Research Grant Evaluation and Strategic Funding Approaches

Date: For Parliament Sitting on 10 September 2024

Title: "Administration and Evaluation of National Research Foundation Grants, and Consideration of ARPA-like Models in Singapore"

MPs Speaking: Ms He Ting Ru, Mr Heng Swee Keat

Ms He Ting Ru asked the Prime Minister and Minister for Finance (a) what percentage of National Research Foundation (NRF) research grants are awarded through grant calls administered directly by NRF instead of through statutory boards; (b) for statutory board-administered grant calls, what percentage of evaluators have significant relevant research experience; and (c) why has an ARPA-like model, where expert programme managers autonomously set strategic research directions and funding, not been adopted.

Mr Heng Swee Keat (for the Prime Minister): Singapore’s Research, Innovation and Enterprise (RIE) investments involve close collaboration among the National Research Foundation (NRF), Government agencies and RIE ecosystem stakeholders. These investments include RIE programmes supported by the National Research Fund, that are administered by NRF and other Government agencies. As at end-FY2023, around 50% of RIE2025 National Research Fund grants are directly administered by NRF.

Government agencies evaluate research grant applications based on the respective RIE programme’s approved eligibility and evaluation criteria. The evaluation processes for technical proposals generally incorporate reviews by experts in the relevant field, such as international peer reviewers or technical evaluation panel members, to support Government agencies’ funding decision-making. 

NRF regularly reviews RIE grant administration modalities, including studying overseas models such as the Advanced Research Projects Agencies (ARPA). ARPA-like models require a strong cadre of expert programme managers and leadership driving the programmes towards specific goals that are often higher-risk or ambitious in nature. We will continue to review where such models are appropriate in Singapore’s context, and deploy such models if useful.