بحران های مالی و سیاست پولی: شواهدی از انگلیس

We analyse UK monetary policy using monthly data for 1992–2010. We have two main findings. First, the Taylor rule breaks down after 2007 as the estimated response to inflation falls markedly and becomes insignificant. Second, policy is best described as a weighted average of a “financial crisis” regime in which policy rates respond strongly to financial stress and a “no-crisis” Taylor rule regime. Our analysis provides a clear explanation for the deep cuts in policy rates beginning in late 2008 and highlights the dilemma faced by policymakers in 2010–11.

The global economic crisis that began in 2007 has presented a series of severe challenges to monetary policy. Deep and rapid reductions in output opened up an output gap of over 5% in many countries. Profound shocks to the financial system disrupted the transmission mechanism linking monetary policy to the real economy and created fears for the stability of the system. Objections have been raised to low and stable inflation being the main aim of monetary policy and dissatisfaction has been expressed with the New Keynesian and DSGE models that provided the theoretical underpinning for that aim. In this context, it would not be surprising if the behaviour of policymakers had changed during the crisis. This paper explores the interest rate setting behaviour of monetary policymakers in the UK using monthly data for the period 1992–2010. We have two main findings. First, although policymaking can be described using a simple Taylor rule in the period before the 2007 financial crisis, the Taylor rule then breaks down. The estimated response of the policy rate to inflation falls markedly and becomes insignificant, while the estimated response to the output gap is sharply reduced. Second, policy rates over the period from 1992 can best be described as a weighted average of two regimes, a “financial crisis” regime and a “no-crisis” regime, where the weights on these regimes reflect the probability of a financial crisis. The no-crisis regime is a conventional Taylor rule, whereas the financial crisis regime has a reduced response to the output gap, a strong response to measures of financial stress but no response to inflation. This model gives a plausible account of UK monetary policy. The no-crisis regime is dominant in 1992–2007, explaining the success of the Taylor rule over that period. But the onset of the major financial crisis in 2007 led to a marked change: the policy rate ceased to respond to inflation and the weight on the output gap fell as financial stress became the dominant influence on UK monetary policy. On this account, the sharp fall in interest rates beginning in late 2008 reflected difficulties in financial markets and the urgent need for policy measures to respond to the crisis. The rapid fall in the policy rate occurred despite inflation being above target and exceeding 3% for much of the crisis period; in our view, this explains the failure of the Taylor rule after 2007. This account also highlights the dilemma facing policymakers in early 2010. The no-crisis regime increasingly pointed to higher policy rates through early 2010 to the end of our sample in July, driven by persistently high rates of inflation. By contrast, our measure of financial stress pointed to a continuation of the financial crisis and so argued for a continuation of the policy of exceptionally low interest rates. The paper is structured as follows. In Section 2, we estimate a simple Taylor rule representation of monetary policy. We show that estimates of this model using a sample that ends in 2007 conform to expectations with a response to inflation in excess of unity and a strong response to the output gap. Estimates that use the full sample that ends in 2010 are very different. Although the crisis period represents less than 20% of the sample, the estimate on inflation becomes insignificant and the point estimate is negative. The response to output remains significant but is more than halved. We detect a structural break around the time the crisis began. In Section 3, we investigate whether the addition of measures of financial stress to a Taylor rule gives more satisfactory estimates. Some writers, most prominently Curdia and Woodford (2009), have suggested that including the determinants of credit spreads in a policy rule may be optimal in the presence of financial frictions. We use two measures of financial stress as determinants of credit spreads. Our first measure is a composite index of financial stress compiled by the IMF, providing a broad spectrum measure of stress across money, foreign exchange and equity markets in the UK. Given that the recent crisis originated in the US, our second measure is the US Financial Stress Index provided by the Federal Reserve Bank of Kansas City.1 We find that inclusion of these measures in a Taylor rule does not give satisfactory estimates. We continue to detect a structural break and still observe a marked reduction in the estimated response to inflation after 2007. From this we conclude that no model with a constant response of interest rates to inflation, the output gap and financial stress can explain UK monetary policy over 1992–2010. This evidence suggests that models of monetary policy must allow for changes in the behaviour of policymakers. Accordingly in Section 4, we develop a model in which the policy rate is a weighted average of two alternative policy regimes and where the weight of these regimes reflects the probability of a financial crisis. We find that this model provides a satisfactory explanation of UK monetary policy and that the estimates are econometrically superior to those of the constant parameter policy rules considered above. We find a strong response to inflation in the no-crisis regime but no response in the financial crisis regime. Our estimates suggest a strong response to the output gap in the no-crisis regime and a much weaker response in a crisis. We find a strong response to measures of financial stress in the crisis regime but none in the no-crisis regime. Section 5 concludes the paper.

This paper explores the interest rate setting behaviour of monetary policymakers in the UK over the period 1992–2010. We have two main findings. First, although interest rate setting behaviour can be described by a simple Taylor rule in the period before 2007, the Taylor rule then breaks down. Second, policy rates over the period from 1992 are best captured using as a nonlinear model in which policy switches between a no-crisis regime that resembles a simple Taylor rule and a crisis-regime in which there is not response to inflation but instead a powerful response to financial stress. Our estimates raise the question of whether UK policymakers “abandoned” inflation targeting during the financial crisis. Although our estimates are suggestive, our model is not able to address that issue. Changes in the coefficients of a monetary policy rule can be due to a change in policymakers’ preferences or from a change in macroeconomic structure described by captured by aggregate demand and supply relationships. Although our results suggest there was a change in the monetary policy rule during the financial crisis, they tell us nothing about the causes of that change. Assessment of this requires joint estimation and analysis of aggregate demand and supply relationships alongside a monetary policy rule. That is a topic for future research.