Resource Allocation as Revealed Priorities and Capabilities

In the absence of binding plans, the budget is the only document where priorities become concrete. Serbia's aggregate spending levels should produce more convergence than they do: the problem lies in the functional mix and the poverty of instruments. Security functions absorb a materially larger share of GDP than in the EU, while education and health are underfunded. The instruments through which money flows collapse into simple transfers, wage-bill financing, and centrally managed flagships. Private-sector support favours large investors over SMEs by a ratio exceeding 9:1. Education spending is overwhelmingly absorbed by wages, with almost no programmatic content. Budget reserves function as a parallel allocation channel, reinforcing centralisation. The pattern is consistent with a governance model that prefers central control over dispersed implementation capacity.

V.1 Revealed priorities in aggregate spending

Aggregate spending levels do not explain Serbia's convergence gap, but the functional mix does. General government spending stands below both the EU average and the regional average, yet Serbia allocates a materially larger amount to security functions and a substantially smaller share to human-capital functions. As Table 4 shows, Serbia spends about 6.6% of GDP on public order, safety, and defence combined, compared with about 4.3% in the EU. Health and education together amount to about 10% of GDP in Serbia, compared with about 12% in the EU. These gaps are large enough to affect productivity and social mobility over time. Since the pattern is stable across years, it is best read as a persistent allocation choice rather than a temporary response to shocks.

The persistent underfunding of education and health relative to security functions is itself likely to constrain convergence. It is difficult to build human capital at the rate needed to support rising investment when the spending mix works against it. The substantial spending on security partly reflects the fact that it is easier to administer centrally, but more likely represents an actual policy preference. One reason why transparent wage reform has proved difficult is precisely the wedge between wage levels in security and other public-service sectors. Without a binding development anchor, short-horizon choices are easier to reproduce in successive budgets (European Commission, 2024; SIGMA, 2025).

Table 4: Functional Structure of General Government Expenditure (COFOG), % of GDP

Source: Ministry of Finance of the Republic of Serbia; Eurostat; authors' compilation. Note: Sub-functions shown for Serbia 2024 only. The relatively high share of general public services partly reflects the inclusion of budgetary reserves within this category.¹³

The absence of a prioritisation pipeline and the crudity of programming explain why spending at these levels produces limited results. No development plan, no ranked project list, no medium-term anchor — spending decisions cannot build synergies or select for maximum efficiency of individual projects. This issue is explored further in Chapter VII. Second, although Serbia underwent a long process of introduction of programme-based budgeting (completed in 2015), this was a heroic effort considering the absence of defined goals and the impossibility of exercising judgement documented in Chapter IV. Not surprisingly, different analyses of the process indicate that while the formal structure is in place, programme budgeting practices still "lack substance in some key elements" (PEFA, 2021), including the elements needed to show that results are being measured and evaluated (Transparency Serbia, 2025). We analyse it through the spending structure and find that programmes are realised through necessarily crude instruments. These instruments are visible in the budget, and we examine them in the following section.

V.2 The mechanics of spending: four delivery archetypes

In a well-functioning programme budget, each programme would be defined by a clear objective, a delivery mechanism, and measurable outputs. The spending structure of such programmes typically combines staff costs for programme management, subsidies as incentives, material costs reflecting outside expertise or implementation capacity, and in many cases supporting capital expenditure. In Serbia, programme design often collapses into a limited set of delivery archetypes reflecting the described planning and capability constraints. Only a narrow, well-established set of mechanisms are administratively feasible. Most programmes rely on transfers or subsidies that can be executed through standardised rules (32 and 23 respectively in Table 5), or on a small number of centrally managed capital projects (16). At the same time, large service systems are financed mainly through wage bills, which leaves limited room for performance-based funding. Reserve instruments provide an additional channel for reallocations that can substitute for planned programmes (discussed in Section V.5).

In Table 5 central budget programmes are classified by their dominant delivery mechanism reflected in the 2026 draft budget.¹⁴

Table 5: Delivery Archetypes of Budget Programmes in Serbia

Source: Authors' classification of programmes in the Draft Law on the Budget of the Republic of Serbia for 2026, based on dominant economic category and delivery logic (Government of the Republic of Serbia, 2025).

Agriculture is the most salient example of transfer dominance. Incentives in agriculture and rural development absorb EUR 962 million in the 2026 central budget, with direct payments accounting for 98% of programme spending and 4% of the entire central budget. Funding is determined annually through a government decree listing measures and maximum amounts, turning the "programme" into a schedule of entitlements rather than a results-oriented framework. Annual allocations become subject to repeated negotiations and protest cycles, undermining policy predictability.

The transfer-dominated structure leaves little room for investment-oriented spending. Rural development measures represent a far smaller share of the agricultural budget than direct payments, and within the decree they largely finance advisory and information services at up to 100% of eligible costs, again booked as subsidies rather than asset creation. This imbalance helps explain why agriculture can absorb material public resources while capital investment in the sector remains marginal in the central budget. It also explains why delivery problems persist: even in transfer-heavy schemes, eligibility verification and control still require reliable registries and administrative functions, but here they remain underdeveloped.

The poverty of policy instruments goes a long way toward explaining why implementation of the long-term national agricultural strategy has been poor. A new strategy for 2025–2034 is still not adopted as of December 2025. The EU's CAP provides a benchmark for predictable, investment-oriented support through multiannual strategic plans with payments linked to registries and compliance checks. In Serbia, the 2025 EU report records delays in building CAP-related systems and low IPARD absorption. The State Audit Institution reports that over EUR 58 million invested in irrigation systems between 2021 and 2024 has not delivered expected results due to planning delays (State Audit Institution, 2025). The result is a policy that stabilises incomes but struggles to finance productivity-enhancing assets.

V.3 Private sector support: centrally managed capital and negotiated subsidies

Private-sector support is dominated by centrally managed investment promotion grants and tax incentives triggered by large thresholds — instruments that by design favour investors with scale and negotiating capacity. Table 6 summarises the allocation pattern: FDI-focused instruments account for approximately EUR 434 million, SME-focused schemes for EUR 40 million — a ratio of over 10:1. Moreover, the SME assistance is fragmented into a large number of programmes, reducing the size and material impact of each one.

Table 6: Structure of State Aid Programmes by Beneficiary Type (2023)

Source: Authors' compilation based on state aid reporting and budget documentation.

Even the flagship SME upgrading instrument, the equipment procurement programme, has been scaled down sharply and has not matured into a stable upgrading pipeline. The official programme documentation shows a falling grant envelope over the last cycle: RSD 3.1 billion in 2022, RSD 1.4 billion in 2023, and RSD 0.8 billion in 2024. That shift matters because equipment upgrading is the core channel through which productivity diffusion should occur in a tradables-oriented economy, especially under the cost pressures described earlier in the report. A reduced and unstable envelope turns upgrading into a lottery over annual calls, rather than a predictable instrument that firms can plan around. The evidence from the preceding analysis suggests not merely administrative incapacity but an active preference for instruments that favour scale, centralised management, and political visibility — a pattern consistent with the governance dynamics documented in Chapter IV.

V.4 Education: wage-bill driven sectors

Education is one of the clearest examples of a service system that is financed mainly to keep existing institutions running, rather than to drive improvement in performance. Within the republican budget, wages account for 64.4% of 2024 execution and 73.4% of the 2026 budget in the core education envelope, while core capital spending amounts to only 3.4% and 3.9%, respectively. Even after adding the identifiable state investment project for the renewal and construction of education and science facilities, capital rises only to 4.8% of spending in 2024 and 5.5% in 2026, while wages still absorb 63.5% and 72.3%. Almost the entire nominal increase between 2024 and 2026 is explained by wage growth: total education spending rises by RSD 112.8 billion, of which RSD 111.1 billion is wage growth (authors' calculations based on Ministry of Finance of the Republic of Serbia, 2025).

The programme structure points in the same direction. Most spending is concentrated in broad lines that finance the routine operation of institutions rather than specific improvements in learning, teaching, or infrastructure. In 2026, 96.2% of primary education spending is concentrated in the generic line for the realisation of primary education. In higher education, about 95.2% of spending is concentrated in support lines for universities and higher schools, while only a small share goes to infrastructure modernisation, openness, doctoral and master support, or broader development measures (authors' calculations based on Ministry of Finance of the Republic of Serbia, 2025). The budget therefore finances education mainly as an operating structure, not as a set of policy instruments aimed at raising system performance.

Higher education shows most clearly how additional resources are being absorbed. Spending rises from RSD 68.5 billion in 2024 execution to RSD 96.1 billion in the 2026 budget, but the composition shifts further toward wages rather than investment. The wage share rises from 61.2% to 73.1%, while capital falls from 2.9% to 0.2% of programme spending (authors' calculations based on Ministry of Finance of the Republic of Serbia, 2025). The budget therefore becomes more expensive to maintain without becoming materially more investment-led or more programmatic.

The main weakness lies in the logic of allocation, not in the formal placement of particular investment lines. Serbia's school funding system is divided between the Republic and local self-government, and local governments finance, among other items, capital expenditure in school education (Eurydice, 2023). OECD likewise defines public spending on education as a general-government concept that includes central, regional, and local financing, as well as transfers between these levels (OECD, 2023). The stronger point, therefore, is that even after visible education-related investment is added back into view, the envelope remains heavily payroll-led and weakly programmatic. European Commission guidance identifies strategic planning, project selection, medium-term budgeting, implementation, monitoring, and the integration of capital and recurrent expenditure planning as core features of effective public investment management (European Commission, 2021). That kind of integration is difficult to see in the current education budget structure, where most spending remains concentrated in broad operating lines rather than in a clearer set of delivery instruments.

V.5 Budget reserves: discretion through in-year reallocations

Budget reserves illustrate how discretion enters fiscal policy when programme discipline is weak. The Budget System Law allows in-year reallocations through the current budget reserve for unforeseen expenditures and a permanent reserve for emergencies. In practice, extensive reliance on reserve decisions turns the reserve into a parallel budgeting channel: spending is authorised by government decision after the budget is adopted, often without clear ex ante criteria and with limited explanation of why the expense could not be planned within regular appropriations (Transparency Serbia, 2024).

Oversight analyses indicate that Serbia has used the current reserve for more than exceptional needs. Transparency Serbia finds that reserve decisions have often financed items that recur from year to year, pointing to the reserve substituting for regular programme planning (Transparency Serbia, 2024). The Fiscal Council reports that in 2023 the government adopted around 280 decisions on the use of the current reserve and that spending reached about 2.5 times the amount planned in the original budget (Fiscal Council of the Republic of Serbia, 2024). These decisions covered a wide range of purposes, from subsidies and infrastructure projects to transfers to local governments and one-off payments, with limited information on selection criteria.

When reserve spending becomes routine, the enacted budget becomes a starting point for in-year political allocation rather than a credible plan. Heavy reliance on reserves reinforces centralisation: in 2024, more than 70% of executed reserves were channelled through the General Secretariat of the Government (NKEU, 2025). The executive bypasses programme logic, raising oversight costs and weakening incentives to improve planning. A simple monitoring step would be to report reserve spending by beneficiary and policy purpose, and to treat repeated reserve financing of the same items as a trigger for redesigning programme appropriations.

The instrument poverty visible in each of these spending domains is consistent with a governance model that prefers centrally managed, discretionary allocation. Where Serbia has faced a choice between building dispersed implementation capacity and retaining central control, it has consistently chosen control. This is a revealed preference, not merely an inherited constraint. A government could lack a national development plan and still operate functioning programmes, gradually increasing their depth and sophistication rather than further hollowing them out — as Serbia's trajectory over the past decade suggests has occurred. Serbia's distinctive problem is that the concentration of discretionary authority at the political apex leaves the current system with no incentive to develop programme capacity at all.