Monetary policy rules in theory and in practice: evidence from the UK and the US

Abstract

Given the large amount of interaction between research on monetary policy and its practice, this article examines whether some simple monetary policy rules that have been proposed in the academic literature, part of which has originated from within central banks, provide a reasonable characterization of actual policy in the United Kingdom and the United States. The article finds the simple rule that describes best the actual US monetary policy is a speed limit rule with dynamics, whilst for the UK it is a forward-looking rule. The simpler dynamics in the UK's monetary policy rule are reflective of the lower persistence of inflation as a result of its policy of inflation targeting.

Acknowledgements

I would like to thank an anonymous referee for helpful comments.

Notes

¹Given the assumption of the natural rate of unemployment (and interest) the real sector would inevitably always return to its natural rate, so that attempts at economic stabilization could at most only help to bring this about at a faster rate.

²Also called New Neoclassical Synthesis models (Goodfriend and King, 1997)

³On this, see Carare and Tchaidze (2005) and Minford et al. (2002).

⁴See for example, Kobayashi (2005).

⁵Nikolov (2002).

⁶This lack of specificity implies that it follows neither an instrument rule nor a target rule.

⁷Carare and Tchaidze (2005) discuss additional dangers in the use of Taylor rules, especially when used for policy recommendations.

⁸Nikolov (2002) states that Taylor rules are used at the Bank of England as an indicator of the stance of current policy; Yellen suggested that following a Taylor-type rule would represent good policy, which is remarkable when the alternative policy being considered was one of strict (over the medium term) inflation targeting, as argued by Broaddus (Federal Reserve Board 1995, pp. 39−44).

⁹ Indeed, given that much research on monetary policy is conducted within central banks, the fact that some of the proposed rules analysed in this article were published near the end of the sample period does not preclude their relevance.

¹⁰See Woodford (2003, pp. 252−61).

¹¹The general condition can be found in Woodford (2003, p. 255).

¹²This would enable one to consider an operational monetary policy rule, as in McCallum and Nelson (1999).

¹³An insightful analysis of such preferences in a monetary policy context can be found in Nobay and Peel (2003).

¹⁴The coefficient on the output gap (incorrectly defined as hp detrended output) was also high, at 0.93.

¹⁵All data were taken from the Federal Reserve Bank of St Louis.

¹⁶The backward-looking rule is estimated by least squares, with Newey−West SEs in parentheses.

¹⁷Obviously, more complicated rules are likely provide a more accurate description of actual interest rate behaviour, but this article focuses on only a selection of simple rules with theoretical foundations.

¹⁸Adam et al. (2005) reported a change in regime pre- and post-independence of the Bank of England; however, Lord George, in an interview for the Financial Times on 2 May 2007, has stated that it was the introduction of inflation targeting and the general economic consensus in the UK that mattered most.

¹⁹Although this is common practice, as mentioned earlier one should be aware that this is not the equivalent measure from a theoretical output gap measure.

²⁰In the estimations, inflation is then RPI minus the official inflation target. Modifying this to the gap using first RPIX and then CPI inflation, the Bank's official target, has not effect on the results.

²¹However, it is not significantly above unity.

²²In effect, Equation 6 can be seen as a re-parameterisation of an autoregressive distributed lag (ARDL) model embodying an error correction model. Hence it can be interpreted as a Taylor rule but with richer dynamics.

²³Their estimation was conducted by least squares on the grounds that given the lags in the monetary transmission mechanism, reverse causation from interest rates to inflation and output was unlikely.

²⁴An important methodological issue concerns the stationarity properties of the interest rate. For the United States, both the ADF and Phillips−Perron test reject stationarity at the 1% level, but not at the 5% level. For the United Kingdom, although the ADF test rejects stationarity at the 1% level, the Phillips−Perron gives the same results as for the United States. Hence the results are not conclusive, but are indicative of substantial persistence in the series. Moreover, there are strong theoretical reasons to believe that the interest rate is a stationary series, especially in the case of an inflation targeting regime.

²⁵This is measured as the civilian unemployment rate in the present article.

²⁶However, as Walsh also points out, this rule performs substantially worse than the optimal rule under more general misspecification.

²⁷Previous empirical support for a speed limit rule was found by Peel et al. (2004).

²⁸A potential explanation is that the nonlinearity that best describes the data may be the one used by Taylor and Davradakis (2006) – a threshold model – and not Equation 5.

²⁹See Benati (2007).