. Export projects, by contrast, come with a named investor, a groundbreaking date and a job number in the press release. The measurement asymmetry, not the economics, is what keeps the money on one side.

The instruments are unglamorous and well known: land and property registration, municipal services, cold chains, road maintenance, skills certification, access to working capital, and the removal of regulatory penalties on formalisation. None produces a ribbon-cutting. Together they determine whether output convergence turns into living-standard convergence.

5.3 The shared layer

Both tracks run on the same foundations: reliable electricity, connectivity, basic and technical skills, and functioning logistics. An earlier report showed how reliability rather than price is the binding constraint for the industries that could employ people, and how outages cost firms in the countries with the weakest grids more than a tenth of their sales. That layer is the highest-return investment in this report precisely because it is the only item that serves both tracks.

6. What the Polycrisis Specifically Changes

Not the direction of the argument, but the cost and the timing of everything in it.

Three effects are worth separating from the general gloom.

It raises the price of the enabling layer. Capital costs more in vulnerable economies, climate risk is priced into sovereign borrowing, and emergency spending crowds out the grid, the road and the school. The compounding here is unhelpfully precise: the countries that most need the shared layer are the ones being charged the most to build it.

It shortens the windows. As an earlier report argued, crises force replacement decisions and the choice falls to whatever is available, financeable and permitted in the moment. That applies to skills and connectivity as much as to power: a disruption is an opportunity to re-equip only if the alternative is already in place.

It changes what the export track has to satisfy. Border carbon rules, supply-chain due diligence and buyer standards add compliance requirements that are, in effect, a second entry gate — and one that rewards countries with clean power and functioning measurement systems. That is a constraint for some and an opening for others, but it is not neutral.

There is also a fourth effect worth naming, because it bears directly on Section 3.3. The conditions that let Bangladesh and Vietnam industrialise included duty-free or preferential access to large markets. As trade policy fragments and preference schemes are renegotiated, that access becomes less predictable — which raises the risk premium on precisely the investments that would absorb the most labour. A garment buyer deciding where to place a five-year order is pricing tariff risk alongside wages.

None of this alters the two-track conclusion. It raises the return to the shared layer, shortens the time available to build it, and makes the difference between the two kinds of convergence wider than it would otherwise have been.

7. Recommendations

Separated by track, because that is the point.

7.1 Measure all three convergences

Publish absorption alongside output. A development plan should report the share of the workforce in employment above a stated productivity threshold, not only GDP per capita and export growth.
Stop crediting export sectors with employment they do not create. Job figures attached to special economic zones and services campuses should be audited against payroll data, not projected from investment value.

7.2 Run the export track honestly

Select on endowment, not fashion. Clean power, minerals, growing seasons, language and time zone are real advantages; a national ambition to host data centres without the electricity to run them is not.
Build the compliance capability early — emissions measurement, standards certification, traceability — because market access increasingly depends on it.

7.3 Take the absorption track seriously

Target productivity in non-tradables: distribution and cold chains, construction methods and materials, urban transport, care services. These reach more workers per unit of public spending than any tradable sector can.
Remove the penalties on formalisation — registration cost, tax complexity, arbitrary enforcement — so that firms can grow without becoming targets.
Treat urban services as economic policy. Reliable water, waste, transport and power in the cities where informal work concentrates raise the productivity of that work directly.

7.4 Development partners

Fund the shared layer first. Power reliability, connectivity and skills serve both tracks and are chronically underfunded relative to project-level interventions.
Resist the enclave temptation. A successful export enclave is easier to finance, evaluate and photograph than a distribution system; the evaluation framework should not reward that asymmetry.

8. Conclusions

Convergence is still possible. Convergence of the kind that transformed societies is not, on the escalators currently available.

The question in this report's title has a two-part answer. On output, emerging economies can still converge — some are, and the routes are known: tradable services, processed minerals, green industry, high-value agriculture. On absorption, the arithmetic does not work. Ten to twelve million young Africans enter the labour market each year against roughly three million formal jobs, informality has barely moved in twenty years, and no realistic expansion of the tradable escalators changes those numbers by an order of magnitude.

What made manufacturing exceptional was never its productivity alone. It was that a single investment raised productivity, was tradable, and hired large numbers of people whose schooling had ended early. Rodrik's finding — peaks at 30 to 37 per cent of the workforce in the early industrialisers, against manufacturing shrinking at US$2,000 a head in India and US$3,000 in China — is a statement about that combination disappearing, not about factories disappearing.

The useful response is to stop asking one strategy to do two jobs. Run the export track for foreign exchange and capability, and judge it on those. Run an absorption track aimed at the productivity of the sectors where people already work, and judge it on employment and household incomes. Build the shared layer of power, connectivity, skills and logistics that both require, which the polycrisis is making more expensive and more urgent at the same time.

The alternative is the pattern already visible in the data: rising output, a handful of impressive export sectors, and an employment structure that looks much as it did twenty years ago. That is convergence on the measure that is easiest to report, and it is not what anyone means by development.

References

Every quantitative claim above is attributed inline. The principal sources are collected here.

World BankGlobal Economic Prospects, January 2026 — virtually half of developing economies not narrowing the income gap since 2019; per capita income at about 12 per cent of the advanced-economy level; per capita growth about a percentage point below the 2000–2019 average. Full report here.
Dani RodrikPremature Deindustrialization, NBER Working Paper 20935 — manufacturing employment peaking at 30–37 per cent of the workforce at US$11,000–14,000 per head in early industrialisers, against deindustrialisation beginning at about US$5,000 in Brazil, US$3,000 in China and US$2,000 in India. See also the working paper version.
UNCTADDeveloping economies surpass $1 trillion mark in digitally deliverable services exports — 56 per cent of global services exports, US$4.1 trillion from developed economies and an estimated US$1.3 trillion from developing ones, growing at 9 and 12 per cent respectively. On concentration, see Digitally deliverable services boom risks leaving least developed countries behind.
Recent industrialisersBangladesh's ready-made garment sector — about four million workers and US$38.5 billion of exports in 2024, roughly four-fifths of export earnings — per WTO and industry reporting; Vietnam's manufacturing workforce of about 12 million, some 23 per cent of employment, and its industrial-zone FDI employment, per Vietnam Briefing.
African Development BankYouth and Jobs for Youth in Africa — 10 to 12 million young people entering the labour market annually against roughly three million formal jobs.
ILOAfrica: informality regional statistical profile — informal employment at about 85 per cent of total employment in Africa in 2024 against a world average of about 58 per cent, and little change since 2005.
H HeuristicsPremature Deindustrialization and Alternative Pathways to Middle Income — the companion essay setting out the alternative-pathway argument this report builds on. Related reports in this series: climate-resilient industrialization pathways; convergence through crisis; and navigating the transition.

Metadata

Keywords: economic convergencepremature deindustrializationstructural transformationemployment absorptiondigitally deliverable servicesinformalityyouth employmentdevelopment pathwayspolycrisisindustrial policyservices exportsproductivitylabour marketsAfrica jobs
Topics: Emerging Markets Human Capital Technology Institutions & Governance
JEL classification: O14, O11, J21, O33, F63 — industrialization and manufacturing; macroeconomic analyses of economic development; labour force and employment; technological change and diffusion; economic development and globalization
Data and method: This report synthesises published research and institutional data rather than producing new modelling. The premature-deindustrialization evidence is from Dani Rodrik's NBER working paper and associated publications; convergence and growth figures from the World Bank's Global Economic Prospects, January 2026; services trade data from UNCTAD's reporting on digitally deliverable services; labour-market entry estimates from the African Development Bank; and informality shares from ILO modelled estimates for 2024. All figures were verified against their sources in September 2026. Employment-absorption comparisons between sectors are directional rather than precise: employment data for tradable services are inconsistently measured across countries, and the report says where a figure is an order-of-magnitude comparison rather than a like-for-like statistic. The three-measures framing in Section 1, Table 1's escalator comparison and the two-track proposal in Section 5 are the author's analysis, assembled with the cited evidence already known, and are offered as a planning frame rather than a forecast. Figure 6 is a conceptual schematic. The report is analytical rather than predictive.
Report: H Heuristics Digital Report № 2026-18 · Published 17 September 2026
Licence: CC BY-NC-ND 4.0
Cite as: Hunter Hughes (2026). Convergence Without Absorption: Can emerging economies catch up when the escalator that carried workers has narrowed?. H Heuristics Digital Report 2026-18. https://digitalreports.hheuristics.com/reports/convergence-without-absorption/

↑ Back to top

[1] World BankGlobal Economic Prospects, January 2026 — virtually half of developing economies not narrowing the income gap since 2019; per capita income at about 12 per cent of the advanced-economy level; per capita growth about a percentage point below the 2000–2019 average. Full report here.

[2] Dani RodrikPremature Deindustrialization, NBER Working Paper 20935 — manufacturing employment peaking at 30–37 per cent of the workforce at US$11,000–14,000 per head in early industrialisers, against deindustrialisation beginning at about US$5,000 in Brazil, US$3,000 in China and US$2,000 in India. See also the working paper version.

[3] UNCTADDeveloping economies surpass $1 trillion mark in digitally deliverable services exports — 56 per cent of global services exports, US$4.1 trillion from developed economies and an estimated US$1.3 trillion from developing ones, growing at 9 and 12 per cent respectively. On concentration, see Digitally deliverable services boom risks leaving least developed countries behind.

[4] Recent industrialisersBangladesh's ready-made garment sector — about four million workers and US$38.5 billion of exports in 2024, roughly four-fifths of export earnings — per WTO and industry reporting; Vietnam's manufacturing workforce of about 12 million, some 23 per cent of employment, and its industrial-zone FDI employment, per Vietnam Briefing.

[5] African Development BankYouth and Jobs for Youth in Africa — 10 to 12 million young people entering the labour market annually against roughly three million formal jobs.

[6] ILOAfrica: informality regional statistical profile — informal employment at about 85 per cent of total employment in Africa in 2024 against a world average of about 58 per cent, and little change since 2005.

[7] H HeuristicsPremature Deindustrialization and Alternative Pathways to Middle Income — the companion essay setting out the alternative-pathway argument this report builds on. Related reports in this series: climate-resilient industrialization pathways; convergence through crisis; and navigating the transition.