Policy design under research outcome unpredictability: A portfolio-theory model for Japan's research funding
Year: 2025
Type: preprint
Abstract: This study applies portfolio theory to research funding policy under the assumption that research outcomes are inherently unpredictable. Using a mathematical model, we demonstrate that Japan’s “selection and concentration” strategy since 2004 has reduced effective research capacity despite increased headline R&D expenditure. Comparative analysis between 2004 and 2024 shows that expected impact has shrunk to about 40% of the baseline due to reduced research time, weakened operational grants, and unstable career paths. The decline in Japan’s share of Top 10% highly cited papers corroborates the model’s prediction. By integrating theoretical reasoning, empirical evidence, and international comparison, this paper provides a comprehensive proof that portfolio-style baseline funding is the only rational design for sustaining long-term research performance. Furthermore, we show that breakthrough probabilities remain inherently low across scientific domains, typically only a few percent (OECD MSTI 2024; NISTEP Indicators). Even if ex ante screening accuracy is in the realistic range of 60 ± 10% (i.e., 50–70%; Bornmann & Daniel 2008), realized success rates rise only marginally—from about 6% at 50% accuracy to about 11% at 70%, and even under an optimistic 80% assumption only to about 17%. These results confirm that concentration strategies cannot engineer breakthroughs, and that diversified baseline funding with restored research time and operational support is the only utility-dominant path for sustaining high-impact outcomes. The analysis formalizes research utility as U(w) = E[R] + λ·Breakthrough − γ·Var, demonstrating that only diversified baseline funding maximizes expected impact while controlling risk.
Author: STUDENT, By
Language: English
Subfield: Information Systems and Management
Field: Decision Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7108451962