The Infrastructure of Sustainable Economic Convergence

Between 2000 and 2024 the OECD's average GDP per capita, at purchasing power parity and in constant prices, grew by 1.22% a year. Of the 110 economies that started below half that average, 35 grew more than two points a year faster. They hold 4.1 billion people, led by China and India, and include Viet Nam, Bangladesh, Ethiopia, Indonesia, the Philippines, Türkiye, Poland and most of Central Asia. Another 49, with 2.1 billion people, grew faster than the OECD but by less, among them Nigeria, Pakistan, Brazil, Egypt and Kenya. And 26, with half a billion people, fell further behind, including South Africa, Sudan, Iraq, Angola, Madagascar, Zimbabwe and Haiti.

If convergence were automatic, the poorest economies would be in the fast group. They are not. Plotting each economy's growth gap against its starting income gives a flat cloud: starting income explains none of the variation in catch-up growth (R² of 0.00, with a slope indistinguishable from zero). This is a familiar result in growth economics — unconditional convergence has been weak for most of the post-war period — and it is why economists talk instead of conditional convergence: poorer economies grow faster only once the conditions for growth are in place.

Holding two power measures constant changes the picture completely. Adding electricity access in 2000 and the share of firms reporting outages, the slope on starting income becomes strongly negative and the fit rises to 39%. Among economies with similar power, each halving of starting income goes with about one percentage point a year of faster growth. Switch the scatter below to "adjusted for power" to see the hidden convergence appear. The map shows where it happened: fast catch-up across East and South Asia and Central and Eastern Europe, slow catch-up and falling behind across much of Africa, the Middle East and Latin America.

The difference in starting electricity access between the three groups is stark. In 2000, the median fast converger already gave 94% of its people electricity; the median slow converger 70%; the median economy that later fell behind 34%. Firms tell the same story about reliability. In World Bank Enterprise Surveys between 2006 and 2025, a median of 40% of firms in fast-converging economies reported experiencing electrical outages, against 65% among slow convergers and 74% among those falling behind. Power consumption per person grew by a median of 3.1% a year in the fast group and 1.8% in the group falling behind, where several economies used less electricity per person in 2023 than in 2000.

On its own, electricity access in 2000 (with starting income held constant) explains 29% of the variation in catch-up, and outage exposure 25%. Together they are the core of the conditional convergence model: each 10 points of access is associated with 0.3 points a year of faster catch-up, and each 10 points more of firms suffering outages with 0.3 points a year less. The association keeps its sign and similar size without China and India and without Europe and Central Asia; within sub-Saharan Africa alone (40 economies) both power measures remain at the edge of significance; and over the shorter period 2010–2024 access remains strongly significant while outage exposure weakens. It is the most robust result in this dashboard.

The exceptions prove the rule in an interesting way. Seven economies converged fast despite starting with less than half their people on the grid — Bangladesh, Cambodia, Ethiopia, Laos, Myanmar, Nepal and Rwanda. All of them expanded access dramatically: Bangladesh from 32% to 100%, Cambodia from 17% to 99%, Rwanda from 6% to 72%. Expansion alone was not enough, though. The group falling behind also raised access by a median of 23 points, but from a lower base, with less reliable supply and often amid conflict.

Catch-up growth has usually run through trade: exporting manufactures, joining supply chains and importing machines and inputs. That makes logistics — ports, roads, customs, shipping connections and the firms that move goods — a natural candidate for a gate. The World Bank's Logistics Performance Index scores countries from 1 to 5 on customs, infrastructure, shipments, competence, tracking and timeliness, and UNCTAD's Liner Shipping Connectivity Index measures how well a country is linked to global container shipping.

In this sample, logistics separates the bottom from the rest more than it separates fast convergers from slow. The median LPI score over 2007–2012 was 2.59 in the fast group, 2.57 in the slow group and 2.38 among economies falling behind. Shipping connectivity followed the same pattern, with medians of 23 and 20 for the converging groups against 12 for those falling behind. Controlling for starting income, a higher LPI score is associated with faster catch-up, but the association is weaker than for power (it explains about a tenth of the variation), and it fades to the edge of significance alongside power.

The micro-evidence is clearer. In a study of 98 countries, Djankov, Freund and Pham found that each extra day a product is delayed before shipping reduces trade by more than 1%, equivalent to moving about 70 kilometres further from trading partners, with larger effects for time-sensitive goods such as fresh produce. Lifelines put the cost of transport disruptions to firms' capacity use at $107 billion a year. For economies trying to break into export markets, logistics decides which products are viable at all.

Heat is a direct drag on work. In a panel of Indian factories, Somanathan and co-authors found that plant output falls by about 2% for each degree Celsius of warming, mainly because workers are less productive and more often absent; climate control at the workplace reduced the same-day productivity loss. The International Labour Organization estimates that 2.41 billion of the world's 3.4 billion workers are likely to be exposed to excessive heat at some point in their work, and projected in 2019 that heat stress would cost 2.2% of total working hours worldwide in 2030, equivalent to 80 million full-time jobs.

The convergence record shows the same gradient. The median fast converger has a mean annual temperature of 12°C and a heat-exposure index (cooling degree days from monthly means) of 410; the slow convergers and those falling behind sit at 23–24°C and more than 2,000. Controlling for starting income, each additional degree of mean temperature is associated with 0.13 points a year slower catch-up. Part of this is geography: the fast group includes many cool transition economies of Europe and Central Asia. But the more interesting finding is what happens when power is added. The heat penalty shrinks and loses statistical significance.

That is what would be expected if heat is largely overcome through electricity. Cooling workplaces, keeping medicines and vaccines cold, and refrigerating food all need reliable power. The IEA estimated in 2018 that only 8% of the 2.8 billion people living in the hottest parts of the world owned an air conditioner, against more than 90% of households in the United States and Japan. UNEP and FAO found that developing countries refrigerate only about 20% of the perishable food they produce, against 60% in developed countries, and that the lack of refrigeration caused the loss of 526 million tonnes of food — 12% of global production — in 2017.

Internet use in 2010 — the earliest year with broad coverage in this sample — was a median of 30% in the fast-converging economies, 11% in the slow and 5% among those falling behind. By 2024 the medians were 84%, 63% and 46%, and fixed broadband subscriptions per 100 people were 20, 5.8 and 0.65. Controlling for starting income, internet use in 2010 is strongly associated with catch-up on its own, explaining about a fifth of the variation. When electricity access and reliability are included alongside it, its coefficient falls to near zero.

That does not make connectivity irrelevant. It means that, across countries, connectivity and power have spread together — the internet cannot run without electricity, and the economies that built reliable grids were also the ones that connected first. Within countries, careful studies isolate connectivity's own effect. Hjort and Poulsen used the staggered arrival of submarine cables along Africa's coast to show that the arrival of fast internet raised employment rates in connected towns across 12 African countries, with more firm entry, higher productivity and more exporting.

Connectivity is therefore best understood as a multiplier. It makes the power gate more productive by enabling digital payments, remote services, logistics tracking and access to markets and information. Its constraint today is less coverage than use: the ITU reports that mobile broadband remains unaffordable in about 60% of low- and middle-income countries, and that about 2.2 billion people were still offline in 2025, nearly all of them in those countries.

Each gate looks important when tested alone. Holding starting income constant, electricity access, outage exposure, logistics performance, internet use and heat exposure are all associated with catch-up growth, and all but logistics strongly so. But the gates are highly correlated with each other: the economies with reliable power also tend to have better logistics, more internet use and cooler climates. Testing them one at a time therefore credits each with effects that the others share.

Putting all five measures into one model, with starting income, separates them. Electricity access remains strongly significant (t-statistic 4.3) and outage exposure significant (−2.3). Logistics performance weakens to the edge of significance (1.7), and internet use (0.8) and heat exposure (−1.3) lose it. The five together explain 48% of the variation in catch-up across 100 economies with complete data, against 0% for starting income alone. The chart compares each gate's statistical strength alone and together.

The reading is not that logistics, cooling and connectivity do not matter: the micro-evidence says they do. It is that across economies their effects are bundled with power and cannot be separated from it with this data. Power is the gate that holds independently, and the others mostly operate through it or alongside it. That ordering is the basis for the sequence proposed in section 8.

The calculator applies the conditional convergence model — catch-up growth as a function of relative income, electricity access and outage exposure — to any of the 110 economies. It first compares the economy's actual catch-up over 2000–2024 with what the model predicted from its starting position, which shows whether it outperformed or underperformed economies with similar power. It then projects relative income from 2024 to 2050 using today's electricity access and outage exposure, and lets you change both.

The projection updates relative income year by year, so catch-up slows as an economy gets closer to the frontier, as the model implies. Moving the sliders shows what universal access and more reliable supply would be worth if the cross-country relationship held for that economy. For most economies in the slow and falling-behind groups, closing the difference between today's outage exposure and the fast convergers' median of 40% is worth about a point of growth a year or more — in Nigeria, where 85% of firms report outages, about 1.5 points. Compounded over 25 years, that is a large difference.

The numbers are illustrations, not forecasts. The model explains 39% of the variation across economies, so any single economy can differ greatly from it; the relationship is an association that may partly reflect institutions, peace and policy rather than power itself; and it assumes the OECD keeps growing at its 2000–2024 rate. Use it to compare scenarios for the same economy, not to predict its future income.

The evidence supports a sequence rather than a list. Power comes first: universal access and, as important, reliable supply for firms, because outage exposure is as strongly linked to slow catch-up as missing connections. Logistics comes next where growth depends on trade: customs, corridors and port connections decide whether manufactured and perishable exports are viable. Cooling and connectivity are multipliers that make powered economies more productive in a hotter, more digital world. None of the four works well in isolation, but only power appears as a precondition for the others.

This is also where the sustainability of convergence is decided. For most of the 20th century, opening the power gate meant building coal and oil plants, and convergence and emissions rose together. That trade-off has weakened sharply. IRENA found that in 2024, 91% of newly commissioned utility-scale renewable capacity produced electricity more cheaply than the cheapest new fossil-fuel alternative; new solar power averaged $0.043 per kilowatt-hour, 41% below the cheapest fossil option, and onshore wind $0.034. The IEA's 2025 roadmap puts the investment needed for universal electricity access at about $23 billion a year over 2025–2035.

The same logic applies to the multipliers. Efficient cooling equipment can cut future cooling energy demand by about 45%, which matters most for the hot economies whose grids are least able to absorb a surge in air conditioning. Cold chains powered by renewables reduce food loss without adding emissions. Electronic customs and digital logistics reduce the time costs that delay exports. The cost of capital, not technology, is now the main obstacle to building clean infrastructure in poorer economies, which makes concessional and blended finance central to sustainable convergence.

All series except temperature are from the World Bank World Development Indicators API, downloaded in October 2026: GDP per capita at PPP (constant 2021 international dollars), electricity access, electric power consumption per person, firms experiencing electrical outages (Enterprise Surveys), the Logistics Performance Index, the Liner Shipping Connectivity Index, internet use, fixed broadband subscriptions and population. The frontier is the OECD-members aggregate. The sample is the 110 economies with more than one million people in 2024, GDP data for 2000 and 2024, and less than half the OECD's GDP per capita in 2000.

The growth gap is average annual growth in 2000–2024 minus the OECD's 1.22%. Groups are fast (more than 2 points), slow (0–2) and falling behind (0 or less). Electricity access and power use are measured at 2000; internet use at 2010; outage exposure averages all surveys in 2006–2025, and logistics averages 2007–2012 in the models, so both are partly contemporaneous with growth. Heat exposure sums, over months, days times the excess of the 1991–2020 monthly mean temperature (ERA5 via Our World in Data, area-weighted) above 18°C. Regressions are ordinary least squares with robust standard errors.

The "adjusted for power" scatter subtracts each economy's fitted electricity and outage effects relative to the sample means. The calculator uses the three-variable model and updates relative income annually from 2024. All results are associations: conflict, institutions, geography and reverse causality are not separately identified. The map uses Natural Earth 1:50m boundaries via world-atlas; boundaries do not imply endorsement. Related reports: convergence without absorption, convergence through crisis and the Global South resilience stack.