اثر زمان واقعی مداخله بانک مرکزی در بازار یورو

This paper investigates the real-time effects of sterilized foreign exchange intervention using official intraday intervention data provided by the Danish central bank. Our analysis employs a two-step weighted least squares estimation procedure. We control for macro surprises, address the issue of endogeneity, and carry out an array of robustness tests. Only when the direction of intervention is consistent with the monetary policy stance do we find that intervention exerts a significant influence on exchange rate returns.

This paper investigates the real-time (intraday) effects of Danish central bank intervention in the Euro (EUR) market over the 1 January 2002 to 31 December 2004 time-period.2 The interventions under study are carried out under the provisions of the Exchange Rate Mechanism (ERM II).3 Proprietary data on official, intraday intervention transactions, provided by the Danish central bank (Danmarks Nationalbank, henceforth DN), along with indicative 5-minute spot DKK/EUR exchange rate quotes facilitate our investigation. Our investigation employs the time-series econometrics of Andersen and Bollerslev (1998) and Andersen et al. (2003). It is very rare that a central bank makes official intraday intervention data available for research.4 In fact, until now only the Bank of Canada and the Swiss National Bank have made such data available. It has, however, been more than a decade since the Bank of Canada and the Swiss National Bank last intervened.5 By contrast, Denmark is currently pursuing an active intervention policy, i.e. DN carries out sterilized interventions on a discretionary basis when deemed necessary. The DN intraday intervention data, therefore, constitutes a unique opportunity to learn about the real-time effects of foreign exchange intervention carried out by a central bank that is currently intervening. We investigate whether sterilized intervention is effective in influencing exchange rate returns and whether it works differently when backed by consistent monetary policy (e.g. sterilized intervention sales of domestic currency during a period of domestic monetary policy easing). As discussed in Ghosh (2008), industrialized countries that intervene often use monetary policy (e.g. changing interest rates) towards achieving domestic goals such as controlling inflation or stimulating growth, while they use sterilized intervention (e.g. sell domestic currency and use open market operations to off-set any impact on money supply or policy rates) towards achieving exchange rate goals. In such a context, sterilized rather than unsterilized intervention is used since monetary and exchange rate goals are not always compatible and, as a result, it is not uncommon to observe sterilized interventions that are monetary policy inconsistent.6 In the case of Denmark, however, the primary objective of both monetary and exchange rate policy is to keep the DKK/EUR exchange rate within its deviation band. Therefore, the Danish monetary policy follows the monetary policy of the ECB, and only deviates when deemed necessary to keep the exchange rate within the band.7 The active intervention policy is pursued when intervention is deemed sufficient to influence the exchange rate as desired and a monetary policy move is deemed unnecessary. Even though monetary and exchange rate goals in the Danish context are compatible, monetary policy inconsistent interventions can occur. To illustrate, suppose during a period of, say, sustained relative monetary policy tightening the DKK appreciates just enough to prompt the DN to intervene by selling DKK, but not enough to make a relative lowering of interest rates seem necessary, the resulting intervention sales of DKK are monetary policy inconsistent. The period under study encompasses a structural break and two distinctly different sub-periods in terms of monetary policy. The official Danish monetary policy rate, the discount rate, as well as the key ECB interest rates, including the ECB deposit and lending rates, were all held constant during the first sub-period, whereas during the second sub-period these rates were lowered in tandem by a total of 125 basis points. However, the Danish deposit and lending rates were lowered by a total of 140 basis points during the second sub-period when the ECB deposit and lending rates were lowered by a total of 125 basis points (see Danmarks Nationalbank, 2002, Danmarks Nationalbank, 2003b and Danmarks Nationalbank, 2004). Accordingly, the first sub-period is associated with an unchanged Danish monetary policy stance while the second sub-period constitutes a period of independent Danish monetary policy easing in relative terms by a total of 15 basis points.8 It follows that the interventions that occur during the second sub-period are either monetary policy consistent or inconsistent. We investigate whether monetary policy consistency matters for the effectiveness of intervention by estimating the effect of intervention separately across the two sub-periods and separately across consistent and inconsistent interventions.9 Our main result is that only when the direction of intervention is consistent with monetary policy do we find evidence that intervention influences exchange rate returns. This suggests that even though sterilized intervention is by construction technically detached from monetary policy it is, in reality, not an independent policy instrument but only effective when consistent with monetary policy. We also assess whether consistency with either official or what we refer to as “de facto” exchange rate policy matters for the effectiveness of intervention. An intervention aimed at bringing the exchange rate closer to the center of the ERM II deviation band is considered to be consistent with official exchange rate policy, whereas an intervention aimed at pushing the exchange rate further away from the center is considered to be inconsistent with official exchange rate policy. As a measure of “de-facto” exchange rate policy consistency we consider the position of the exchange rate at the time of the intervention relative to the unconditional full-sample mean of the exchange rate. Contrary to our findings regarding monetary policy, we find no evidence that exchange rate policy consistency is a condition for effective intervention. In addition, we address the issue of endogeneity. Doing so reveals that some endogeneity is present even in our intraday analysis of intervention. However, we also show that the resulting simultaneity bias is too small to affect our results. We also extend our analysis to incorporate Danish, German, and Euro-area macro surprises. This allows us to get a sense of the relative influence of intervention. We show that the magnitudes of the coefficient estimates associated with scaled and thus comparable macro surprises and interventions are similar, thereby illustrating the importance of taking into account interventions when estimating exchange rate models. Since the Danish ERM II intervention experience pertains to maintaining the exchange rate within a narrow deviation band, our findings are applicable to Denmark and other countries that intervene to keep their respective exchange rates in narrow bands. Additional research is warranted in order to shed light on whether our results also pertain to countries intervening to keep their respective exchange rates in wide deviation bands and to countries with flexible exchange rates. The rest of the paper is organized as follows. The next section provides an overview of the institutional aspects regarding ERM II and DN intervention. 3 and 4 present the data and the econometric methodology, respectively. Section 5 discusses the results. Section 6 presents several robustness checks. Section 7 concludes.

This paper investigates the real-time (intraday) effects of Danish intervention in the EUR market over the 1 January 2002 to 31 December 2004 period, using proprietary intraday intervention data provided by the Danish central bank and the WLS time-series econometrics of Andersen and Bollerslev (1998) and Andersen et al. (2003). The Danish ERM II intervention experience pertains to maintaining the exchange rate within a very narrow deviation band. Therefore, our findings are not necessarily applicable to countries intervening to keep their respective exchange rates in wide deviation bands or to countries with flexible exchange rates. We test for and find a structural break in the data in August 2002, coinciding with the Danish deposit and lending rates being lowered while the ECB deposit and lending rates were held constant. Based on the structural break we separate our data into two sub-samples that are distinctly different in terms of the stance of the Danish monetary policy. Sub-sample 1 is associated with an unchanged Danish monetary policy stance while sub-sample 2 constitutes a period of Danish monetary policy easing. This facilitates an investigation of whether monetary policy consistency of sterilized intervention matters for whether sterilized intervention is effective. Our results show that only when intervention is consistent with monetary policy does intervention exert a significant influence on exchange rate returns. We also assess whether consistency with either official or a measure of “de facto” exchange rate policy matter for the effects of intervention. We find no evidence that exchange rate policy consistency is a condition for effective intervention. Furthermore, we address the issue of endogeneity by estimating a central bank reaction function in order to capture the expected component of the intervention variable. In turn, we use the residuals of the reaction function estimation as a proxy for unexpected intervention that we then use for obtaining an estimate of intervention that is free of simultaneity bias. We find that some endogeneity is present even in our intraday analysis of the effectiveness of intervention, yet we also find that the resulting simultaneity bias is too small to affect the results. In order to compare the influence of intervention relative to the influence of macro surprises as well as to ensure that our estimated effects of intervention are not tainted by the coincidental arrival of macro news, we also include time-stamped Danish, German, and Euro-area macro surprises in our estimations. Our results show, not surprisingly, that some but not all of the macro surprises influence the DKK/EUR exchange rate. More importantly, the magnitude of the coefficient estimates associated with scaled and thus comparable macro surprises and interventions show that the relative influence of intervention is comparable to the relative influence of most macro surprises. Our main result suggests that monetary policy consistency is necessary for intervention to be effective. In other words, even though it is technically possible to use sterilized intervention to pursue an exchange rate goal that is not supported by monetary policy, doing so is unlikely to significantly influence the exchange rate level.