A number of measures of underlying inflation are regularly published in the Monetary Policy Report. These are often shown as a band in which the measure showing the highest and lowest rate of increase respectively for each month constitutes the band's upper and lower limit. The band includes both measures where certain predetermined components have been excluded (for example the CPIF excluding energy) and measures that use statistical methods to reduce the significance of those goods and services that have historically varied the most. The measures of inflation are also published here on the Riksbank's website every month, normally on the same day as new outcomes for the CPI are published by Statistics Sweden.
In the CPIF excluding energy and perishable goods, the prices for meat, fish fruit and vegetables have also been excluded in addition to those for energy. The measures calculated using statistical methods are based on the CPIF divided up into 70 sub-groups and calculated as annual percentage changes.
One way of calculating underlying inflation is to adjust for prices that vary considerably in a given month. In Trim85, 7.5 per cent of the highest and the lowest rates of price change each month have been removed for the sub-groups included in the CPIF. Thus, 85 per cent of the total weighting in the CPIF remains in this measure.
The Trim1 measure (or weighted median inflation) is similar to Trim85, except that only the median price change is used, which means that only 1 per cent of the CPIF's total weighting remains.
Another way of calculating underlying inflation is to give prices that only vary a little a higher weighting. In UND24, all of the sub-groups are retained but are given different weighting than the one they have in the CPIF. Goods and services whose annual percentage price change has varied quite substantially over the past 24 months are given a lower weighting and vice versa.
KPIFPV is calculated by estimating autoregressive equations (of the order 1) for each of the group's annual percentage change. The estimated coefficients are then used as weights for each component respectively. The higher the estimated coefficient, i.e. the more prolonged a price change is, the higher the weight.
A third way of calculating underlying inflation is to estimate common trends in various sub-groups of the CPIF. KPIFPC is a "static factor" estimated with the aid of a principal component analysis. All the sub-groups of the CPIF are included in the calculation. As a final step, the component is standardised so that it has the same mean value as the CPIF.
All measures are calculated based on sub-groups in the CPIF
14 September 2018
Please note that with effect from publication of the CPIF for August 2018, all measures are calculated based on sub-groups in the CPIF. Prior to that, Trim85 and Und24 calculated for sub-groups in the CPI were published as the CPI was the Riksbank's previous target variable. The interest expenditure index has been replaced by the capital stock index.