Optimal Allocation of Research Funds under a Budget Constraint

Area of Science:

• Health economics
• Decision analysis
• Research methodology

Background:

• Health economic evaluations often incorporate expected value of sample information (ENBS) to guide implementation and research decisions.
• Decision-makers face the challenge of allocating limited research budgets across multiple studies, similar to managing implementation spending portfolios.

Purpose of the Study:

• To develop and demonstrate a budget-constrained portfolio optimization framework for selecting research studies and determining optimal sample sizes.
• To maximize the total population expected net benefit of sampling (ENBS) across a portfolio of research investments under a fixed budget.

Main Methods:

• Employed a portfolio optimization framework to decide which studies to fund and at what sample size, considering a fixed research budget.
• Formulated the objective to maximize the sum of ENBS across all funded studies.
• Illustrated the framework with a stylized example and a real-world application using 6 published cost-effectiveness analyses.

Main Results:

• For studies receiving investment, the optimal sample size is reached when the marginal population ENBS divided by the marginal cost of sampling is equal across all studies.
• Budget-constrained optimal sample sizes are generally smaller than those maximizing individual study ENBS, enabling more studies to be funded.
• The portfolio optimization approach demonstrated the potential for a higher total ENBS compared to unconstrained optimization.

Conclusions:

• Budget constraints fundamentally alter optimal research investment decisions, necessitating a portfolio approach rather than maximizing individual study ENBS.
• A portfolio optimization strategy can lead to a greater overall expected net benefit of sampling.
• The optimal sample size for research is determined by the maximum willingness to pay for incremental information within budget limitations.