Why monetary policy matters: New UK narrative evidence

James Cloyne, Patrick Hürtgen 15 May 2014

a

A

In recent decades, central banks around the world have predominantly used interest rates as their main monetary policy instrument. And while the zero lower bound has necessitated a range of unconventional monetary policies, many central banks clearly still intend to use interest rates as their preferred tool as their economies recover. A range of empirical estimates have emerged from the academic literature over several decades putting the effect on prices and output of a one percentage point increase in interest rates between 0.5% and 1%. A notable exception — the so-called narrative method pioneered by Romer and Romer (2004) — has found considerably larger effects and drawn much attention in both policy and academic circles. To our knowledge, no other studies have attempted to corroborate these findings by employing a similar methodology elsewhere. In the current environment, it is a particular pertinent time to provide new estimates for the effect of interest rate changes on the macroeconomy, and this column describes our new findings.

Identifying the effects of changes in monetary policy is challenging and requires tackling at least three issues.

  • First, monetary-policy instruments, interest rates, and other macroeconomic variables are determined simultaneously as policymakers respond to macroeconomic fluctuations and intend their decisions to affect the economy.
  • Second, policymakers are likely to react to expected future economic conditions as well as current and past information.
  • Third, policymakers base their decisions on real-time data, not ex-post data often used in the empirical literature.

While common methods often address the first issue above, the other two points are typically overlooked. In Cloyne and Huertgen (2014), we tackle all three issues head-on for the United Kingdom, applying the identification strategy of Romer and Romer. We provide new narrative-based estimates for the UK and also contribute to the wider debate between competing approaches.

Identifying the effects of policy using real time and forecast data

In applying this method, we first need to disentangle the policymakers’ intended changes in the policy target rate from cyclical movements in short-term market interest rates. A particular advantage of the UK monetary framework is that the Bank of England’s policy rate — Bank Rate — is the intended policy target rate and is readily available from the Bank’s website.

In a second step, we purge the target series of discretionary policy changes that were responding to fluctuations in macroeconomic conditions that are observed or expected by policymakers. A key advantage of this methodology is that we control for real-time data and forecasts actually available at the time of, and prior to, the policy decision and thus we tackle the three challenges mentioned above. For example, forecasts can be seen as excellent summary statistics for all the information available to policymakers, and they allow us to control for policy reactions designed to offset future business cycle movements. Our new shock series is the residual from this exercise. We use historical sources to construct an extensive new dataset of historical Bank of England forecasts, private sector forecasts, and real-time data. Furthermore, we show that our use of forecasts solves some key problems affecting other approaches, as we’ll discuss in more detail below.

New estimates for the United Kingdom

We first use our new series of policy changes in a simple auto-regressive distributed lag framework. A one percentage point increase in the interest rate leads to a peak decline in output of 2.3%, and a drop in inflation of 1.5 percentage points, as shown in Figure 1. Simulations using this method model a 1% shock to the change in the policy target, implying a permanent shock to the policy target rate. These results are similar to the magnitudes obtained using a quarterly version of the data with GDP.

Figure 1 Results from a simple autoregressive distributed lag model

To compare our results to the Romers’ estimates for the United States, we extend their shock series up to 2007 (to exactly compare the same sample period, i.e. from 1975 to 2007 for both countries). Figure 2 shows that our findings are remarkably similar to those of the US using the same methodology. In both countries, the peak decline of industrial production is reached after around two years. Interestingly, the dynamics and the magnitude of the response of consumer prices is almost exactly the same, being relatively small in the first two years but then falling significantly.

Therefore, in contrast to more traditional ‘recursively’ identified vector autoregressions (VARs) used in the literature, we find evidence for a decline in prices and inflation rather than a counterintuitive increase. The latter is often referred to as the so-called ‘price puzzle’ – an increase in prices following a monetary tightening which has generated much debate in the literature. Excluding forecasts from our analysis reintroduces this puzzle – clearly highlighting the importance of their inclusion.

Figure 2 The similarity of UK and US estimates

How big are the macroeconomic effects?

To be more comparable with this wider literature, we use our new shock measure in a vector autoregression. Romer and Romer (2004) and Coibion (2012) have dubbed this the ‘hybrid VAR’ approach. Figure 3 presents the main results of this column.

  • In response to a one percentage point increase in the interest rate, inflation falls by up to 1 percentage point. Industrial production exhibits a peak decline of -0.6%.

Again, in a quarterly specification we show that the effect is very similar using GDP. Our new estimates are, therefore, quite comparable with other VAR studies (e.g. Christiano, Eichenbaum, and Evans 1999, Uhlig 2005, Bernanke, Boivin, and Eliasz 2005, Barakchian and Crowe 2013), but somewhat smaller than the results in Coibion (2012).

So, are the effects of policy changes large or more modest? How could we compare the previous two approaches? A key difference between these methods is the persistence of the policy experiment being simulated. Once we consider a comparable policy exercise, the two sets of results, at least for the UK, become very similar as illustrated by the blue line on Figure 3. And, reassuringly, the implied effect on the actual policy rate is also similar across both methods in Figure 3. We are, therefore, able to square our new results across different methods, and show they are in line with the typical magnitudes found elsewhere in the literature.

Figure 3 Reconciling different approaches

Conclusion and outlook

This column presents new estimates of the effects of monetary policy for the UK by applying the identification methodology of Romer and Romer (2004). While numerous studies employ more conventional VAR methodologies, to our knowledge, there has been no other application of this narrative strategy to corroborate the large effects found for the US economy. Moreover, there is comparatively little evidence of the macroeconomic effects of monetary policy for the UK, and our new research (described in Cloyne and Huertgen 2014) fills this gap.

Our UK findings are in line with US estimates by Romer and Romer. But when we consider a comparable policy experiment, as in the wider UK literature, we find that interest rate changes that are unwound over time lower inflation by 1 percentage point and output by around 0.6%.

In addition to new estimates, our research also provides a novel forecast and real time dataset which is crucial for our results. These data provide a fruitful resource for future research.

This column, therefore, provides fresh narrative evidence on the effects of monetary policy and aims to help inform the policy discussion as economies move away from the zero lower bound.

Authors’ note: The views expressed are the authors’ and not necessarily those of the Bank of England.

References

Barakchian, S. Mahdi and Christopher Crowe (2013), “Monetary policy matters: Evidence from new shocks data,” Journal of Monetary Economics, 60, 950–966.

Christiano, Lawrence, Martin Eichenbaum, and Charles Evans (1999), “Monetary Policy Shocks: What Have We Learned, and To What End?”, in John B Taylor and Michael Woodford (eds.), Handbook of Monetary Economics, Elsevier Science, pp. 65-148.

Bernanke, Ben S, Jean Boivin, and Piotr Eliasz (2005), “Measuring the Effects of Monetary Policy: A Factor-Augmented Vector Autoregressive (FAVAR) Approach”, Quarterly Journal of Economics, 120(1):387-422.

Cloyne, James and Patrick Huertgen (2014), “The macroeconomics effects of monetary policy: a new measure for the United Kingdom”, Bank of England Working Paper No. 493.

Coibion, Olivier (2012), “Are the Effects of Monetary Policy Shocks Big or Small?” American Economic Journal: Macroeconomics, 4, 1–32.

Romer, Christina D. and David H. Romer (2004), “A New Measure of Monetary Shocks: Derivation and Implications”, American Economic Review, 94:1055-1084.

Uhlig, Harald (2005), “What are the effects of monetary policy on output? Results from an agnostic identification procedure,” Journal of Monetary Economics, 52, 381–419.

a

A

Topics:  Monetary policy

Tags:  monetary policy, UK, interest rates changes

Senior economist, Monetary Assessment and Strategy Division, Bank of England

Postdoctoral researcher, Institute for Macroeconomics and Econometrics, University of Bonn