Pension Scheme Efficiency in Kenya

• Efficiency of Kenyan pension schemes is measured using DEA models that assess cost efficiency by comparing administrative and investment costs against ROI and asset growth.
• Empirical studies show that strong governance, particularly increased employee trustee representation and robust risk management, significantly enhance scheme performance.
• Collectivised post-retirement investment designs demonstrate policy potential by lowering initial capital requirements by about 20% for equivalent pension adequacy.

Pension-scheme efficiency in Kenya comprises the technical and organizational capacity of retirement funds to maximize member benefits for given administrative and investment costs, subject to regulatory, actuarial, and governance constraints. In the Kenyan context, efficiency is increasingly scrutinized due to demographic transitions, rising pension assets, and heightened member expectations for old age income security. Research leveraging quantitative methodologies—including Data Envelopment Analysis (DEA), panel regression, and dynamic governance models—has identified cost minimization, governance structures, and risk management as pivotal drivers of efficiency in Kenyan pension schemes (Namagwa, 14 Nov 2025, Namagwa, 14 Nov 2025). Furthermore, optimal fund designs based on collectivised risk-sharing frameworks yield quantifiable gains compared to individual arrangements (Armstrong et al., 2019).

1. Definitions and Quantitative Measurement

Efficiency in Kenyan pension schemes is operationally measured using a constant-returns-to-scale DEA model, where each scheme functions as a decision-making unit (DMU). Inputs are annual administrative costs $C_{admin,it}$ and investment management costs $C_{inv,it}$, while outputs are net return-on-investments $ROI_{it}$ and annual asset growth $\Delta A_{it}$. The DEA efficiency score $EF_{it}$ reflects the maximal attainable output-to-input ratio for each scheme $i$ in year $t$, identifying frontier and below-frontier DMUs (Namagwa, 14 Nov 2025). Supplementary ratios include:

• ROI-to-cost ratio: $$Efficiency_{it}^{ROI} = \frac{ROI_{it}}{C_{admin,it}+C_{inv,it}}$$
• Cost-efficiency metric: $$CostEff_{it} = \frac{C_{admin,it}+C_{inv,it}}{\Delta A_{it}}$$

The sectoral mean DEA score in current analyses hovers around 0.75, indicating appreciable slack relative to the best-practice frontier (Namagwa, 14 Nov 2025).

2. Empirical Dataset, Variables, and Panel Structure

Studies utilize a panel spanning seven years (2015–2021) across 128 registered pension funds, yielding 896 observations with scheme-year granularity. Key variables encompass board composition (proportions of top-management, employee-elected, female, and independent trustees), a risk-management index (1–5 scale based on disclosure and oversight processes), regulatory compliance scores (document count), DEA efficiency scores, and controls for scheme size, age, asset mix, and regulatory category. Data is sourced from audited reports, board documents, and Retirement Benefits Authority filings (Namagwa, 14 Nov 2025, Namagwa, 14 Nov 2025).

3. Governance, Risk Management, and Efficiency: Statistical Models

Regression models estimate the relative impact of governance structures and risk management on efficiency. The baseline panel-fixed-effects specification is:

$$Efficiency_{it} = \alpha + \beta_1\,Gov_{it} + \beta_2\,X_{it} + u_i + \varepsilon_{it}$$

where $Gov_{it}$ denotes a governance attribute (e.g. employee representatives $Mb_{it}$; independent trustees $Im_{it}$), and $X_{it}$ are controls. All predictors are log-transformed.

Analysis further deploys a mediation framework in which governance variables influence efficiency both directly and indirectly via risk management:

1. Governance $\rightarrow$ Risk management: $$RiskMgmt_{it} = \gamma_0 + \gamma_1\,Gov_{it} + ...$$
2. Governance, Risk management $\rightarrow$ Efficiency: $$Efficiency_{it} = \delta_0 + \delta_1\,Gov_{it} + \delta_2\,RiskMgmt_{it} + ...$$ The indirect effect is quantified as $\gamma_1 \times \delta_2$ (Namagwa, 14 Nov 2025).

4. Key Results: Employee Representation, Risk Management, and the “Self-Cleaning Mechanism”

Panel regression finds a strong positive effect ($\beta_2=1.023$, $p<0.001$) of employee trustees on efficiency, while independent trustees impart a negative effect ($\beta_4=-0.574$, $p=0.018$). Robust risk-management infrastructure similarly enhances efficiency ($\beta_5=0.328$, $p=0.025$); regulatory compliance is statistically insignificant (Namagwa, 14 Nov 2025).

The mediation model confirms that employee representation benefits scheme efficiency both directly and via improved risk management ($\gamma_1=0.0923$; $\delta_2=0.4161$), with the indirect effect accounting for ≈4% of the total impact (Namagwa, 14 Nov 2025). Employee trustees channel member-focused insights into risk-responsive strategy only when supported by robust risk management frameworks. Board decisions thus translate into cost-effective, well-hedged portfolios.

Theoretical extensions posit a dynamic “self-cleaning mechanism” operating through democratic trustee elections. Formally, scheme efficiency $EFF_{t+1}$ in each cycle is:

$EFF_{t+1} = \alpha\,EFF_t + \delta\,M_t + \nu_t$

where $M_t$ is the fraction of board member-elected trustees. Electoral cycles progressively eliminate underperforming trustees, aligning board incentives with member interests and raising long-run efficiency (Namagwa, 14 Nov 2025).

5. Solutions for Structural Efficiency: Collectivised Post-Retirement Investment

Independent research on collectivised post-retirement investment quantifies absolute efficiency gains in pooled schemes, surpassing annuities and individual DC pots. In such frameworks, assets of deceased members are redistributed among survivors, eliminating idiosyncratic longevity risk as $n \to \infty$. Mathematical optimization demonstrates that the initial capital requirement for collective structures is ≈20% lower for equivalent utility outcomes compared to individual arrangements (Armstrong et al., 2019).

Parameterization for Kenya involves recalibration:

• Mortality: Kenyan life tables or UN estimates
• Financial: local bond yields, equity returns, inflation
• Adequacy: local replacement rate targets
• Administration fees: scheme-level cost data

Efficient collective DC design utilizes

$\eta = \frac{K_{annuity}}{K_{collectivised}} - 1$

where $\eta$ quantifies the cost reduction from risk pooling and dynamic payout optimization. This suggests a concrete policy opportunity: collectivised schemes, particularly if configured with democratic governance and professional risk oversight, can deliver higher pension adequacy at lower sponsor cost (Armstrong et al., 2019).

6. Policy and Regulatory Implications

Recommended actions include:

• Mandating formal risk-management frameworks (board-level risk committees, standardized disclosures)
• Ensuring sustained and substantial employee trustee representation (minimum 25–40% seats via member election)
• Using risk-management KPIs in trustee evaluations
• Conducting RM audits in regulatory reviews
• Introducing election protocols and public performance reporting for trustees
• Reviewing and training independent trustees to reduce misalignment
• Rationalizing regulatory compliance toward outcome-based supervision

These reforms are projected to advance Kenyan scheme efficiency toward the global frontier (DEA $\approx$ 1), maximizing member outcomes and sectoral sustainability (Namagwa, 14 Nov 2025, Namagwa, 14 Nov 2025).

7. Limitations and Prospects for Further Research

Subjective RM index scoring and DEA’s constant-returns assumption warrant methodological scrutiny. Instrumental-variable or system-GMM approaches may better adjust for endogeneity in governance selection. The impact of external service providers and broader scheme inclusion (public/unlisted domains) remains underexplored. Adoption of objective risk metrics (e.g., VaR, stress-testing) could refine RM measurement. Further simulation and data collection tailored to Kenya’s unique demographic and institutional parameters will enhance model precision and policy relevance (Namagwa, 14 Nov 2025, Namagwa, 14 Nov 2025, Armstrong et al., 2019).