Serbia’s Ministry of Finance is preparing to make public enterprises pay for state guarantees attached to their borrowing, introducing a cost for a form of sovereign support that has often been treated as effectively free. The change appears modest—a revision of only two provisions of the Public Debt Law—but it could alter the financing discipline of some of the country’s largest energy, transport and infrastructure companies.
Under the draft amendments, the existing provision that the Republic of Serbia “may” charge a fee for issuing a guarantee would be replaced by a mandatory rule: a fee will be charged for every state guarantee. The finance minister would determine the conditions, collateral requirements, application procedure and level of commission.
The amendment is still in the consultation phase preceding a formal public debate. No fee schedule has been published, leaving the commercial effect dependent on whether the government applies a nominal administrative charge or a risk-based premium that reflects the probability that taxpayers will ultimately repay the guaranteed loan.
That difference is decisive. A small uniform fee would provide limited budget revenue but leave the underlying incentive structure largely unchanged. A properly calibrated guarantee premium could make public enterprises confront the real cost of their credit risk, reduce unnecessary borrowing and distinguish financially viable capital projects from investments that proceed primarily because lenders are protected by the sovereign balance sheet.
Serbia had €41.14bn of public debt at the end of May 2026, equivalent to 43.7 per cent of GDP. Direct state obligations accounted for approximately €39.48bn, while indirect obligations—principally debt covered by public guarantees—stood at about €1.66bn.
The 2026 budget shows a similar guarantee exposure of €1.669bn, comprising approximately €462.2mn owed to domestic creditors and €1.207bn to foreign lenders. The principal beneficiaries include Elektroprivreda Srbije, Srbijagas, Elektrodistribucija Srbije, Elektromreža Srbije, Putevi Srbije, Srbija Kargo, Srbijavoz, Infrastruktura železnice Srbije, Železnice Srbije, Jugoimport SDPR, as well as the cities of Belgrade and Novi Sad.
These guarantees do not necessarily generate an immediate budget payment. They remain contingent obligations as long as the original borrower services the loan. Yet they are economically connected to the state from the moment they are issued because the guarantee transfers part or all of the borrower’s credit risk to the Republic of Serbia.
A bank may be willing to lend to a weak public enterprise at a lower interest rate because repayment is backed by the government. The company captures the benefit through cheaper financing, while the state—and ultimately taxpayers—assumes the downside. Until now, that transfer of risk has not always been matched by a transparent price.
The scale of the risk became clearer in 2025, when Serbia repaid RSD25.6bn of principal on guaranteed loans after the original debtors failed to meet their obligations. At an exchange rate near RSD117 per euro, this represents approximately €219mn transferred from the central budget to creditors on behalf of public-sector borrowers.
This is much larger than the figure of almost RSD2bn sometimes cited for formally “activated guarantees”. The apparent discrepancy arises from accounting and budget presentation. Some repayments are recorded as expenditures for servicing guaranteed obligations even when the legal or statistical classification of activation is narrower. For taxpayers, the distinction is secondary: cash that would otherwise finance public services or reduce borrowing is being used to repay debts incurred by public enterprises.
The Fiscal Council previously estimated that around 70 per cent of the budgetary amount allocated to guaranteed-debt repayment was connected to Srbijagas, with the remaining 30 per cent related primarily to railway companies, including Železnice Srbije, Infrastruktura železnice Srbije, Srbija Kargo and Srbija Voz.
At the end of September 2025, the stock of guaranteed debt was approximately €1.85bn, around €200mn lower than at the end of 2024. EPS accounted for about €585mn, Srbijagas for approximately €470mn, and the railway group for slightly less than €190mn.
By the end of March 2026, the guaranteed-debt stock stood at RSD199.3bn, of which RSD122.6bn, or more than 61 per cent, related to EPS and Srbijagas. The concentration demonstrates that Serbia’s guarantee risk is closely linked to energy policy rather than being evenly distributed across the public sector.
EPS is financially and operationally different from Srbijagas or the railways. It has substantial generation assets, electricity revenue and the capacity to generate profit, although its cash flow remains exposed to hydrology, coal production, wholesale power prices, regulated tariffs and capital expenditure. Srbijagas has historically carried a more complicated mixture of commercial obligations, regulated prices, supply-security responsibilities and legacy liabilities. Railway companies perform public-service functions that are unlikely to become fully commercial without continuing budget support.
A uniform commission would ignore these differences. The appropriate guarantee fee for a profitable transmission-system operator should not be identical to the fee charged to an enterprise with recurring losses, weak liquidity and a record of transferring debt service to the budget.
A credible system should calculate the premium using the borrower’s standalone credit quality before state support. The assessment would include leverage, cash-flow coverage, liquidity, profitability, currency exposure, maturity, collateral and the economic life of the financed asset. It should also account for the expected recovery value if the guarantee is called.
The guarantee’s structure matters as much as the borrower. A limited guarantee covering a defined share of a senior loan presents less risk than an unconditional guarantee covering principal, interest, default interest, fees and other costs. A guarantee supporting a revenue-generating transmission asset has a different profile from one attached to rolling stock or a project dependent on future budget subsidies.
The state-aid dimension is another reason for the legislative change. The proposed amendments explicitly require guarantee decisions to take account of Serbia’s state-aid control rules. A state guarantee offered without an appropriate premium can confer an economic advantage on the borrower because it allows financing on terms unavailable in the market.
For public enterprises engaged in commercial activities, that advantage may constitute state aid. As Serbia aligns its legal system with the European Union, it must demonstrate that public guarantees are transparent, proportionate and, where necessary, priced consistently with market risk.
A market-equivalent fee does not eliminate state aid automatically, but it provides evidence that the enterprise is paying for the credit enhancement received. It also reduces the competitive distortion between state-controlled businesses and private companies that must pay bank-guarantee fees, provide collateral or accept higher interest rates without access to the sovereign balance sheet.
The potential revenue is not negligible, although the calculation depends on how the fee is structured. Applying an illustrative annual commission of 0.5 per cent to the existing €1.67bn guarantee stock would generate around €8.3mn a year. A 1 per cent fee would produce approximately €16.7mn, while a 2 per cent average charge would yield about €33.4mn.
These figures are only scenarios. The eventual law may apply primarily to new guarantees rather than the entire existing portfolio, and fees may be charged once at issuance rather than annually on the outstanding balance. Some loans from international financial institutions may also be subject to specific agreements or policy considerations.
The main fiscal benefit will not come from commission income. Even a 1 per cent annual fee on the entire guarantee portfolio would recover less than one-tenth of the approximately €219mn of guaranteed principal paid by the state in 2025. The stronger benefit would come from reducing future defaults and discouraging projects whose repayment capacity exists only because the government stands behind them.
Pricing changes behaviour when the charge is sufficiently material. A public company paying a higher premium because of weak leverage or poor debt-service coverage has an incentive to improve its balance sheet before seeking another guarantee. Management must incorporate the commission into project returns and operating budgets rather than presenting the state guarantee as a costless financing instrument.
The fee can also improve project selection. The current law permits guarantees only for loans financing capital investment and prohibits them for routine operations or liquidity needs. In practice, however, the boundary can become blurred when a project does not generate sufficient cash and the enterprise relies on budget support to service its debt.
A capital asset is not automatically a financially viable investment. A railway project may have substantial economic and social value but limited direct revenue. A gas project may improve supply security but depend on regulated tariffs. An electricity-grid investment may generate regulated income over several decades, while a poorly designed industrial project may never cover its financing cost.
Guarantee pricing should force these differences into the approval process. Projects that perform a public-service function can still receive support, but the subsidy and fiscal exposure should be recognised openly rather than hidden through guaranteed borrowing.
The proposal also has implications for lenders. A sovereign guarantee reduces the bank’s need to analyse the underlying borrower because repayment depends ultimately on the state. Mandatory fees will not change the guarantee’s legal strength, but they could reduce the volume of guaranteed borrowing or lead enterprises to seek unsecured financing where they have sufficient standalone credit quality.
That would be a healthy development for stronger public companies. EMS, for example, operates regulated transmission infrastructure and may be capable of raising financing based on its own cash flow for appropriate projects. EPS, following its conversion into a joint-stock company and recent profitability improvement, may progressively move towards borrowing structures in which state support is limited rather than automatic.
Removing guarantees abruptly would increase interest costs and could delay critical energy and infrastructure investments. The better approach is gradual differentiation: commercially robust enterprises should borrow increasingly on their own credit, while state guarantees remain available for projects with demonstrable public value and limited alternative financing.
For weak enterprises, a higher guarantee fee creates a difficult circular effect. The companies most likely to default are also least able to pay a substantial risk premium. Charging them a market-based fee may worsen their cash flow, while charging a nominal fee fails to compensate the state for the risk.
This is where ownership policy and public-service contracts become essential. A railway company required to operate unprofitable passenger services cannot be evaluated as though it were a private freight operator. The state should compensate clearly defined public-service obligations directly and assess the company’s remaining commercial activities separately.
Similarly, Srbijagas should not carry policy costs indefinitely through its balance sheet if tariffs, strategic storage, emergency supply or social measures are set by the government. Mixing commercial operations with unpriced public-policy obligations makes it difficult to determine whether a guarantee supports an economically weak enterprise or finances a government policy that has not been transparently budgeted.
Guarantee commissions will work only when accompanied by better disclosure. Serbia’s monthly public-debt reports show the aggregate difference between direct and indirect obligations but do not provide a complete company-by-company schedule of guarantees, maturities, currencies, creditors, interest rates and repayment performance.
The annual budget contains more detail, but the information remains fragmented. Investors and taxpayers cannot easily determine which guaranteed loans are performing, which companies are reimbursing the state and which obligations are likely to migrate onto the budget.
A stronger system would publish the original guarantee amount, outstanding balance, beneficiary, creditor, project purpose, currency, maturity, fee rate, risk classification and any payments made by the state. It would also disclose recoveries from the borrower after a guarantee is activated.
Recovery is central to fiscal discipline. When Serbia repays a guaranteed loan, the payment should create a receivable from the public enterprise rather than becoming an unexplained permanent subsidy. The government should report how much of these claims has been recovered, restructured, converted into equity or written off.
Without recovery data, the public cannot determine whether guarantee activation is temporary liquidity support or an irreversible transfer from the budget. A commission helps compensate for expected losses, but it cannot replace enforcement of the state’s claim against the original debtor.
The reform arrives while Serbia’s headline debt ratio remains comparatively moderate. Public debt of 43.7 per cent of GDP is below the Maastricht reference value of 60 per cent and lower than in many EU economies. The government has also retained a 3 per cent of GDP fiscal-deficit ceiling for 2026.
Sovereign investors, however, evaluate more than the reported debt ratio. They examine guarantees, state-enterprise liabilities, public-private partnerships, local-government obligations and other commitments capable of becoming central-government debt. A country with moderate recorded debt but weak control of contingent liabilities can still face a higher sovereign-risk premium.
Transparent risk-based guarantee pricing could therefore contribute to lower borrowing costs at the sovereign level. It would signal that Serbia is measuring the value of its support, reducing moral hazard and aligning public-company financing with state-aid principles.
The effect on credit spreads would be gradual. Investors will not reprice Serbian bonds because of a single legislative wording change. They will look for evidence that fees are actually collected, weak borrowers are charged more, guarantee applications can be rejected and activated payments are recovered.
The same evidence will matter to rating agencies and international financial institutions. Serbia’s energy and transport companies have repeatedly been identified as major fiscal-risk channels. A guarantee framework that distinguishes commercially sustainable investment from implicit operating subsidies would strengthen the credibility of public-finance management.
The danger is that the amendment becomes primarily a revenue measure. Charging all enterprises a small standard commission would allow the government to claim that guarantees are no longer free, while doing little to change borrowing decisions or compensate for expected loss.
The fee must be linked to risk and updated over the life of the guarantee. A company whose financial condition deteriorates after the guarantee is issued should face closer monitoring, additional collateral requirements or restrictions on further borrowing. A company that improves its credit metrics should benefit through a lower premium on future guarantees.
Guarantee approval should also include stress testing. Energy companies should be assessed under adverse electricity, gas and carbon-price scenarios. Transport companies should be tested against construction delays, lower traffic, cost overruns and weaker budget compensation. Foreign-currency loans should be evaluated against dinar depreciation where revenues are primarily domestic.
For major capital projects, the state should calculate the expected loss as the probability of default multiplied by the loss after recoveries. This would allow the guarantee commission and budget reserve to reflect the economic value of the risk rather than an arbitrary administrative rate.
The draft amendment is consequently more important than its limited legal scope suggests. Serbia is beginning to acknowledge that a sovereign guarantee is a financial asset transferred to a public enterprise and a contingent liability imposed on the budget. It has a measurable value and should carry a price.
The credibility of the reform will depend on the commission schedule, the transparency of individual guarantees and the government’s willingness to refuse support where the project or borrower cannot justify the risk. With €1.67bn of guaranteed debt outstanding and RSD25.6bn of principal repaid for borrowers in a single year, symbolic pricing would be insufficient.
A guarantee fee that reflects credit quality would not end Serbia’s exposure to public enterprises, nor should it prevent the state from financing essential infrastructure. It would make the cost visible at the point when the borrowing decision is taken, rather than years later when a repayment falls onto the budget.
