The Use of Proxy Means Testing (PMT) to Determine Premiums Payable to the Social Health Insurance Fund (SHIF) by the Informal Sector in Kenya

The Kenya Kwanza administration Bottom-up Economic Transformation Agenda (BETA) prioritizes Universal Health Care (UHC). Under BETA, the government is committed and determined to realize the constitutional right to health enshrined by article 43 of Kenya’s constitution. The administration has set an ambitious goal of delivering a UHC system within its first two years of governance.

By enacting the Social Health Insurance Fund (SHIF) Act of 2023, the Kenya Kwanza healthcare reform has repealed the National Hospital Insurance Fund (NHIF) and in its place established the Social Health Authority (SHA). SHA manages three healthcare funds:

1. The primary healthcare (PHC) fund: A fully, publicly financed fund that reimburses preventive, promotive, outpatient & basic diagnostic services at a controlled tariff at Level 2, 3 and select Level 4 healthcare facilities.
2. The Social Health Insurance fund (SHIF): A mandatory contributory social health insurance fund (SHIF) that pays for health services on referral basis at Levels 4-6 healthcare facilities.
3. The Emergency, Critical and Chronic Illnesses (ECCI) fund: A publicly financed fund for medical conditions such as cancer, diabetes, strokes & accident rehabilitation, pandemics that have hitherto not been adequately covered by insurance.

To access the two publicly financed funds, a Kenyan citizen must be a valid contributor to SHIF. To implement SHIF for all eligible Kenyans, the Kenya Kwanza administration has proposed to develop and deploy a proxy means testing (PMT) tool that targets informal sector contributors to the SHIF.

The intention of PMT seems to be to effectively target the poor within Kenya’s informal sector for government social health insurance subsidies or assignment of indigent status while effectively excluding the non-poor from such support. This aggressive approach to shielding the government from making premium payments inadvertently increases the risk of exclusion errors, that is, the risk that poor households will be wrongly classified as "well off" and thus denied benefits. The consequence is that PMT is intended to make the SHI Fund the largest of the three SHA funds and to minimize the resources allocated by the State to health programs, specifically, the publicly funded PHC and ECCI Funds.

This adverse result, the exclusion of genuinely poor families, will result from several in-built weaknesses of PMT such as, imperfect information, measurement errors, inappropriate or insufficient indicators, design errors in the regression model used, the dynamic nature of poverty, the inflexibility of PMT and administrative inefficiencies.

Imperfect information

PMT relies heavily on observable household information to estimate household consumption or income. Some of the household characteristics that PMT may gather include,

• Household Composition: The size and composition of the household, including the number of adults and children. Number of children under 5 years old.
• Housing Quality: Type of toilet facility, access to piped water or electricity, type of flooring material.
• Asset Ownership: Ownership of a television, refrigerator, bicycle, motorcycles and cars. Ownership of livestock and agricultural equipment in rural areas.
• Education: Education level of the household head or spouse, school enrolment status of children in the household.
• Employment: Employment status and sector of employment (e.g., formal vs. informal).
• Health indicators: ?Health status of household members, including the presence of chronic illnesses or disabilities. Access to healthcare services and health insurance coverage
• Location: Urban or rural residence.

However, some of these characteristics often do not adequately capture the true economic status of a household because of temporary income shocks or unreported informal earnings such as remittances from children to retired elderly parents.

Each of the chosen characteristics is assigned a weight based on what is considered their predictive power and the combined score is used to estimate whether a household is poor or non-poor. Regardless of the mix of household characteristics that are included in the model and the weight assigned to each characteristic PMT invariably runs into headwinds of failing to properly identify genuine beneficiaries of social interventions, that is, the poor.

Challenges of Proxy Means Testing (PMT) in Generating Accurate Information

1. Measurement Errors: Errors in data collection on household characteristics (e.g., misreporting, data entry mistakes, survey inconsistencies) lead to inaccurate consumption or income estimates by PMT.
2. Inappropriate or Insufficient Indicators: Indicators chosen for PMT may not fully capture poverty determinants in specific contexts. For example, assets or household demographics relevant in one region might not indicate poverty in another. How comparable are indicators like owning a bodaboda motorcycle in Nairobi versus Busia? Similarly, does house walling type in Samburu accurately reflect poverty as it does in Lamu?
3. Dynamic Poverty: Poverty fluctuates dynamically; households move in and out due to economic cycles, health crises, or employment changes. PMT relies on static data, often failing to reflect current economic conditions of households experiencing recent changes. Kenya lacks consistent employment status reporting mechanisms essential for PMT's dynamic accuracy.
4. Inflexibility of PMT: Even with improved employment reporting, PMT models are rigid once implemented. They struggle to adapt to economic shifts or changes in poverty demographics, resulting in outdated targeting. Legal requirements, like Kenya's gazetting of the Means Testing Tool, further complicate updates, creating administrative hurdles.

Why we should expect Implementation Challenges with Proxy Means Testing

The background information we have to consider is as follows.

• The SHIF Act 2023 caps spending on administrative expenses at 5% of the annual expenditure of the Fund and
• KNBS is chronically underfunded and short-staffed whereas the demands of SHIF’s PMT for regular accurate consumption data will place additional strain on meagre resources.

Proxy Means Testing (PMT) may not always be the most practical or efficient method for targeting social assistance in low- and middle-income countries. In our context, the Ministry of Health relies heavily on tools developed by the Kenya National Bureau of Statistics (KNBS), particularly using the latest Kenya Continuous Household Survey 2021 (KCHS 2021). Funding for KCHS 2021 comes from the Government of Kenya, the World Bank, and other multilateral partners.

To highlight KNBS's capacity gaps in developing a robust PMT tool for social health insurance, consider that KCHS 2021 spanned 12 months and involved a workforce consisting of 99 interviewers, 16 supervisors, and 5 Survey System Administrators. The interviewer team comprised 61 non-KNBS interviewers and 38 KNBS interviewers from county offices.

An additional challenge in implementation involves conflicting definitions of a household between KCHS 2021 and the SHIF Act 2023. According to KCHS 2021, a household is eligible if the dwelling is occupied, even if no interview takes place. This definition differs significantly from that in the SHIF Act 2023, which defines a household as a social unit comprising an eligible contributor and their beneficiaries or those sharing similar socioeconomic needs related to consumption and production. The SHIF Act 2023 also considers all Kenyans above 25 years old as distinct households.

Challenges in Implementing Proxy Means Testing (PMT)

Several factors contribute to the high cost and complexity of implementing Proxy Means Testing (PMT):

• Data Collection, Processing, and Management:

PMT requires detailed household surveys to gather data on observable characteristics. Conducting these surveys demands significant financial, human, and logistical resources. Regular surveys are also essential to maintain accuracy, leading to ongoing costs. Continuous database maintenance and updates to reflect household changes will further escalate costs over time.

• Training and Capacity Building:

KNBS lacks sufficient internal enumerators for the scale of targeting envisioned by the SHIF Act 2023. Implementing PMT effectively will also demand increased expertise in econometrics, data analysis, and database management, necessitating more investments in capacity building efforts. While this may create job opportunities for skilled Kenyans, the limited healthcare funding we have must prioritize service provision.

• Model Development and Calibration:

Developing the initial PMT model involves selecting suitable indicators, estimating regression models, and validating results. Support from donor agencies like Palladium International under the USAID-funded PROPEL Health project aids MOH in identifying parameters to minimize inclusion and exclusion errors. However, this process requires time, a constraint for MOH and its partners. For example, Palladium's consultant recruitment advert in May 2024 outlines extensive requirements, including delivering statistically significant parameters for a means testing tool by July 1st, 2024—the start date of the new social health insurance scheme. This timeline is impractical for onboarding informal sector Kenyans under PMT-determined premiums.

• Social Acceptance:

Proxy Means Testing (PMT) rules are intricate, this was highlighted when the Cabinet Secretary for Health issued a gazette notice outlining the PMT formula. The notice received minimal public feedback, largely due to its technical nature.

The complexity of PMT rules can make it difficult for the architects of the reforms to explain to the public how the method determines their premiums. This lack of clarity is likely to leave beneficiaries suspicious about the criteria used for assessing their payments. This challenge poses risks for this ambitious social initiative. Establishing and maintaining public trust in the PMT system will demand enhanced community engagement and education efforts.

To further illustrate how trust issues will emerge consider this, the initial PMT model is based on the Kenya Continuous Household Survey 2021 (KCHS 2021), which categorizes consumption data across 50 study domains—national, urban, rural, and the 47 counties. Consequently, the PMT model statistically estimates a household's SHIF premium based on its domain, that is, whether the household is urban, rural, or in a specific county.

The public will soon realize that the premium they are paying has been arrived at solely from mathematical estimates, rather than an actual survey of their circumstances. No matter how accurate they may be, premiums arrived at from mathematical estimates will face opposition on political and social fronts. Public discontent will necessitate SHA or MOH to establish a robust verification and appeals process so as to maintain public confidence. However, these mechanisms will inevitably add to administrative costs.

Conclusions

Proxy Means Testing (PMT) is a method used to identify poor households by estimating household consumption through a regression-based approach using observable characteristics. While PMT effectively identifies non-poor households (increasing contributors to SHIF), it also tends to exclude genuinely needy families, leading to higher exclusion errors. PMT is complex and expensive to administer, requiring substantial investments in trained personnel, monitoring, calibration, data analysis, and storage capabilities. Its reliance on abstract characteristics and mathematical estimates complicates acceptance, necessitating further investments in community engagement.

For a lower middle-income country like Kenya, the costs of PMT can outweigh the benefits intended for the very poor, fostering resistance to social safety nets. Given Kenya's strained administrative capacity and resources, simpler, transparent, and equally effective targeting methods should be explored to reach those in need more effectively.

Dr Brian Makamu Lishenga is the National Chairman of the Rural & Urban Private Hospitals Association of Kenya (RUPHA) Rural & Urban Private Hospitals Association of Kenya - RUPHA Dr. Abdi Mohamed Dr Amit N. Thakker ,EBS Stephanie Koczela Ministry of Health, Kenya