Effectiveness of intervention in a small emerging market: an event study approach

Abstract

This article addresses the effectiveness of intervention using daily data from a small open economy for which intervention constituted an integral part of policy making. A matched-sample test of equality of means before and after intervention events, shows that the sterilized interventions by the central bank were effective for both purchases and sales of US dollars, but with associated fiscal costs. These results, which are robust to alternative event-window definitions and to alternative criteria for measuring ‘success’, suggest that the authorities were successful in keeping the exchange rate within a ‘target’ corridor. With many small emerging market economies seeking to balance the twin objectives of maintaining competitiveness while containing imported inflation, these results present an interesting case study which suggests that intervention can be an appropriate policy tool in some small open and emerging market economies.

Keywords:

• intervention
• event studies
• central bank
• exchange rate

JEL Classification::

• E58
• F31
• G14

Acknowledgements

We are grateful for helpful comments from Joshua Aizenman, Lucio Sarno, Ronald McDonald, José Sanchez-Fung and Ebrima Faal, none of whom is responsible for any remaining errors. The views expressed in this article are the sole responsibility of the authors and do not reflect the views or policies of the International Monetary Fund.

Notes

¹ Alternative narrower definitions of the start of an intervention period, used for robustness checks, included (i) the net of the interventions for the day changes its sign compared with the previous intervention day (net buy, net sell, net zero); (ii) the type of intervention changes compared with the previous intervention day (pure sell, pure buy, or sell and buy).

² To correct for the bias (greater rejection of the null of no difference between means) arising from the excess kurtosis (nonnormality) in the sample data, we use White's heteroscedasticity-consistent (robust) SEs in an auxiliary regression to estimate the significance of the mean of the sample matched pair differences.

³ Preliminary research, using GARCH regression models, indicate the importance of accounting for asymmetry in policy reaction functions. In particular, dummy variables for foreign exchange sales and purchases explain movements in the mean of the exchange rate, after controlling for own-lag effects, and large sales and purchase interventions (greater than two SDs) reduce significantly the volatility of exchange rate movements, but with differential impacts.