A central bank that is to steer inflation ahead in time seeks to exclude temporary price variations when setting policy rates. At a given point in time, it is not that easy to determine which price changes are permanent and which changes are temporary. Indicators of underlying inflation can be useful in this context. In this article, we make an empirical evaluation of various indicators of underlying inflation in Norway. Our conclusion is that there is no one indicator that is a perfect measure of underlying inflation at all times. A central bank should therefore follow developments in several indicators of underlying inflation.
Low and stable inflation is a central objective of monetary policy in many countries. In countries where monetary policy is operated using an explicit inflation target, a quantified inflation target is often linked to the consumer price index (CPI). In Norway, for example, the Regulation on Monetary Policy of 29 March 2001 states that "the operational target of monetary policy shall be annual consumer price inflation of close to 2.5 per cent over time".
In periods, the CPI may be influenced by temporary changes in one or several prices. This is illustrated in Chart 1, which shows the year-on-year rise in the CPI in Norway in the period January 1993 to December 2005. The degree of variability in the rise in prices was particularly high between 2001 and 2004, primarily reflecting pronounced changes in VAT rates and wide variations in electricity prices. These factors only had a short-term impact on headline inflation.
[GRAPHIC 1 OMITTED]
In interest rate setting, the central bank seeks to ignore such short-term price variations. The Regulation on Monetary Policy in Norway also states that in general "direct effects on consumer prices resulting from changes in interest rates, taxes, excise duties and extraordinary temporary disturbances shall not be taken into account". At a given point in time, it is not that easy to determine which price changes will persist and which changes will only have a temporary effect on headline inflation. Indicators of underlying inflation that seek to remove temporary noise and show the more persistent trend in price developments may be useful in this context. A number of central banks therefore follow developments in indicators of underlying inflation.
In this article, we first take a close took at the uses of indicators of underlying inflation and the definition of "underlying inflation" in the literature. Section 3 provides a brief overview of the various methods proposed for estimating underlying inflation. The main contribution in this article can be found in Section 4, which presents an empirical evaluation of various indicators of underlying inflation for Norway. (2) We evaluate both new indicators presented in this article and indicators that are already in use in Norges Bank. The final section provides a summary of this article.
2 Uses of indicators of underlying inflation
Indicators of underlying inflation can be used for different purposes. The indicator's purpose may have implications for its construction and properties. An indicator of underlying inflation can be constructed with a view to evaluating monetary policy. Such an indicator should not incorporate prices that the central bank has little scope for influencing. In Norway, for example, electricity prices are largely determined by temperature and water reservoir levels, and electricity price swings may result in wide fluctuations in the CPI. A central bank has little scope for countering such fluctuations in the CPI.
The indicators can also be used to assess the inflation outlook. Monthly CPI figures are influenced by short-term fluctuations in certain prices. An indicator of underlying inflation used in this context should only capture persistent changes in inflation. An increase in electricity prices that is perceived as permanent may gradually lead to a higher rate of increase in other prices because producers seek to compensate for higher electricity costs (second-round effects) or because it has an influence on economic agents' inflation expectations. Such a change in electricity prices should not be disregarded when making inflation forecasts.
It has been argued that an indicator of underlying inflation is not necessary when the central bank is conducting a forward-looking monetary policy. Temporary disturbances will not affect inflation 2-3 years ahead, and projections for underlying and headline inflation will have converged. (3) The projected level of inflation 2-3 years ahead will, however, depend on how high the central bank judges the "persistent" part of inflation to be when the projection is made. Indicators of underlying inflation are useful in determining the correct starting point for the projection.
An indicator of underlying inflation can also be a useful tool in justifying and explaining the conduct of monetary policy to the general public. If the rise in the CPI temporarily deviates from the inflation target, an indicator of underlying inflation can contribute to preventing doubts as to the central bank's commitment to reaching the target. It will minimise the risk that temporary shocks to price trends influence inflation expectations. At the same time, when the general public uses an indicator that varies less than the CPI as a reference, this may contribute to more stable inflation expectations. An indicator of underlying inflation that is established and well-known by the general public can have this function.
Disagreement about what underlying inflation really is, is one reason why indicators of underlying inflation have different purposes. In the literature, definitions vary. Eckstein (1981) defined underlying inflation as "trend increase in the cost of factors of production". Underlying inflation is the level of inflation prevailing when the economy is in long-term equilibrium, i.e. in the absence of shocks and when actual output is equal to potential output. This definition of underlying inflation is closely linked to economic agents' long-term inflation expectations. Inflation caused by cyclical factors is not considered as a component of underlying inflation according to this definition.
Quah and Vahey (1995) defined underlying inflation as the component of inflation that is due to shocks that do not affect output in the long run. The definition in Quah and Vahey (1995) includes Eckstein's (1981) concept of underlying inflation, but also incorporates price rises caused by cyclical factors. Inflation caused by permanent supply-side shocks is not included in underlying inflation.
Other definitions of underlying inflation are more related to how one should in practice choose the best underlying indicator among several candidates. Bryan et al. (1997) defined underlying inflation as the indicator that tracks a moving average of headline inflation most closely. Smith (2004) defined underlying inflation as the indicator that is the best forecaster of inflation. This definition was inspired by Blinder (1997), who defined underlying inflation as the "persistent component" of inflation.
3 Different measures of underlying inflation
Numerous methods for constructing an indicator of underlying inflation have been proposed. This reflects varying concepts of underlying inflation, but it is also because indicators of underlying inflation can have different purposes. The various methods can be broadly divided into four groups:
* Exclude fixed components from the CPI. This is the most common method, and the calculation in Norway of the CPI-ATE (the consumer price index adjusted for tax changes and excluding energy products) is an example. The choice of components to be excluded can be based on statistical criteria--for example, exclude the most volatile price index components--or can be based on a greater element of discretion. In a number of countries, food and energy prices are excluded. This is often justified by the highly volatile nature of these prices, and that variations are caused by supply-side changes rather than changes in demand. A simple example is that bad weather can lead to a sharp rise in prices of certain food products.
* Exclude different components of the CPI from one period to another. The choice of the components that are excluded can be based on statistical criteria or discretion. The indicators trimmed mean and weighted median, which are regularly published by Norges Bank, are two examples where components are excluded based on statistical criteria. Up to 1997, the central bank of New Zealand made adjustments for various shocks on a discretionary basis. But this approach was discontinued as the central bank considered it awkward to estimate the indicator monetary policy would be judged by. (4)
* Reweighting of CPI components. In this method, no component is completely excluded from the estimation of underlying inflation, but the weight that is assigned to each sub-group will depend on the subgroup's historical time-series properties. An example is the indicator published by the Bank of Canada, which gives less weight to prices for goods and services that have shown wide historical variations. (5)
* Model-based methods. These methods are somewhat different from those above. An example is the indicator of core inflation that was proposed by Quah and Vahey (1995). This indicator is based on a structural VAR model of inflation. Another example is the indicator that was introduced by Cristadoro et al. (2005). They use a factor model to calculate underlying inflation based on developments in a large number of time series.
4 Empirical evaluations of different indicators
Since there is no agreement on the definition of underlying inflation, there is no agreement on the weight to be given to various criteria when evaluating different indicators. Roger (1998) and Wynne (1999) discuss several criteria that an indicator of...