Survey of consumer expenditure

Total consumption expenditure

Consumer expenditure is the average annual consumption in NOK for a household, according to COICOP (Standard for classification of individual consumption by purpose. See detailed description under “Standard classifications”). Consumer expenditure does not include expenses for direct taxes, social security premiums, real investments, and contractual savings. The share of consumer expenditure indicates the share of total consumption for a specific commodity group or service. Thus, the total consumption constitutes 100 percent.

Household is the unit in the household budget survey. A household includes all persons who are permanently resident at the same address and who have a common food budget.

Income after tax is the total income where fixed tax and negative transfers (pension premiums in employment and paid child support within the public scheme) are deducted.

Income information is obtained from the register, with status from one year before the statistical year. We group the respondents by income group (quartiles):

• First quartile: lowest 25 percent of the income distribution

• Second quartile: next lowest 25 percent of the income distribution

• Third quartile: next highest 25 percent of the income distribution

• Fourth quartile: highest 25 percent of the income distribution

Income per consumption unit / equivalent income To be able to compare the after-tax income level between different households (as an approximation to compare living standards), it is common that in addition to household income, the number of people in the household is also taken into account. This is done by dividing the total household income after tax by the number of consumption weights or consumption units in the household. The number of consumption units is calculated using so-called equivalence scales. The consumption units take into account that households with many people need higher income than households with few people to have a corresponding standard of living, and that households with many people will have economies of scale when it comes to several goods (e.g., TV, washing machine, newspaper, broadband connection, electricity expenses, etc.). There are several types of equivalence scales used in different contexts. In income and wealth statistics, the so-called EU scale (see below) is mainly used.

Consumption units calculated according to the EU scale assign the first adult in the household a weight=1, then the next adults a weight=0.5 and children under 17 years a weight=0.3. According to this equivalence scale, for example, a household of two adults and two children must have a household income that is 2.1 times as high as a single person to have the same economic welfare.

The low-income EU scale The annual low-income threshold is set to 50 or 60 per cent of the median after-tax income per consumption unit. When calculating persistent low-income over a three year-period, the low-income threshold is set to 50 or 60 per cent of the average median during the same time period. When calculating persistent low-income over a four year-period, persons with income below the annual low-income threshold the current year, and below the low-income threshold in at least two of the previous three years, are regarded as having persistent low income. After-tax income per consumption unit equals total household taxable and non-taxable income, minus taxes, divided on the number of consumption units in the household. The number of consumption units is calculated by using the 'modified' OECD scale or the EU scale, where the first adult is given a value of 1, any additional adult is given the value of 0.5, and each child is given a value of 0.3. The number of consumption units in a household consisting of two adults and two children is thus 2.1, according to this method

Centrality is an index that distributes municipalities based on proximity to workplaces and service functions, without using urban areas in the classification. The centrality distribution follows the standard for centrality, which categorizes all municipalities from 1 (most central) to 6 (least central). A complete list of which municipalities belong to which centrality category can be found in the list under ‘standard classifications

Classification of Individual Consumption According to Purpose (COICOP) 2018

Classification of households

Standard for centrality

Name: Survey of consumer expenditure
Topic: Income and consumption

Division for Income and social welfare statistics

Results are on national level, there are tables on centrality

The survey of consumer expenditure are published at least every five years.

Data from the survey of consumer expenditure is reported to Eurostat. Microdata is made available to researchers and students through Eurostat. Comparable results from all the European countries are presented on Eurostat’s websites.

Statistics Norway stores collected and revised data securely, in line with current legislation for data processing. Anonymized files are available to researchers and students through Sikt - the Knowledge Sector’s service provider.

The main purpose of the consumer expenditure statistics is to provide a detailed overview of private household consumption divided by categories for goods and services. Statistics Norway (SSB) has published hpusehold budget statistics since 1958. In the period 1974-2009, data for survey of consumer expenditure statistics were collected annually without major content changes. Before 1974, nationwide surveys were conducted in 1958, 1967, and 1973. After 2009, the statistics were published in 2012 and 2022. In 2022, the statistics underwent a major restructuring that breaks the time series. From 2025, the statistics will be published according to the European Parliament and Council Regulation (EU) 2019/1700 and will be published at least every five years.

Important external users include the government, media, and researchers in areas such as economics, living conditions, consumption, and diet. In addition, the statistics serve as a basis of information for others with an interest in private consumption and changes in consumption patterns.

Statistics Norway uses the statistics internally for the revision of the national accounts, for weights for the Consumer Price Index (CPI) at the detailed commodity and service levels, and in analysis and research at Statistics Norway.

No external users have access to the statistics and analyses before they are published and accessible simultaneously for all users on ssb.no at 08:00 am. Prior to this, a minimum of three months' advance notice is given in the Statistics Release Calendar. This is one of Statistics Norway’s key principles for ensuring that all users are treated equally.

The consumer expenditure statistics are based on sample surveys where households report their own consumption. From this, household consumption is calculated by commodity and service classification COICOP.

The statistics for consumption in households publish reported and calculated housing expenses for tenants and owners, both consumption measured in Norwegian Kroner and as a share of total consumption. For owners, housing consumption is not based on actual reported housing costs, but a housing consumption is calculated, equivalent to what it would cost to rent a corresponding home that one owns. See definition under “04.2.1.0 Imputed rentals of owner-occupiers for main residence”. This differs from housing expenses as calculated in the statistics "Housing conditions, survey on living conditions", which is based on an survey where one asks about actual paid housing expenses such as interest, maintenance, common costs, electricity, insurance, etc.

Calculating housing consumption as done in the HBS statistics means, among other things, that one gets a higher housing consumption for owners with homes with a high value and low loan burden, than if one had calculated actual housing costs..

Voluntary survey

The statistics are developed, prepared, and disseminated under the Act of June 21, 2019, no. 32 on official statistics and the Central Bureau of Statistics (Statistics Act, lovdata.no). The statistics are part of the national program for official statistics, main area Income and consumption, sub-area Income and wealth.

From 2025, the consumer expenditure statistics will be conducted under the European Parliament and Council Regulation (EU) 2019/1700, which establishes a common framework for European statistics based on surveys aimed at individuals and households. This framework regulation came into force in November 2019 and applies to consumption statistics from 2025.

The observation unit of the statistics is private households in Norway. Households are defined as cost households, meaning that the household consists of persons who have a common address and food budget.

The data sources are a representative sample of Norwegian households that report private consumption, as well as information retrieved from Statistics Norway’s administative registers.

Income information is retrieved from Statistics Norway’s income register, which is mainly based on information from the tax return.

Household information and other background variables such as gender and age are retrieved from the population register. Educational level is retrieved from the education register.

The gross sample is 12,000 households in 2022. The sample is drawn from Statistics Norway’s cost household register and stratified/distributed by household type and county.

Data collection

Data collection takes place throughout a calendar year from January 1 to December 31. Data collection is done through three different sources:

• Participants are recruited and report individual expenses and background information through telephone interviews (Computer Assisted Telephone Interview CATI).
• After the first interview, participants report all their purchases in an assigned reference week from Monday to Sunday. Participants’ reference weeks are evenly distributed throughout the year. Data is reported continuously in an app specially developed to collect consumption data.
• Participants fill out a form about major purchases in the last 12 months, as well as fixed monthly expenses in a web form in the app.

Cost households are drawn for the sample. The analysis unit is households. Information about occupation, industry, income, education, household information, age, and housing is linked from administrative registers.

See more about data collection in the documentation of the survey: Household budget survey 2022, Documentation Report (upcoming)

Editing

By editing, we mean control, review, and modification of data.

Both during the interview and when reporting expenses in the app/questionnaire, there are controls to prevent errors. However, this does not prevent all misregistrations. In the processing of incoming data, additional checks are set up to identify outliers or possible incorrect registrations.

Imputation Values are imputed for missing reported information. For missing information on expenses in the form, where the respondents state that they have an expense, but answer "don't know" to the question about the value, average or median values are calculated based on the distribution for the total sample. In some cases, extreme values (outliers) are also observed, which are assumed to be the result of reporting errors or other measurement errors. These also receive imputed values.

Missing values of the goods registered in the diary can, for example, be due to poor pictures of receipts or the household has only reported a total sum for a purchase without specifying the contents. Registrations of receipts with missing or deviant values are automatically edited or sent for manual control. If errors are detected in the manual check, these are edited to contain the same information as the receipt image. In cases where the household has only reported a total sum for a purchase without further specification of the content, algorithms are used to assign correct item codes and prices based on similar purchases in similar stores.

Duplicates occur where respondents have reported the same expense in both the diary and the questionnaire. If a household has reported the same commodity group in both the questionnaire and the diary, as a general rule, only the expense reported in the questionnaire is retained if the question explicitly concerns the relevant commodity group. This also applies to situations where the household has not reported a value in the questionnaire but has still reported in the diary. In cases where the question in the questionnaire does not explicitly ask about the expense, but the household has still registered the same commodity type in both places, we remove the expense from the diary only if it meets a set of rules.

Editing of COICOP groupsThe goods and services registered in the diary are coded to COICOP groups based on, among other things, the name of the commodity/service. This classification is done both manually and by using algorithms. To identify systematic errors in COICOP coding, random checks are carried out. Based on these random checks, minor and major edits of the COICOP classifications are performed.

Calculations The figures in the statistics bank are published as estimates for average consumption expenditure in Norwegian Kroner and consumption expenditure as a share of total consumption.

To correct for bias in the net sample compared to the population, dropout weights are created. That is, we let responses from persons with characteristics that are underrepresented in the net sample count more, while responses from persons with characteristics that are overrepresented count less. The dropout weight adjusts for bias in who responds compared to the population the statistics are intended to cover.

The figures are calibrated against register information with a combination of age, student/non-student, income, education, county, and family size. In addition, adjustments are made for biases in dropout after the reporting period.

Not relevant

SN has worked out guidelines for coupling of different data sources for statistical purposes. The guidelines are based on SNs authorisation given by the Data Inspectorate for person registers, and the Statistics Act. According to these guidelines responses given in surveys can only serve for the purpose of making statistics. i.e. information concerning groups of people will be given, not for individuals. When survey data files are coupled to registers, encryption techniques are used in order to ensure that it is impossible to identify persons from the survey or register information in the coupled data file.

The statistics underwent a major change/restructuring in 2022, which will result in a break in the time series.

The most important change from 2012 is the new classification of goods and services from the previous COICOP 1999 to UN COICOP 2018. The biggest changes in the new classification is that the previous group 12 "Other goods and services" is divided into two new groups: 12 " Insurance and financial services" and "13 Other goods and services". In addition, former group 08 "Postal and telecommunications services" and group 09 "Culture and leisure" have changed content between them. For example, IT equipment has now been moved from group 09 to a more refined group 08 for "Information and communication".

In addition, a change in method has been implemented for group 04.3 "Maintenance, repair and security of housing": Compared to previous years, only minor expenses have been entered for maintenance and renovation in 2022, while larger expenses are considered an investment. This has led to a relatively large decrease in this item. The change thus affects the other main groups of consumption (shares go up).

The restructuring also involves a change from telephone/visit-assisted interviews and handwritten booklets for daily purchases to self-administered online forms and digital reporting of purchases in an app, with the option of optical character recognition (OCR) of purchase receipts.

The reference period for reporting current expenses has also been changed from two weeks in previous surveys to one week from 2022.

Changes have also been made to the way weights are calculated. The main differences compared to previous weights are that previously the net sample was weighted to correspond to the gross sample, while today it is weighted against the population. In addition, the new dropout weights include more population characteristics than before.

Measurement and processing errors

In any survey, both in censuses and sample surveys, there will be incorrect answers. The errors can occur both in connection with the collection and during processing.

Collection errors

Data collection in the HBS is done through telephone interviews (CATI), self-administered web questionnaire, and by recording purchases in an app through a reference week. In the telephone interview, the interviewers are in direct contact with the respondents. An important advantage of this collection method is that it is easier to avoid misunderstandings of questions or registration of invalid values.

In the self-administered web fquestionnaire, the understanding and interpretation of the questions are to a greater extent left to the respondents. This can affect the quality of the answers, for example, by misunderstanding questions. When asked questions that people find complicated, there is a greater risk of incorrect answers. Collection errors can also occur because certain questions are perceived as sensitive. It can be, for example, purchases that one does not want to report. When households are asked to report larger purchases as far back as 12 months, it is also reasonable to assume that not all purchases are remembered, and that we therefore get an underreporting of such purchases. In the HBS, we know through comparisons with other sources that there is underreporting of expenses on some goods and services. Since we do not know which households underreport and the size of this underreporting, it is not possible to correct for such measurement errors. We have therefore chosen to use the expense values that households actually report.

Processing errors are deviations between the value that is recorded and the value that is eventually reported. Such errors can occur, for example, during derivations (re-recording). The goods and services recorded in the diary are coded to COICOP groups based on, among other things, the name of the goods/service. This classification is done both manually and by using algorithms. Both manual coding and coding using algorithms are subject to uncertainty, especially for goods and services that occur rarely or where the text makes it impossible to classify into the correct COICOP group. The goods and services that are coded by algorithms are sent for manual coding at low prediction probability. To identify systematic errors in COICOP coding, samples are taken.

Dropout error

The gross sample is drawn to reflect the population, but when the dropout is not the same in all groups, the net sample will no longer be fully representative. This bias will vary with grouping and which variable one is looking at. To correct for the biases in the net sample relative to the gross sample, the numbers in the tables are weighted.

Sampling error

There are several types of uncertainty or errors associated with the results of a survey. One type will be sampling variance due to the fact that it is measurements of a sample instead of a total count. The size of the sampling variance (standard deviation) depends, among other things, on the size of the sample, the length of the registration period, and the way the sample is drawn.

In the tables, we report standard errors as a measure of error margin. The assumption behind this measure is that the average figures for consumption expenditure are only subject to sampling error. That is, the measure does not take into account other types of errors, such as measurement errors, processing errors, and coverage errors. In cases where these errors are large, the standard error may not necessarily give a correct picture of the actual error.

Not relevant