FINANCIAR
Almost half of Romania's best-performing companies have no bank loan at all
Almost half of Romania's best-performing companies — 46% in March 2026 — had no active bank loan, according to the National Bank of Romania's Financial Stability Report.
This is not an eligibility problem. The same firms — 27,800 companies making up 3% of all active businesses but more than half of gross value added and half of all jobs — have, on the central bank's own estimate, room for another 321.5 billion lei of sustainable financing. The thesis of this article is that the blockage lies neither in the banks nor in the firms taken separately, but in an arrangement in which the state has become the banking system's best customer. The dominant scenario for the next two quarters is that the arrangement does not change.
1. The lowest financial intermediation in the EU — and a contradiction inside the source describing it
Romania's non-government credit — the loans banks extend to firms and households — stood at 22% of GDP at the end of March 2026, against a European average of 66%. The corporate and household components are almost equal: roughly 11% of GDP each.
The figure is quoted by the financial press from a central bank report, and the Romanian article carrying it runs a headline saying that „only Ireland is below us". The body of the same article says something else: „Romania remains the economy with the lowest level of non-government credit in GDP in the European Union".
The contradiction is not a newsroom detail. A reader who keeps the headline walks away believing Romania has somebody beneath it, and therefore that the situation is comparable to another member state's. The central bank's own text leaves no room for that: last place, no parentheses.
Ireland probably enters through a mix-up of series. Irish GDP is inflated by the intellectual property of multinationals registered there, which makes any „X to GDP" ratio come out artificially small. On resident bank-credit series, Romania is still last.
2. The mechanism: why a government bond beats a corporate loan
The causal chain has three links, each with separate evidence.
The cause. Romania ran the European Union's largest government deficit in 2025: 7.9% of GDP, after 9.3% in 2024. The 1.4-percentage-point correction was the EU's largest fiscal consolidation that year and, at the same time, insufficient: the EU average was 3.1%. Public debt climbed to 59.3% of GDP, from 54.8% in 2024.
The mechanism. A state borrowing on that scale has to place paper, and the natural buyer is the domestic banking system. Banks' total exposure to the state — securities plus lending to government — reached 28% of banking assets in 2025, the highest level in the Union; Poland sits at around 25%, Hungary at around 20%. Government securities alone accounted for 222.4 billion lei at the end of 2025, or 22.3% of total assets — 13% more than in 2024. In the first quarter of 2026 they passed 230 billion, three times the level of a decade ago.
The effect. A government bond carries a sovereign guarantee, a favourable risk weight in capital requirements, immediate liquidity and zero cost of analysing the borrower. A loan to a mid-sized firm needs analysis, collateral, monitoring and provisions. At comparable yields, the choice is not a dilemma.
The weakest link in the chain is the last one. That banks buy a lot of government paper is measured; that they would have lent that money to firms in the absence of the paper is an inference, not an observation. Nobody can point to the loan that was never made.
3. The counter-hypothesis that deserves to be taken seriously: the firms are not bankable
The alternative explanation says the low level of lending is not a blockage but a rational equilibrium: banks do not lend because the firms do not support a loan. It has serious empirical backing and cannot be dismissed cheaply.
In 2025, 22.5% of active firms had negative equity, up from 20.27% in 2021. The median cash reserve fell from 57.5 days of activity in 2021 to 36.3 days in 2025, while the median collection period rose from 36.2 to 48.5 days. Over the same span, the combined turnover of active firms grew 42.6% in nominal terms. Firms are selling more and holding less cash.
The central bank describes the same reality from the inside: one third of firms have net assets below half their subscribed capital, which in balance-sheet language means a capitalisation deficiency. And the structure of corporate financing shows that intra-group loans and trade credit make up a third of the sources treated as corporate funding, while bank and non-bank financial-institution credit stays below a tenth.
In other words: Romanian firms are not free of debt. They are indebted elsewhere — to suppliers and to their own shareholders — where there is no risk analysis, but also no long maturities and no negotiated interest rate.
The recapitalisation the economy would need to become bankable again is estimated at almost 32 billion euros, roughly 160 billion lei. That is not a sum an interest-rate decision solves.
4. The figure that separates the two explanations
If the problem were purely the quality of the firms, the good firms would have credit. They do not.
The 27,800 high-performing companies identified by the central bank are exactly the segment any bank wants on its books: 3% of active firms, more than 50% of gross value added, 50% of jobs. Of these, 46% had no active bank loan in March 2026.
The firms in that group that do use bank financing look different from those that do not: gross value added 22% higher, turnover more than 50% higher, net profit around 20% higher. Correlation alone does not prove causation — firms that are already better find credit more easily — but the order of magnitude of the gap is hard to explain by selection alone.
This is where the two hypotheses part. A demand blockage explains weak firms without credit; it does not explain why almost half of the top of the economy sits outside the banking system, with sustainable lending potential estimated at 321.5 billion lei — roughly 68% of the entire stock of non-government credit outstanding today.
5. The price of money: 8.5% in lei, 3.79% in the euro area
The total stock of non-government credit stood at 472.4 billion lei in July 2026, 7.6% more than a year earlier. The figure looks like a recovery. It is not: adjusted for inflation, the stock fell 0.5% against July 2025. Nominal growth does not move the ratio to GDP.
Cost explains part of the inertia. The National Bank's policy rate stands at 6.50% a year and has been held at every board meeting through 2025 and 2026. The next decision is scheduled for 24 September 2026. New leu-denominated corporate loans have settled, on figures presented publicly by the central bank's leadership, at an average of about 8.6% over the past three years, excluding fees.
In the euro area, the composite cost-of-borrowing indicator for new loans to non-financial corporations was 3.79% in June 2026. The gross gap is roughly 4.7 percentage points.
The comparison is not between identical assets — a leu loan carries currency risk and a country premium that a euro loan does not. The practical effect on an entrepreneur's decision is the same regardless: at 8.5%, the hurdle rate on an investment rises far enough that many projects no longer clear it.
Borrowers responded predictably: they moved demand into foreign currency. The foreign-currency component of non-government credit reached 156.6 billion lei in July 2026 — 33.2% of the total, up 16.6% in a year. The leu component, 315.8 billion, grew 3.6%.
The four-to-one ratio between those two growth rates is an interest-rate arbitrage, not a preference. It shifts currency risk onto balance sheets that mostly have no euro revenue.
6. What would unblock the mechanism — and what would lock it in
Unblocking has one necessary condition, and it has nothing to do with banks: the state has to ask for less. Romania has been under the excessive deficit procedure since 2020, and the EU Council set 2030 as the exit deadline, with annual ceilings on net expenditure growth. The 2026 target is a further consolidation of about 2.4 percentage points, towards a deficit of around 6% of GDP.
If the trajectory holds, issuance falls, yields come down, and corporate credit becomes competitive again in bank allocation. If it does not hold, the mechanism reinforces itself: a larger deficit, higher yields, an allocation even more favourable to the state.
The second condition sits on the firms' side and has a longer horizon than any interest-rate cycle. Without capitalisation, a large segment of the economy stays outside credit no matter how cheap it gets.
7. Four scenarios for the next four quarters
- Dominant scenario (30–45%) — status quo with nominal growth. The credit stock keeps rising 6–9% a year in nominal terms but stagnates or falls in real terms, and the ratio to GDP does not move. Driver: the state's borrowing demand stays large. Constraint: the expenditure ceilings under the excessive deficit procedure. Indicator: the central bank's monthly non-government credit statistics.
- Plausible scenario (25–35%) — monetary easing without a lending recovery. Inflation allows a policy-rate cut, but exposure to the state stays above 25% of assets. Indicator: board communiqués and the structure of banking assets.
- Tail scenario (15–25%) — portfolio quality deteriorates. Growth in foreign-currency lending meets a depreciation, and the non-performing loan ratio rises. Indicator: the NPL ratio reported for the corporate segment.
- Unlikely scenario (under 10%) — rapid unblocking. Fiscal consolidation runs ahead of target, yields fall sharply and banks rotate into corporate lending in under four quarters. Indicator: the share of government securities in banking assets dropping below 20%.
| Prediction | Horizon | Probability | How to check it | Verification source |
|---|---|---|---|---|
| Non-government credit stays below 25% of GDP | 31.12.2026 | 70–85% | Non-government credit stock over nominal GDP; falsified at ≥25% | NBR / INS |
| Banks' government-securities exposure stays above 20% of total assets | 30.06.2027 | 65–80% | Financial Stability Report; falsified below 20% | NBR |
| The 2026 ESA deficit exceeds 6.0% of GDP | 31.03.2027 (Eurostat notification) | 45–60% | Spring fiscal notification; falsified at ≤6.0% | Eurostat |
| The central bank cuts the policy rate by at least 25 basis points | 30.06.2027 | 35–50% | Board communiqués; falsified if the rate stays at 6.50% | NBR |
| The foreign-currency share of the corporate loan portfolio exceeds 55% | 31.12.2026 | 40–55% | Currency breakdown of credit to non-financial corporations | NBR |
Method, definitions and figures
Non-government credit. The term the National Bank of Romania uses for the stock of loans extended by resident credit institutions to the domestic non-bank sector — firms, households, non-monetary financial institutions. It excludes lending to general government, direct cross-border loans, leasing outside the banking system, trade credit between firms and intra-group loans. Those last four are exactly the channels Romanian firms use heavily, which is why „firms under-borrowed from banks" and „indebted firms" can both be true at once.
Why two different figures for the same ratio. Two public pairs circulate for „non-government credit / GDP": 22% for Romania against a 66% EU average (the Financial Stability Report, March 2026) and 40% against 106% (a speech by a deputy governor of the central bank, 8 September 2026). They do not reconcile arithmetically and have not been reconciled by the issuer either. This article uses the first pair as its spine, because its perimeter is explicitly defined, and cites the second only for the bank-exposure figure, which appears only there. Anyone reusing these numbers must say which series they come from.
Derived figures. The 4.7-percentage-point rate gap is the plain difference between 8.5% (new leu loans to firms, Romania) and 3.79% (composite cost-of-borrowing indicator for new loans to non-financial corporations, euro area, June 2026). These are different currencies: the comparison measures the effective cost of local financing, not a monetary-policy differential adjusted for currency risk. The ratio of the 321.5 billion lei lending potential to the 472.4 billion lei stock gives roughly 68%; the numerator refers to March 2026 and the denominator to July 2026, so the ratio is an order of magnitude, not a measurement.
Nominal and real. The stock growth rates quoted as annual percentages are nominal, meaning they include inflation. The real adjustment used here is the one the central bank publishes alongside its monthly monetary indicators, not one computed in this newsroom. The distinction flips the sign of the conclusion: +7.6% nominal and −0.5% real describe the same month.
What the probabilities mean. The ranges in the table and in the scenarios are editorial estimates expressed in bands of at least 10 percentage points, following probabilistic calibration practice. They do not come from a validated statistical model and have not been tested against a history of resolved forecasts. They work as risk markers; they are not measurements.
What is not comparable. „To GDP" ratios are not comparable across states with very different GDP structures — the Irish case, where multinationals' intellectual property inflates the denominator. The negative-equity figures come from an analysis of trade-register data with a different perimeter from the central bank's reporting, and cannot be added to it.
Limits of this analysis
The analysis cannot establish causation in the decisive link: that the money placed in government securities would have reached firms in the absence of the deficit. Credit that was not extended cannot be observed, and separating a supply constraint from low-quality demand cannot be done from aggregated public data.
The figures attributed to the central bank come from financial-press reporting and from a public speech by its leadership, not from a direct extraction of the primary publications; the discrepancy described in the method section remains unresolved as of publication. The average rate on new leu loans (8.5–8.6%) is a publicly reported range, not a value read from an official series.
The diagnosis would be invalidated by any of the following: a fall in banks' exposure to the state below 20% of assets without a corresponding reduction in the deficit; corporate credit accelerating past 15% a year with the ratio to GDP rising; or a revision of the central bank's series moving Romania out of last place in Europe. A full re-evaluation follows the next Financial Stability Report and the Eurostat fiscal notification of spring 2027.