اثرات روزانه مداخله بانک مرکزی روی گسترش نرخ ارز

We investigate the intraday effects of intra-marginal intervention in a horizontal band on the exchange rate spread. Official intraday data on Danish intervention transactions in the ERM II, the Exchange Rate Mechanism of the European Union, facilitates our analysis. We show that intervention purchases and sales both exert a significant influence on the exchange rate spread, but in opposite directions. Intervention purchases of the small currency, on average, narrow the spread while intervention sales of the small currency, on average, widen the spread. This is a novel finding that differs from those of existing studies that find intervention always widens the exchange rate spread and increases market uncertainty.

Many studies have investigated the intraday effects of foreign exchange intervention.1 Few studies have examined the intraday effects of intervention using accurate, official intraday intervention data.2 Even fewer studies have analyzed the intraday effects of intervention on exchange rate spreads. These studies examine the spread effects of intervention in a crawling peg (Melvin et al., 2009) and the spread effects of intervention in floating exchange rates (Chari, 2007; Pasquariello, 2007).3 All of these studies find that foreign exchange intervention is associated with an intraday increase in the bid-ask exchange rate spread. In this paper we investigate how unannounced intra-marginal intervention in a horizontal band such as the ERM II, the Exchange Rate Mechanism of the European Union, influences the bid-ask exchange rate spread. Official intraday data on intervention transactions in the Danish Krone-Euro (DKK/EUR) market provided by Danmarks Nationalbank (DN), the Danish central bank, facilitates our analysis.4 It is interesting to investigate how the exchange rate spread is influenced by intervention since doing so can reveal if intervention increases or reduces foreign exchange market uncertainty regarding whether a currency is seen as properly priced.5 It is particularly interesting to analyze the effects of intervention on exchange rate spreads using the Danish intervention data for three reasons. First, the Danish intervention data presents a very rare opportunity for learning about the influence of intervention on exchange rate spreads using official, time-stamped data provided by a currently intervening central bank.6 Second, no previous study has analyzed the effects of intervention in a horizontal band, such as the ERM II band, on exchange rate spreads. Third, analyzing the Danish experience of intra-marginal interventions in the horizontal ERM II band can bring insights of relevance to Denmark and to other EU member states currently participating in the ERM II as well as to the EU member states that are not in the Euro-zone and are not participating in the ERM II, but are expected to participate in ERM II at a later date in order to fulfill the exchange rate criterion necessary for adopting the EUR.7 In our context of unannounced intervention by a small central bank aimed at maintaining a small currency in a horizontal band around a major currency, such as maintaining the DKK against the EUR in the ERM II, we cannot assume that the effects of intervention purchases and sales are necessarily symmetric and that foreign exchange intervention is always associated with an intraday increase in the bid-ask exchange rate spread.8 First, it is reasonable to assume that as the intervention occurs, foreign exchange market customers observe only a large order flow but cannot discern the origin of the trade.9 Therefore, an unannounced intervention by the small central bank, whether a sale or a purchase, is initially understood by customers only as a large trade and, accordingly, interpreted as informative regarding pressure, either upwards or downwards, on the small currency. An unannounced intervention by the small central bank, however, is unlikely to be interpreted as informative regarding pressure on the large currency.10 Second, downward pressure on a currency in a horizontal band is potentially more concerning than upward pressure, especially if the downward pressure adversely affects the credibility of the currency and its exchange rate regime.11 This concern is particularly relevant for the DKK, which has a relatively recent history of instability and frequent devaluations prior to participation in the ERM I, the predecessor of the ERM II. The fact that the DKK is continuously maintained at an appreciated rate relative to the central rate corroborates that downward pressure is viewed as more concerning than upward pressure. Reduced credibility will manifest itself in the form of increased uncertainty and thus a widening of the exchange rate spread. By contrast, a large purchase of the small currency may be interpreted as a confirmation of the credibility of the small currency and the exchange rate regime, thereby leading to reduced uncertainty and thus a narrowing of the exchange rate spread. In other words, the effects of unannounced intervention in a horizontal band could be asymmetric across purchases and sales. To assess the intraday effects of intervention on the exchange rate spread in the horizontal ERM II band, and to formally test whether intervention purchases and intervention sales influence the spread asymmetrically, we estimate time-series models of the DKK/EUR exchange rate bid-ask spread, calculated from indicative 5-minute spot bid and ask DKK/EUR prices purchased from Olsen and Associates, with intervention purchases and sales entering as separate explanatory variables. Our sample covers the 1 August 2002 to 31 December 2004 period. We carry out our baseline estimations using OLS with heteroskedasticity and autocorrelation consistent (HAC) standard errors and covariances. As a methodological robustness test we also estimate models using the weighted least squares (WLS) procedure developed by Andersen and Bollerslev (1998). Consistent with the existing studies of the intraday effects of intervention on the exchange rate spread in the context of a crawling peg or floating exchange rates, our results confirm that intervention significantly influences the spread. However, in contrast with these studies, we show that interventions in a horizontal band do not necessarily increase the spread. Instead, we show that intervention purchases of the small currency, on average, reduce the spread, while only intervention sales, on average, increase the spread. This finding conforms to the suggestion that uncertainty in the market regarding the exchange rate decreases when interventions can be interpreted as upward pressure on the small currency, while uncertainty in the market increases when interventions can be interpreted as downward pressure on the small currency. Our results hold up against an array of robustness checks, including employing a different econometric procedure and controlling for coincidental arrival of macro news. The rest of the paper is organized as follows. Section 2 briefly outlines the traditional intervention transmission channels and their relevance to the Danish interventions. Sections 3 and 4 present the data and the empirical model, respectively. Section 5 discusses the results. Section 6 presents several robustness checks. Section 7 concludes.

The existing literature on intervention and spreads is very scarce, and it does not consider intervention in a horizontal band such as the ERM II. Our study is the first to analyze the intraday effects of intra-marginal intervention on the exchange rate spread in the context of a horizontal ERM II band. Proprietary data on official Danish intervention transactions in the DKK/EUR exchange rate market facilitates our analysis. It is interesting to investigate how the exchange rate spread is influenced by intervention because doing so can reveal whether intervention affects foreign exchange market uncertainty. It is particularly interesting to study the Danish intervention experience in ERM II since it can provide valuable insights of relevance to not only Denmark but also to the currently 8 EU member states that have not yet adopted the EUR but are legally required to do so at some point in time. These 8 EU member states, therefore, are either already participating in the ERM II or expected to participate in the ERM II at a later date prior to joining the Euro-area. Our time-series estimations of the 5-minute DKK/EUR exchange rate bid-ask spread provide two insights. First, we show that failure to allow for the possibility of asymmetric affects across intervention purchases and intervention sales can lead to failure to detect the influence of intervention. This is a result of general interest as it illustrates the necessity of considering asymmetries when assessing the influence of large trades, such as intervention transactions, in foreign exchange rate markets. Second, we show that intervention purchases and sales in a horizontal ERM II band both significantly influence the spread, but the direction of the intervention matters for the resolution, or creation, of uncertainty in the market. Specifically, we show that intervention purchases of the ERM II currency, on average, reduce the spread while intervention sales of the ERM II currency, on average, increase the spread. This is a new and interesting result that differs from those of existing studies that find intervention to increase market uncertainty regardless of the direction of the intervention. It is also a finding of practical relevance to authorities with the mandate to carry out interventions in the context of a horizontal band. Contrary to findings of studies analyzing intervention in floating exchange rates that, regardless of the exchange rate level or volatility effects of intervention, show that intervention is costly because it increases market uncertainty, we show that this is not necessarily the case. Instead, depending on the direction of intervention, we show that intervention can, in fact, reduce market uncertainty. Lastly, our result has implications for hedge funds and other speculators that may profit when spreads widen or narrow with intervention in a predictable manner. The institutional framework surrounding intervention in a horizontal ERM II band is markedly different from that of intervention in other exchange rate regimes such as, for example, floating rates. Therefore, we cannot infer that our results describe the intraday effects of intervention on the exchange rate spread in general, i.e. regardless of the specifics of the exchange rate system within which the interventions occur. Rather, our study encourages additional research in order to assess the extent to which our findings also describe the influence of intervention on the exchange rate spread in the context of other exchange rate systems.