Designing welfare-maximising financing for publicly-provisioned digital child-health platforms: A mixed-methods policy simulation from Thailands KhunLook application

This paper employs a mixed-methods policy simulation on Thailand's KhunLook digital health platform to demonstrate that a means-tested subsidy with a flat fee for the top 60% of users optimizes the trade-off between welfare, equity, and fiscal sustainability compared to full public provision or market-based pricing.

The Gist

Imagine the Thai government has built a magical, free smartphone app called KhunLook. It's like a digital "Pink Book" that helps parents track their baby's growth, remember vaccination dates, and get advice from doctors. For years, it's been free for everyone, paid for by the government and donors. But now, those donors are leaving, and the government needs to figure out how to keep the lights on without turning off the app for the families who need it most.

This paper is like a financial weather forecast. Instead of just guessing, the author ran a sophisticated computer simulation to test five different ways to pay for the app, asking: "Which plan keeps the most people happy, helps the poorest families the most, and still leaves enough money in the bank to keep the app running?"

Here is how the paper breaks it down, using simple analogies:

The Setup: The "Willingness to Pay" Map

The researchers started with a map of what Thai parents are actually willing to pay. They surveyed 680 parents and found a clear pattern:

• Richer families are willing to pay more (like buying a premium subscription).
• Poorer families are willing to pay much less.
• If you charge everyone the same high price, the poor families get locked out.
• If you charge nothing, the government runs out of money.

The Five Contenders (The Financing Scenarios)

The paper tested five different "pricing strategies" in the simulation:

1. The "Free-for-All" (Status Quo): The government pays for everything.
   • Result: Everyone gets the app (100% uptake), and total happiness (welfare) is at its highest. But, the government loses money every year. It's like a bakery that gives away free bread to everyone but goes bankrupt because no one pays for the flour.
2. The "Flat Fee" (One Price for All): Everyone pays the same amount (about 341 Baht).
   • Result: The government makes money, but the poorest families can't afford it and quit. The app becomes a club for the rich. This creates a huge gap between rich and poor users.
3. The "Freemium" (Basic Free, Premium Paid): The basic features are free, but you pay extra for "super" features.
   • Result: In this specific simulation, almost no one bought the premium upgrade. The extra features weren't valuable enough to most people to justify the cost, so this plan didn't raise any extra money.
4. The "Fine-Tuned Tier" (Pay by Income Bracket): The government tries to charge every income group a slightly different price (e.g., Q1 pays 0, Q2 pays 342, Q3 pays 395, etc.).
   • Result: This sounds fair on paper, but it's a logistical nightmare. The simulation showed that the middle-income families got priced out because the specific price for their group was too close to what they could actually afford. It also requires a complex system to check everyone's income, which Thailand doesn't currently have for this app.
5. The "Means-Tested Subsidy" (The Winner): The government uses an existing list of poor families (the "Welfare Card" register) to give free access to the bottom 40% of households. The top 60% (the middle and rich) pay a flat fee of 395 Baht.
   • Result: This was the champion.
     • Fairness: It keeps the app free for the poorest, so they don't get locked out.
     • Money: It raises enough money from the top 60% to cover the costs and even build a surplus of nearly 200 million Baht over five years.
     • Simplicity: It uses a list the government already has. They don't need to build a new system to check incomes; they just check the Welfare Card.

The Big Takeaway

The paper argues that simplicity beats precision.

Trying to be too clever and charging every single income group a different price (Scenario 4) actually hurts the middle class and is hard to manage. Instead, using a "coarse" but simple rule—Free for the bottom 40%, Flat Fee for the top 60%—works best.

It's like a community pool:

• If you charge everyone the same high fee, the poor can't swim.
• If the city pays for everyone, the city goes broke.
• The winning strategy is: Free entry for families on the welfare list, and a standard ticket price for everyone else.

Why This Matters

The paper concludes that for Thailand's KhunLook app, this "Means-Tested Subsidy" is the best path forward. It captures 85% of the happiness of the fully free model but generates a financial surplus to keep the app alive for the future.

The author notes that this logic could work for other countries, but only if they have a similar "Welfare Card" list to easily identify who is poor. If a country doesn't have that list, they might have to stick to the "Free-for-All" model, even if it costs the government more.

In short: Don't overcomplicate the pricing. Use the existing list of poor families to give them free access, charge the rest a fair flat fee, and you get a system that is both fair and financially sustainable.

Technical Summary

1. Problem Statement

National digital health platforms for maternal and child health in low- and middle-income countries (LMICs) are scaling rapidly, often outpacing the evidence on sustainable financing models. As initial donor and research grants expire, governments face a critical dilemma: how to transition from fully public (free) provision to a cost-recovery model without compromising equity, user uptake, or fiscal sustainability.

The specific case study is KhunLook, a Thai mobile application serving as the digital successor to the state-issued "Pink Book" (Maternal and Child Health Handbook). While the platform has high engagement (550,000+ downloads), its long-term funding is uncertain. The paper seeks to answer: What pricing structure maximizes total social welfare while ensuring fiscal sustainability and equitable access across income groups in a middle-income setting?

2. Methodology

The study employs a mixed-methods policy simulation framework combining quantitative welfare economics with qualitative institutional analysis.

A. Quantitative Component: Welfare Simulation

• Data Source: Secondary analysis of willingness-to-pay (WTP) data from a survey of $n=680$ Thai parents and relatives (Lounkaew et al., 2025).
• Parametric Inputs:
  • Regression Models: Coefficients from Ordinary Least Squares (OLS) and Unconditional Quantile Regression (UQR) at the 25th, 50th, and 75th percentiles.
  • Imputation: Individual-level WTP was not available. The authors imputed quintile-level mean WTP using monotone-cubic interpolation (PCHIP) based on published percentile anchors.
  • Population Simulation: A synthetic population of 250,000 active caregivers was drawn from truncated Normal distributions centered on the imputed quintile means, with dispersion derived from the OLS $R^2$ (0.711).
• Scenarios Simulated: Five financing models were compared over a 5-year horizon (discounted at 3%):
  1. S1 (Status Quo): Fully public provision (Free for all).
  2. S2 (Flat Fee): Market price set at the OLS mean WTP (341 THB).
  3. S3 (Freemium): Free basic tier; premium tier priced at the 75th percentile WTP (441 THB).
  4. S4 (Fine-Grained Tiering): Income-based pricing with distinct fees for each quintile (Q1=0, Q2=342, Q3–Q4=395, Q5=441).
  5. S5 (Means-Tested Subsidy): Free for the bottom 40% (Q1–Q2); flat fee of 395 THB (median WTP) for the top 60% (Q3–Q5).
• Metrics Calculated:
  • Total Welfare ($W$): Sum of Consumer Surplus (CS) and Net Producer Surplus (PS).
  • Fiscal Sustainability: Net Producer Surplus (Revenue minus Implementation Cost).
  • Equity: Uptake gaps between quintiles and the Wagstaff Concentration Index (CI) of consumer surplus (negative values indicate pro-poor distribution).
  • Deadweight Loss (DWL): Welfare lost due to non-adoption.

B. Qualitative Component: Institutional Feasibility

• Method: Inductive thematic analysis of Thai digital health policy documents, the Civil Registration Act, and historical accounts of the "Pink Book."
• Purpose: To constrain the set of feasible scenarios and interpret simulation results against real-world administrative capacity and political constraints (e.g., the existence of a state welfare card register, the tradition of free public health goods).

3. Key Contributions

1. Prescriptive Policy Design: Moves beyond descriptive valuation (estimating what users will pay) to prescriptive design (recommending how to price) by integrating welfare simulation with institutional constraints.
2. Administrative Simplicity vs. Precision: Challenges the intuition that fine-grained, income-specific pricing is superior. The study demonstrates that coarse means-testing (using existing registries) can outperform fine-grained tiering on both welfare and equity grounds because it avoids pricing out "middle" income groups who are sensitive to price tiers near their own WTP.
3. Transferable Framework: Provides a replicable framework for other LMICs to design financing for digital public goods, specifically where a social registry exists to identify low-income populations.

4. Key Results

• Scenario S1 (Free): Maximizes total welfare (437.4 million THB) but results in a persistent fiscal deficit (implementation costs are not recovered).
• Scenario S2 (Flat Fee): Generates revenue but creates severe inequity. It excludes the poorest quintile entirely, resulting in a strongly pro-rich concentration index (+0.460) and significant deadweight loss.
• Scenario S3 (Freemium): Failed in this context. At the base case, no user's residual valuation for the premium tier exceeded the price, leading to zero premium uptake.
• Scenario S4 (Fine-Grained Tiering): Generated high deadweight loss (98.7 million THB). By pricing tiers too close to the specific mean WTP of middle quintiles, it caused partial market exit.
• Scenario S5 (Means-Tested Subsidy) – The Recommended Design:
  • Welfare: Achieved 372.9 million THB in total welfare (85% of the S1 maximum).
  • Fiscal Sustainability: Generated a net producer surplus of 198.6 million THB over five years, covering implementation costs and funding future investment.
  • Equity: Distributed consumer surplus toward lower-income groups (Concentration Index = -0.258).
  • Feasibility: Leverages Thailand's existing State Welfare Card register to identify the bottom 40% at near-zero marginal administrative cost.
• Sensitivity Analysis: The ranking held across all 12 sensitivity specifications (varying costs, discount rates, user base size, and price elasticity). S5 remained the optimal fiscally sustainable design in 100% of runs.

5. Significance and Implications

• For Thailand: The study provides a concrete, evidence-based pathway for the Ministry of Public Health to transition KhunLook from donor-dependent to fiscally self-sustaining without violating the equity principles of the Universal Coverage Scheme. It recommends a specific policy: Free access for the bottom 40% and a flat 395 THB/year fee for the top 60%.
• For Global LMIC Context: The paper argues that in countries with:
  1. Steeply rising WTP with income,
  2. Existing social registries for means-testing, and
  3. A public-health mandate for maternal/child care,
     Coarse means-testing is superior to complex tiered pricing. The administrative simplicity of using an existing registry outweighs the theoretical efficiency gains of fine-grained price discrimination.
• Methodological Insight: The study highlights that welfare simulations ignoring institutional realities (e.g., the lack of infrastructure to verify income quintiles at the point of service) yield unimplementable recommendations. The integration of qualitative policy analysis is essential for generating actionable economic advice.

Conclusion: The paper concludes that a means-tested subsidy model (S5) is the optimal solution for KhunLook, balancing high welfare retention, fiscal self-sufficiency, and pro-poor equity, while relying on existing administrative infrastructure to minimize implementation costs.