HEALTH FINANCING AND ACCOUNTING

A POSTGRADUATE DIPLOMA PUBLICATION

A Comparative Quantitative Analysis of Kenya, Poland, and Alberta

By Dominic Okoro

New York Center for Advanced Research (NYCAR)

Research Division — Health Financing and Public Financial Management

Institutional Review · July 2026

Publication No.: NYCAR-TTR-2026-RP074

DOI: https://zenodo.org/records/22028924

Peer Review Status

This postgraduate publication has undergone independent peer review conducted under the joint editorial framework of the New York Center for Advanced Research (NYCAR) and The Thinkers’ Review. Independent reviewers assessed the research for academic coherence, source integrity, financial and methodological rigor, scientific voice, and APA 7th edition alignment. Each quantitative model was independently re-derived, every cited source independently verified, and the work cleared for release only on the basis of that independent assessment.

The cover carries independent peer review because the research compares public financial management performance across separate national and subnational health budgets reported under different accounting bases.

Abstract

Budget Execution Gaps in Primary Healthcare Financing: A Comparative Quantitative Analysis of Kenya, Poland, and Alberta examines the distance between what health systems approve and what they actually spend. The research treats budget execution as a management control problem rather than an accounting formality, because an approved allocation that is not converted into service delivery buys nothing, and an overrun that is settled by in-year revision destroys the meaning of the approval. The central problem is not the size of the health budget. It is credibility: whether the number voted at the start of the year predicts the money that reaches primary care facilities by the end of it.

The evidence base is read through published budget implementation reports, statutory financial statements, ministry annual reports, and payer financial plans, with every reported ratio recomputed from the source denominators. Execution rates are calculated against original approved budgets and scored against the Public Expenditure and Financial Accountability PI-1 credibility bands. Within-year absorption is modeled against a linear execution benchmark. The movement from approved plan to realized cost is decomposed into its reported components. The quantitative layer is deliberately modest: it is built to expose where budgetary claims exceed budgetary residue, not to manufacture precision that public filings cannot support.

Execution failure proves to be bidirectional and structurally different in each system. Kenya’s 47 counties spent KSh 124.0 billion of an approved KSh 218.99 billion development budget, an execution rate of 56.6 percent, against recurrent execution of 90.6 percent; national health sector absorption ran at 84.3, 84.8, and 82.1 percent across three consecutive years. Poland’s National Health Fund realized costs of PLN 190.7 billion against an initial plan of PLN 164.7 billion, an execution rate of 115.8 percent driven by a PLN 26.0 billion in-year revision, with free medicines executing at 300 percent of plan. Alberta Health delivered a derived execution rate of 102.4 percent. Only Alberta falls inside the PI-1 A band; Kenya and Poland both score D, from opposite directions. Mean absolute deviation from the approved budget across the five headline measures is 17.8 percentage points, with a spread of 59.2 points.

The core argument is that underspending and overspending are symptoms of the same failure, which is the decoupling of the approved budget from the operating calendar. Kenya’s counties spent 54.1 percent of the entire year’s development budget in the final quarter, at 3.5 times the rate of the preceding nine months, which is a procurement and disbursement timing failure rather than an absence of need. The research finds that credible health financing does not depend on the accuracy of the original estimate. It depends on whether cash release, procurement, and commitment control are synchronized to the fiscal year, and on whether the entity carries obligations it has not recognized. Kenyan county pending bills of KSh 176.9 billion stand at 186.2 percent of the unspent development balance, which means the apparent underspend conceals a net liability rather than an idle surplus.

Keywords: budget execution; budget credibility; absorption rate; primary healthcare financing; public financial management; PEFA PI-1; National Health Fund; county government financing; pending bills; comparative health systems.

Table of Contents

List of Tables

Table 1: Case evidence matrix — included sources and their reporting basis

Table 2: Budget execution audit — outturn against approved budget with PEFA PI-1 scoring

Table 3: Within-year absorption schedule — Kenya county development budget

Table 4: Plan-to-outturn decomposition — Polish National Health Fund, 2024

Table 5: Quantitative model audit — equations, variables, and quality limits

Table 6: Composite credibility index and allocative comparison

Table 7: Implementation control schedule — actions, owners, and verification

Table 8: NYCAR quality-control checklist

List of Figures

Figure 1: Budget execution deviation with PEFA PI-1 credibility bands

Figure 2: Within-year development budget absorption in Kenya’s counties

Figure 3: Plan-to-outturn decomposition of the Polish National Health Fund, 2024

Chapter 1: Context, Research Problem, and Professional Significance

The management problem

The analysis places budgetary intentions beside audited outturns, disbursement records, and the procurement calendar that determines whether the two ever meet.

A health budget is a promise expressed in currency. It states what a government intends to buy on behalf of its population during a defined period, and it carries legislative authority precisely because that statement is supposed to be reliable. Budget execution is the process by which the promise becomes a purchase. Where execution is weak, the approved figure loses its meaning as a planning instrument, and every downstream activity that depends on it — facility staffing, commodity procurement, contractor engagement, service expansion — inherits the unreliability.

This matters more in primary healthcare than almost anywhere else in public finance. Primary care is delivered through a large number of small facilities with limited financial autonomy, short commodity cycles, and almost no reserve capacity. A tertiary hospital can absorb a delayed disbursement across a large balance sheet. A dispensary with an empty commodity shelf cannot. The consequence of a late or unspent allocation at primary level is not an accounting variance; it is a stockout, an unfilled post, or a facility that closes early.

The measurable issue across the three systems examined here is credibility, not adequacy. Every system studied could argue that its health budget is too small relative to need, and each would have a case. That argument is not the subject of this research. The subject is whether the approved number predicts the spent number. A weaker system treats the approved budget as an opening negotiating position, subject to in-year revision whenever pressure builds. A stronger one treats it as a constraint that management is accountable for meeting from both directions. That difference is small in language and very large in operating consequence.

The anchor figures used throughout this research define the scale of the problem. Kenya’s 47 county governments spent KSh 124.0 billion of an approved KSh 218.99 billion development budget in FY 2024/25 (Office of the Controller of Budget, 2025b). Poland’s National Health Fund realized costs of PLN 190.7 billion against an initial approved plan of PLN 164.7 billion. Alberta approved a health operating budget of CAD 24.5 billion for 2023-24 and forecast an outturn of approximately CAD 25.1 billion. These are not decorative figures. They define the range of execution performance that public health financial management currently produces, and they bound what a manager working inside any of these systems can reasonably expect from an approved allocation.

Published evidence and institutional mechanics

The three systems were selected because they fail differently, and because the difference is instructive. Kenya operates a devolved model in which 47 county governments receive an equitable share from the national exchequer and administer health as a devolved function. Poland operates a single-payer model in which one central fund, the Narodowy Fundusz Zdrowia, contracts for all publicly financed services. Alberta operates a ministry-and-authority model in which a provincial department funds a small number of large delivery organizations, with ministry funding accounting for the large majority of each organization’s budget (Alberta Health Services, 2024). Devolved, single-payer, and integrated-provincial are the three dominant structures in public health financing, and each generates a characteristic execution pathology.

In Kenya the pathology is underspending. The Office of the Controller of Budget (2025b) reported that several county governments recorded development absorption below ten percent, and attributed the shortfall to weak own-source revenue and delays in the release of exchequer funds by the National Treasury. Both causes sit upstream of the spending unit. A county health department cannot procure against money it has not received, and the procurement cycle for capital items does not compress to fit whatever part of the year remains once the cash arrives.

In Poland the pathology is overspending resolved by revision. The National Health Fund recorded a 2024 loss of PLN 7.7 billion, fully covered by drawing down the reserve fund, which is now effectively exhausted (Narodowy Fundusz Zdrowia, 2025). The difference between initially planned and final healthcare expenditure in the Fund’s financial plan was PLN 26 billion in 2024 and more than PLN 23 billion in 2023. A fund that was previously largely self-financing has become deficit-driven, and the approved plan has become a document that is revised rather than met.

In Alberta the pathology is neither, and the case functions as a control. Aggregate execution sits close to the approved figure. The interesting question in Alberta is therefore allocative rather than aggregate: whether a system that reliably spends what it approves is approving the right composition. Budget 2024 allocated CAD 475 million to primary care modernization against CAD 6.6 billion for physician compensation and CAD 4.4 billion for acute care (Government of Alberta, 2024a), which is a statement about priority that no execution ratio will reveal.

An approved budget that is routinely revised is a forecast, not an authorization.

Aim, objectives, and research questions

The aim of this research is to measure and compare budget execution performance in health financing across Kenya, Poland, and Alberta, and to determine whether execution failure in these three systems arises from a common mechanism despite presenting in opposite directions.

1. To compute budget execution rates for health expenditure in each system against the original approved budget, using published financial reports.

2. To score those rates against an established budget credibility standard.

3. To model within-year absorption against a linear execution benchmark, and to quantify the timing component of the execution gap.

4. To decompose the movement from approved plan to realized cost where in-year revision occurs, and to identify the reported drivers.

5. To test whether the apparent underspend in the devolved case represents an idle balance or an unrecognized liability.

6. To derive implementation controls for health finance managers and measurement standards for oversight institutions.

Five research questions follow: what execution rate does each system achieve; how do those rates score against a credibility standard; how is spending distributed within the fiscal year; what drives the movement from plan to outturn where revision occurs; and whether unspent balances are matched by unrecognized obligations.

Research hypotheses

H1:

Health budget execution rates in the three systems fall within the PEFA PI-1 A band of 95 to 105 percent of the approved budget.

H2:

Within-year absorption in the devolved case follows a linear execution path.

H3:

Unspent development balances in the devolved case represent an idle surplus rather than a net obligation.

Professional significance

For health finance managers, the research separates two questions that budget commentary habitually merges: whether the allocation was adequate, and whether the allocation was executed. These are independent, and the second is the one a manager controls. A department that absorbs 56 percent of its capital budget has a management problem regardless of whether the budget was generous or mean, and arguing for a larger allocation while absorbing half the existing one is a weak position at the negotiating table.

For oversight institutions, the research demonstrates that a single annual execution ratio conceals the information that matters. A department reporting 56.6 percent absorption and one reporting the same figure with an even quarterly distribution have very different problems, and only the within-year series distinguishes them.

The scope is confined to publicly financed health expenditure reported in official documents. Private and out-of-pocket spending is excluded. Development and recurrent classifications are analyzed separately where the sources permit, because they behave differently and pooling them conceals the capital execution problem. The comparison is across systems reporting on different accounting bases, in different currencies, and with different fiscal calendars, and no currency conversion is performed at any point.

The chapter treats the approved budget as a control document, not as a statement of aspiration.

Chapter 2: Literature, Theory, and Evidence Base

Budget credibility as a concept

Budget credibility is the degree to which a government’s actual revenues and expenditures correspond to the amounts approved in its enacted budget. The concept acquired formal operational status through the Public Expenditure and Financial Accountability framework (PEFA Secretariat, 2019), which scores aggregate expenditure outturn against the original approved budget and assigns ratings on the tolerance the deviation falls within. The framework’s underlying logic is that a budget which cannot predict spending cannot discipline it, and that the credibility of the aggregate is a precondition for any meaningful discussion of composition or efficiency.

The literature distinguishes several failure modes. Aggregate deviation captures whether total spending matches total approval. Compositional deviation captures whether the spending landed in the sectors and programs it was approved for, and can be severe even where the aggregate is perfect. Timing deviation captures whether spending occurred when planned, and is the least reported of the three despite being the most tractable for management action. This research addresses aggregate and timing deviation directly and treats composition through the allocative comparison in Chapter 5.

The direction of deviation is treated asymmetrically in most commentary, and the asymmetry is not well justified. Overspending attracts attention because it produces a deficit that must be financed. Underspending attracts less, and is sometimes presented as fiscal prudence. In service delivery terms the second is frequently worse, because an overrun at least purchased something while an underspend purchased nothing and surrendered the appropriation. The International Budget Partnership’s (2021) analysis of Kenyan county budgets makes this point through a specific and useful distinction: counties spent 93 percent of the funds actually issued to them, against only 61 percent absorption of approved development budgets across the same period.

The spending units were not the binding constraint. The disbursement was.

Theoretical perspectives

Principal-agent theory.

The budget is a contract between a legislature acting as principal and an executive acting as agent, and execution variance is a measure of contractual non-performance. The theory predicts that where monitoring is weak and the cost of deviation is low, agents will deviate, and it predicts that deviation will be larger for expenditure classes where performance is hardest to observe. Capital spending fits that description precisely: a delayed road or a half-built health center is easier to explain away than an unpaid salary, which is why development execution is worse than recurrent execution in every system examined here.

Cash rationing and the fiscal transmission chain.

In systems where the treasury releases cash periodically rather than granting spending units access to their full appropriation, execution becomes a function of release timing rather than of managerial capacity. This mechanism dominates the Kenyan case and explains why the same departments that absorb only 26 percent of their development budget in nine months absorb 93 percent of what they actually receive. The appropriation and the cash are different instruments, and only the second can be spent.

Soft budget constraint theory.

Where an entity expects that overspending will be covered by a superior authority, it will not treat the approved budget as binding. Kornai’s formulation was developed for state enterprises in centrally planned economies, and it transfers directly to a single-payer health fund that can draw on a reserve or receive a state subsidy when costs exceed plan. The Polish case exhibits the mechanism in its textbook form: a reserve fund absorbed a PLN 7.7 billion loss and is now practically exhausted, which means the constraint is about to harden whether or not the Fund is ready for it.

Commitment control and the recognition problem.

An expenditure that has been committed but not paid does not appear in the execution ratio, which means that a system with weak commitment controls can report an underspend while accumulating arrears. This is the single most important qualification on any absorption figure, and it is the reason the pending bills analysis in Chapter 5 is not an aside but a test of whether the headline number means what it appears to mean.

Measurement standards and their limits

The PEFA PI-1 indicator scores aggregate expenditure outturn against the original approved budget on a four-point scale. The A rating requires that the outturn fall between 95 and 105 percent of the approved figure in at least two of the last three years; B widens the band to 90 and 110 percent; C to 85 and 115; and D applies to anything outside those bounds. The framework is used here as a scoring standard rather than as a full assessment, since a formal PEFA assessment requires an evidence set well beyond published financial reports.

Three limitations of the standard bear on this research. It is symmetric, treating a ten point underspend and a ten point overspend as equivalent, which the service delivery consequences do not support. It is annual, and therefore silent on within-year distribution. And it is applied to the aggregate, so a system can achieve an A rating while the composition beneath it has shifted substantially. Each of these is addressed by an additional analysis in Chapter 5 rather than by abandoning the standard.

A further measurement problem concerns the choice of denominator. Execution can be computed against the original approved budget, against a revised or supplementary budget, or against funds actually released. These produce very different numbers, and the difference between them is itself diagnostic. Computing against a revised budget flatters any system that revises freely, which is why this research uses the original approved figure as the primary denominator throughout and reports the alternative denominators separately.

Empirical evidence by system

Kenya.

The Office of the Controller of Budget publishes budget implementation review reports for national and county government under a constitutional mandate, and these constitute the primary evidence base. For FY 2024/25 the 47 counties spent KSh 124 billion on development against an approved development budget of KSh 218.99 billion, and KSh 346.98 billion on recurrent expenditure at an absorption rate of 90.6 percent. Personnel emoluments accounted for 63.59 percent of recurrent expenditure. The Public Finance Management Act of 2012 requires that at least 30 percent of a county budget be directed to development (Republic of Kenya, 2012), a requirement that the aggregate execution record does not meet in practice.

The within-year record is more informative than the annual total. At the half-year point counties had spent KSh 33.60 billion on development, 16 percent of the annual development budget, itself an improvement on the 12 percent recorded at the same point in the previous year. By nine months, cumulative development spending had reached KSh 56.87 billion, an absorption rate of 26 percent, with development accounting for only 20 percent of total county spending against 80 percent recurrent. Health’s share of county budgets rose from 23 to 25 percent of expenditure over the period reviewed by the International Budget Partnership, so the sector was gaining allocative ground while the execution problem persisted.

At national level the picture is comparable. The State Department for Public Health and Professional Standards recorded the highest absorption among health entities in the first quarter of FY 2024/25 (Office of the Controller of Budget, 2025c) at 33 percent for development and 22 percent for recurrent, while the State Department for Medical Services recorded 17 and 16 percent respectively. Health sector budget absorption reported in the medium-term expenditure framework documentation stood at 84.3, 84.8, and 82.1 percent (National Treasury of Kenya, 2024) for FY 2021/22, 2022/23, and 2023/24, a consistent shortfall of roughly one sixth of the approved budget sustained across three years.

Poland.

The National Health Fund operates as a single central payer receiving all health insurance contributions and financing hospitals, clinics, and pharmaceutical programs. The structural consequence is that any deficit in the Fund transmits immediately to patient access, because the system has no alternative payer and no parallel stabilization mechanism (World Health Organization Regional Office for Europe, 2024).

For 2024 the Fund reported realized total costs of PLN 190.7 billion, more than PLN 26 billion above the costs incurred in 2023, and a loss of PLN 7.7 billion which was fully covered by the reserve fund. The loss was PLN 1.7 billion smaller than planned and smaller than the 2023 loss, but the reserve that absorbed it has been practically zeroed. Administrative costs consumed less than 0.7 percent of the payer’s budget, a figure low enough that several parliamentary deputies raised it as a concern in committee rather than as an achievement.

The composition of realized cost places hospital treatment at 51.58 percent, primary care at 11.15 percent, and outpatient specialist care at 9.74 percent. Two components drove the overrun. Planned outlays for services performed over contracted limits rose to PLN 6.3 billion in 2024, against PLN 2.2 billion in 2023 and PLN 760 million in 2022. The cost of free medicines for children and seniors reached more than PLN 3 billion against a planned PLN 1 billion, following the extension of the program to children and adolescents and the lowering of the senior age threshold from 75 to 65. The second is a policy decision executing at three times its own plan.

Primary care reform provides the allocative counterpoint. Entrusted budgets were introduced into Polish primary healthcare in July 2022 to expand diagnostic access and specialist consultation at the primary level. By 2024 the share of primary care physicians contracting for coordinated care had reached 40.2 percent nationwide, and by 2025 only 43.1 percent, with variation between voivodships running from 24.8 percent upward. Uptake has effectively stalled. A reform can be funded and still not execute, and the constraint here is provider participation rather than money.

Alberta.

Alberta publishes ministry annual reports under the Financial Administration Act and the Sustainable Fiscal Planning and Reporting Act, with audited consolidated financial statements and a comparison of actual performance results against the business plan. Budget 2023 provided a health operating budget of CAD 24.5 billion (Alberta Health, 2024). Budget 2024 set operating expense at CAD 26.2 billion, described as an increase of CAD 1.1 billion or 4.4 percent over the 2023-24 forecast, which places that forecast at approximately CAD 25.1 billion.

Reported 2023-24 spending on major cost drivers comprised CAD 5.0 billion on hospital services (Alberta Health, 2024), CAD 6.4 billion on physician compensation and development, and CAD 2.8 billion on drugs and supplemental health benefits. Approximately CAD 735 million was spent on health capital projects. Provincial per capita health spending in 2022-23 was CAD 5,476, below the Canadian average of CAD 5,749 and below the average of British Columbia, Ontario, and Quebec at CAD 5,748.

Primary care received CAD 243 million over three years under Budget 2023 to develop new models and stabilize the system, of which CAD 125 million was directed to implementing recommendations from the Modernizing Alberta’s Primary Health Care System initiative. Budget 2024 raised the primary care allocation to CAD 475 million, including CAD 300 million for Primary Care Networks and CAD 200 million over two years to improve access to family physicians. Against a CAD 26.2 billion operating budget this is 1.81 percent, and the comparison with Poland’s 11.15 percent primary care share is the allocative finding that the aggregate execution ratio conceals.

Gaps and conceptual framework

The gaps are of a piece. Comparative execution analysis across health systems at different income levels is rare, and where it exists it usually compares like with like. Within-year absorption is seldom modeled against an explicit benchmark despite the data being published quarterly in several jurisdictions. The relationship between reported underspending and unrecognized arrears is acknowledged in the public financial management literature but rarely quantified against the same denominator. And execution analysis is almost always conducted on the aggregate, leaving the primary care share unexamined.

The conceptual framework treats the approved budget as an authorization that must pass through three gates before it becomes service delivery. The first is cash release, controlled by the treasury or contribution flow. The second is commitment, controlled by procurement and contracting. The third is payment, controlled by the accounting function. Execution failure can occur at any gate and presents identically in the annual ratio, which is why the framework predicts that the same headline number can conceal entirely different problems and why the diagnostic value lies in the within-year and arrears analyses rather than in the ratio itself.

Read also: Healthcare Practice and Strategic Management in Barbados

Chapter 3: Methodology, Data Integrity, and Analytical Boundaries

Philosophy, design, and justification

The research adopts a post-positivist position. It assumes that budget execution is a measurable property of a public financial management system, while accepting that measurement is conditioned by accounting basis, classification convention, and the choice of denominator, and that any single ratio is provisional. The approach is deductive: hypotheses derived from the framework in Chapter 2 are tested against extracted figures. The reasoning is quantitative throughout, and the interpretation is explicitly forensic about what a published ratio can and cannot support.

The design is a comparative secondary analysis of published financial reports. It is not an audit, since no underlying transaction records were examined, and it is not a formal PEFA assessment, since the evidence set required for one extends well beyond published documents. It is best described as a structured execution audit: reported figures are extracted, ratios are recomputed from source denominators rather than accepted as printed, and comparison is made only where the underlying constructs are equivalent.

This design was selected because the substantive question concerns whether published budgetary claims survive recomputation, which is answerable from public filings and is not answerable by primary data collection at postgraduate diploma scale. The design also makes the comparability problem visible: incomparabilities appear in the extraction matrix rather than being dissolved into a single cross-country index.

Sources, inclusion criteria, and extraction

The sources fall into three categories. Statutory budget implementation reports supplied the Kenyan evidence, principally the county and national government budget implementation review reports issued by the Office of the Controller of Budget under Article 228(6) of the Constitution, supplemented by medium-term expenditure framework documentation from the National Treasury. Payer financial statements and committee proceedings supplied the Polish evidence, principally the National Health Fund’s 2024 activity report and financial statements as approved by the Health and Public Finance Committees. Ministry annual reports and fiscal plans supplied the Alberta evidence, principally the Health annual report prepared under the Sustainable Fiscal Planning and Reporting Act and the associated budget documents.

Figures were included where the source was official or reported the official figure directly; where both a numerator and a denominator were recoverable, or where a published ratio could be reconciled against a stated total; and where the expenditure class was identifiable as development, recurrent, or aggregate. Figures were excluded where the denominator could not be established, where the reporting period could not be matched across the comparison, and where the figure related to private or out-of-pocket expenditure.

For each included figure the following fields were extracted: jurisdiction; reporting entity; fiscal year and period covered; expenditure classification; approved budget; revised budget where stated; funds released where stated; actual expenditure; reported execution rate; currency; and accounting basis. Where a source published a rate without the underlying values, the values were reconstructed by applying the rate to the stated total and the reconstruction is flagged in the audit note column of Table 2.

Variables and analytical procedures

The primary variable is the budget execution rate, defined as actual expenditure divided by the original approved budget, expressed as a percentage. The original approved figure is used as the denominator throughout because a revised denominator flatters any system that revises freely, which is the precise behavior under examination in the Polish case. Secondary variables are the within-year cumulative absorption rate, the plan-to-outturn revision, the pending bills stock, and the primary care share of total health expenditure.

Five analytical procedures were applied, all computed in Python 3 using NumPy and SciPy, with the analysis script supplied as a companion file so that every figure can be recomputed independently.

Execution ratio computation.

Each ratio was recomputed from the source numerator and denominator rather than accepted as published. Where the recomputed value differed from the published value, the difference is reported in the audit note. Deviation from the approved budget is reported in percentage points rather than as a ratio, because percentage points are additive and directly interpretable by a budget holder.

Credibility scoring.

Each execution rate was scored against the PEFA PI-1 bands: A for deviation within 5 percentage points, B within 10, C within 15, and D beyond. The scoring is applied symmetrically as the framework specifies, and the asymmetry of the underlying service delivery consequences is addressed in the discussion rather than by adjusting the scale.

Within-year absorption modeling.

Cumulative absorption at six, nine, and twelve months was compared against a linear execution benchmark under which a constant share of the budget is spent in each period. The execution deficit at each observation is the difference in percentage points. A least-squares line was fitted through the cumulative series to establish whether the observed path is linear, and the coefficient of determination is reported.

Plan-to-outturn decomposition.

Where in-year revision occurred, the movement from approved plan to realized cost was decomposed into reported components, with the unexplained portion carried as a residual and labeled as such. No component was inferred where the source does not state it.

Arrears reconciliation.

The pending bills stock was expressed against two denominators: the annual development budget, and the unspent development balance. The second is the diagnostic ratio, because a value above 100 percent establishes that the entity owes more than it failed to spend, which converts an apparent surplus into a net obligation.

Data integrity, ethics, and analytical boundaries

Public figures are retained in their reported currencies throughout. No cross-currency conversion is performed at any point, because conversion at any single exchange rate would impose a false precision on figures drawn from different fiscal years and would invite comparisons of magnitude that the research does not support. All comparison is made on ratios, which are currency-free.

Fiscal calendars differ across the three systems. Kenya operates a July to June year, Alberta an April to March year, and the Polish National Health Fund a calendar year. The within-year analysis is conducted within the Kenyan calendar only, and no attempt is made to align periods across jurisdictions, because alignment would require assumptions about intra-year distribution that the sources do not support.

Accounting bases also differ. Alberta reports on an accrual basis under Canadian public sector accounting standards. Kenyan county reporting is substantially cash-based, which is precisely why the pending bills stock is material and why an absorption figure computed on a cash basis overstates fiscal performance. The Polish Fund reports on a basis that recognizes contracted liabilities. These differences are not reconciled; they are stated, and their direction of effect is noted wherever a comparison touches them.

The research analyzes published aggregate financial data, involves no human participants, and required no institutional review board approval. All figures are attributed. No value has been estimated, simulated, or imputed to fill a gap, and where a required figure could not be recovered from public sources the absence is stated as a limitation rather than filled by assumption. One derived figure appears in the analysis, the Alberta 2023-24 outturn, which is reverse-engineered from the stated percentage increase in the subsequent budget; it is labeled as derived at every point of use.

Internal validity is limited by reliance on self-reported official figures that this research did not audit. External validity is limited by the selection of three jurisdictions chosen for structural contrast rather than by sampling. Construct validity is limited by classification differences, particularly the boundary between development and recurrent expenditure, which is drawn differently in each system.

The methodology accepts a narrower set of comparisons in exchange for comparisons that survive recomputation.

Read also: Managed Care Models In Healthcare By Cynthia Anyanwu

Chapter 4: Case Evidence and Published-Data Record

Table 1: Case evidence matrix — included sources and their reporting basis

Public figures are retained in their reported currencies. Cross-currency conversion is avoided to preserve source integrity.

The management problem

The source sequence matters because structure, not size, determines how a budget fails.

Two features of the matrix govern everything that follows. The reporting instruments are not equivalent: a constitutionally mandated implementation review report, a payer’s audited financial statement, and a ministry annual report answer different questions under different standards. And the accounting bases differ in a direction that matters for the argument, since a cash-basis absorption figure will always look better than the same system’s accrual position where arrears are accumulating.

The research therefore separates the ratio from the reporting basis in the same way a forensic reading separates a claim from its residue. A 90.6 percent recurrent absorption rate computed on a cash basis and a 102.4 percent accrual execution rate are both meaningful, and they are not the same kind of statement. Where the comparison crosses that boundary, the direction of the distortion is stated rather than adjusted away.

Published evidence and institutional mechanics

The Kenyan record is the most granular of the three because the Controller of Budget reports quarterly under a constitutional mandate. For FY 2024/25 the 47 counties held an approved development budget of KSh 218.99 billion and spent KSh 124.0 billion, while recurrent expenditure reached KSh 346.98 billion at a reported absorption rate of 90.6 percent. Within recurrent spending, personnel emoluments accounted for 63.59 percent and operations and maintenance for 36.41 percent, which establishes that the recurrent budget is substantially a payroll and that its high absorption rate is therefore close to automatic.

The quarterly series is where the diagnostic value sits. At the half year, development spending stood at KSh 33.60 billion, sixteen percent of the development budget then in force (Office of the Controller of Budget, 2025a). The Controller of Budget recorded Mandera at the highest half-year absorption of 32 percent, followed by Narok at 30, Garissa at 28, Uasin Gishu at 27, and Marsabit at 26; at the other end, Baringo and Tana River recorded 7 percent each, Taita-Taveta, Kisumu, Nairobi City, and Nyeri 6 percent each, and Elgeyo-Marakwet, Lamu, Nakuru, and Kitui 5 percent each. By nine months, cumulative development spending had reached KSh 56.87 billion, an absorption rate of 26 percent, with total county spending of KSh 286.49 billion of which only 20 percent was development. The Controller attributed the shortfall to weak own-source revenue and to delays in the release of exchequer funds by the National Treasury.

Two further Kenyan figures complete the picture. County pending bills stood at KSh 176.9 billion as of 30 June (Office of the Controller of Budget, 2025b), an obligation that does not appear in any absorption ratio. And the International Budget Partnership’s multi-year analysis found that counties spent 93 percent of development funds actually issued to them against only 61 percent absorption of approved development budgets, with total spending against issues at 97 percent. The spending units were converting almost everything they received.

The constraint was upstream.

The Polish record inverts the problem. The National Health Fund realized total costs of PLN 190.7 billion in 2024, more than PLN 26 billion above 2023, and recorded a loss of PLN 7.7 billion fully covered by the reserve fund, which is now practically zeroed. The loss was PLN 1.7 billion lower than planned and lower than the previous year’s, so on the Fund’s own terms 2024 was an improvement. The difference between initially planned and final healthcare expenditure in the financial plan was PLN 26 billion in 2024 and more than PLN 23 billion in 2023, which means the plan has been revised upward by roughly one sixth in each of two consecutive years.

The reported drivers are specific. Planned outlays for services performed over contracted limits reached PLN 6.3 billion in 2024 against PLN 2.2 billion in 2023 and PLN 760 million in 2022, a near-eightfold increase across two years. The cost of free medicines for children and seniors reached more than PLN 3 billion against a planned PLN 1 billion, following the extension of the program to children and adolescents and the reduction of the senior threshold from 75 to 65 years. Both are policy decisions taken after the plan was approved, and both were financed by revising the plan rather than by constraining the decision.

The Alberta record functions as the control case and is the least eventful, which is the point. Budget 2023 provided CAD 24.5 billion in health operating expense. Budget 2024 set operating expense at CAD 26.2 billion, described as CAD 1.1 billion or 4.4 percent above the 2023-24 forecast, placing that forecast at approximately CAD 25.1 billion. Reported 2023-24 spending comprised CAD 5.0 billion on hospital services, CAD 6.4 billion on physician compensation and development, and CAD 2.8 billion on drugs and supplemental benefits, with approximately CAD 735 million on health capital projects and CAD 159.4 million on the Connect Care clinical information system against a total expected project cost of CAD 1.47 billion.

Alberta’s per capita position provides the efficiency counterpoint. Provincial per capita health spending in 2022-23 was CAD 5,476, against a Canadian average of CAD 5,749 (Government of Alberta, 2024b) and a British Columbia, Ontario, and Quebec average of CAD 5,748, with Alberta recording the lowest average annual per capita spending growth of the comparator group over five years. A system can therefore execute its budget accurately and spend less per head than its peers at the same time, which disposes of any suggestion that execution accuracy is purchased through generosity.

The chapter closes without a comparative verdict, because the ratios have not yet been recomputed against a common denominator.

Chapter 5: Quantitative Model, Execution Analysis, and Math Audit

The math uses direct ratios recomputed from source denominators. No execution rate below is accepted as published without reconciliation against its own numerator and denominator.

Table 2: Budget execution audit — outturn against approved budget with PEFA PI-1 scoring

Hypothesis H1 stated that execution rates in the three systems would fall within the PEFA PI-1 A band of 95 to 105 percent. H1 is rejected for Kenya and Poland and accepted for Alberta. Of the nine measures audited, one falls in the A band, two in the B band, and six in the D band.

The direction of failure is the finding that the aggregate scoring obscures. Kenya misses the approved budget by not spending it; Poland misses it by spending beyond it. Both receive a D under a symmetric standard, and the standard is correct to treat both as credibility failures, but the management response required is opposite in each case. A Kenyan county needs faster cash release and earlier procurement. The Polish Fund needs commitment control and a harder constraint on in-year policy expansion.

Two rows will not sit quietly. The Kenyan recurrent rate of 90.6 percent scores a B and looks respectable beside the development rate, but recurrent spending is 63.59 percent payroll, and payroll executes itself. A 90.6 percent absorption rate on a budget that is substantially salaries is closer to a measure of establishment vacancy than of financial management. The exchequer issues row, at 93.0 percent, is the most diagnostic number in the table: measured against money actually received rather than money approved, county performance improves by 32 percentage points, which locates the constraint upstream of the spending unit with some precision.

Figure 1: Budget execution deviation with PEFA PI-1 credibility bands

Deviation is plotted rather than the raw ratio so that the two directions of failure are visible on a single axis. Only Alberta lands inside the A band.

Table 3: Within-year absorption schedule — Kenya county development budget, FY 2024/25

Hypothesis H2 stated that within-year absorption would follow a linear execution path. H2 is rejected. A least-squares line fitted through the cumulative series returns a coefficient of determination of .921, which appears high, but the fitted line implies a negative absorption of 28.0 percent at the start of the year and a full-year completion of 53.2 percent, neither of which is admissible. The apparent linearity is an artifact of three points; the underlying path is convex.

The table carries one correction that the source reporting does not make itself. The Controller of Budget published the six-month figure as 16.0 percent, computed against a development budget of KSh 211.53 billion, while the nine-month and year-end figures are computed against KSh 218.99 billion (Office of the Controller of Budget, 2025a, 2025b). The development budget was revised upward by KSh 7.46 billion, or 3.5 percent, part-way through the year. Table 3 restates all three observations against the year-end denominator so that the series is internally consistent, which moves the six-month figure to 15.3 percent. Computing the headline annual rate against the original KSh 211.53 billion instead of the revised figure raises execution from 56.6 to 58.6 percent and leaves the PEFA rating at D. That a denominator moved mid-year inside a budget credibility analysis is not an inconvenience. It is the phenomenon under study, appearing in the measuring instrument.

The substantive finding is in the final column. Of the KSh 124.0 billion spent on development across the entire year, KSh 67.13 billion, or 54.1 percent, was spent in the final quarter. The fourth-quarter spending rate runs at 3.53 times the average rate of the preceding three quarters. This is not an absence of need and it is not an absence of capacity, because the same units demonstrably spent at more than three times their earlier rate once resources permitted.

It is a timing failure, and timing failures at this magnitude have well-documented consequences: procurement compressed into a closing window is procurement conducted under pressure, with the attendant risks to competition, specification quality, and delivery verification.

The execution deficit peaks at nine months rather than at year end, at 49.0 percentage points against 43.4 at close. A county finance officer reviewing performance in April sees a system further behind than it will finish, which creates precisely the incentive to spend at any cost in the closing weeks that the fourth-quarter figure records.

Figure 2: Within-year development budget absorption in Kenya’s counties

The shaded area is the cumulative execution deficit. The convexity of the actual path, not its endpoint, is the management finding.

Table 4: Plan-to-outturn decomposition — Polish National Health Fund, 2024

Reported components are stated as published; the residual is derived by subtraction and is labeled as derived.

The total upward revision of PLN 26.0 billion represents 15.8 percent of the initial approved plan. Two reported components account for PLN 6.1 billion, or 23.5 percent of the revision. The remaining PLN 19.9 billion is carried as a residual because the sources consulted do not disaggregate it, and it is labeled as derived wherever it appears. A research design that assigned that residual to named causes would be manufacturing detail, and the honest presentation is to show that three quarters of the revision is not publicly decomposed.

The two-year pattern matters more than either year alone. Upward revisions of PLN 26.0 billion in 2024 and PLN 23.0 billion in 2023 represent 15.8 and 16.6 percent of their respective plans.

A single large revision is an event. Two consecutive revisions of similar relative magnitude constitute a practice, and a plan that is revised by one sixth as a matter of routine is not functioning as an authorization.

The reserve fund is the mechanism that made the practice sustainable and is the reason it is about to stop. The 2024 loss of PLN 7.7 billion was fully covered by the reserve, which is now practically exhausted.

This is the soft budget constraint hardening in real time: the entity that could previously absorb a plan overrun internally will, from the next cycle, have to seek external cover or reduce commitments.

Figure 3: Plan-to-outturn decomposition of the Polish National Health Fund, 2024

Reported and derived components are distinguished in Table 4. The residual is the largest single element, which is itself a transparency finding.

Table 5: Quantitative model audit — equations, variables, and quality limits

Table 6: Composite credibility index and allocative comparison

The credibility index is simply one hundred less the absolute deviation, which converts the two directions of failure onto a single scale where higher is better. Alberta scores 97.6, Kenyan recurrent 90.6, Poland 84.2, Kenyan health sector 82.1, and Kenyan county development 56.6. The index is a presentational device rather than a validated instrument, and it is offered as such.

The final column carries the allocative finding, and it does not align with the execution finding at all. Poland, which fails the credibility test by overspending, directs 11.15 percent of realized cost to primary care, equivalent to PLN 21.26 billion. Alberta, which passes the credibility test comfortably, allocated CAD 475 million to primary care modernization in Budget 2024, 1.81 percent of the operating budget, against CAD 6.4 billion on physician compensation and CAD 5.0 billion on hospital services in the preceding year. The two figures are not strictly comparable, since the Albertan figure is a modernization allocation rather than the whole of primary care spending, and much of physician compensation funds primary care delivery. The comparison is offered as an indication of stated priority rather than as a like-for-like share.

The point stands nonetheless: execution accuracy and allocative priority are independent properties. A system can spend precisely what it approved and still have approved a composition that under-weights primary care, and no execution ratio will ever detect that.

Arrears reconciliation and sensitivity analysis

Hypothesis H3 stated that unspent development balances in the devolved case represent an idle surplus rather than a net obligation. H3 is rejected, and the rejection is the sharpest quantitative result in the research.

The unspent development balance for the 47 counties is KSh 94.99 billion, the difference between an approved KSh 218.99 billion and a spent KSh 124.0 billion. County pending bills stood at KSh 176.9 billion as of 30 June. Pending bills therefore represent 186.2 percent of the unspent balance and 80.8 percent of the entire annual development budget.

The counties owe KSh 81.9 billion more than they failed to spend.

This inverts the natural reading of the absorption figure. A 56.6 percent execution rate invites the interpretation that money sat unused, and that interpretation supports a policy response about capacity building and procurement training. The arrears position establishes that the aggregate is a net liability, and it supports a different response about commitment control and the recognition of obligations at the point they are incurred rather than at the point they are paid. The two responses have almost nothing in common.

The reconciliation carries an important qualification, stated here rather than buried. Pending bills are a stock accumulated over multiple years, while the unspent balance is a single-year flow, so the ratio compares quantities of different dimension. It is reported as indicative of the direction and rough magnitude of the problem, not as an exact accounting identity, and a proper reconciliation would require the age profile of the pending bills stock, which the sources consulted do not publish.

Sensitivity checks were run on the three assumptions most likely to carry the result. Substituting exchequer issues for the approved budget as the Kenyan denominator raises development performance from 56.6 to 93.0 percent and moves the PEFA score from D to B, which quantifies exactly how much of the apparent failure is attributable to the release mechanism rather than to the spending unit. Recomputing the Polish execution rate against the revised rather than the initial plan produces a rate at or near 100 percent, which is precisely why the revised denominator was rejected in the methodology. And treating the Alberta outturn as equal to the approved budget rather than to the derived forecast changes the execution rate from 102.4 to 100.0 percent and leaves the A rating unchanged, so the single derived figure in the research does not carry the Alberta conclusion.

One check could not be performed. No source consulted publishes Kenyan county health sector execution separately from total county execution at the same level of disaggregation as the development and recurrent split, so the health-specific within-year absorption path cannot be isolated from the all-sector path. The analysis therefore applies the all-sector development profile to the health sector on the assumption that health capital spending follows the same disbursement calendar, and that assumption is stated rather than tested.

Summary of hypothesis testing

Chapter 6: Governance, Institutional Mechanics, and Assurance Analysis

One mechanism, two directions

The central interpretive claim of this research is that Kenya’s underspending and Poland’s overspending are the same failure observed from opposite sides. In both cases the approved budget has ceased to govern the operating calendar. In Kenya the calendar is governed by exchequer release, so spending occurs when cash arrives rather than when the plan says. In Poland the calendar is governed by service demand and in-year policy decisions, so spending occurs as commitments accrue rather than as the plan permits. Neither system is executing its budget; both are executing something else and reporting against the budget afterward.

The evidence for this reading is strongest in the Kenyan alternative denominator. Measured against approved budgets, county development absorption is 61 percent over the multi-year period and 56.6 percent in FY 2024/25. Measured against funds actually issued, it is 93 percent. A 32 percentage point improvement from changing the denominator establishes that the spending units convert almost everything they receive and that the binding constraint sits in the transmission chain. The corresponding Polish evidence is the two-year revision pattern: PLN 26.0 billion and PLN 23.0 billion, 15.8 and 16.6 percent of their respective plans. A plan revised by one sixth in consecutive years is not being executed either; it is being retrofitted.

Alberta’s position confirms the reading by contrast. A ministry-and-authority structure concentrates both the appropriation and the disbursement in the same institutional layer, which removes the transmission problem that dominates the Kenyan case, and operates against an accrual reporting standard with audited statements that make in-year revision visible in a way that a cash-basis system does not. The result is an execution rate of 102.4 percent, achieved while spending less per capita than the national average and less than the average of the three largest comparator provinces.

Why development execution fails first

Every system examined executes recurrent spending better than capital spending, and the gap is large. Kenyan counties absorbed 90.6 percent of recurrent budgets against 56.6 percent of development budgets, a 34 percentage point difference within the same entities in the same year. The Kenyan national health departments show the same pattern inverted at the margins but the same order of magnitude.

The reasons are mundane, and each carries its own control implication. Recurrent budgets are substantially payroll, and payroll executes automatically once posts are filled; 63.59 percent of Kenyan county recurrent spending is personnel emoluments. Capital spending requires a procurement cycle whose duration is largely fixed and does not compress: a tender that takes four months takes four months regardless of when the cash arrives, which means a disbursement received in month eight cannot produce a completed capital project by month twelve. And capital underperformance is politically cheaper than payroll underperformance, because an unbuilt facility generates no immediate constituency while an unpaid salary generates one instantly.

The managerial consequence follows directly. Capital execution is determined before the fiscal year begins, by whether procurement was initiated against the approved budget in advance of cash receipt.

A health department that waits for the first disbursement before starting a tender has already lost the year. This is the single most actionable finding in the research, and it does not require additional resources to implement.

The year-end spending surge as an assurance risk

The finding that 54.1 percent of Kenyan county development spending occurred in the final quarter, at 3.53 times the rate of the preceding nine months, is a governance problem before it is a financial one. Procurement conducted under closing-window pressure is procurement in which competition is curtailed, specifications are drawn from whatever is available, delivery verification is compressed, and the incentive to spend displaces the incentive to spend well.

The pattern also interacts with the arrears position in a way that compounds both. A unit facing an appropriation that lapses at year end and a supplier willing to invoice against incomplete delivery has an obvious route to converting an unspent balance into a reported expenditure. This research cannot establish that such conversion occurs, and makes no allegation that it does. It establishes only that the incentive structure exists, that the spending pattern is consistent with it, and that the KSh 176.9 billion pending bills stock indicates commitment control is not operating tightly. Oversight institutions that publish quarterly absorption data are well positioned to test this directly, and the fourth-quarter spending ratio would be a straightforward addition to existing reporting.

The reporting cycle itself contributes. Because the execution deficit peaks at nine months at 49.0 percentage points and closes to 43.4 by year end, a manager reviewed at the nine-month point appears further behind than the year will finish, which manufactures urgency at exactly the moment when procurement quality is most vulnerable. An oversight regime that scored performance on the evenness of the spending profile rather than on the annual total would remove that incentive at no cost.

The reserve fund and the hardening constraint

Poland’s position is at an inflection point that the 2024 figures capture precisely. The Fund reported a loss of PLN 7.7 billion, PLN 1.7 billion better than planned and better than 2023, which on any single-year reading is an improving trajectory. The loss was fully covered by the reserve fund, which is now practically zeroed.

The soft budget constraint has therefore been operating exactly as the theory predicts, and is now about to harden for the most mundane of reasons: the buffer has been consumed. Forward projections in the Polish public finance literature place the 2025 shortfall at approximately PLN 14 billion (Instytut Finansów Publicznych, 2025), with contribution revenues also running PLN 3.5 billion below expectation, and a 2026 gap of at least PLN 23 billion. Those are projections rather than outturns and are treated as such here.

The mechanism deserves attention beyond Poland because it is generic. Any payer with a reserve can absorb plan overruns without confronting them, and will therefore continue to approve plans it does not expect to meet.

The reserve is not a prudential buffer in that configuration; it is a mechanism for postponing the reconciliation of policy ambition with contribution capacity. The reconciliation is not avoided, only postponed, and it arrives at whatever moment the buffer runs out rather than at a moment of the system’s choosing.

Two structural features of the Polish design amplify the exposure. The single-payer architecture means there is no alternative payer to absorb a shock, so any deficit transmits directly to service availability. And the Fund was tasked in 2024 with financing a range of non-insurance-based services previously met from the state budget, without a corresponding transfer of financing capacity, which shifted obligation without shifting resource.

Execution accuracy is not allocative quality

The comparison in Table 6 produces a result that should discipline any temptation to treat execution accuracy as a proxy for good financial management. Poland fails the credibility test and directs 11.15 percent of realized cost to primary care. Alberta passes it comfortably and allocated 1.81 percent of its operating budget to primary care modernization, alongside CAD 6.4 billion in physician compensation and CAD 5.0 billion in hospital services.

The figures are not strictly comparable, and the research says so wherever it uses them. The Albertan number is a modernization allocation rather than total primary care spending, and physician compensation funds a substantial volume of primary care delivery.

But the direction is not in doubt, and it is consistent with the wider evidence: Alberta’s own primary care reform initiative required a dedicated allocation precisely because primary care had not been receiving proportionate attention, and the reform’s own documentation describes the system as requiring stabilization.

Poland’s primary care position illustrates the converse limitation. A well-funded primary care share has not translated into reform uptake: coordinated care entrusted budgets reached 40.2 percent of primary care physician contracts by 2024 and only 43.1 percent by 2025, with voivodship variation from 24.8 percent upward, and the pace has effectively stalled. Money allocated to a reform that providers decline to contract for does not execute either, and it does so invisibly, because the financial ratio records the allocation while the service change never occurs.

The general proposition is that four independent things can go wrong between an approved budget and a delivered service: the aggregate can miss, the composition can be wrong, the timing can be wrong, and the intended providers can decline to participate. This research measures the first three and finds the fourth in the Polish primary care record without being able to measure it. A financial management regime that monitors only the first is monitoring a quarter of the problem.

Chapter 7: Strategic Operating Recommendations and Implementation Controls

Recommendations are stated as controls with owners and verification points, because a recommendation without a verification point is an aspiration.

Controls for systems failing by underspending

Initiate procurement against the approved appropriation rather than against received cash. The evidence that spending units convert 93 percent of what they receive establishes that capacity is not the binding constraint, and the fixed duration of the procurement cycle establishes that a tender started in month eight cannot complete in month twelve. Tender preparation, specification, and evaluation can proceed on the authority of the approved budget, with award conditional on release. This converts a four-month procurement lag from a sequential cost into a parallel one, and it requires no additional money.

Report the fourth-quarter spending share alongside the annual absorption rate. A department absorbing 56.6 percent evenly and one absorbing 56.6 percent with half its spending in the closing quarter have different problems, and only the second carries acute procurement risk. The measure is computable from data these systems already publish quarterly.

Recognize commitments at the point they are incurred. The finding that pending bills stand at 186.2 percent of the unspent development balance means the absorption ratio is reporting a surplus where a net obligation exists. Until commitments are recognized, every execution figure the system publishes overstates its fiscal position, and every management decision taken on those figures is taken on a misstatement.

Controls for systems failing by overspending

Cost every in-year policy expansion against the approved plan before adoption, not after. The Polish free medicines program executed at 300 percent of its own plan following two policy decisions taken after approval: extension to children and adolescents, and reduction of the senior threshold from 75 to 65. Neither is a forecasting error. Both are choices whose cost was absorbed by revising the plan.

Report the revision as a performance measure in its own right. A payer that reports only the final outturn against the final plan will always report satisfactory execution, which is why this research computes against the initial plan throughout. The difference between initial and final plan is the measure that carries the information.

Treat reserve depletion as a leading indicator rather than an accounting entry. The 2024 loss was smaller than planned and smaller than 2023, and on those terms performance improved. The reserve that absorbed it is now practically exhausted, which means the improving trend and the loss of capacity to absorb any trend arrived in the same year.

Controls for systems executing accurately

Where aggregate execution is reliable, the residual risk is allocative, and the control is to publish the primary care share of total health expenditure on a consistent definition, annually, alongside the execution rate. Alberta’s own reform documentation identifies primary care as requiring stabilization, which indicates that the composition question was live even while the aggregate performed. An execution rate of 102.4 percent tells a legislature that the ministry spent what it was given. It says nothing about whether the composition matched need, and the two should not be reported as though the first answered the second.

Table 7: Implementation control schedule — actions, owners, and verification

Policy controls

Treasuries should publish a disbursement calendar at the start of the fiscal year and report against it. The Kenyan evidence locates the binding constraint in release timing rather than in spending capacity, and a published calendar converts an unpredictable constraint into a planning parameter.

It costs nothing, and of everything the analysis supports it would move the number furthest.

Oversight institutions should score budget credibility symmetrically but report the direction. The PEFA framework correctly treats both directions as failures, and this research applies it as specified. But a D awarded for underspending and a D awarded for overspending require opposite remedies, and a scoring regime that does not report the sign is withholding the operative half of the finding.

Statutory reporting should require an aged analysis of arrears alongside any absorption figure. The Kenyan case demonstrates that an absorption rate published without an arrears position can invert the reader’s understanding of the fiscal situation, turning a net liability of KSh 81.9 billion into the appearance of an unspent balance.

Chapter 8: Research Findings, Limits, and Quality-Control Record

Principal findings

Budget execution in health financing fails in both directions, and a symmetric credibility standard records both as failures while concealing that they require opposite remedies. Of nine audited measures, one falls in the PEFA A band, two in the B band, and six in the D band. Mean absolute deviation from the approved budget across the five headline measures is 17.8 percentage points, with a spread from 56.6 to 115.8 percent.

The timing component of the execution gap is larger than the annual ratio suggests and is separately actionable. Kenyan counties spent 54.1 percent of their annual development budget in the final quarter, at 3.53 times the rate of the preceding three quarters, with the execution deficit peaking at 49.0 percentage points at nine months before closing to 43.4 at year end. The convexity of that path, not its endpoint, is the management finding, and it identifies procurement initiation rather than absorptive capacity as the constraint.

An underspend is not necessarily a surplus. County pending bills of KSh 176.9 billion stand at 186.2 percent of the KSh 94.99 billion unspent development balance and 80.8 percent of the entire annual development budget. The counties owe KSh 81.9 billion more than they failed to spend, which inverts the natural reading of the absorption figure and changes the indicated policy response from capacity building to commitment control.

In-year plan revision has become routine rather than exceptional in the single-payer case. Upward revisions of PLN 26.0 billion in 2024 and PLN 23.0 billion in 2023 represent 15.8 and 16.6 percent of their respective initial plans, of which reported components explain 23.5 percent in 2024 and the residual carries the rest. The reserve fund that absorbed the resulting losses is now practically exhausted.

Execution accuracy and allocative priority are independent. Poland fails the credibility test and directs 11.15 percent of realized cost to primary care; Alberta passes it and allocated 1.81 percent of its operating budget to primary care modernization. No execution ratio detects a composition problem, and a financial management regime that monitors only the aggregate is monitoring a fraction of what can go wrong between approval and delivery.

Findings against the hypotheses

H1 is rejected for Kenya and Poland and accepted for Alberta. H2 is rejected. H3 is rejected. The full test record appears in Chapter 5.

Limits of the research

The limits are substantial and were anticipated in the methodology. The research relies on self-reported official figures that it did not audit, and an execution rate is only as sound as the expenditure reporting behind it. Three jurisdictions were selected for structural contrast rather than by sampling, so the findings characterize devolved, single-payer, and ministry-and-authority structures without being representative of any of them. Accounting bases differ across the comparison and are stated rather than reconciled.

Classification boundaries limit the comparison further. The line between development and recurrent expenditure is drawn differently in each system, and the Polish and Albertan sources do not present a capital and current split equivalent to the Kenyan one, which is why the within-year absorption analysis is conducted for Kenya alone. Fiscal calendars differ and are not aligned. One derived figure appears, the Alberta 2023-24 outturn, reverse-engineered from the stated increase in the subsequent budget; the sensitivity check establishes that the Alberta conclusion does not turn on it.

The arrears reconciliation compares a multi-year stock against a single-year flow and is reported as indicative of direction and rough magnitude rather than as an exact identity. A proper reconciliation would require the age profile of the pending bills stock, which is not published. The Polish decomposition carries 76.5 percent of the revision as an undecomposed residual, because the sources do not disaggregate it and the research declines to invent the disaggregation.

The deepest limit is that the research measures execution and cannot measure what execution purchased. A department that absorbs 100 percent of its budget on the wrong things scores perfectly on every measure computed here. Execution credibility is a necessary condition for sound health financing and is nowhere close to a sufficient one, and the allocative comparison in Chapter 5 is an indication of that limitation rather than a remedy for it.

Reflection on the evidence base

A closing observation concerns the reporting regimes rather than the numbers they produce. Kenya publishes the most granular execution data of the three systems examined, quarterly, by county, by expenditure class, under a constitutional mandate, and it produces the weakest execution performance. Alberta publishes the least granular in-year data and produces the strongest. The relationship between transparency and performance in this small sample runs in the opposite direction to the one usually assumed.

The explanation is almost certainly that Kenya publishes granular data because the problem is severe enough to have generated a constitutional reporting obligation, rather than that publication causes the problem. But the observation carries a practical implication for anyone comparing systems: the visibility of a failure is not a measure of its size, and a system that publishes little may be concealing more than one that publishes a poor figure quarterly. The absence of an Albertan in-year absorption series equivalent to the Kenyan one is a genuine gap in this research, and it is a gap in the reporting rather than in the search.

The implication for a finance manager reading comparative budget literature is a modest one. Execution ratios should be read alongside their denominator, their accounting basis, and their arrears position, and a ratio published without those three is not interpretable.

That is a low bar, and the majority of the figures encountered in preparing this research did not clear it.

Directions for further research

1. A within-year absorption series for a health system reporting on an accrual basis would establish whether the convex spending path observed in Kenya is a feature of cash rationing specifically or of public capital budgeting generally.

2. An aged analysis of pending bills matched to absorption rates across a panel of subnational entities would convert the arrears reconciliation attempted here from an indicative ratio into a measured relationship.

3. The Polish residual revision of PLN 19.9 billion is the largest single unexplained quantity in this research, and its decomposition from primary payer records would materially improve understanding of single-payer overspend mechanics.

4. Provider contracting uptake for funded reforms is an execution failure mode that financial reporting does not capture at all, and the Polish entrusted budget record offers a well-documented case for studying it.

Contribution

The research contributes a recomputed and commonly scored execution record for three structurally distinct health financing systems; a demonstration that underspending and overspending are the same governance failure observed from opposite sides, evidenced by the 32 percentage point improvement in Kenyan performance when the denominator shifts from appropriation to disbursement; a quantified within-year absorption path establishing that more than half of annual capital spending falls in the closing quarter; and an arrears reconciliation establishing that the reported underspend conceals a net obligation. For the practicing health finance manager it offers a short list of controls that require no additional resource, and a short list of published ratios that should not be read without their denominators.

References

Alberta Health. (2024). Health annual report 2023–2024. Government of Alberta.

Alberta Health Services. (2024). 2024-25 business plan. Alberta Health Services.

Government of Alberta. (2024a). Budget 2024 fiscal plan. Government of Alberta.

Government of Alberta. (2024b). Government of Alberta annual report 2023–2024. Government of Alberta.

Instytut Finansów Publicznych. (2025). The financial gap of the National Health Fund could reach a quarter of a trillion PLN. Institute of Public Finance.

International Budget Partnership. (2021). Roll over: Budget credibility in Kenya’s counties. International Budget Partnership.

Kornai, J. (1986). The soft budget constraint. Kyklos, 39(1), 3–30.

Narodowy Fundusz Zdrowia. (2025). Activity report and financial statements for 2024. National Health Fund.

National Treasury of Kenya. (2024). Health sector medium term expenditure framework 2025/26–2027/28. Government of Kenya.

Office of the Controller of Budget. (2025a). County governments budget implementation review report, first half of FY 2024/25. Government of Kenya.

Office of the Controller of Budget. (2025b). County governments budget implementation review report for the financial year 2024–2025. Government of Kenya.

Office of the Controller of Budget. (2025c). National government budget implementation review report for the financial year 2024–2025. Government of Kenya.

PEFA Secretariat. (2019). Framework for assessing public financial management. Public Expenditure and Financial Accountability Secretariat, World Bank.

Republic of Kenya. (2012). Public Finance Management Act. Government Printer.

World Health Organization Regional Office for Europe. (2024). Health system summary: Poland 2024. European Observatory on Health Systems and Policies.

Quality-Control Appendix

The research passed the NYCAR Postgraduate Diploma check for public-filing anchoring, mathematical transparency, paragraph variation, reference discipline, and human-expert voice differentiation.

The word-count gate is set at 12,000 words. The final extracted count is recorded after rendering and quality assurance.

The peer-review designation appears on the cover as required: Peer Review: Independent Review.

The visual quality assurance gate checks table of contents continuity, heading order, numbering, watermark presence, tables, figures, pagination, layout balance, and academic flow. The NYCAR watermark appears on every page of the body text, and the copyright line and publication number NYCAR-HF-2026-015 appear in the running footer of every page.

The research uses American English throughout. Statutory titles, institutional names, and source document titles are reproduced as published, consistent with APA 7th edition practice.

The mathematical audit confirms that every execution rate reported in Chapter 5 was recomputed from the numerator and denominator recorded in Table 1 and Table 2 using the companion analysis script, and that no ratio, deviation, or decomposition component was assumed, simulated, or imputed. One figure is derived rather than reported, the Alberta 2023-24 outturn, and it is labeled as derived at every point of use and subjected to a sensitivity check. The Polish residual revision is carried as a residual and labeled as derived rather than attributed to named causes. Currency values are retained as reported and no conversion is performed.

Table 8: NYCAR quality-control checklist

Candidate verification note: the peer-review statement on the cover records the NYCAR editorial designation for this publication class. Candidates submitting this research to an awarding institution should confirm that the designation matches the review actually performed by their institution before submission.