Monetary policy and interest rates under inflation targeting in Australia and New Zealand

Abstract

One advantage cited for formal inflation targeting is that by anchoring inflationary expectations this policy framework would aid in the pricing of long-term securities. Long-term interest rates would become less sensitive to temporary shocks to the economy including policy-induced changes in short-term interest rates. This paper examines the experience in this regard of New Zealand and Australia, two countries that have been inflation targeters for many years. Our results are consistent with inflationary expectations having become more firmly anchored after the move to inflation targeting in each country. There is no evidence that the credibility of the inflation-targeting regime in either country weakened during the recent world financial crisis.

Keywords:

• inflation targeting
• monetary policy
• long-term interest rate

Acknowledgements

The authors are grateful to two anonymous referees for numerous helpful comments.

Notes

1. Rogoff, Kenneth. (2005, April 23). A case for inflation transparency. Financial Times, p. 13.

2. Bernanke (2004, p. 166). Bernanke cites Gurkaynak, Sack, and Swanson (2005) for empirical evidence that under current Federal Reserve procedures, ‘private agents’ views of long-run inflation are not strongly anchored’ (p. 425). English, Lopez-Salido, and Tetlow (2013) describe the Federal Reserve's subsequent move toward ‘flexible inflation targeting’.

3. See also Bernanke et al. (1999, p. 223). There has been convergence in the interpretation of inflation targeting in New Zealand and Australia. By 2002, the Reserve Bank of New Zealand's Policy Targets Agreement specified the inflation target range as ‘1% to 3% over the medium term’. We will return to this question of convergence in procedures at a later point.

4. Data descriptions and sources are given in Table 6.

5. Here we date the beginning of the crisis with the Lehmann Brothers bankruptcy and AIG bailout in September of 2008. There was certainly stress in financial markets prior to this.

6. On the operation of Reserve Bank policy in the pre-2000 period of inflation targeting, see Guthrie and Wright (2000). Guender and Wu (2010) provide evidence that the change in operating procedure in 1999 increased the predictability and decreased the volatility of New Zealand interest rates at maturities out to 180 days.

7. We have also broken the sample period at the points where IT countries have started targeting inflation. For that break, New Zealand and Australia show declines in the variance of the 10-year interest rate that exceed other inflation targeters in our sample.

8. These macroeconomic variables are also subject to innovations caused by real shocks. The effects of these are not clearly affected by the change in policy regime.

9. The analysis does not extend to possible changes in the underlying error processes for the shocks and therefore to the sources of risk (or term) premia in long-term rates. Wright (2011) constructs a data set and examines term premia for 10 countries, including Australia and New Zealand, for the post-1990 period. He finds term premia to have declined globally and attributes the decline in part to ‘declining inflationary uncertainty amid substantial changes in monetary policy frameworks of several central banks’ (p. 1515). Bauer, Rudebusch and Wu (2014) reach somewhat different conclusions using the same data though they still find term premia declining, Wright (2014) replies.

10. The exchange rate and commodity price variables were included to control for shocks that affect interest rates in addition to monetary policy. The selection of commodity prices rather than consumer prices is due to the unavailability of monthly data for the consumer price index for both countries for our sample period. Commodity prices are highly correlated with consumer prices. The correlation coefficients (quarterly log-levels) for Australia and New Zealand are 0.98 and 0.66, respectively, for the whole sample.

11. To save space, we do not include tables of unit root tests.

12. The test was originally proposed in Chow (1960). For the application to VARs, see Canova (2007).

13. The VAR specifications were also estimated with a time trend included. The pattern of impulse responses is robust to this change.

14. Confidence bands for the estimated impulse responses are calculated using a bootstrap method of 2000 draws to compute the standard errors.

15. The level of significance is 5% throughout the analysis of impulse response functions.

16. The effect on the commodity price index for a small country such as New Zealand would be expected to be only via the effect on the exchange rate. The positive effect is consistent with the negative effect on the exchange rate. Still both the effects are counter intuitive. Both reflect the ‘exchange rate puzzle’ as discussed in Kim and Roubini (2000) and Chen and Rogoff (2002).

17. Data for the 2-year interest rate for Australia do not go far enough back to estimate the VAR for the pre-inflation targeting period. Thus, the 1×1 cell in Figure 4 is blank.

18. A specific comparison is to results from VAR analyses for the United States summarized in Berument and Froyen (2009, Table 1). An update of these estimates to provide evidence on changing responses of long-term rates to change in the federal funds rate is precluded by the fact that since December 2008 the target federal fund rate has been at the effective lower bound of 0.25%.

19. An interbank rate futures market does exist for Australia. The data does not extend far enough back for our purposes. Moreover, that approach would complicate comparisons to the New Zealand experience. Smales (2012) investigates the response of interest rate futures to Reserve Bank of Australia target rate announcements for the period 2004–2010.

20. See Guender and Rimer (2008) for a detailed description of the Reserve Bank's procedures.

21. Note that the policy variable here is the official cash rate rather than the actual overnight cash rate. Discrete policy changes are changes in that rate. Definitions and data sources for all interest rates are given in Table 6.

22. These sample periods are chosen for purposes of comparison with New Zealand. Estimates that begin with the adoption of inflation targeting in October 1994 have quite similar implications to those in Table 5.

23. In the case of Australia, these effects also lose statistical significance at longer terms to maturity.

24. The Federal Reserve announces changes in the target federal funds rate in the early afternoon before the bond markets close. Therefore, here the changes in the market rates are from day (t – 1) to (t), when t is the date of the target rate change. The sample period for the United States in Table 5 ends in August 2008 for purposes of comparison. If the sample is extended to December of 2008 when the federal funds rate hits the effective zero bound, the responses of longer term interest rates are higher. They are still lower than in Australia or New Zealand for 2-, 5- and 10-year rates.

25. Swiston (2007); Table 1) studies the United States for the period March 2000–June 2006 using the method of Kuttner (2001) which relies on data from the federal funds futures market to measure unanticipated changes in the federal funds target rate. His estimates of the effect on longer term market interest rates of surprise changes in the policy rate are quite similar to those in Table 5 (Panel C), especially for longer maturities.