microstructure framework, intervention is expected to be stabilizing if implemented in the volatile state and de-stabilizing if implemented in the tranquil state. This implies that 1 0 and 2 0. If intervention increases the probability of remaining in the volatile regime (2 0), as well as the tranquil regime (1 0), then intervention is only effective in the tranquil regime, paradoxically when it is generally not needed. In this case, it would have limited effectiveness as a policy instrument in the foreign exchange market. The two-state Markov switching framework is therefore open-ended and compatible with any view of the impact of intervention. In fact it is also compatible with the view that intervention has no impact in either regime(1 2 0).
2.3 Data and Intervention Dynamics
We use daily data on exchange rates for Croatia, Iceland, Jamaica and Australia.
The data set on Croatia includes 1636 observations covering the period January 3, 2000 to April 10, 2006. Iceland’s data set includes 1573 observations covering the period January 4, 2000 to April 28, 2006. Jamaica’s data set includes 1161 observations covering the period February 7, 2002 to September 28, 2006. Finally, the data set on Australia covers 1434 observations over the period January 3, 2000 to June 30, 2005.
The exchange rate for Croatia, Iceland and Jamaica is defined as the midpoint between the weighted average bid and ask prices of the domestic currency per unit of the intervention currency. The exchange rate for Australia is defined as the midpoint between the weighted average bid and asks prices of the intervention currency per unit
the United Sates dollar while Croatia’s intervention currency is the Euro. We use the actual daily sales and purchases of the intervention currency by the central bank against the domestic currency as our intervention variable. The policy interest rates utilized in this study are annualized rates and include the inter-bank rate for Croatia, the 3-month repurchase rate for Iceland, the month reverse repurchase rate for Jamaica and the 3-month treasury bill rate for Australia.
All of the exchange rate and interest rate data for Croatia, Iceland and Australia were sourced from Datastream. Intervention data were sourced from the respective central banks. Interest rate and exchange rate data for Jamaica were sourced from the Bank of Jamaica. A briefly description of the intervention dynamics for the countries reviewed are outlined below.
The Croatian foreign exchange market is relatively small and under-developed.
It is driven by links to the Euro rather than the US dollar, with approximately 67% of Croatian reserves denominated in Euros. The only market makers in the Croatian market are commercial banks. The Croatian central bank (Croatian National Bank, CNB) intervenes relatively infrequently (5%) in the spot market using partially sterilized interventions.
Table 2.1: The Dynamics of Intervention
Variable Australia Croatia Iceland Jamaica
Sales Days Purchases Days
Total Interventions Days Total Sample Days
371 23 29 222
535 52 511 n/a
906 75 540 222
1434 1636 1573 1161
% Intervention Days
% Intervention Sale Days
% Intervention Purchase Days
0.63 0.05 0.34 0.19
0.25 0.01 0.02 0.19
0.37 0.04 0.32 n/a
Average Inter. (US$M) Average Inter. Sales Average Inter. purchases Max. Inter. Sales Max. Inter. Purchases
19.5 2.4 1.2 2.3
7.2 0.7 0.3 2.3
12.3 1.7 0.9 n/a
176 148 42.2 26.1
458 179 82.5 n/a
Note: % intervention days are calculated as intervention days divided by total sample days.
The CNB used mostly buying interventions in this period because buying foreign exchange in the first years of Croatian independence was necessary to build up the stock of international reserves. The only year in the reviewed period where the CNB used selling operations to support the Kuna was 2003 when a banking crisis caused instability and capital outflows (see Table 2.1). Croatia has a managed floating exchange rate regime but the exchange rate exhibits relatively low flexibility. This is most likely due to the CNB’s objective of managing the Kuna within a tight band against the Euro17.
The Icelandic foreign exchange market is also a small, relatively underdeveloped market. The majority of trades are conducted in US dollars on the spot market. The Icelandic foreign exchange is built on an inter-bank market with the major commercial banks as market makers. In the period reviewed, the Icelandic Central Bank (ICB) interventions were relatively infrequent, occurring only on 34% of the days in the sample, with 32% of interventions being intervention purchases. The ICB intervened with predominantly selling operations in the 2000 and 2001, reflecting the fact that it was “leaning against the wind” in a period characterized by outflows of foreign currency and uncertainty leading to a weakening of the Krona. In the period 2002 to 2006 the ICB intervened mostly with frequent and relatively small buying operations, with the aim to build up foreign exchange reserves. The ICB does not fully and immediately sterilize its foreign exchange interventions as do central banks in some developed markets18.
The Jamaican foreign exchange market is based on an inter-bank system, where commercial banks are the major market makers. Traders can directly observe trade volumes and prices on a real time basis through the electronic gateway for auctions trade and foreign exchange management (E-GATE). In the period under review the Bank of Jamaica (BOJ) intervened relatively infrequently on 19% of the sample days.
They intervened selling only in the period under review, indicating that the BOJ was generally “leaning again the wind” in a period characterized by episodic instability in the market. The average intervention was relatively small. The BOJ sterilizes most of its foreign exchange interventions. There are instances, however, where full sterilization
17 See Egert (2007) and Chmelarove and Schnabl (2006) for more analytical reviews of intervention in the Croatian foreign exchange market.
would move the money base away from the target set by the BOJ and in these cases the intervention would not be fully sterilized, leading to intervention reinforcing monetary policy measures (Robinson and Robinson, 1997).
The Australian foreign exchange market in April 2004 accounted for approximately 3.4% of average daily global foreign exchange turnover in traditional instruments. In terms of actual practice, the Reserve Bank of Australia (RBA) conducts all its interventions in the spot Australian dollar/US dollar market and immediately sterilizes these operations. A broad overview of the actual intervention operations of the RBA is outlined in Table 2.1. This shows that the RBA intervened on 906 days out of the 1434 sample days, with intervention purchases (535 days) being more frequent than intervention sales (371 days). This suggest that the RBA was “leaning against the wind” in this period in which the Australian dollar was appreciating19.
The dynamics of intervention in these markets clearly reflect different intervention objectives and differences in the size and sophistication of the markets.
There are some similarities across jurisdictions, such as, interventions were executed predominantly in the spot market and most central banks attempted to sterilize their interventions. There are important differences though which reflect the differences in the degree of development of the market and the intervention objectives, as well as differences in intervention frequency, surprisingly with Australia, the developed market, recording the highest intervention frequency (64%) relative to Iceland (34%), Jamaica (19%) and Croatia (5%). The fact that in some developing markets interventions are not immediately and/or fully sterilized might lead to monetary/interest rate policy dominating direct interventions.