Order flow and the real: indirect evidence of the effectiveness of sterilized interventions

This study presents indirect evidence of the effectiveness of sterilized interventions in Brazil based on the complete records of daily customer order flow data reported by Brazilian dealers, as well as foreign exchange intervention data over a time span of 10 years (2002-2011). We find that the effect of USD sales by end-users on the BRL/USD was much stronger on days in which the BCB did not intervene in the spot foreign exchange market. The regressions suggest that a 1% appreciation of the Real would have required the sale of 2.0 bn USD by final customers on days in which the Central Bank refrained from intervening. This compares to required sales of 5.5 bn USD on days in which the Central Bank was present in the market. This large effect, in spite of the fact that the median intervention amounted to only 140 mn USD, can be interpreted as evidence for the indirect damping channel. Furthermore, we find that order flows coming from outside of the financial sector have a (considerably) stronger effect on the BRL/USD exchange rate than those coming from financial customers. We argue that some studies may have failed to find significant effects of BCB interventions due to a problem of reverse causality, as in a regime of discretionary interventions the decision to intervene is often taken during trading hours.

JEL classification: E52, E58, F31

Keywords: order flow, intervention, exchange rate