Cookies on this website
We use cookies to ensure that we give you the best experience on our website. If you click 'Continue' we'll assume that you are happy to receive all cookies and you won't see this message again. Click 'Find out more' for information on how to change your cookie settings.
Dates: 2012-2013
Funding: Coeliac UK  & HERC funding
Information: Mara Violato, Alastair Gray

A common problem in health economics, public health and epidemiology is to estimate the effect of a treatment, intervention or policy programme on outcomes. In situations where the assignment to treatment changes discontinuously with respect to a threshold value of a running variable, quasi-experimental variability in data has often been exploited to identify local causal effects whenever randomised experiments were impractical.

In this study, we revise some of the more commonly applied parametric methods (before-after, interrupted time series, regression discontinuity) used in the literature for estimating (local) causal treatment effects in discontinuity designs with a discrete running variable and propose some methodological improvements to address issues of efficiency and heterogeneity in the estimation of effects.

Firstly, in cross-sectional settings, we propose the use of wild-bootstrapped clustered-standard errors to scale down overstated precisions of estimates induced by conventional standard errors and to allow efficiency gains over the currently used clustered-standard-errors, even in cases when the number of clusters is small. Secondly, the extension of the technique to a panel data context induces further efficiency improvements and allows controlling for unobserved individual and time-invariant heterogeneity. Methods and their extensions are then compared empirically by applying them to the study of the impact of diagnosis of coeliac disease on health care costs in the UK using the General Practice Research Database. This is a question of considerable relevance in many evaluations of cost effectiveness.