Prices and price indices;Income and consumption;Svalbard

Consumer price index for Svalbard2012/2013


About the statistics


Name and topic

Name: Consumer price index for Svalbard
Topic: Prices and price indices

Responsible division

Division for Price Statistics

Definitions of the main concepts and variables

Price refers to the actual retail price of goods and services offered to households. For Svalbard this means prices not included indirect taxes, fees and subsidies.

Expenditure or budget shares are proportional to the consumption of a certain good and total consumption in a household. Expenditure shares in the average household are obtained from the household budget survey.

A Laspeyres price index is a price index where the base-period weights remain fixed. A chained Laspeyres price index is an index linked by Laspeyres indices with different sets of weights. New sets of weights are incorporated into the index every year.

A Fisher price index is a geometric mean of a Laspeyres and Paasche price index. A Paasche price index also uses fixed weights but unlike the Laspeyres price index, the weights are from the actual current period. The Fisher price index is used in the CPI for the index of motor vehicles and indices of alcoholic beverages sold through the State wine and liquor monopoly.

COICOP (Classification of individual consumption by purpose) is a consumer classification developed by EUROSTAT. The classification criterion is the end purpose of the consumption.

Relative price level is understood as the ratio of the price of an individual product in one country to the price of that same product in some other country at a given point in time. Results from this price level survey give the expression of the differences in the price level that a Svalbard household is set against an equivalent household in mainland Norway.

Standard classifications

COICOP classes are used when calculating and subsequently publishing the Consumer Price Index.

Administrative information

Regional level

Regional level CPI. The Price level survey includes only Longyearbyen in Svalbard. In the mainland the whole country is covered.

Frequency and timeliness

As a result of insufficient weight information neither the CPI nor the price level survey for Svalbard has been produced since 2013. For more information, see http://www.ssb.no/inntekt-og-forbruk/artikler-og-publikasjoner/forbruksundersokelsen-pa-svalbard-2012 (only available in Norwegian). The CPI Svalbard is originally annual, published by mid-December every year, about 3 months after the collection of data. The price level survey was meant to be published every 5 years from 2010, with results available by March, about 6 months after the survey period.

International reporting



Data on micro level are stored mainly in excel. Index series and catalogues are stored in Unix and in fame.


Background and purpose

The purpose of this survey is to measure the actual price development of the consumption of the Norwegian households in Svalbard. The survey was first conducted in 2001 on commission from Svalbard Samfunnsdrift, and financed through contributions from various parties. The finance has later been covered by the government commission to Statistics Norway. The survey has been annual except from the year 2007, when it was estimated based on the Price level survey in 2007 and published together with the CPI for Svalbard 2008. The CPI for Svalbard became official statistics in 2009, and has been published annually since then.

The price level survey aims to measure the price level in Svalbard compared to mainland Norway in a specific period. The earliest statistic comparing price level differences between Svalbard and the mainland was in 1990, followed by similar surveys in 2001, 2007 and 2010.

Users and applications

The survey is used by the public sector, both on the mainland and Svalbard. Other important users are the local business community and the population on Svalbard. The CPI is also used as a deflator for the index of retail sales.

Coherence with other statistics

The CPI for Svalbard is based on the model for the mainland. In principle, the same foundation of principles, methods and classification are used. There are only minor methodical differences between the two surveys, and that enables comparisons between the two surveys. Part of the data in the CPI for Svalbard is collected from the CPI for the mainland and from the Internet, since some of the goods are bought on the mainland and over the Internet and consumed on Svalbard.

The consumption expenditures (weights) are based on data obtained from the annual household budget survey for Svalbard (HBS), and the last one was conducted in 2007.For the CPI Svalbard 2013, the weights are based on the HBS for Svalbard 2007, and projected with the HBS for mainland Norway 2012, together with other relevant sources.

Price level survey uses representative goods catalogue for mainland Norway from the CPI as the starting point. Price material that represents the mainland Norway’s prices is identical with the CPI materials for the mainland. Weights used in the computation are from the latest results of Svalbard and mainland Norway’s household budget survey (HBS).

Legal authority

The Statistics Act of June 16, 1989 number 54, §§2-2 and 2-3.

EEA reference

No EU regulations.



The population is defined as all goods and services offered to households in Svalbard. Prices and consumption expenditures of a sample of goods and services are measured. The consumption expenditures (weights) are based on data obtained from the annual household budget survey (HBS), while prices are collected from a total selection of companies in Longyearbyen.

Svalbard households have expenditures on products and services purchased and consumed on the mainland, such as housing and public transport, but these are excluded from the CPI. Purchases of products and services are included in the CPI if the consumption happens on Svalbard, such as white goods and clothing.

The price collection on Svalbard is restricted to prices of products and services in Longyearbyen.

Statistics Norway is trying to measure the price development of an average Svalbard household.

The price level survey uses the same population and weights defined for the CPI purposes.

Data sources and sampling

The CPI is based on the following sources: price collection on Svalbard, Internet, electronic data from companies, telephones, and budget shares from the household budget survey.

In the price level survey, supplementary price collection is done in the mainland to ensure that the basket of good measured is more or less comparable. The choice of stores is drawn from the CPI sample for the mainland.

A sample of about 600 goods and services is selected. The size of goods and services is clearly less compared to the sampling on the mainland. This is due to a lack of observation in either the reference period or in the existing period of the prices, plus goods and services from the mainland being removed because of little representation on Svalbard.

The CPI for Svalbard is based on the same goods and services used by the CPI for the mainland. The goods and services in the sample are selected based on information from the household budget survey and industry information, but have also adapted the consumption pattern on Svalbard. The CPI for Svalbard includes goods such as snow scooters, weapons and scooter suits. At the same time it has been necessary to remove goods and services that are unimportant in the consumption for the Svalbard households.

The sample of goods and services is mainly kept the same, but is regularly updated with new important products that are in the market, while outdated ones are removed.

The CPI is mainly based on sample principle, and carried out by selecting a sample of companies according to statistical methods. The area on Svalbard is restricted despite an increasing number of companies, and that makes it possible to carry out almost a complete selection among the companies. It is possible to obtain information from all shops in Longyearbyen that offer goods and services aimed at private consumption. The advantage of a complete selection of shops is that we can avoid sources of error on registrations and error sources.

Collection of data, editing and estimations

The main part of the prices of goods is collected by questionnaires in the third week of September every year, and is carried out by people from Statistics Norway. The prices of services are mainly gathered in Oslo through the Internet, e-mail and telephone. Statistics Norway receives scanner data with prices from the convenience chains. The population on Svalbard purchases clothing via mail order, and the survey therefore includes prices from a sample of mail order companies.

Data collection for the price level survey will be every 5 years and coordinated with the data collection for CPI Svalbard and CPI for mainland Norway.

The price data is considered carefully and systematically both before and after the calculation of the CPI. The process of revision is much smaller for Svalbard compared to the mainland, since the amount of data is considerably less for the CPI for Svalbard. The majority of the revision happens at the same time as the data collection. The use of price collectors gives the benefit of asking questions and clarifying extreme values while gathering the data.

In the CPI for Svalbard, the data on micro level is controlled for data registration errors and other obvious errors, and extreme values are identified, controlled and either approved or rejected. Companies are contacted in the event of special uncertain price changes.

Price level survey has additional controls to ensure that criteria for comparability and equi- representativity are met. Prices are to be reported for comparable products or products with identical or equivalent physical and economic characteristics. This is to ensure that the differences in prices between Svalbard and the mainland for products reflect actual price differences and not a reflection of quality differences. It is ideal that brand and models are specified however if this is not possible, generic specifications are also employed. Comparing products that are not equally representative for the two areas will result in biased price relatives, thus the need for equi-representativity. This means that in both areas, we should be able to price that number of representative products commensurate with the heterogeneity of the products and the price levels covered by the basic heading and its expenditures on the basic heading. Comparability and representativity requirements may present problems since consumption patterns vary. It should be noted that should there be a need for compromise, comparability is favored over representativity. A lack of representativity may lead to biased results but without comparability, there will be nothing to compare. These specifications require then that the choice of goods and services should be equally representative for both areas.

The CPI for Svalbard is based on the same model as for the mainland. There are only some small differences in the method and this gives the possibility to compare the two CPIs. The calculation of the CPI for Svalbard is less complicated compared to the survey for the mainland since no stratification is necessary for geographical areas.

Indices at micro level are calculated by a non-weighted geometric mean of the price observation in the existing period (for instance October 2009) compared with the same mean in the period of reference (for instance October 2008). Price indices are calculated for all goods and services. Further aggregation up to consumption level and further to total indices is in accordance with Laspeyres price indices formula with expenditure shares from Svalbard.

Fisher price indices are geometric means of Laspeyres and Paasche price indices. The indices for alcoholic beverages in the CPI are calculated by Fisher price indices together with the lowest consumption level for the indices of food and non-alcoholic beverages.

The CPI for Svalbard initially calculates short-term indices with September as a basis, then chained to the long-term series with March 2001 as 100. The indices are expressed in index form and as change-rates expressed as a per cent.

Price level survey uses two price sets and two sets of expenditure share for Svalbard and mainland Norway in the aggregation of the price material such that goods and services with higher expenditure share gets the corresponding share in the main results. To be able to come to the price relative for each good, an index is made based on the average prices of a commodity in an area which is set to 100 and the average price of the other area is computed based on this. A price relative for a group of goods is computed based on the weighed arithmetic average of the price relatives per commodity in this group. Computation starts from the lowest level and is aggregated using the weight information step by step until the All-item index is computed. To illustrate this, computation can start from the representative good white bread, which is aggregated with the other price index for the different types of bread towards a price relative for the consumption group bread. The results are then used to aggregate up to the next level which is bread and other grain products and further aggregated until the main consumption group food. The resulting index at the main consumption group is then weighed together with the results from other main groups like clothing, footwear, and transport towards the total all item index.


Data collected from firms and households are subject to secrecy and are to be kept or destroyed in a secure manner. Any use of the data must be in accordance with the rules set out by the Data Inspectorate.

Comparability over time and space

The CPI for Svalbard was established in March 2001. There are no retrograde series farther back than 2001. Since 2001, the CPI for Svalbard has been published every year, with the exception of 2007 when it was estimated based on data from the price level survey. The CPI for Svalbard has been Official Statistics since 2009, and published annually on 15 December.

Series are published down to a 3-digit COICOP class level, in addition to selected 4-digit groups.

Independent price level surveys between Svalbard and mainland Norway was done in 1990, 2001, 2007 and 2010. The goal of the price level survey is basically to compare the price between areas in a specific time frame. Changes in markets and consumption patterns over time means that there has to be changes in methods earlier used in the survey to ensure that the comparability criteria is met in such a survey. Such change can affect the comparability of results through time.

Accuracy and reliability

Sources of error and uncertainty

Svalbard is a small area with a restricted offer of goods and services; there are challenges of lacking prices in either the period of reference or the existing period. Various errors will occur in statistical surveys. This leads to greater uncertainty in the prices for Svalbard than for the mainland. Sometimes the price indices are based on only one observation. The individual price observation thereby has greater significance in the price index for Svalbard compared to the mainland.

Errors can arise during the collection of prices and preparation of data. The CPI on the mainland can often have errors in the collection routines and the preparation carried out by the companies that report the prices. The CPI for Svalbard uses physical price-collectors, but errors in the registration or lack of description can still arise. Despite the data being controlled and analysed, errors can occur in the data.

Total and partial non-response are important sources of misalignment in the data. Total non-response is very low in the CPI for Svalbard. The use of price-collectors instead of postal questionnaires, combined with a restricted area, results in all the companies being visited and the price information gathered. Total non-response in the CPI for Svalbard happens when companies have closed down since the last survey or are closed during the collection period. Sometimes price information is impossible to gather at Svalbard, but it is often gathered afterwards via telephone or e-mail. The biggest challenge to CPI for Svalbard is that non-response can lead to a product area having no prices.

Partial non-response happens when goods are out of production or sold out. For the mainland, the prices are imputed mechanically for both total and partial non-response. Because there are few prices, non-responses in the price index for Svalbard are not imputed unless they relate to whole product groups.

The CPI for Svalbard has a a complete selection among the companies, so sample errors doesn't occur.

An important assumption is to calculate indices that make it possible to compare the prices of goods and services that are collected in both periods. An annual CPI complicates this as opposed to a monthly CPI. In particular, the sample of goods and services may have been replaced within the relevant periods. This can lead to problems in replacing goods and services, and possible quality changes. The longer the period is between two measurements, the greater the uncertainty in the result.

Traditionally non-sampling errors in the CPI are divided into three main types of measurement errors;

a) Income effects which influences consumer behaviour through time;

Households through time face changes in income, which also affect the expenditure shares of different goods and services. To be able to set proper weights in CPI, in each period, the weights in the CPI should reflect the expenditure shares. Therefore the weights are updated yearly. There have not been any calculations done to measure errors caused by non-representative weights.

b) Price effects caused by changes in relative prices;

The price relationship between different goods and services changes over time. This changes also the expenditure share of households and causes the same measurement challenge as the income effects.

c) The unsatisfactory treatment of quality changes;

Statistics Norway has not accomplished separate calculations of these measurement errors in the Norwegian CPI. Calculation of measurement errors and analyses done in USA, Canada, Sweden and Great Britain estimate the measurement errors somewhere between 0.4 to 1.1 per cent measured as annual growth rate. However there are uncertainties in these estimates. It’s likely to assume that the Norwegian CPI overestimates the development in the cost of living, but the level is likely to be less than one per cent measured as annual change.

Sources of error in price level surveys.

The quality and how expenditure share and price data of the goods and services in the survey represents the price level between Svalbard and mainland Norway affects the precision of the results.

Goods and services chosen in the comparison should be equally representative for both Svalbard and mainland Norway’s households. This is to avoid biased results wherein one assumes that there is a lower price level rather than assuming that the goods and services are not representative. The basket of goods and services in the survey is base don CPI for mainland Norway’s representative good catalogue. The design of the catalogue basically to measure the development of the cost of living of private households in mainland Norway and not for measuring price levels. To meet the requirements of equi-representativity, there has been some adjustment done in the choices of goods and the specification for the goods and services that are known to be of importance to Svalbard households.

There are uncertainties connected to the computation of average prices. In a price level survey, efforts are made to compare similar goods and services. Due to the small amount of choices of goods and services in Svalbard, we have a limited set of price observations, thus there is uncertainty in the computation of the average prices per good. The CPI’s representative goods catalogue on the other hand have a number of goods with general specifications like Men, winter jacket. Average prices for a good in Svalbard and the mainland can also have different qualities. A typical winter jacket used in Svalbard has a quality that can withstand a colder climate compared to a typical winter jacket in the mainland. Thus winter jackets in Svalbard have extra features and will consequently have higher prices. In this case, the definition of the principle of comparability of similar item is broken. A number of manual controls have been done to limit this kind of uncertainty. A larger choice of goods and more price observations contributes also smoothen out the mistake which is caused quality difference of the goods and services.