Budgetary allocation and health for all: Computational methods for population well-being

This is our first attempt to bring the policy priority inference framework to a domain-specific application. Here, we study the problem of reaching a well-being economy in England using a unique dataset with manually mapped data on budget lines and outcome indicators.

We study the budgetary contraction that occurred in England between 2013 and 2019 compiling extensive data on population outcomes and local government expenditure. Overall, we collect information on 322 unique indicators, spanning 14 health categories. Out of those indicators, 93 had information stratified by the Index of Multiple Deprivation, that we use to assess impact differences between the top and the bottom of the deprivation distribution. We employ a computational model of policy prioritization that describes the (vertical) causal chains linking public spending to indicators measuring the performance of government programmes across multiple policy dimensions. It accounts for key aspects of public governance and incorporates the complex interlinkages between policy indicators through a network of conditional dependencies. Through the model, we devise an intuitive counterfactual scenario where public spending is increasing, which is estimated using Gaussian Processes. Finally, we measure the effect of budget shrinking through an impact metric developed to assess alternative budgetary allocations. We find a significant impact on 50 indicators spanning several health categories. Among those with the greatest impact, several belong to the wider determinants of health (e.g., Education, Mental Health, Social Care). The impact does not substantially differ between deprivation deciles.