Development policy in the Pacific and Southeast Asia is only as strong as the analysis that follows data collection, a point brought into sharp relief at a panel on December 4, 2025, at the Australasian AID Conference. Chaired by Professor Beth Webster, director of the Melbourne Institute of Applied Economic and Social Research, the session showcased three pieces of empirical work from the Institute that argue policymakers must move beyond national averages to target the people and places most at risk.
Webster framed the discussion as a complement to the conference keynote on the science of scale, emphasizing that qualitative and quantitative evidence are mutually reinforcing: qualitative work identifies problems and pathways, while quantitative analysis reveals how widespread problems are and whether policies actually work. The three presentations that followed illustrated that logic in practice across Timor-Leste, Papua New Guinea and Fiji.
Diana Contreras Suárez presented new analysis on child stunting in Timor-Leste, where roughly one in two children under five is stunted. Using machine learning and advanced econometric techniques applied to a rich dataset collected in partnership with The Asia Foundation, Contreras Suárez disaggregated stunting by severity. Her results indicate that maternal underweight is a particularly strong predictor among children who are severely stunted, whereas household food insecurity raises the risk across the entire distribution. Importantly, micronutrient supplementation emerged as especially protective for children facing the worst outcomes, suggesting that universal programs may miss opportunities to prioritise those most in need.
Christopher Hoy turned the analytical lens to tax policy in Papua New Guinea. Drawing on detailed survey and administrative data, reinforced by web-scraped market information and AI tools to follow the chain from tax exemptions to prices and household outcomes, Hoy’s work shows that exemptions on basic food items — commonly justified as pro-poor — have been regressive in practice. His findings challenge assumptions that headline policy measures automatically benefit poorer households and underline the need for routine ex-post evaluation using granular data.
Kushneel Prakash argued that better analysis also begins with asking different questions about what development data should capture. His research in Fiji urges policymakers to measure wellbeing beyond GDP and consumption: trust, social cohesion, cultural identity and uncertainty are central to resilience in contexts where concepts such as vanua — the Fijian relation between people, land and identity — shape everyday life. Prakash warns that development programs that neglect these social and cultural dimensions risk eroding the very social fabric that sustains communities.
Taken together, the session’s papers underscore a regional shift from simply collecting more data to extracting the right insights from it. That shift echoes calls made at earlier regional forums for stronger links between research and policy in Fiji and the Pacific, and follows recent moves by national statistical offices to adopt new survey methods and real-time monitoring. The immediate policy takeaway is clear: interventions must be both evidence-based and tailored — combining broad measures where appropriate with targeted actions for the most vulnerable, and expanding metrics to reflect local notions of wellbeing.
If development policy is meant to improve lives, the conference’s latest offerings argue, then the evidence base must do more than count people. It must point to who is being left behind, why, and which policy levers will actually reach them.