Changes to the CPI - updated successively



Information about changes in the production of the CPI is posted successively. Updated 10 February 2021. 

Changes in the CPI (and HICP) from 2021

Statistics Norway introduces a multilateral calculation method in the price index for food and non-alcoholic beverages as of CPI and HICP January 2021 publication. The main purpose has been to implement a calculation method that works across different consumer groups in the CPI, takes into account expenditure shares at detailed level and captures new items as quickly as possible. Multilateral methods are internationally considered as the best approach for optimal utilization of scanner data in price indices. Multilateral method deviates from traditional price statistics methodology in that more than two periods are used in the calculations.

Multilateral methods originate from comparisons of price levels across countries, which are then adapted to comparisons over time. A multilateral calculation method can be seen as a framework that affects the choice of index formula, product definition, length and splicing of time windows and aggregation structure. Statistics Norway has chosen to introduce a GEKS framework based on bilateral Törnqvist price indices, which utilize rolling 25-month time windows which are linked together in the middle of the time window.

At the same time changes are done to the aggregation structure at detailed level in the price index for food and non-alcoholic beverages. Until now, the price for a given item in a given store has been compared with the same item in the same store over time. From now on, the price for a given item will be aggregated over individual stores within the same retail chain and at the same time fixed annual weights are introduced between the chains. Higher aggregation to consumer groups follows same procedure as earlier.

Analyzes carried out for the period 2017-2020 show that the multilateral calculation method contributes to pull the price growth for food and non-alcoholic beverages slightly up throughout the period as a whole. For sub-groups at detailed level, there may be greater differences between the old and the new method.

In addition, the COVID-19 pandemic has had implications for the annual updates of weights in the CPI and HICP which are updated every January based on household final consumption expenditures from annual National Accounts (NA). The weights are kept constant throughout a whole year. The sample of goods and services in the CPI is also updated.

Measures taken by the Government to limit the COVID-19 outbreak has caused households to adjust their consumption habits to the constraints and disrupted household consumption patterns. Consequently, the pandemic has had a significant impact on household consumption expenditures during 2020. Changes seen in consumption behaviour are believed to be a combination of both permanent and temporary nature. Given the situation of the pandemic in the beginning of 2021, it is likely that the pandemic will affect the household consumption patterns also in 2021, however the severity is uncertain.

Annual NA is regarded as the most comprehensive data source reflecting the household consumption expenditures. The latest NA figures at the level of detail needed in the compilation of weights, are however lagged by two years. In normal years this is unproblematic as changes in consumption from one year to another are typically small, and the two year lagged data are considered representative for the year in question. However, given the sudden changes in consumption caused by the pandemic this assumption no longer holds. Following standard procedure of updating the weights, the pandemic would not have been reflected in the weights for 2021. It is recommended to deviate from the standard procedures by using more recent data, such as household consumption expenditures for at least the available first three quarters of 2020 (quarterly NA). Note however that the quarterly NA are less detailed than the annual NA.

When deriving weights for 2021 in the CPI, Statistics Norway has used an average of the household consumption expenditures in 2019 and 2020 based on annual NA 2019, as well as quarterly (Q1-Q3) and monthly NA (until November) 2020. Thus, the weights take into account both the changed consumption pattern caused by the pandemic, but also the uncertainty in the level of more long-lasting changes. For the HICP, the household consumption expenditures up until November 2020 are used. This is in line with the guidelines from Eurostat, as well as the regulation of the HICP, and ensures comparability of the HICP across countries.

The impact of these procedures will be an HICP reflecting the entire changed household consumption patterns in 2020, while the CPI weights will be less influenced by the consumption pattern caused by the pandemic.

Supplement information COVID-19 pandemic 2020

As of mid-March 2020 government restrictions due to the outbreak of the COVID-19 pandemic affected the consumer areas covered by the CPI. Read more about the international recommendations for appropriate methods on(?) how to deal with this extraordinary situation in the article How does the corona situation affect CPI.

The restriction at national level continued throughout most of April. The article Corona consequences for CPI addresses how the affected data collection influenced the compilation of the CPI for April, and the principles of how similar large shifts in consumption has been treated throughout the pandemic.

The national broadcasting license fee in the Consumer price index from 2020

The national broadcasting license fee in Norway is abolished from 2020. The license fee will be removed from the Consumer price index (CPI) and the Harmonized index (HICP) as well. For more information, see The national broadcasting license fee in the Consumer price index.

New base year in the CPI from 2017

From the January index, published on 10 February 2017, the CPI base year is changed to 2015 (2015=100). This also applies to CPI-AT, CPI-ATE and other series. For more information, see New reference year, 2015=100, in the CPI

As of January 2017 the selection of goods is enhanced for price indices of non-food from grocery outlets, - a result of better utilization of scanner data from grocery chains. The price index of airline fares has been extended with prices for 2 more departure days in connection with weekends per month. In addition , the price index for passenger transport by railway is greatly expanded both in terms of the measurement period and range. 

ECOICOP and new edition of CPI by delivery sector from 2016

The CPI adopted Eurostat’s new detailed 5-digit consumer classification, ECOICOP in January 2016. Statistics Norway has published unofficial 5 and 6-digit COICOP indices for the consumer group Food and non-alcoholic beverages up until January 2016. As a result of a discrepancy between the unofficial and the new official ECOICOP, some previously published indices are no longer available. This results in new names for some of the published groups, see the new names here

A new version of the price development in the CPI by the delivery sector has been implemented. This is more in line with the development in the economy and will enable a more detailed analysis of the origins of price impulses. It will be possible to compare goods versus services and Norwegian goods versus imported goods. Here is the detailed list of representative goods and services in the CPI by delivery sector. The list will not be automatically updated 

The period of measuring prices on groceries is enhanced from one week to three weeks. This applies to scanner data supplied by the grocery chains. 

Changes in the CPI from 2015

As of January 2015 CPI is no longer labelled «per 15». However, data collection periods and statistical methods have not been changed. Also from the same time, Financial Intermediation Services Indirectly Measured (FISIM) has been taken out of the CPI. 

Changes in the CPI from 2013

In the price index of food and non-alcoholic beverages, which is entirely based on scanner data from the Norwegian grocery market, the calculation method at elementary level will be changed as of January 2013. For more information (in Norwegian), see the following article published in Økonomiske analyser 6/2012. In the price indices of actual and imputed rentals for housing, changes are made in the data collection as well as in the method for measuring the monthly price changes. 

Changes in CPI in 2012

In the price index for food and non-alcoholic beverages, which is entirely based on retail scanner data from the Norwegian grocery market, a method for dealing with missing values is introduced. This change is made due to indications of a downward bias partly as a result of an increased sales activity in combination with missing values. Introducing a method for dealing with missing values contributes to reduce the bias in the index. Seasonal items only available during a certain period of the year, have until now been excluded from the price index of food and non-alcoholic beverages due to lack of satisfactory treatment. From January 2012, seasonal items are included in the index as a result of the introduction of estimations for missing values.

In the price index for medical products the product range will be greatly expanded. Statistics Norway has received scanner data on medical products for several years, but has only used data for some representative items. From January 2012, the index will include almost all the products that consumers purchase. As before, the index will be calculated using an unweighted geometric average at elementary level.

In addition, smaller adjustments are made in the price index of airline fares. As of January index 2012, prices of an additional departure day are registered each month, at the same time the number of price measurements prior to the departure is reduced.

The publishing levels are not affected. 

New source for weights in the CPI from 2011

As from the January index in 2011 (published on 10 February 2011), a new weighting will be applied to the consumer price index based on the national accounts' figures for household consumption. Read article on new source on weighting

Changes in housing component of the CPI from 2006

As of July index, published the 10th of August 2006, changes are made in the housing component of the CPI. A new sampling method is established as well as changes in the computation method for estimating changes for actual and imputed rentals for housing.

New sampling method

One of the challenges in creating rental statistics is the lack of existing registers of dwellings for rental as well as of tenants. As of July index 2006, the sample of dwellings for rental is established through selecting addresses from the Central Population Register (DSF). To exclude homeowners from the sample, the addresses are connected to the Ground Property, Address and Building Register (GAB). Finally to reduce the portion of cooperative dwellings, the sample is connected to register data of large cooperative firms.

This sample is the foundation for the Rental Market Survey 2006 as well as the monthly estimations of rental development for actual and imputed rentals for housing in the CPI. The total sample amounts to approx. 5 100 tenants. The computations of the monthly price development will, as of July index, be based on a third of the total sample i.e. 1 700 tenants. A third of the sample will be rolled each month such that tenants participate in the survey every third month. The sample is stratified by geographical areas and by dwelling size, such as number of rooms.

Adjustment of computation method for actual and imputed rentals for housing

As of the July index, the computation method is adjusted. Both actual and imputed rentals for housing are based on actual observed rentals in the rental market. Since the characteristics are quite different in the rental and the owner-occupied market, as of July index, corrections are made to account for these differences by introducing weight shares both for geographical areas and dwelling size such as number of rooms. There will be no breaks in the official time series of actual and imputed rentals for housing, but in detailed levels there are changes due to new stratifications.

New weight foundation for imputed rentals for housing

In August each year the weights in the CPI is updated based on expenditure shares from the Household Budget Survey (HBS). From August 2006 to July 2007, the weights are based on the HBS from 2003, 2004 and 2005. As of August index, to be published on the 11th of September 2006, the Rental Market Survey 2005 will be the foundation for the estimation of expenditure shares of homeowners in the 2005 HBS and the weight shares in the CPI. To achieve the best approximation of owner-occupied housing the sample is restricted to only professionally based rentals, rejecting rentals containing different forms of regulations and subsidies.

Improved sub-index for food and non-alcoholic beverages from 2005

The CPI for August 2005 will include an improved sub-index for food and non-alcoholic beverages that solely is based on bar code data. The choice of method and computation for the improved index will be thoroughly documented later. The change will, however, involve some practical adjustments in the CPI.

The purpose of the change in the sub-index is to exploit all the information available to create an improved and more robust index. There will be no breaks from previous series and the official groupings, coicop level 4, 3 and 2, will remain identical. The changes appear on a more detailed level, and the greatest change is that the representative goods in the sub-index disappear completely.

The sub-index for food and non-alcoholic beverages previously consisted of approximately 250 representative goods, of which a large number were published monthly in Table 4 with monthly index figures and average prices for the reference year 1998. As of the August 2005 release the computation of the index based on the 250 representative goods will be replaced by an index based on almost all the prices within the food and non-alcoholic beverages grouping. The bar code data reports contain more than 300 000 price observations for 14 000 goods every month. This information enables an improved and more robust index for food and non-alcoholic beverages.

A consequence of the change is that the indices at the most elementary level no longer will be computed at representative goods level, but at a more aggregated coicop 6 level. As a result, Table 4 will disappear, but an alternative will be considered as soon as the computation process has been implemented.

CPI adjusted series

Read more on adjustments in CPI in the article CPI adjusted for tax changes and excluding energy products (Economic Survey 1/2002).