Matching estimates of the impact of over-the-counter emergency birth control on teenage pregnancy.

In this paper, we demonstrate how matching estimators can be used to evaluate policy interventions which are implemented in relatively few regions at different times. Our technique is based on translating calendar time into 'experimental time' to provide a common starting point for entry by different areas into the scheme. Such an approach is likely to have many applications, in particular to cases of state- or country-level interventions for which only aggregate data are available. We illustrate the technique using the case of free over-the-counter access to emergency birth control for teenagers at pharmacies in England. We construct matching estimates of the impact of this scheme on the under-18 conception rate in local authorities. Irrespective of either the matching or the adjustment procedure, we find no evidence that over-the-counter emergency birth control schemes lead to lower teenage pregnancy rates. Copyright 2006 John Wiley & Sons, Ltd.