Are funding rule instrumental variables plausibly exogenous for estimating the health effects of English NHS expenditure?

<p dir="ltr">Estimating the causal effect of local NHS expenditure on health outcomes in England often involves the use of instrumental variables (IVs) to address endogeneity due to unobserved confounding and reverse causality. Earlier studies used census-derived IVs with limited theoretical support for their exogeneity. More recent work uses NHS funding rule components, such as the age-cost index, market forces factor, and distance-from-target index, as instruments. While these are theoretically stronger, their validity still depends on the assumption that they only affect mortality through expenditure.</p><p dir="ltr">This study evaluates the robustness of IV-based estimates of healthcare expenditure on all-cause, cancer, and circulatory disease mortality using 2014/15 Clinical Commissioning Group expenditure data. We apply a sensitivity analysis that relaxes exclusion restriction assumptions, using methods from Conley et al. (2012) and Claxton et al. (2015), and simulate outcome elasticities under plausible deviations from exogeneity.</p><p dir="ltr">Findings show that relaxing this assumption yields modest bias and tends to underestimate the effect of expenditure on mortality. The structural uncertainty is also relatively modest, and most estimated elasticities remain statistically significant. These results provide reassurance that, for policy purposes, the original IV regression models offer useful evidence on the marginal productivity of NHS expenditure.</p>