This paper investigates the effectiveness of sterilized foreign exchange interventions by exploiting a discontinuous policy rule used by the Central Bank of Colombia. We use a unique data set that comprises tick by tick intervention and order book data, daily capital in- and outflows, and balance sheet information of financial institutions. We apply regression discontinuiTY methods to identify the surprise component of rule-based interventions and use this variation to measure how they affect exchange rates and capital flows. Our findings indicate that interventions had significant effects on the exchange rate, albeit short-lived (2–3 weeks). Moreover, capital controls amplify the effect of intervention, though some effect remains even in the presence of free capital flows. A methodological contribution of the paper is to extend regression discontinuiTY designs to a time-series environment and to show how these techniques can be used to identify and estimate local non-linear impulse response functions. A clearly defined policy rule and high frequency data are crucial in exploiting local variation around the policy cutoff.
