Central bankers should not be forward-looking in times of crisis

Paul De Grauwe, Yuemei Ji 05 May 2020

a

A

The unpredictability of the Covid-19 pandemic has created extraordinary uncertainty about the future course of the economy. Forecasts made today about future output and inflation risk are completely off the mark tomorrow. This uncertainty raises the following questions: Should central banks be forward-looking when deciding about monetary policies? Is it not better that they just take into account what they observe today to decide about the monetary policy actions? Major central banks currently appear to be following the latter approach. Under extreme uncertainty it is probably better not to rely on one’s forecasts about the future.

Can this cautionary principle be generalised? That is what we tried to do in our latest CEPR Discussion Paper (De Grauwe and Ji 2020). We use a behavioural macroeconomic model that produces endogenous business cycle movements driven by animal spirits, i.e. self-fulfilling changes in market sentiments (see De Grauwe 2012, De Grauwe and Ji 2019, Hommes and Lustenhouwer 2019).

An important characteristic of this model is that it generates ‘tranquil’ and ‘turbulent’ periods. Most of the time the economy is in tranquil periods. During turbulent periods, the model produces extreme behaviour of output and inflation, associated with the occurrence of extreme values (fat tails) in the distribution of these variables. These occur regularly but in an unpredictable fashion. We call these turbulent periods, ‘crisis periods’.

We exploit this aspect of our model to generate tranquil and crisis periods and to analyse how two different Taylor rules perform during these periods. In the first Taylor rule regime the central banks look only at current values of output gap and inflation. In the second Taylor rule regime the central bank uses forecasts of the output gap and inflation.1

Our first finding is that in tranquil periods when market sentiments (animal spirits) are neutral, forward-looking and current-looking Taylor rules produce similar results, in terms of their capacity to reduce the volatility of output and inflation. Empirical studies confirm that in normal times it is difficult to find that one of the two rules outperforms the other (see Taylor and Williams 2010).

Our second finding, however, is that when the economy is in crisis periods, a central bank that bases its interest rate decisions on forecasted values of output and inflation introduces more variability in these variables than a policy focusing on currently observed values. We interpret this result as follows. In crisis periods dominated by booms and busts in economic activity (fat tails and extreme values in output and animal spirits), forecast errors made both by private agents and by the central bank become very high. As a result, a forward-looking central bank will make many policy moves that turn out to be wrong. Put differently, the forward-looking central bank will make many policy mistakes that have to be reversed, thereby exacerbating the volatility of the output gap and inflation. Thus, when the forward-looking Taylor rule is used, the quality of policymaking declines during crisis times, leading to greater variability of output and inflation.

We checked for this interpretation by calculating the forecast errors made by agents (and by the central bank) under the two Taylor rules during tranquil and crisis periods. During tranquil periods, market sentiments (animal spirits) are neutral, whereas during crisis periods they take on extreme values. We plot the squared forecast errors of the output gap against the animal spirits in Figure 1. We find that when animal spirits are close to zero (tranquil times) the forecast errors are similar in magnitude in the two Taylor rule regimes. As animal spirits increase (in absolute values) the forecast errors increase and more so under the forward-looking Taylor rule. 

This leads to the following insight. Extreme moods of optimism and pessimism are not the result of rational expectation but of the fact that increasing numbers of agents tend to extrapolate what they observe today, resulting in a boom in the optimistic case or a decline in the pessimistic case. It is then better for the central bank to use currently observed output and inflation to set the interest rate, rather than to try to outwit these agents by reacting to the forecasts of output and inflation. Given the extreme volatility of these variables when animal spirits are intense, the forward-looking central bank will make many policy errors that have to be corrected afterwards.

Figure 1 Squared forecast errors output gap and animal spirits

Our third finding is that several factors affect the relative performance of forward-looking and current-looking Taylor rules. In particular, larger exogenous shocks tend to produce stronger and more uncertain effects on output and inflation. They also lead to more frequent crisis periods, making a forward-looking policy less attractive.

We illustrate this feature in Figures 2 and 3. We simulated our model for increasing standard deviations of demand and supply shocks and computed the corresponding standard deviations of output gap and inflation under the two Taylor rule regimes. We find that when the demand and supply shocks are relatively small, both the forward-looking and current-looking Taylor rules produce similar results. When the standard deviations of the shocks exceed a threshold value (approximately 0.5) we observe that the forward-looking Taylor rule produces stronger increases in the volatilities of output gap and inflation than the current Taylor rule.

Figure 2 Standard deviation output and shocks

Figure 3 Standard deviation inflation and shocks

In a last experiment, we derived the policy choices (trade-offs) central banks face with the use of the forward-looking and current-looking Taylor rules. These policy choices arise as a result of changing the intensity with which central banks stabilise the output gap. This intensity is measured by the output stabilisation coefficient in the Taylor rule. We computed the standard deviations of inflation and output gap for increasing values of this coefficient. We obtain a highly non-linear relationship shown in Figure 4. Let us first concentrate on the downward sloping part of the curve. This is obtained for values of the output stabilisation coefficient in the Taylor rule exceeding a threshold value of approximately 0.5. For these high values of the output stabilisation coefficient the central bank faces a conventional, negatively sloped trade-off between inflation and output volatility. As the intensity of output stabilisation increases, we move up along the negatively sloped trade-off. Thus. in this region, more output stabilisation comes at the cost of more inflation volatility. When the economy is on the negatively sloped segment, tranquil periods are the rule and there are very few crisis periods. We observe that when we are on this segment of the trade-off, both Taylor rules produce similar results in terms of output and inflation volatility.

Let us now concentrate on the upward sloping segment of the policy choices. This is obtained when the output stabilisation coefficient is below the threshold value of 0.5. Then, crises occur more frequently. For example, points A and A’ on the uppermost curves (corresponding to the forward-looking Taylor rule and the current-looking Taylor rule, respectively) are the situation when the central bank does not stabilise output (i.e. the output coefficient is close to 0). When this coefficient increases (i.e. the central bank increases its output stabilisation effort) we move down along these curves. This move leads to a ‘win-win’ scenario: by increasing output stabilisation, the central bank reduces the fat tails in the distribution of the output gap and as a result reduces both the volatility of output and inflation. In this scenario, more output stabilisation can be achieved without cost in terms of more inflation volatility. 

We can now contrast the results obtained in the two Taylor rule regimes during crisis periods. We observe from Figure 4 that point A is located above and to the right of A’. This means that in the forward-looking Taylor regime the volatility of output and inflation is higher when the central bank does not do much output stabilisation. As explained earlier, this creates an environment of very high forecast errors. The central bank is also subjected to these high forecast errors. As a result, when the central bank uses a rule based on forecasting output and inflation it will make large policy errors. This is less the case when the central bank uses a rule based on currently observed values of output and inflation. This also means that in crisis scenarios the need for more active output stabilisation is stronger in the forward-looking Taylor regime than in the current-looking Taylor regime. 

Figure 4 Output-inflation trade-off

Our results generalise an intuition we have when the economy experiences great and uncertain volatility. The recent COVID-19 crisis is an example. This crisis is characterised by historically large and uncertain disturbances. Under these conditions, no rational central bank would base its monetary policy decisions on a forecast of future output and inflation. The future is too uncertain to do this. When uncertainty is extreme, prudent central banks will be guided by what they observe, and not by unreliable forecasts. There is empirical evidence that central banks actually are often not forward-looking (Taylor and Williams 2010). Our model provides the theoretical justification for this.

References

Belke, A and J Klose (2011), “Does the ECB rely on a Taylor rule during the financial crisis? Comparing ex-post and real time data with real time forecasts”, Economic analysis and policy, 41(2):147-171.

Blattner, T S and E Margaritov (2010), “Towards a robust monetary policy rule for the euro area”, ECB Working Paper No. 1210. Available at SSRN.

De Grauwe, P (2012), Lectures on Behavioral Macroeconomics, Princeton University Press.

De Grauwe, P and Y Ji (2019), Behavioural Macroeconomics. Theory and Policy, Oxford University Press.

De Grauwe, P and Y Ji (2020), “Should Central Banks Be Forward-Looking”, CEPR Discussion Paper 14540.

Gorter, J, J Jacobs and J De Haan (2008), “Taylor rules for the ECB using expectations data”, Scandinavian Journal of Economics 110(3): 473-488.

Hommes, C and J Lustenhouwer (2019), “Infation targeting and liquidity traps under endogenous credibility”, Journal of Monetary Economics 107: 48-62.

Orphanides, A (2001), “Monetary policy rules based on real-time data”, American Economic Review 91(4): 964-985.

Orphanides, A and J C Williams (2007), “Inflation targeting under imperfect knowledge”, In F Mishkin and K Schmidt-Hebbel (Eds.), Monetary policy under inflation targeting, Santiago, Chile: Central Bank of Chile.

Taylor, J B and J C Williams (2010), “Simple and robust rules for monetary policy”, In Handbook of monetary economics Volume 3: 829-859, Elsevier.

Endnotes

1 There is a large emprical literature on the use of Taylor rules, both forward and current-looking ones. For an overview of the literature see Taylor and Williams (2010). See also Orphanides (2001), Gorter et al (2008), Belke and Klose (2011), and Blattner and Margaritov (2010).

a

A

Topics:  Covid-19 Global crisis Monetary policy

Tags:  central bank, monetary policy, self-fulfilling cycle, animal spirit, Taylor rule

John Paulson Chair in European Political Economy, London School of Economics, and former member of the Belgian parliament.

Lecturer, University College London

Events

CEPR Policy Research