CASA NEXUS / Insights
We publish what we find, before we are asked to.
CASA NEXUS runs an independent research programme alongside its client work. When a decision is consequential enough and public enough, we evaluate it and publish the result, whether or not anyone commissioned us to.
Our policy research is open access. If analysis is going to shape a national decision, the people affected by that decision should be able to read it.
- Publisher
- CASA NEXUS Limited
- Framework
- SSII
- Series
- Four, running
- Titles
- Eight, open access
- Access
- Open, for public distribution
- Citation
- Permitted with attribution
Four volumes issued 28 July 2026 · Open access
Sixty years of Kenyan government, scored on one instrument that does not move
Four administrations, from independence to 2022, evaluated against the five pillars and twenty-five dimensions of the SSII Framework. The weight vector, dimension anchors and evidence rules were fixed at Volume I and have not changed since, which is what makes the volumes comparable to one another rather than four separate opinions.
Each volume is self-commissioned. There is no external client, no funder constraint on publication, and no finding on individual culpability. Every result carries two numbers: a Strategic Integrity Score for the performance, and an Evidence Confidence Index that governs which verbs the report is permitted to use about it.
The Protection, Inclusion and Legitimacy pillar scored 28.00 in both Volume I and Volume II, rose to 40.00 in Volume III, and reached 52.00 in Volume IV.
Protection performance is a variable of the Kenyan state, not a constant of it.
| Vol. | Administration | SIS | Reportable band | ECI | PILI | Safeguarding Veto |
|---|---|---|---|---|---|---|
| I | Kenyatta1963–1978 | 42.20 | Low–Moderate | 76.70 Subst. | 28.00 | Fired |
| II | Moi1978–2002 | 31.60 | Low–Moderate | 79.30 Subst. | 28.00 | Fired |
| III | Kibaki2002–2013 | 52.80 Moderate on arithmetic | Capped to Low–Moderate | 77.80 Subst. | 40.00 | Fired · decisive |
| IV | Uhuru Kenyatta2013–2022 | 46.80 | Low–Moderate | 77.00 Subst. | 52.00 | Fired · not decisive |
Detailed Evaluation Reports, in full
The Kenyatta Administration, 1963–1978
The volume that fixed the instrument. It scores the executive government from 12 December 1963 to 22 August 1978 across twenty-three distinct upstream provenance chains, drawing on a truth commission, a public land commission, a parliamentary select committee, a national statistical series and a peer-reviewed economic literature.
The material record is strong and not in dispute: life expectancy rose from 50.8 to 58.2 years, infant mortality fell from 100.7 to 72.5 per 1,000 live births, primary gross enrolment moved from 60.3 per cent in 1970 to 95.6 per cent in 1978, and real GDP compounded at 6.80 per cent a year. The Safeguarding Veto fired all the same, and the report leads with that. Four rival explanations for the growth survive testing, so the evaluation says the evidence supports and declines to say caused.
The Moi Administration, 1978–2002
The lowest composite in the series, resting on the highest evidence rating in it. The period was examined by a judicial commission into Goldenberg, a judicial commission into the ethnic clashes, a parliamentary select committee, a public land commission, a truth commission, and a household survey programme with no equivalent under the previous administration. The weak result rests on the stronger record.
Real GDP per capita compounded at minus 0.11 per cent a year across 1979 to 2002. Child survival improved while adult life expectancy fell. Against that, the administration legislated a two-term limit in 1992, was bound by it in 2002, lost decisively and handed power to the opposition, the first such transfer in the country’s history.
The Kibaki Administration, 2002–2013
The highest arithmetic result in the series, and the first volume in which the Safeguarding Veto changes the published verdict rather than restating a band the arithmetic had already reached. That the cap is decisive here is itself recorded as a finding.
The material record is the strongest in the series: life expectancy recovered the entire loss of the preceding twenty-four years, under-five mortality fell 40.8 per cent, primary enrolment rose 22.2 points after fee abolition, and HIV prevalence fell from 7.8 to 5.2 per cent. The 2010 Constitution, approved by roughly 67 per cent of voters, is assessed as the single largest positive institutional change in the fifty years the series covers to that point.
The Uhuru Kenyatta Administration, 2013–2022
The first evaluand in the series scored against a constitution its predecessor created rather than one inherited from the colonial settlement, which makes every finding here also a finding about how the 2010 Constitution performed under sustained pressure from the executive it was designed to constrain.
Access to electricity rose from 40.1 to 76.0 per cent of the population, the largest single infrastructure movement in the series, and measured inequality fell for the first time. Public debt rose from about KSh 1.79 trillion to about KSh 8.7 trillion over the same period, and debt service as a share of exports moved from 4.7 per cent in 2013 to 38.4 per cent in 2019. PILI scores 52.00, the first pillar score above 50 in four administrations.
March 2026 · Open access
What the Evidence Shows, What the Gaps Mean, and What Ignoring Them Will Cost Kenya
An independent SSII Framework baseline evaluation of the National Infrastructure Fund Act, 2026, including a Social Return on Investment tracking architecture and a scenario analysis of what a KSh 5 trillion fund delivers when its design gaps are left unaddressed.
Kenya built at an unprecedented pace between 2013 and 2025 and paid for it largely through external borrowing. The NIF replaces that borrowing model with an investment portfolio model, pooling privatisation proceeds, pension capital, development finance, and climate finance into a professionally managed vehicle. Our evaluation asks whether the design can carry the ambition.
The Fund is credible, the policy logic is sound, and the governance architecture, though imperfect, carries more safeguards than most comparable funds launched on the continent at this stage.
But 53 out of 100 is a Bronze score. And a Bronze fund cannot build a Gold economy.
Everything we have published, in full
From Revenue Rage to Fiscal Reckoning
A policy analysis of revenue mobilisation, expenditure credibility, and macroeconomic constraint. The paper makes a simple argument: public debate is over-concentrated on one instrument when three should be in view. The Finance Bill is a revenue tool, not the fiscal system itself; it sits downstream of the Budget Policy Statement, the Medium-Term Expenditure Framework, and sector ceilings set months earlier.
Debt service absorbed about 69 per cent of ordinary revenue in FY 2024/25, several times above the IMF Debt Sustainability Framework thresholds for low-income countries, which range from 14 to 23 per cent. At the margin, new revenue is servicing debt rather than expanding productive capital. The analysis closes with targeted recommendations for Treasury, Parliament, the Controller of Budget, civic actors, development partners, and the Central Bank.
Kenya NIF: What the Evidence Shows
The first independent evaluation of the National Infrastructure Fund, completed on the day the Act was signed into law. Scores the Fund at 53 out of 100 on the SSII instrument, sets out where the design is strong and where it leaks, and introduces an SROI tracking architecture that did not previously exist in the Kenyan policy debate on the Fund.
The INGO Reckoning: A Post-Trump Adaptation. Save Lives or Shape Out
A critical policy review of humanitarian actor strategies at the convergence of conflict, climate, and compounding crisis. Its position statement is blunt: the international humanitarian system is at a reckoning that is not merely a funding crisis, and not merely a political rupture, but a structural failure. Written for INGO leadership teams deciding what their organisation is actually for in the next funding era.
The Aesthetics Trap: Why the Gen-Z Entrepreneurial Ethic Is a Structural Risk to Real Growth in Emerging Economies
A Kenya case study. Kenya’s Gen-Z cohort is the most digitally native generation in the country’s history, and potentially its most economically misaligned. The thesis argues that a dominant entrepreneurial ethic anchored on social media virality, platform aesthetics, TikTok commerce, and influencer identity carries consequences for productive capacity that the growth statistics have not yet caught up with.
Four standing series
Independent Evaluation Series
Full-length Detailed Evaluation Reports on evaluands that no one has commissioned us to examine, scored against the SSII Framework on a weight vector fixed at the first volume and held constant thereafter. The Kenya Presidencies is the current programme in this series: five volumes covering the executive government of Kenya from independence to the present, of which four are issued. Comparability across volumes is the point of the series, so the instrument is registered before the first score and never adjusted to suit a later result.
Policy & Evaluation Series
Independent evaluations of public instruments: funds, acts, and national programmes, scored against the SSII Framework and published open access. This is the series that produced the NIF baseline. We publish these on our own initiative, without commission, when the decision is consequential and public enough to warrant it.
White Paper Series
Numbered analytical papers on the structural questions underneath the headlines: fiscal architecture, revenue and expenditure credibility, growth composition, and macroeconomic constraint. Each paper closes with recommendations addressed to named institutions rather than to no one in particular.
Policy Nexus Analytical Review
Strategy reviews written for organisational leadership rather than for the policy record: what a shift in the funding, political, or climate environment means for how an institution should be built. The INGO Reckoning is the current review in this series.
The Strategic Systems Integrity & Impact Framework
A performance score on its own is an unfalsifiable assertion. An evidence score on its own is a methods note.
Conventional evaluation in this region carries three structural faults. It is written for compliance rather than for a decision. It splits the analysis across criteria that never speak to one another. And it describes a problem where a Minister or a board needed a course of action, a cost, and a risk.
SSII starts from a different premise: programmes succeed or fail on five interacting system layers, political, institutional, programmatic, protective and economic. Those layers are read simultaneously rather than in sequence, because a programme can be exemplary on one and failing on another, and only the whole picture supports advice.
What comes out is not a rating. It is a pair of numbers, a coverage grade for what the record could carry, a set of costed reform scenarios, and an explicit licence governing what the evaluation is permitted to claim.
Neither number is ever published without the other
Strategic Integrity Score
How well did the system work?
The weighted performance of the five pillars, on a scale of 0 to 100. No evidence multiplier touches it, so the number answers one question and one question only. It carries a rating band from Critical through to High Integrity, and each band carries a stated decision consequence: suspend, restructure, reform, monitor, or scale.
SIS = (PIAI × W1) + (IPGI × W2) + (IOII × W3) + (PILI × W4) + (CESVI × W5)Evidence Confidence Index
How well can we prove it?
The strength of the record behind the finding, on the same weights and the same scale. The impact pillar draws on the EQI ladder, which rates counterfactual strength; the other four draw on the ESI ladder, which rates documentary provenance from a single uncorroborated source up to an adjudicated one. Disaggregation completeness caps the protection term.
ECI = (EPIAI × W1) + (EIPGI × W2) + (EIOII × W3) + (EPILI × W4) + (ECESVI × W5)Reported together, in the same sentence or the same visual field: Moderate–High (SIS 77.7) · Substantiated (ECI 78.8). The earlier SSII composite folded evidence quality into a single figure as a multiplier, which meant a well-run programme evaluated with a modest design was reported as though it had performed worse. Separating the two questions removed that distortion, so scores carried over from the earlier composite are not directly comparable with these.
Twenty-five dimensions · Simultaneous lenses, not sequential steps
Political & Incentive Alignment
Who holds power over this programme, what are they actually rewarded for, and does the reform survive contact with that arithmetic? Political economy analysis runs at every stage, with a coalition stability index that deducts up to fifteen points where the coalition carrying a reform is fragile.
- Power mapping
- Stakeholder incentives
- Elite capture risk
- Regulatory feasibility
- Conflict sensitivity
Institutional Performance & Governance
Whether the machinery can carry what has been promised. Institutional diagnostics, process tracing and governance benchmarking against CHS, Sphere, the Grand Bargain and OECD standards, plus a digital governance integrity score and a measure of how fast the institution can actually change its mind.
- Leadership accountability
- Financial stewardship
- Procurement integrity
- Data governance
- Decision-making velocity
Impact & Outcome Integrity
Whether the result happened, and whether this programme is why. The counterfactual is pre-registered in the design memo before any data moves. A ripple effects scan is mandatory, because a programme that hits its own targets while destabilising the market around it has not succeeded.
- Outcome performance
- Contribution analysis
- Unintended consequences
- Systemic spillovers
- Sustainability probability
Protection, Inclusion & Legitimacy
Do no harm compliance, safeguarding architecture, and whether the people the programme exists for regard it as legitimate. Survivor-centred protocols, perception surveys and an intersectionality analysis across six identity dimensions. This is the pillar that carries the veto.
Carries the safeguarding veto- Do no harm compliance
- Safeguarding architecture
- Gender-transformative impact
- Survivor-centred integrity
- Community trust & accountability
Cost-Effectiveness & Strategic Value
Cost per outcome against the counterfactual cost, and what else the same money could have bought. Value for money on the four Es plus a fifth, ethics; social return modelled dynamically over a ten-year horizon; portfolio attribution on fully-loaded rather than headline cost.
- Cost per outcome
- Cost-consequence matrix
- Portfolio efficiency
- Counterfactual cost modelling
- Opportunity cost mapping
The weights above are the default vector. They shift by context and the shift is declared before scoring begins: a government reform evaluation lifts institutional performance to 0.30, a humanitarian emergency lifts impact and protection to 0.30 each, an ODI-grade policy evaluation lifts political alignment and institutional performance to 0.25. The vector must sum to 1.00 and is fixed at Sprint 0, which is what makes a series such as The Kenya Presidencies comparable volume to volume rather than four separate opinions.
What the evidence permits the evaluation to say
Most frameworks handle weak evidence by quietly discounting the score. SSII does the opposite. The score stands, and the Evidence Confidence Index instead fixes the strongest verb the evaluation may use, in every product it touches: the report, the dashboard, the executive memo, the policy brief, the broadcast script, the social post.
A verb stronger than the licence permits is a quality assurance failure and blocks release, whatever the evidence appears to show. It is the reason our published evaluations say contributed to where a consultancy report would say caused, and the reason we will occasionally publish a finding as an evidence gap rather than a result.
| ECI | Band | Strongest permitted verb | What that means in practice |
|---|---|---|---|
| 85–100 | Adjudicated | demonstrates · caused · established | Causal language is on the table. Reached by experimental or strong quasi-experimental design, or by an adjudicated documentary record. |
| 70–84 | Substantiated | contributed to · indicates · the evidence supports | The finding is documented rather than alleged. Attribution is defensible; causation is not asserted. |
| 55–69 | Indicative | is associated with · suggests · consistent with | A pattern is visible in the record. Rival explanations have not all been excluded. |
| 40–54 | Provisional | the available record shows | Descriptive only. No attribution of any kind is permitted. |
| 0–39 | Insufficient | Not reportable | Cannot be published as a scored finding. Reported as an evidence gap, with a named retrieval route and an owner. |
Where the framework overrides the arithmetic
The safeguarding veto
A do no harm or safeguarding dimension scored 1 or 2, or a protection pillar below 50, caps the reportable band at Low–Moderate whatever the composite says. It cannot be waived by a client. It is why Volume III of The Kenya Presidencies publishes as Low–Moderate on an arithmetic of 52.80.
Performance is not coverage
A dimension score of 1 must mean the system failed. It may never mean the record is silent. Silence is logged separately as a coverage grade, so a programme that delivered poor value for money is never conflated with one whose value for money nobody ever computed.
The evaluand tier
Before anything is scored, the evaluation declares what it is scoring: a donor as steward, an intermediary channelling funds, an implementer delivering to people, or a chain of all three. A chain is reported as a tier matrix. It is never blended into one number, because a weakness has to be locatable.
Three independent chains
Triangulation is counted by upstream origin, not by document count. Six publications tracing back to one commission of inquiry are one chain, not six. A dimension that cannot reach the standard is withheld on the authority of the QA lead, with the route that would close the gap named in writing.
SSII is crosswalked to the standards our clients are already accountable to: the six OECD-DAC evaluation criteria, the UNEG Norms and Standards that govern UN country team work, and the AfrEA African Evaluation Principles, whose Made in Africa commitments are closest to how the framework was designed in the first place. For each engagement the crosswalk is extended to the commissioning institution's own accountability framework, so the evaluation speaks the client's language and not only the sector's.
New research, sent when it publishes.
We publish a handful of times a year, and only when we have something worth your time. Leave your email and you will get each paper as it goes out, alongside our capability statement.
The method behind these papers is the method we bring to your programme.
The NIF evaluation and a capacity assessment of a single county-level organisation run on the same framework. The scale changes; the standard does not.