jonungfig2.jpg
VoxEU Column Monetary Policy

Inflation targets and the benefits of an explicit tolerance band

Inflation-targeting central banks commonly fail to hit their official inflation targets, so targets are combined with a tolerance band which is either implicit or explicit. Taking the Swedish Riksbank as an example, this column argues that adopting an explicit tolerance band would better communicate to the public the central bank’s lack of full control over the rate of inflation and thus foster public confidence in monetary policy, and it would also increase the central bank’s ability to stabilise the economy. The width of the band can be derived from the historical inflation outcome. 

Since the early 1990s, a large number of central banks have adopted inflation targeting, with the operational goal for monetary policy commonly set as a single number. The official target for inflation is often combined with an explicit tolerance band within which inflation is allowed to fluctuate. In a recent survey, Hammond (2012) finds that 22 of 29 inflation-targeting central banks have such an explicit band. Notable exceptions are the ECB, the US Federal Reserve, the Bank of England, and the Swedish Riksbank.

Should the band be an implicit or an explicit one? Here, we provide an answer based on the unique Swedish experience. The Riksbank announced a tolerance band in 1993 and then abolished it in 2010. By comparing the first period with an explicit band with the second period without a band, we are able to draw conclusions concerning the pros and cons of an explicit tolerance band relevant for inflation targeting in other countries.

The significance of tolerance bands is by no means limited to the case of inflation targets. Regardless of whether the policy goal is a fixed (pegged) exchange rate, a monetary target, or an inflation target, tolerance intervals are important to indicate how large the deviations are from the point target that the central bank is willing to accept.1

Arguments for an explicit band

There are several reasons for adopting an explicit band (for a detailed discussion of the pros and cons of an explicit band, see Andersson and Jonung 2017).

First is the flexibility concerning the choice of price index. Every central bank is faced with a choice among several price indices when quantifying the inflation target. There can be substantial differences between different indices such as the GDP deflator and the consumer price index (CPI). Because no price index gives an exact measure of the ‘true’ overall rate of inflation, the central bank needs flexibility to respond to inflationary or deflationary pressures in the economy that may not be adequately captured by the adopted price index. A tolerance interval gives the central bank room to respond to price movements, even when these do not show up in the adopted price index, without abandoning the inflation target.

Second, measurement errors occur in the price index. All price indices are associated with measurement errors. With an explicit tolerance band, central banks can ignore minor changes in the rate of inflation that fall within the error margin, and instead focus on major developments that causes inflation to fall outside the band.

Third, model uncertainty is inevitable. Central banks use econometric models for forecasting and framing monetary policy. Regardless of the model adopted, no econometric model is an exact map of the world. Over time there have been periods of systematic forecast errors in central bank models. An explicit tolerance interval is a method to admit ex ante that even if the econometric model forecasts an inflation rate equal to the target, in real life inflation is likely to deviate from the target ex post due to limitations of the model.  

Fourth, similarly forecasting uncertainty arises from unforeseen shocks. Unexpected shocks impact inflation during the forecasting horizon, causing actual inflation to deviate from the forecasted path. Failing forecasts due to unexpected disturbances constitute a strong argument for an explicit tolerance band.

Additionally, uncertainty in the measurement of inflation expectations occurs. Data for inflation expectations play an important role in the preparation of monetary policy decisions. Inflation expectations are usually measured by interviews of representative samples of households and firms. Their numerical replies are associated with a high degree of uncertainty (Jonung 1987). This uncertainty represents an additional argument for an explicit tolerance band.

Finally, facilitating central bank communication concerning the precision of monetary policy is a consideration.

Experience shows that monetary policy will never hit the target all the time. A tolerance band is a simple method to inform the public of this lack of precision. The experience of the Riksbank clearly demonstrates the importance of communicating the limitations of monetary policy. The Riksbank abolished in 2010 its explicit tolerance band of +/-1% and replaced it with an implicit tolerance band that was stated as “wider”. However, the abolishment, in combination with the fact that the Riksbank did not clearly announce that it remained obliged to maintain the overriding objective of price stability, turned into a communication disaster for the Riksbank.

Without the explicit tolerance band, the media and critics began to focus solely on the specific number of 2%. Any deviation from this exact number was interpreted as a monetary policy failure. The debate became obsessed with the number of 2.0, despite the fact that the Riksbank had never announced that the rate of inflation should be exactly at 2.0%. The lack of an explicit band undermined the Riksbank’s communication strategy and thus its credibility.

Arguments against a tolerance band

The main argument against an explicit tolerance interval is that such a band may introduce uncertainty about the willingness of the central bank to actually stick to the target. Thus, inflationary expectations will be less anchored and inflation uncertainty will be higher. This line of reasoning does not take into account the fact that central banks without an explicit band still have an implicit, and thus unidentified, one, nor that inflation will at times deviate from the target no matter how hard the central bank is trying to hit the target.

Figure 1 illustrates one-year-ahead inflation expectations for households, firms, and labour organisations in Sweden. The white area represents the period when the Riksbank had an explicit tolerance band and the grey area the period with an implicit tolerance band. Judging from Figure 1 there are no signs that inflation expectations are less volatile or more anchored at 2% after 2010 than before 2010.

Figure 1. Inflation expectations (%) one year ahead for households, firms and labour market organisations, 1995Q1 to 2016Q4

The explicit tolerance interval that was in place between 1995 and 2010 apparently did not increase uncertainty compared to the period after 2010. In fact, there is some evidence that inflation expectations deviated more from the target after the explicit band was abolished. We cannot rule out the possibility that the removal of the explicit band actually caused inflation expectations to deviate more from the target than before, due to the strong critique of the Riksbank that emerged after the explicit band had been abolished.

We conclude from this episode that the argument that an explicit band increases inflation uncertainty – given that the band is not too large and is well communicated – does not carry much empirical weight in the Swedish context.

Selecting the proper size of the tolerance band

In our opinion, there are stronger reasons for an explicit tolerance band than for an implicit and consequently unknown band. A key question then emerges: how large should the band be?

We find it hard to believe that an optimal band size can be derived from a theoretical model. The choice of numerical value for the band is influenced by so many factors – as described above in our list of pros and cons – that it is difficult to capture them in a single model. In addition, a band derived from a theoretical model is likely to vary over time as new data are fed into the model. There is a clear advantage for the communication strategy of a central bank to have a constant size of the band, not one that changes over time.

Instead, we rely on our reading of the historical record. Thus, we focus on the actual behaviour of the rate of inflation under inflation targeting. The past performance of the Riksbank is shown in Figure 2. CPI-inflation has varied between -2 and +4% for the period 1995-2016. Inflation according to the Riksbank’s estimate of core inflation, CPIF, has varied between 0 and 3.5%.2 The old band introduced in 1993 of +/- 1% is clearly too narrow for the Swedish economy. This was the key reason behind the Riksbank’s decision to abolish it.

History suggests that a band of +/-2% would be more appropriate. For a small open economy like Sweden with an export sector amounting to half of GDP that is thus strongly exposed to external shocks, this size of the band makes sense, judging from the inflation variability of the past. Similarly, Haldane (1997) suggests a band of 0-4% around a single target of 2% inflation for the Bank of Norway.

Figure 2. CPIF and CPI inflation (%), and a tolerance band of +/- 2%

There is hardly any evidence that the Swedish economy has suffered from inflation volatility of +/- 2% in the past. Still, if, the Riksbank manages in the future to hold the actual rate of inflation close to the target for a prolonged period, the explicit band may be reduced to its original size of +/- 1%.

Conclusions

We suggest that an explicit tolerance band has major advantages compared to an implicit band. First of all, it facilitates central bank communication. The central bank can be honest about its actual capacity – or, more frankly, its inability – to fine-tune monetary policy towards the inflation target and about various trade-offs between policy goals in the short and long run. Being honest is the most important element in building trust in monetary policy. Second, it gives the central bank additional flexibility to respond to developments that may threaten future economic and financial stability although they are not reflected in the adopted price index for the inflation target. Third, it makes it easier to evaluate monetary policy. These lessons from the Swedish experience carry over to inflation targeting central banks in other countries.

The size of the explicit band is key issue to which there is no easy answer. The proper width of the band is likely to vary from country to country. For Sweden, evidence from inflation targeting suggests that a new band should be set broader than the old band. This broader band will ensure that the Riksbank will reap the benefits of a tolerance band without jeopardising the benefits of the inflation target.

References

Andersson, F N G, and L Jonung (2017), “How tolerant should inflation targeting central banks be? Selecting the proper tolerance band – Lessons from Sweden”, Department of Economics, Lund University Working Paper 2017: 2.

Bundesbank (1995), Die Geldpolitik der Bundesbank, Deutsche Bundesbank: Frankfurt am Main.

European Commission (2017), “ERM II – the EU’s Exchange Rate Mechanism”.

Haldane, A G (1997), “The monetary framework of Norway”, p. 67-108 in A B Christiansen and J F Qvigstad (eds.), Choosing a Monetary Policy Target, Scandinavian University Press: Oslo.

Hammond, G (2012), “State of the art of inflation targeting”, CCBS Handbook no 29. Feb. 2012 version, Bank of England.

Jonung, L (1986), “Uncertainty of inflationary perceptions and expectations”, Journal of Economic Psychology 7, 315-325.

Endnotes

[1] For example, the European Exchange Rate Mechanism II allows exchange rates to fluctuate up to +/-15% around the target point (European Commission 2017). The Deutsche Bundesbank allowed variations in money growth of +/- 1% during its monetary targeting regime in the 1980s and early 1990s (Bundesbank 1995).

[2] The Swedish CPI includes the cost of owner-occupied homes. When the Riksbank changes its repo-rate, the mortgage rate responds and the cost of owner- occupied homes changes. CPIF excludes the interest rate effect by assuming a fixed mortgage rate. 

2,730 Reads