The growth nobody felt: what Kazakhstans economy shows in mid-2026

In the second quarter the statistics produced a figure that should have been good news: GDP growth for the first half of 2026 of 4.1%. Manufacturing added 9.8%, construction 15.2%. Private investment rose 21.4% and now accounts for 87% of all fixed capital investment. This is precisely the structure that has been promised for years: not the state pulling the economy along, but private money.

Now the second figure, from the same statistics. In 2025, the year GDP grew 6.5%, real household incomes fell by 1.1%. In the first quarter of 2026 real wages fell a further 2.3% — 3.3% on another measure, the worst result among the Eurasian Economic Union states. Retail trade growth halved, from 6.3% to 2.8%.

Both figures are correct. Both come from the same source. And everything that matters about current economic policy sits between them.

What has genuinely worked

Let me start with what is working, because the list is shorter and rarely stated in full.

The disinflation is real. From a peak of 12.9% in September 2025, inflation slowed for nine consecutive months to 10.3% in June 2026. The National Bank held the rate at 18% through March and again on 24 April, stating plainly that there was no room to cut. It cut only on 5 June, to 17%, once the trend was established. Holding a high rate under political pressure is an unpopular decision and the right one.

The consumer credit bubble was deflated without a bust. Unsecured lending grew at over 30% a year for five straight years; by 1 June 2026 growth was 9.9%. Non-performing loans stayed in the 3–4% range. That is rare: bubbles of this kind usually end not in a slowdown but in a wave of defaults.

Diversification is measurable. Crude oil fell to 44.4% of exports in the first quarter, against 52% a year earlier — a historic low. Copper cathodes gained 62%, copper concentrate 84.6%, silver tripled, wheat rose 53.7% by value.

Here a caveat is needed that the reports omit: oil’s share fell partly because output fell, not only because other exports grew. Oil production dropped 8.4% in the first half, to 45.7 million tonnes, and the annual plan was cut from 100.5 to 96–98 million tonnes. Part of the diversification is not the non-resource sector growing but the resource sector shrinking.

Where the figure parts company with the substance

The headline budget deficit is 2.5% of GDP, and every fiscal rule was formally met in 2025: spending grew by exactly 8.0% against a limit of 8.0%; public debt was 22.8% against a ceiling of 27.2%.

The problem is that this deficit does not describe the whole of the state’s economy.

In February 2026 the IMF published a calculation that everyone making decisions ought to read: the combined assets of Samruk-Kazyna and Baiterek grew from roughly 15% of GDP in 2013 to 54% of GDP in 2025. The expansion of these holdings’ balance sheets has averaged around 20% of the country’s entire budget expenditure over five years. Had that activity been booked as budget spending, the non-oil deficit would have been 3.5 percentage points of GDP higher.

And the IMF’s conclusion on 2026: the planned fiscal consolidation “will be largely offset by expanding quasi-fiscal activities by SOEs.”

Samruk-Kazyna’s own 2025 reporting confirms it: assets up 8%, debt up 21%, net profit down 3%, no dividends paid to the state on the 2024 results, and a further billion dollars borrowed in June 2026. Assets under management of roughly $88 billion — more than the National Fund holds.

A fund spent faster than it fills

In 2025, 5.25 trillion tenge was withdrawn from the National Fund. Tax receipts into the fund from the oil sector were 3.73 trillion. Outflows exceeded oil inflows by roughly one and a half times — for the second year running.

In June 2026 the fund’s foreign currency assets fell a further $899 million, to $65.1 billion. The stated target is $100 billion by 2030. On this balance it can be reached only through investment returns: oil no longer fills the fund faster than it is spent.

In fairness: for 2026 the government abandoned the targeted transfer and cut planned withdrawals from roughly 5.3 to 2.77 trillion tenge. That is a real decision and it points the right way. The question is whether it survives August.

Separately, the “National Fund for Children” programme. Three years have accumulated $370 per child. Actually paid out across all applications by the start of 2026: $31.6 million, against a fund of $65 billion. As a political symbol the programme works. As an economic instrument it is a rounding error.

Three consequences already visible

The state is becoming its own economy’s principal borrower. The external debt of state-controlled organisations grew from $16.9 to $25.2 billion in a year — faster than the government’s own debt. Half of the new borrowing in 2025 went to refinancing the old. When a quasi-state sector holding assets worth half of GDP competes for money with private business at a 17% policy rate, long money does not appear in the country. Fitch attributes the persistent inflationary pressure specifically to quasi-fiscal operations — meaning the state holds a high rate against inflation it is itself substantially creating.

Debt service already exceeds healthcare. In the 2026 budget: debt service 3.5 trillion tenge, healthcare 2.7 trillion, education 811 billion. This is neither a forecast nor a risk — it is the budget law in force. Each further year of this structure narrows the room for everything else.

The gap between the statistics and life is itself the mechanism by which trust is lost. Not press criticism, not the opposition. A person told about 6.5% growth while their real wage falls draws a conclusion not about the economy but about whether they are being lied to. From 1 January VAT rose from 12% to 16% and the registration threshold was halved; from 1 April the utility tariff moratorium ended, with increases forecast at 20–30% a year. When, against that background, 0.1% real income growth is called “wonderful news,” a resource is being spent that no budget decision restores.

What cannot be solved from Astana

There is one risk that economic policy does not touch at all. The CPC pipeline carries roughly 80% of Kazakhstan’s oil exports. Its terminal at Novorossiysk was attacked on 17, 19 and 20 July, and loading was suspended twice in four days. Throughput through the terminal fell from 36.0 to 33.4 million tonnes over the half-year.

The Middle Corridor cannot physically substitute for it: roughly 4.5–4.7 million tonnes a year against some 70 million tonnes of oil exports. The Caspian is falling by about 30 cm a year, and ferry traffic on the Baku–Kuryk route has already contracted 22%. Add that the sanctioned Lukoil is a shareholder in Tengiz, Karachaganak and the CPC simultaneously.

Every 2026 growth forecast — from Fitch’s 4% to the Eurasian Development Bank’s 5.5% — assumes oil flows normalise. If they do not, all of them are wrong.

What I do not know

Nobody outside government knows the true consolidated deficit of the state sector: no consolidated balance sheet is published. Whether the disinflation survives the pass-through of VAT and tariffs is an open question: the National Bank expects 5.5–7.5% in 2027, the Asian Development Bank 9.5%. A gap of 2–4 percentage points between two honest institutions on the central macro variable means nobody currently has a confident answer.

Whether the rise in the SME share to 40.9% of GDP is real is also a question. That figure has already exceeded the target set for 2029, four years early, while the number of registered business entities rose 20.4% in six months. No decomposition into “new activity” and “re-registration under the tax regime” is published.

My view

Economic policy over the past two years did the hard and useful work in the monetary sphere and declined the hard work in the structural one. The rate was held, the credit bubble was deflated, the rules were met — and at the same time the quasi-state sector was grown to half of GDP, financing through it everything the budget preferred not to show.

You can live this way for a long time. But the method has an arithmetic limit, and it is nearer than it looks: when debt service exceeds healthcare and the fund is spent faster than it fills, the room for manoeuvre narrows every year on its own, without any crisis at all.

The World Bank put it gently and precisely: the slowdown reflects “the limits of demand-led growth in the absence of deeper structural transformation.” I will put it harder. Growth a person cannot feel is not growth but redistribution. And it can be explained away with statistics exactly as long as people are willing to treat statistics as a description of their own lives.