Sterilised Foreign Exchange Intervention in Practice

  • Felix Hüfner
Part of the ZEW Economic Studies book series (ZEW, volume 23)


We now turn to the evidence for the use of foreign exchange intervention in inflation targeting countries. So far, we have seen, that UIP does not hold in practice and that central banks regularly sterilise their foreign exchange operations (thus, they can be regarded as an additional policy tool). In this final chapter we aim to fulfil two tasks: to determine to what extent interventions are used at all and, secondly, to analyse if they are used as an additional policy instrument. Our indicator for the first issue are changes in foreign exchange reserves. As an indication for the use of interventions as additional policy tool we look at each central bank’s policy reaction function and at the volatility of its short-term interest rate.


Exchange Rate Interest Rate Monetary Policy Central Bank Foreign Exchange 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 96.
    The only exceptions for industrial countries, to our knowledge, are the US Federal Reserve Bank, the Bank of Japan, the Reserve Bank of Australia and to some extent the Bundesbank. However, all data is published with a time lag of usually six months. 97 There are also some papers that use actual daily intervention data from the US Federal Reserve or the Bundesbank to study the effects of interventions (see e.g. Dominguez and Frankel, 1993 a).Google Scholar
  2. 98.
    See for example Szakmary and Mathur, 1997; Kearney and MacDonald, 1986; Takagi, 1991; Taylor, 1982; Ghosh, 1992.Google Scholar
  3. 99.
    Publications of reserve data are also interpreted as an intervention proxy by foreign exchange market participants. For example, Agence France Press news states in their news release of Bank of England reserve data: “The level of reserves is watched as an indication of the extent of intervention by the Bank of England on foreign exchange markets.“ (Agence France Press (AFX) news, August 3, 1993, “British MO money supply rises 1.5% in July”).Google Scholar
  4. 100.
    See BIS, 1998.Google Scholar
  5. 101.
    Special Drawing Rights (SDR) are an international reserve asset created by the IMF in 1969. Its value is determined from a basket of four currencies (Euro, Japanese yen, pound sterling, and U.S. dollar). Today’s role of SDRs is relatively limited. They serve as a unit of account of the IMF as a potential claim on the freely usable currencies of IMF members (holders of SDRs can exchange their SDRs for these currencies) (See
  6. 102.
    See e.g. Krugman and Obstfeld, 1991.Google Scholar
  7. 103.
    According to Roger (1993) “[…] the most appropriate definition of the cost of holding reserves is the net social opportunity cost — that is, the best alternative social yield on the resources tied up in reserves minus the actual yield on reserves.” (p. 32). In New Zealand the costs for maintaining a reserve level of 4 to 5 billion NZ$ are about 5 million NZ$ per year (this includes financing of the reserves, personnel, systems and overhead expenses) (see Archer and Halliday, 1998).Google Scholar
  8. 104.
    The importance of reserves as signalling confidence to financial markets has been highlighted during the Asian crisis. One of the triggers of the crisis in Thailand was the fact that financial markets were in doubt about the true level of foreign exchange reserves at the Bank of Thailand (see Blustein, 2001, for a detailed account of the Asian crisis).Google Scholar
  9. 105.
    Strictly speaking this represents the balance sheet of a country’s monetary authorities (the central bank and exchange-stabilisation authorities) as in many countries the foreign exchange reserves belong to a special fund that is owned by the government (e.g. the US Exchange Stabilisation Fund or the Exchange Equalisation Account in the United Kingdom).Google Scholar
  10. 106.
    In 1996 the BIS held US$ 91 billion of foreign exchange reserves (about 7% of world-wide foreign exchange reserves) for around 120 central banks (see Deutsche Bundesbank, 1997: 189).Google Scholar
  11. 107.
    A further reason for fluctuations stems from losses on the bond holdings if bond prices decline. However, since reserves are due to liquidity concerns primarily invested in short-term bonds, fluctuations in bond prices should not have an important effect. Nevertheless, this point might gain in importance since the investment performance of the reserve portfolios become more and more important for central banks.Google Scholar
  12. 108.
    The IMF does not give further details about this phenomenon. The increase might be due to an increasing number of emerging market currencies that are held in the reserve portfolios of developing countries.Google Scholar
  13. 109.
    See Dal Bosco, 1998:295.Google Scholar
  14. 110.
    See IMF International Financial Statistics Yearbook, various issues.Google Scholar
  15. 111.
    See also Hüfner (2000) for an analysis of the relationship between the Bank of England’s’ reserve data and IFS data.Google Scholar
  16. 112.
    However, there also exist more sophisticated strategies. The Swedish central bank, for example, holds an investment and a liquidity portfolio of reserves with different maturities (see Ragnartz, 1999).Google Scholar
  17. 113.
    Neely (2000) assumes that all reserves are held in US$ (p. 28) — a rather unrealistic assumption. This might also contribute to the results he obtains for the correlation coefficients after the adjustment for external factors.Google Scholar
  18. 114.
    Furthermore, his results for Germany might be biased since his German data only include interventions undertaken in the US market and not interventions in European currencies. Unless all interventions in the EMS were undertaken in the US$-market this measure might be biased.Google Scholar
  19. 115.
    For an investigation of the Bank of England’s’ foreign exchange reserve changes and the GBP/US$ exchange rate see Hüfher (2000).Google Scholar
  20. 116.
    In his study of the effects of US intervention, Kim (2002) uses both official data and reserves and finds that “many similar results can be obtained by using foreign exchange reserves [instead of official intervention data].” (p. 25).Google Scholar
  21. 117.
    See also the similar definition of Dominguez (1999): “Foreign exchange market intervention is a transaction or announcement by an official agent of a government that is intended to influence the value of an exchange rate.” (p. 5).Google Scholar
  22. 118.
    The costs of the central bank for the sterilisation of these additional foreign exchange reserves were covered by the State.Google Scholar
  23. 119.
    Bloomberg News, February 14, 1994, 1:01 pm, “Bank of England emphasises pound’s role in monetary policy”.Google Scholar
  24. 120.
    If these effects are not corrected then an increase in reserves that is simply due to an appreciation of the reserve currencies in the portfolio, for example, might be wrongly interpreted as a purchase of foreign exchange by the central bank (in order to weaken the domestic currency).Google Scholar
  25. 121.
    See also Takagi (1991) for the estimation of foreign exchange reserves interest earnings. 122 See Bank of England, Minutes of Monetary Policy Committee Meeting 9 and 10 February 2000, available at Some members of the committee view interventions as a more powerful signal than verbal concerns about the exchange rate. However, it turns out that the majority of committee members is concerned that failed intervention operations would seriously damage the credibility of the central bank.
  26. 123.
    Other studies support our finding of high reserve variability for New Zealand: Levy Yeyati and Sturzenegger (1999) find: “[…] many countries that claimed to run a floating rate displayed little exchange rate volatility coupled with intense foreign exchange market intervention, so that in reality they are closer to a fix exchange rate regime. Brazil and New Zealand appear occasionally in this group as well as some Scandinavian economies like Norway, Finland and Sweden.“ (p. 9). Similarly, Calvo and Reinhart (2000) write: “[…] large swings in foreign exchange reserves appear to be commonplace, consistent with a higher extent of intervention in the foreign exchange market — even relative to what is to be expected a priori from a freely floating exchange rate regime. Nor is this exclusively an emerging market phenomenon — Canada’s and New Zealand’s reserve changes are about seven times and five times, respectively, as volatile as that of the United States.“ (p. 20). Both studies do not refer to the statement of the Reserve Bank of New Zealand that it is one of the purest floaters and thus do not attempt to explain this discrepancy further.Google Scholar
  27. 124.
    This interpretation is supported by the fact that there are virtually no news reports of interventions by the Reserve Bank of New Zealand (see Appendix I) — which would be surprising if the central bank is indeed pursuing active intervention operations on a high frequency and scale.Google Scholar
  28. 125.
    This is also reflected by the frequency of news reports about Australian central bank intervention given in Appendix I.Google Scholar
  29. 126.
    Kim and Sheen (2002) find that the Reserve Bank’s interventions were determined by exchange rate trend correction, volatility smoothing, US-Australian interest rate differentials, profitability and reserve inventory considerations. This shows that the Reserve Bank was quite flexible in its approach.Google Scholar
  30. 127.
    In September 1998 the Bank of Canada abolished the intervention programs which triggered intervention in an automatic fashion. Since then, interventions are carried out on a discretionary basis (see Canadian Department of Finance, Annual Report to Parliament on the Operations of the Exchange Fund Account 1999).Google Scholar
  31. 128.
    See Williamson, 1996.Google Scholar
  32. 129.
    Note that this relationship is different to uncovered interest parity: if UIP would hold continuously, then central banks would not need to care about the exchange rate since foreign exchange market efficiency would always be guaranteed.Google Scholar
  33. 130.
    Svensson (2000b), whose model of open economy inflation targeting does not allow for an additional policy tool like sterilised interventions, argues that incorporating the exchange rate into monetary policy under strict inflation targeting leads to more interest rate variability. “[…] under strict inflation targeting […] the direct exchange rate channel offers a potentially effective inflation stabilisation at a relatively short horizon. Such ambitious inflation targeting may require considerable activism in monetary policy (activism in the sense of frequent adjustments of the monetary policy instrument)![…]” (Svensson, 2000b: 4).Google Scholar
  34. 131.
    SeeAdam et al., 2001.Google Scholar
  35. 132.
    See Reserve Bank of New Zealand, 1999.Google Scholar
  36. 133.
    February 1996 is our starting point since the Bank of Canada changed the method of monetary policy implementation which made it easier to determine the Bank’s near term policy stance. Before this date, the Bank rate had been set each week as the average interest rate of auctions of 3-month treasury bills. From 26 February on, the Bank rate equals the upper limit of the operating band for the overnight financing rate which means that policy changes can be identified more precisely than before. According to the Bank of Canada, this measure „eliminates the uncertainty and confusion that had existed previously as to which was the key indicator of the Bank’s policy stance.” (Bank ofCanada Monetary Policy Report May 1996, p. 14).Google Scholar
  37. 134.
    See for example the discussion of the Swedish interventions in 2001 by the Riksbank executive board meeting (Riksbank, 2001b). The controversy among the board members concerning the efficiency of these interventions as well as questions about the relationship between interventions and interest rate policy demonstrates that in the day-to-day conduct of monetary policy the role of interventions is far from self-evident.Google Scholar
  38. 135.
    Taylor did not include the exchange rate into his rule for the United States because „simulations of multicountry models led me to believe that if the central bank reacted too strongly to the exchange rate then inflation-output performance would deteriorate.“ Thus, the Taylor rule should be viewed as a formula specifically designed for the US. As Taylor admits: „However, it is clear that the same conclusion would not necessarily be reached for other countries, especially small open economies.“ (Taylor, 2000b: 16).Google Scholar
  39. 136.
    This follows the procedure of Nelson (2001). Another method often applied to this kind of problem (see e.g. Clarida et al., 1998) is estimation by GMM. However, results with this method proved to be very unstable — not least because of the relatively short sample. Lotti (2000) notes that TSLS produces more stable, although less efficient estimates than GMM. Adam et al. (2001) apply both methods in their study and results obtained with the different procedures do not differ significantly.Google Scholar
  40. 137.
    See Kennedy, 1998.Google Scholar
  41. 138.
    A more detailed description of the TSLS method is given in the appendix.Google Scholar
  42. 139.
    We chose not to use the overnight interest rate, since New Zealand operated, until 1999, a cash target instead of targeting overnight interest rates directly (see Reserve Bank of New Zealand, 1999). As a consequence, New Zealand overnight interest rates in the first nine years of inflation targeting were highly volatile. Thus, we follow Huang et al. (2001), who in their study of New Zealand monetary policy state: “There is thus no equivalent of the Federal Funds Rate to incorporate but it seems sensible to use the rate that the RBNZ [Reserve Bank of New Zealand] itself referred to, namely, the 90-day bank bill rate, as the best indicator of policy in the context of a Taylor rule.” (p. 10). In order to better compare interest rate movements we use the three-month interest rate for all countries.Google Scholar
  43. 140.
    Including the exchange rate in first differences (which would also be stationary) would thus be inferior as it would only measure the central bank’s reaction to shortterm fluctuations.Google Scholar
  44. 141.
    New Zealand industrial production is available only on a quarterly basis, anyway.Google Scholar
  45. 142.
    We also performed the estimates using contemporaneous inflation variables (as in the original Taylor rule) and the results were worse. This supports the fact that monetary policy of inflation targeting central banks is forward-looking, i.e. that interest rate decision are based on forecasted inflation. The finding that the Reserve Bank of New Zealand focuses more on short term than on long term inflation forecasts is supported by Huang et al. (2001:24).Google Scholar
  46. 143.
    This results is in line with the findings of Drew and Plantier (2000) who study interest rate smoothing behaviour in New Zealand, Australia, Canada and the United States.Google Scholar
  47. 144.
    Clarida et al. (2000) show that in the US pre-Volcker period nominal rates were increased by less than the inflation increase. They argue that this caused the macroeconomic instability (i.e. high volatility in both inflation and output development) of the late sixties and seventies compared with the Volcker-Greenspan era.Google Scholar
  48. 145.
    Taken together, our results are supported by related studies like Collins and Siklos (2001). They find for Canada, New Zealand and Australia that “[…] dollar bloc central banks are reasonably well-described as having targeted inflation, pursued heavy interest rate smoothing, and perhaps having little concern over output.” (p. 21).Google Scholar
  49. 146.
    Gerlach and Smets (2000) report similar results: in their study of Australia, Canada and New Zealand they find that the exchange rate coefficient is insignificant in Australia and significant in New Zealand. In contrast to our study, however, they also report a significant coefficient for Canada. It needs to be noted that their sample range is much smaller than ours (theirs ends in 1997:2).Google Scholar
  50. 147.
    Note that the coefficient on the exchange rate does not necessarily have to be negative which would indicate something like a leaning-against-the-wind behaviour, i.e. raising interest rates when the exchange rate depreciates and vice versa. The central bank might instead choose to support an existing appreciation/depreciation trend if it is in line with the achievement of the inflation target.Google Scholar
  51. 148.
    All reserve changes are adjusted for exchange rate valuation effects and interest earnings effects.Google Scholar
  52. 149.
    “In the first half of the 1990s, the exchange rate acted as the principal guide for determining how much policy should be adjusted when the forecast inflation rate deviated from the target range. This was not because significance was attached to the exchange rate level per se, but rather because the exchange rate strongly influences future prices and thus inflation rates in the adjustment period.” (Gerlach and Smets, 2000: 1690).Google Scholar
  53. 150.
    The Monetary Conditions Index (MCI) in New Zealand is a combination of the trade-weighted exchange rate index and the 90 day bank bill rate (a 2% increase (decrease) in the exchange rate index is equivalent to a one percentage point increase (decrease) in 90 day rates). A one-point change on this index is equivalent to a 0.01 percentage change in the 90 day rates in the absence of a change in the TWI. See
  54. 151.
    The Bank of Canada spent about US$ 5,8 billion in August 1998 in support of the Canada dollar (see Bloomberg news, September 30, 1998, “BANK OF CANADA CHANGES INTERVENTION POLICY FOR CANADIAN DOLLAR”).Google Scholar
  55. 152.
    As mentioned earlier there is some discussion about the formal start of inflation targeting. The Reserve Bank defines early 1993 as the start.Google Scholar
  56. 153.
    As mentioned before, the RBA also published time series of its daily intervention operations with a time lag.Google Scholar
  57. 154.
    While this is not yet evident from the reserve developments, it is reflected in news reports (Appendix I).Google Scholar
  58. 155.
    According to George (1998), the Governor of the Bank of England, there have been two periods since the breakdown of Bretton Woods in which the Sterling exchange rate was targeted. The first period extends from 1987 to 1988 when the Bundesbank policy was shadowed in order to import stability. Sterling’s participation in the European Monetary System from October 1990 to September 1992 can be seen as the second attempt to target the exchange rate.Google Scholar
  59. 156.
    As Bernanke et al. (1999) note: “Arguably the United Kingdom’s point target allows some flexibility to policy by permitting some short-run deviations from target without causing the public to focus unduly on whether the official target range has been breached.” (p. 154).Google Scholar
  60. 157.
    In an interview Bank of England Governor Sir Edward George even admitted when speaking about UK inflation rates in the past years: “[…] I’m still surprised that it [inflation] has stayed as close to 2½% as it has: you can’t guarantee that at all.” (Financial Times, 7 May 2002, p. 9).Google Scholar
  61. 158.
    While inflation targeting was officially introduced in 1993, the policy of the Riksbank for 1993 and 1994 was focussed on generally limiting inflationary pressures resulting from the depreciation of the krona (of about 14%) after the fixed exchange rate had been abandoned. Thus, formally the inflation target of 1-3% started in 1995 (see Bernank et al., 1999: 174).Google Scholar
  62. 159.
    From 1933 on the krona was pegged to the UK Pound sterling and from 1939 on it was fixed to the US dollar. Sweden joined the Bretton Woods system in 1951 and after its breakdown in 1973 participated in the European Currency Snake. From 1977Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2004

Authors and Affiliations

  • Felix Hüfner
    • 1
  1. 1.Centre for European Economic Research (ZEW)MannheimGermany

Personalised recommendations