Atlas Uzbekistan: Landlocked Reform and Opening

The population chart is the quiet driver of every other series here: a growth rate more African than post-Soviet, a workforce expanding faster than any European or East Asian peer, and internal migration flowing from the dense Fergana Valley regions toward Tashkent. Life expectancy recovered quickly from the COVID dip. Net emigration, which ran at 50–100,000 a year through the 2000s, has narrowed to under 10,000 as returns nearly balance departures.

On 5 September 2017 the Central Bank let the som fall roughly 50% overnight to the parallel-market rate and made it convertible for current transactions. The break was less sudden than it looked — Mirziyoyev had been Karimov's prime minister since 2003, quietly expanding small-business credit for a decade, so reform went from gradual to revolutionary rather than from nothing. Import liberalization plus the pass-through of a much weaker som pushed inflation to 20% by 2018; a newly serious central bank then ground it down with policy rates held far above inflation, reaching single digits by 2023 and 5.5% by May 2026. What drives inflation now is services — the 2024 spike in the services line is administered-tariff reform, not overheating.

Growth under the old regime was state-directed investment behind closed borders; growth since 2017 is broader, with consumption, exports and private credit all contributing. The economy grew +2.0% in 2020 — one of few positive rates anywhere — and accelerated to 6.3–7.7% in 2023–2025. Dollar incomes finally moved: GDP per capita, stuck near $2,600 for a decade at distorted exchange rates, reached almost $4,000 in 2025 at market rates. The first chart separates the three things that get conflated: real growth, inflation, and their product, nominal growth — som figures roughly triple every five years while real output compounds at ~6.5%.

Since 2017 agriculture has given up almost 11 points of its share of value added (28% down to 17%) while manufacturing gained more than 6 points — the classic development-textbook shift, on an unusually fast clock. The real-output index makes the divergence vivid: since 2016, information & communication output has grown 5.7× (from a small base — it is still only 3% of value added), construction 2.5×, trade & hospitality 2× and manufacturing 1.8×, against +31% for agriculture and +39% for mining. Growth is coming from building, making and connecting things, not from digging more out of the ground.

This is the mechanism behind the structural shift. Modernizing an economy means acquiring capital goods and technology, and for Uzbekistan that increasingly means China: across the ten HS2 categories that carry modernization, China's share of imports rose from 28% in 2017 to 48% in 2025. The granular picture is more striking still — Chinese car exports went from essentially zero to $1.4bn, solar cells and semiconductors 250-fold, EV and storage batteries 180-fold. But the imports are also seeding domestic industry: motor-vehicle output has tripled, electronics has grown five-fold, and the flagship case is cotton — where ending forced labor and building textile clusters is visibly moving the country from spinning yarn to sewing garments. The nuclear plant now breaking ground at Jizzakh is the next chapter of the same story.

The average wage reached about $560 a month in early 2026 — roughly 3× the ~$180 trough right after convertibility repriced the som (pre-2018 dollar figures were an artifact of the administratively strong rate). Tashkent City remains a different economy — its wages run far ahead of the regions, and finance and ICT pay two to three times the wages of education and health. Online job postings, a rough proxy for formal-sector labor demand, are an order of magnitude above their pre-reform level.

Gross tertiary enrollment collapsed to 9% in the Karimov years — among the lowest in the world for a literate middle-income country — then exploded past 40% after private and foreign universities were licensed from 2018. Governance indicators improved from a very low base: regulatory quality and government effectiveness moved most, control of corruption least — the pattern that keeps the reform-reversibility question alive. The same story showed up in the World Bank's Doing Business index, where Uzbekistan climbed from 146th in 2013 to 69th in the final 2020 edition, before the Bank killed the index for rigging other countries' scores.

The browser chart is really a device-affordability story. In 2012 nearly half of web traffic came through Opera — the data-saving Mini browser for cheap phones on pay-per-megabyte connections — with UC Browser playing the same role through the mid-2010s. Chrome's march to 80% tracks real smartphones and affordable data replacing that world, while Safari never breaking 8% says iPhones remain a luxury. On platforms, Facebook's collapse from 86% of social web traffic to 16% and Instagram's rise track a generational handover. One large caveat: Telegram, the true center of Uzbek digital life, is invisible here — it is a messenger, not a website, so page-view statistics cannot see it. And Yandex's sudden 2021 arrival in search, bundled with its super-app and taxi service, is a reminder that Russian platforms still contest this market.

The official foreign category shows 66% of fixed investment in 2025, but that partly reflects a reclassification of previously unallocated funds. The cleaner "of which foreign" measure puts foreign participation at 22% in 2016, a peak of 68% in 2024 and 48% in 2025 — still extraordinary by emerging-market standards (typically under 15%), and overstated relative to balance-of-payments inflows because it counts foreign loans and project finance at project value, not just money that actually crossed the border. Housing tracks the demographics: annual residential commissioning roughly doubled over the decade.

China supplies the machinery, equipment and increasingly the vehicles that the investment boom requires; the deficit with it has widened to roughly $13–18bn a year (post-2022 figures partly reflect an IMF source revision to partner-reported data). Exports are diversifying at the margins — manufactures and food are growing — but gold still dominates export earnings, and services exports have swung toward travel as tourism reopened: visitor numbers passed their pre-COVID peak in 2023. Afghanistan is a quiet story in the partner chart: one of Uzbekistan's few large surpluses. The composition of that China deficit — the modernization goods it buys — is the subject of section V.

Remittances run near $16bn a year on a 4-quarter basis, among the highest in the world relative to GDP. Russia's war in Ukraine, which most observers expected to hurt a remittance economy this exposed, was on net a windfall: inflows jumped about 70% in 2022 on the strong ruble and relocation transfers, ICT services exports quintupled from 2021 to 2025 as Russian tech firms and workers relocated, and the cost was borne in the capital account — bond issuance froze for 27 months — and in an emergency rate hike to 17%. The central bank, which buys domestic gold output with soms, holds over $70bn in reserves, mostly bullion that has repriced sharply upward. And the lending story confounds the Belt-and-Road prior: of $39bn in public external debt at end-2024, the ADB, World Bank and bondholders hold ~$7.5bn each and Japan $5bn, while China's $3.9bn has been flat since 2021. Chinese money increasingly arrives as equity and project finance in energy instead, which creditor statistics don't capture.

Household lending grew fastest from almost nothing. The IMF flagged excessive credit growth as a stability risk as early as 2019; the test came with COVID, when non-performing loans spiked past 6% in 2021 before settling back under 4%. De-dollarization is real and large: the foreign-currency share of loans has fallen from over 60% to under 40%, a direct dividend of credible domestic money.

The country that was a gas exporter is now a net energy importer, buying Russian gas through the same Central Asia pipeline that once ran the other way. It is a textbook resource paradox with four causes at once: the Soviet-era gas fields are three-quarters depleted; decades of near-giveaway domestic prices both encouraged waste and made new drilling uneconomic, so the fields were never replaced; a population growing 2% a year and a booming industry keep demand climbing; and the state gas monopoly lacked the capital to reverse the slide while foreign partners pulled back. The answers, a foreign-financed solar and wind build-out (the IPP chart below) and the first Rosatom nuclear plant, are real but none arrive before 2029. Meanwhile emissions keep climbing with the economy even as energy intensity falls, and Tashkent's air sits far above WHO guideline levels.

Here is the puzzle that should stop any investor. Uzbekistan's economy more than tripled in dollar terms since 2016, but its stock market barely moved: nominal share prices are flat, down roughly two-thirds in real terms. A booming economy with a de-rating market is the textbook signature of undervaluation — and the index has lately begun repricing hard, up nearly threefold off its December-2022 war low. The apparent $36bn market capitalization is misleading, though: it grew 31-fold in som terms while prices fell 26%, so all of it is new listings and state-stake flotations, not investors bidding shares up. And most of that cap is locked government holdings that never trade — turnover is a rounding error against a $147bn economy. Where market development unambiguously happened is debt: Uzbekistan now has $6.8bn of sovereign and $6.8bn of corporate international bonds outstanding, from a standing start in 2019. The gap is the opportunity. A market that de-rated by two-thirds while its economy tripled, now repricing for the first time in years, is a classic value setup. What keeps it cheap is also what makes it hard to own: trading is thin, and most of the float sits locked in state hands.

Caveats worth keeping in mind: pre-2017 dollar conversions use the official rate, which overstated dollar incomes; State Committee visitor counts use a narrower definition than the World Bank arrivals series; IMF DOTS partner trade switches to partner-reported, year-to-date data in 2023 (handled in the build, noted on charts); LinkUp job postings cover online postings only; WGI governance scores are perception-based estimates with wide confidence bands.