RDP 2003-04: Identifying the Efficacy of Central Bank Interventions: Evidence from Australia 2. Review of the Literature on Central Bank Intervention

This section briefly reviews the literature on central bank intervention, focusing on the simultaneous relationship between, and temporal behaviour of, exchange rate returns and intervention. For more general and extensive reviews see Sarno and Taylor (2001), Dominguez and Frankel (1993b) and Edison (1993).

Empirical studies, and statements by central banks, suggest that central banks intervene in foreign exchange markets to slow or correct excessive trends in the exchange rate, i.e. they ‘lean against the wind’, and to calm disorderly markets (for example Lewis (1995a) and Baillie and Osterberg (1997b)).^[2] The survey responses of central banks in Neely (2001) suggest that these factors continue to drive the decision to intervene. A recent study for Australia by Kim and Sheen (2002) has similar conclusions. Importantly, when central banks intervene they trade in blocks throughout the day. As Neely (2001) reports, their subsequent trades are conditional on the response of the exchange rate to their earlier trades.

Two main channels have been suggested through which sterilised intervention can affect the level of the exchange rate: the portfolio balance channel and the signalling channel. Intervention changes the exchange rate through the portfolio balance channel if government bonds are imperfect substitutes, and so the change in the reserve asset holdings of a central bank results in private investors revaluing their portfolios of domestic and foreign assets.^[3] Since interventions are small relative to the stock of outstanding bonds most authors, including Rogoff (1984), have expressed skepticism that intervention could have a large impact through the portfolio balance channel. Not surprisingly, many studies do not find evidence of this channel, and those that do, such as Dominguez and Frankel (1993a), Evans and Lyons (2001) and Ghosh (1992), suggest that it is weak.^[4]

The central bank may also be able to affect the exchange rate by using intervention to credibly disseminate private information about exchange rate fundamentals. This mechanism is described as the signalling channel. Mussa (1981) first suggested this channel with respect to intended changes in monetary policy. The impact of intervention through the signalling channel has often been found to be substantially stronger than through the portfolio balance channel (for example Dominguez and Frankel (1993b)). If the central bank uses intervention to indicate its intended path for interest rates, then this channel can only be an ongoing transmission mechanism if the central bank does follow up with appropriate changes in monetary policy. In this case intervention operating through the signalling channel is not an independent policy tool. Despite the evidence of a signalling channel, Fatum and Hutchison (1999) are unable to find an explicit link between intervention and future monetary policy, while Lewis (1995b) and Kaminsky and Lewis (1996) suggest it occasionally operates in the wrong direction.

Most central banks do not publicly announce their interventions.^[5] In light of the evidence for a signalling channel many authors have questioned this policy.^[6] However, interventions are not completely secret. Dominguez and Frankel (1993b) find that most, and particularly the largest, interventions are reported. Central banks can choose the method of intervention, ranging from direct trades with commercial banks, to indirect trades through brokers, to control the degree of secrecy of their actions.

If intervention operates through the signalling channel, then the exchange rate should react as soon as traders digest the information contained within the intervention. Goodhart and Hesse (1993), Peiers (1997), Dominguez (2003) and Chang and Taylor (1998) find that interventions that are intended to be visible are typically reported by news services within 10 minutes to 2 hours, by which time it is often no longer ‘news’ to traders. Any effect through the portfolio balance channel is also likely to be rapid as bond holders quickly respond to the change in the relative supplies in the highly liquid market for government securities. Indeed, Neely (2001) reports that the majority of central banks believe the full effects of intervention are reflected in the exchange rate within a matter of hours.

Despite the evidence of the rapid response of the exchange rate to intervention, studies using daily data have often abstracted from the endogeneity of intervention and exchange rate determination by only including lagged intervention (for example, Baillie and Osterberg (1997a) and Lewis (1995b)). While intervention may still have an effect on the days subsequent to the initial trades, omitting the contemporaneous intervention prevents measurement of the immediate impact and is likely to bias other coefficient estimates. Other studies that include contemporaneous intervention, such as Kaminsky and Lewis (1996) and Kim, Kortian and Sheen (2000), typically obtain an incorrectly signed contemporaneous coefficient, suggesting that purchases of the domestic currency cause it to depreciate. Seemingly what they capture is the policy function coefficient that represents central banks' tendency to ‘lean against the wind’. The insignificant and incorrectly signed coefficients in many previous studies indicate that an accurate estimation of the impact of intervention on the exchange rate must incorporate the contemporaneous effect, and account for the endogeneity between these variables.

Footnotes

Other reasons occasionally cited by central banks include: to target particular exchange rates, or to support other central banks. We do not explicitly consider the impact of intervention on conditional exchange rate volatility. See Rogers and Siklos (2001) for Australia, or more generally Dominguez (1998) and Bonser-Neal and Tanner (1996). [2]

The portfolio balance channel requires not only that bonds are imperfect substitutes, but also the failure of Ricardian equivalence. [3]

Dominguez and Frankel (1993c) and Evans and Lyons (2001) both assess the portfolio balance channel for unsterilised intervention and so provide upperbound estimates for the impact of sterilised intervention. [4]

One exception is the Swiss National Bank as noted by Fischer and Zurlinden (1999). [5]

Alternatively, Bhattacharya and Weller (1997) and Vitale (1999) develop models of intervention that rationalise secrecy about interventions. [6]