Sickness absence

Name: Sickness absence
Topic: Labour market and earnings

Division for Labour Market and Wage Statistics

Figures of doctor-certified sickness absence are published on national, county and municipality level.

Figures of self-certified sickness absence are published on national and county level.

The statistics are published quarterly and annually. Figures are usually released 9-10 weeks after the end of the quarter.

Figures from the statistics are reported to OECD.

Collected and revised data are stored securely by Statistics Norway in compliance with applicable legislation on data processing.

Statistics Norway can grant access to the source data (de-identified or anonymised microdata) on which the statistics are based, for researchers and public authorities for the purposes of preparing statistical results and analyses. Access can be granted upon application and subject to conditions. Refer to the details about this at Access to data from Statistics Norway.

The system for assuring the quality of Norwegian official statistics is based on quality requirements in the Statistics Act and in the European Statistics Code of Practice. The annual report on the quality of official statistics assesses compliance with the quality requirements for all official statistics as a whole.

The National programme for official statistics sets the framework for the areas Statistics Norway and other public authorities produce statistics on. The programme defines and outlines official statistics.

Further information about the system for quality in official statistics can be found at ssb.no

The statistics intend to shed light on sickness absence trends in total and distributed by industry, sector, hours worked, and various personal variables connected with those on sick leave. These form the basis for developing and evaluating measures aimed at sickness absence.

Publication began in 2001 and replaces the partially summary statistics that previously covered the public sector and parts of the private sector with a comprehensive set of statistics, intended to cover all industries and sectors and help comparison between them.

The production of the statistics is in collaboration with the Norwegian Labour and Welfare Administration (Nav). Doctor-certified sickness absence is based on a Nav register, holding all medical certificates issued by doctors (the sickness certification register). Self-certified sickness absence is based on a sample survey conducted by Statistics Norway (SSB).

Major users of the statistics are the authorities, employers' and employees' organisations, individual companies, and researchers. The statistics played a central role in the evaluation of the Inclusive Working Life Agreement (IA Agreement) before 2025.

Aims that have governed the planning of the statistics are:

1. The possibility for employers and funds-appropriating authorities to make reliable estimates of expenses.
2. The possibility for employers' and employees' organisations to evaluate the efficiency of various measures implemented to reduce sickness absence.
3. The possibility of identifying risk groups to implement preventive measures.
4. The possibility of comparison over time as the basis for research in the area.

No external users have access to statistics before they are released at 8 a.m. on ssb.no after at least three months’ advance notice in the release calendar. This is one of the most important principles in Statistics Norway for ensuring the equal treatment of users.

There is one exception in the sickness absence statistics. The statistics consist of a survey part (on self-certified absence) and a register part on doctor-certified absence (from Nav), where Nav is also a partner in the production. As a result, Nav have access to parts of the statistics scheduled for publication before 08:00 on the release date. See the principles of equal treatment for more information.

Labour force survey (LFS)

The primary purpose of the Labour Force Survey (LFS) is to provide data on people’s participation in the labour market. Information on sickness absence is therefore limited to the reason individuals did not work during the reference week. Everyone who was absent for the entire reference week and reported their own illness as the main reason is recorded as absent due to sickness in the LFS. The survey also collects data on partial absence during the reference week and the reason for the absence. However, they do not publish partial absence.

The level figures in the Labour Force Survey (LFS) and the sickness absence statistics differ. This difference is not due to measurement errors or sampling uncertainty, but to different definitions of the measured variables. The main differences are:

• The LFS only includes sickness absence lasting at least one week, while the sickness absence statistics include all sickness absence.
• The LFS also includes figures for the self-employed and occasional small jobs, where sickness absence is lower than among employees covered by the sickness absence statistics.

Figures on sickness absence for the entire reference week from the Labour Force Survey (LFS) are available in Statbank.

LFS is suitable for comparing sickness absence between countries because it is a standardized survey conducted across countries. However, there are challenges related to sickness absence:

• Different countries have different rules for whether individuals are employed when they have a long-term absence. This can be an issue because interviewers, in the LFS, only ask people classified as employed about sickness absence.
• Differences in sickness insurance regulations can affect whether individuals count as employed.
• Employment protection during sickness absence varies between countries.

Working environment, The survey of living conditions

The Living Conditions Surveys is an annual survey since 1996, with varying themes. Every three years the themes repeat. The main sample for the Living Conditions Survey consists of 5,000 individuals. Interviewers collect data through face-to-face and telephone interviews.

The Living Conditions Surveys include information on work-related health problems and sickness absence for individuals who have had continuous sickness absence of more than 14 days during the past 12 months. The survey also provides a substantial amount of data on people's working environment, both physical and psychosocial.

Other Sickness Absence Statistics

The Confederation of Norwegian Business and Industry (NHO, NAVO, and Finance Norway, previously produced sickness absence statistics for their member companies. The government also used to produce sickness absence statistics for all employees for whom it had employer responsibility. These organizations have now stopped their data collection and receive sickness absence statistics for their units from Statistics Norway (SSB). Furthermore, these statistics only covered parts of the population included in SSB’s sickness absence statistics, making direct comparisons with SSB’s data difficult.

The Norwegian Association of Local and Regional Authorities (KS) previously produced sickness absence statistics for permanent employees in municipalities, county authorities, and other companies within KS’s collective bargaining area. Reporting sickness absence to KS was voluntary. KS has now dropped its data collection on sickness absence and receives statistics for its members from Statistics Norway (SSB). KS’s former sickness absence statistics are not comparable with SSB’s statistics.

The statistics are developed, produced and disseminated pursuant to Act no. 32 of 21 June 2019 relating to official statistics and Statistics Norway (the Statistics Act).

The statistics are part of the national program for official statistics, main area Labour and wages, sub-area work environment, sickness absence and work stoppages.

Not relevant

Quarterly time series starting from the 2nd quarter of 2000.

Population

The population in the sickness absence statistics is based on the International Labour Organization (ILO) definition of employed persons. This is derived from data reported through the a-ordning (English description) by businesses for their employees. The population covers ordinary and maritime employment relationships for all resident employees aged 16 to 69 whose employment is active. An employment relationship counts as active when they receive wages and the start and end dates of the employment fall within the reference period.

The population does not include the self-employed or individuals in compulsory military service. Furthermore, the statistics do not cover absences due to a child’s illness or parental and maternity leave, following the legal definition of self-certified and doctor-certified sickness absence.

It is important to be aware that both the definition of the reference period and the units used in the statistics vary between the counting of lost man-days and the counting of individuals.

• For statistics on sickness absence man-days and the associated sickness absence rate, the units are employment relationships. This entails that each employment relationship sums separately.
• This contrasts with the counting of employees with sickness absence, where the units are individuals. This includes only the employment relationship considered the most important (the main job). For individuals with only one job, there is no difference. For those with multiple jobs, this approach allows for consistency in variables such as industry and municipality of residence (which could potentially change during the quarter).

There are also differences in the reference period between the two statistical variables.

• For sickness absence man-days and the associated sickness absence rate, the reference period refers to the week that includes the 16th of each month in the quarter.
• For the statistics on the number of employees with doctor-certified sickness absence, the reference period refers to as the Tuesday in the week that includes the 16th of the last month of the quarter.

Self-certified sickness absence

In the survey on self-certified sickness absence, businesses report the number of lost man-days for all their employees, broken down by gender. Since the sample survey counts lost man-days per business rather than per individual, we cannot say with certainty that the figures for self-certified sickness absence cover only the defined population. This is not a major issue in practice, as self-certified sickness absence accounts for only a small part of total sickness absence, and the population covers a large share of the working population. Due to difficulties in obtaining reliable information on self-certified sickness absence, the sample survey does not include businesses from the following industries:

• private households with employed persons
• extra territorial organizations
• investment companies
• extraction of crude petroleum and Natural gas
• water transport
• animal husbandry service activities (except veterinary activities).

The production of sickness absence statistics is based on three different data sources. The statistics on self-certified sickness absence come from a sample survey of around 10,000 businesses. Doctor-certified sickness absence is from Nav’s sickness certification register. Information about the number of hours employees are contracted to work, along with other at the employment, individual, and company level, is collected through the a-ordningen (English description).

The survey on self-certificated sickness absence

There is no register of self-certified sickness absence in Norway. For this reason, Statistics Norway (SSB) conducts a sample survey to estimate the number of sickness absence man-days documented with self-certification during the quarter. Although businesses bear the cost of this type of absence themselves, they must keep records of it at the individual level. This is partly because Nav covers the cost of absence after the first 16 days, when self- and doctor-certified absence days can occur together.

Doctor-certified sickness absence

Information on doctor-certified sickness absence comes from Nav’s sickness certification register. The register is based on Nav’s records of Medical Certificate 1A in the sickness benefit system. These records intend to cover all doctor-certified absences due to the individual’s own illness.

A-ordningen

As from January 2015, all information related to employment, individuals, businesses, and enterprises has been collected from the a-ordning (English description). This is a coordinated digital reporting system for employment, income, and tax deduction data given to the Norwegian Tax Administration, Nav, and Statistics Norway (SSB).

It provides, among other things, information on the start and end dates of employment, agreed working hours, employment percentage, industry, and occupation. In addition, SSB supplements this with information from other registers, including the Central Coordinating Register for Legal Entities and The Central Register of Establishments and Enterprises (BoF).

According to § 4 of the a-opplysningsloven (Act on Reporting of Employment and Income Information), employers must report an a-melding at least once per month. The reporting obligation applies when salary and/or expense reimbursements paid to an individual exceed NOK 1,000 per year.

Before 2015, this information came from the Employee Register (Arbeidstakerregisteret). It included employment relationships expected to last at least six days and involved an average of four or more working hours per week. Employers reported new and ended employment relationships to Nav, as well as absences exceeding 12 months and leaves lasting more than 14 Days.

The reporting obligation under the a-ordning is stricter than under the former Employee Register. As a result, employers report more employment relationships after the transition to the a-ordning, even employment relationship without wages. For this reason, the sickness absence statistics use the definition of active employment relationships, which requires that wages in the employment relationship during the period.

The survey on self-certificated sickness absence

Sample selection

Each year, we draw a random sample of approximately 10,000 businesses from the Central Register of Establishments and Enterprises (BoF). This includes stratifying the businesses by:

• industry (2- digit NACE Rev. 2 classification. With merging of too small groups)
• size (number of employees, dived into five groups)
• county of the workplace.

This approach ensures good representation across all industries, among both large and small businesses, and across all counties in Norway.

Since 2015, NORSAMU (using Statistics Norway’s coordinated sampling system) draws the sample, to distribute the overall reporting burden among businesses as fairly and predictably as possible.

All businesses within the same industry and size stratum have the same probability of selection. The selection probability increases with business size, selecting no small businesses (fewer than 4 employees), while selecting all large businesses (more than approximately 150 employees). In this way, the survey covers around 35 percent of all employees, while only questioning about 5 percent of businesses. The sample achieves balance across counties by applying the same national selection probabilities within each county.

Once a year, part of the sample rotates, by removing some businesses and adding others. The largest businesses remain in the sample continuously, while other selected businesses participate for about four years at a time.

The introduction of the industry standard SN2007 led to changes in the stratification of the sample starting in 2009.

Reporting

Questionnaires are sent out during the last week of each quarter to enterprises that have selected establishments under them. The businesses must return the completed forms within a deadline of about four weeks, through Altinn.

In the questionnaire, businesses report how many people, cases, days, and full man-days of self-certified sickness absence their employees have had during the period.

Estimation

Sample data on self-certified sickness absence for men and women are scaled to the population level using stratified model-based rate estimators. These correspond to linear regression models without intercepts, with variance proportional to the number of employed men (women). The explanatory variable is the number of employed men (women) per business, sourced from the a-ordning. The dependent variable is lost man-days due to self-certified absence among male (female) employees. Relationships are estimated via weighted least squares within industry and size strata, separately for men and women. Population totals are predicted by multiplying the known total number of employed men and women in each stratum by the estimated rates. The combined figure is the sum of predicted totals for men and women.

Estimation of small businesses

Since we exclude the smallest businesses from our survey, we assume that average absence per employee is the same in the smallest and second-smallest strata.

The estimation involves multiplying the average rate of the second-smallest group by the number of employees in the smallest stratum.

Estimation and groups:

• Education: Analyses have previously shown a strong correlation between education level and self-certified sickness absence. This information is now included in the estimation method. A measure of the company’s education level is used along with industry information to estimate self-certified sickness absence. Education level is calculated as a dichotomous variable indicating whether more or less than half of the employees have higher education.
• Industry level: Instead of calculating figures at the 2-digit NACE level, small businesses use 19 groups (broader industry categories).
• Sector: Sector based estimation within the industry group "public administration", as variation between sectors within this group was high.
• Outliers: Use of established methods to detect whether a unit is an outlier. For more information, see: Information about the KOSTRA package (GitHub).

Editing

Editing refers to the process of checking, reviewing, and correcting data.

After businesses send the forms to Statistics Norway (SSB) through Altinn, we perform tests to find systematic and logical errors in the reported data. Around 10% of the sample reports their figures automatically through their payroll and personnel systems (LPS), The remaining businesses report manually by completing and sending the form in Altinn. We evaluate the automatically sent figures for systematic errors and the manually sent data we perform logical checks to detect outliers and inconsistencies in the responses. For found errors, we either contact the respondents, remove the business from the sample, or make a discretionary assessment.

A common type of error across SSB’s business surveys is that the respondent does not know which unit they are supposed to report for. This often results from a lack of understanding of the difference between an establishment and an enterprise (legal entity). To minimize this type of error, the number of male and female employees is pre-filled in the form, based on what the business reported to the a-ordning for the first month of the quarter in question.

The Sick Leave Register

Nav owns the Sickness Certification Register for doctor-certified sickness absence. Nav continuously updates all medical certificates issued by general practitioners or hospitals linked to an organization number in the register. Nav performs various checks and corrections on the medical certificates, particularly related to overlapping or inconsistent dates.

A-ordningen

The a-ordning, includes checks at two stages:

• Business rules (validation checks) at the Norwegian Tax Administration’s intake system
• Statistics Norway’s production system for wage and employment data

The deadline for employers to submit reports to the a-ordning is the 5th of the following month. Once the Tax Administration receives the a-melding, they apply a series of business rules to detect errors or omissions in the given information.

In Statistics Norway’s (SSB) production process, we conduct checks and automated procedures to ensure data quality for statistical purposes. A part of this is removing inactive employment relationships. That is jobs without registered wages in the reference week (the week of the month that includes the 16th).

This may apply to seasonal workers who did not work during the reference week (and therefore did not receive wages), cases where end dates for terminated jobs were not reported (in error), and instances where individuals were mistakenly reported as having an active job (e.g., people on substitute lists who did not work during the period).

As of the release of data for the 1st quarter of 2021, a new method for calculating working hours (Norwegian) has been implemented, applied retroactively back to the 1st quarter of 2015. This change primarily involves adjusting agreed working hours/position percentages for jobs where the reported working hours from employers through the a-ordning are missing or incomplete. A more detailed description of the method is available in the article linked above. For the sickness absence statistics, this results in a change in the sickness absence rate through adjustments to both the number of lost man-days (the numerator of the sickness absence rate) and the contractual man-days (the denominator). The effect on the number of self-certified sickness absence man-days is small, but it influences the self-certified sickness absence rate through changes in the denominator (contractual man-days).

To compare quarters in the same year we adjust for seasonal components in the data by using the X-13ARIMA-SEATS software. We adjust indirectly, by separately seasonally adjust the numerator and denominator for gender divided sickness absence rates certified by doctor and for self-certified sickness absence rates by gender.

Including grouping the total sickness absence rate by:

• County (as of 2020)
• Industry (SIC2007)
• Immigration Category (only for doctor-certified sickness absence rate)

The most recent series are seasonally adjusted only from 2015 onwards, following the introduction of the a-ordning.

See the section About Seasonal Adjustment below for more detailed information about the seasonal adjustment of the sickness absence statistics.

Interviewers and everyone who works at Statistics Norway have a duty of confidentiality. Statistics Norway has its own data protection officer.

Statistics Norway does not publish figures where there is a risk of identifying individual data about jobs, persons or businesses.

The suppression method is used in these statistics to ensure this.

More information can be found on Statistics Norway’s website under Methods in official statistics, in the ‘Confidentiality’ section.

The effect of the a-ordning on the sickness absence rate

Introduction of the a-ordning to replace the Register of Employees and Employers (Aa-register), from the 1st quarter of 2015, gives changing figures for the number of employees. This is due to a broader coverage of the group, and a better quality. The transition to the new data has had a small, but still observable effect on the sickness absence rate at the aggregate level. The effect of the transition to the new data is small also when it comes to self-certified and doctor-certified sickness absence separately. But since the self-certificated absence is lower than the doctor-certificated, this small effect provides a greater impact on the numbers of alteration for self-certified absence than for doctor-certified absence, and for the total absence.

Adjustment of the break in the time series due to the a-ordning from 2015 is only done for the series with seasonal and influenza adjusted sickness absence rates.

We have not calculated the size of breaks related to the a-ordning for other variables or groups, but the size may vary between groups. Of importance may be the fact that the a-ordning captures more of the small job relations, than the Aa-register did.

Also, we use direct reporting of position proportion is now, while Aa-register only gave information about contractual working hours, and one assumed that 37.5 hours per week equalled a full position. This may also have affected the figures, especially for industries with a high degree of shift and rotation.

The shift from the Aa-register to the a-ordning at the turn of the year 2014/2015 resulted in a break in the time series. In addition, Statistics Norway (SSB) changed the reference point from the last week of the month to the week containing the 16th. In the 2nd quarter, this change contributed to an increase of approximately 0.2 percentage points.

Change of industry classification

The industry classification SIC2007 was introduced from the publication of the 1st quarter of 2009. Compared with earlier published figures classified by the previous industry standard SIC2002 there will be a break in the time series.

Municipal and regional reform

As of the first quarter of 2020, the statistics are affected by the municipal and regional reform which was implemented on the 1st of January 2020. With the reform, a few enterprises in the former municipalities/counties ceased and employment relationships in these were transferred to existing or newly established enterprises in the new municipalities/regions. We have not produced any new StatBank tables with new regional classification for years prior to 2020.

Sector

• The general government consists of the local and central government.
• The public sector consists of local government, central government, and public corporations.
• In our figures, the private sector and public corporations are combined in most cases, as the public corporations resemble companies in the private sector more than those in the general government.

Industry

All industries are included either directly or by estimation. The following industries are estimated for self-certified sickness absence as they are not included in the sample survey (SIC2007):

Support activities for animal production (01.62), Extraction of crude petroleum and natural gas (06), Water transport (50), Vocational rehabilitation activities for unemployed persons (88.993), Individual adapted work (88.994), Activities of households as employers of domestic personnel (97) and Activities of extraterritorial organisations and bodies (99)

These are not included in the survey due to challenges in obtaining reliable information.

Since 2015, the statistics on sickness absence have been based on medical certificates from Nav and the a-ordning as sources. Both sources are of good quality, but errors may occur. For the a-ordning, for example, there may be errors in the reporting from obligated parties about employees, while for medical certificates, errors may occur during the completion of forms by the person issuing the sick leave. In addition, errors may occur in the processing of data afterward. We correct such errors in the statistics when discovered.

Payment during the employer's period

We are unable to link self-certifications and medical certificates together at the individual level, as we only have data on self-certified sickness absence at the enterprise level for a selection of enterprises. This means that we do not know whether a medically certified absence is the first day the person is sick or, for example, the ninth day. This also means that we cannot accurately distinguish between how many days are paid by the employer and how much is paid by Nav.

Medically certified sick leave is a full count, while for self-certified sick leave, there is a response rate of over 90 percent after the obligation to report was introduced in the 2nd quarter of 2003.

Variance (applies to self-certified sickness absence)

It is possible to calculate two uncertainties for self-certified sickness absence. As this is a sample survey, there is a sampling uncertainty. There is also a bias due to special consideration by removing the smallest businesses to reduce their reporting burden.

The sampling uncertainty is between +/-0.01 and 0.07 percentage points for the business sector, depending on both seasonal variation and industry.

The table below shows the standard error multiplied by 1.96, distributed by gender and industry (SN07) for the sick leave percentage for 2022. You can thus construct your own confidence interval according to gender and industry as desired. You can then say that if all businesses had been surveyed, instead of a sample, one would be within that interval with 95 percent certainty.

In addition to this, there is some uncertainty related to choices in the model. The smallest businesses (fewer than 4 employees) are excluded from the sample but are included in the estimation of the total. Therefore, it is reasonable to assume that the uncertainty for small businesses is greater. Thus, it is not an unbiased estimator, but the bias is quite small because the sample is of a reasonable size, and the consideration of the businesses' reporting burden weighs heavily. In total, uncertainties can lead to an error estimate of up to 0.13 percentage points, but usually around +/- 0.06 percentage points for the smallest businesses.

Nonresponse errors (applies to self-certified sickness absence)

These are errors that may occur when units choose not to respond for various reasons. We have calculated the effect of nonresponse errors over time, and we found no systematic biases due to different nonresponses.

Processing errors

Processing errors are errors that occur during SSB's data processing. Typical examples of this are assessing correct figures as incorrect and are erroneously corrected. Electronic reporting from Nav and the a-ordning for medically certified sickness absence reduces this risk.

Coverage errors

Coverage errors occurs when the population we measure deviates from the one we want to measure. Around 5 percent of all sick notes in 2022 could not be linked to an employment relationship. Analyses suggest that the vast majority are notes that do not apply to reportable employment relationships and missing from the data set, which leads to some under coverage.

Up to and including 2014, we calculated sickness absence man-days using sickness absence man-days and man-days, combined with a calculated the percentage of full-time equivalent. We used data from the LFS for weather the employment relationship was full-time, short, or long part-time. This generalization resulted in errors at the individual level but less so at the aggregated level.

From 2015, the a-ordning has provided information both job percentage and the number of hours worked per week for each employee. In some cases, data quality remains a challenge, particularly for hourly workers, where reporting errors are common and standard values such as 0 or 100 are sometimes used incorrectly.

To increase data quality has Statistics Norway developed a method that test the relationship between reported job percentage and salary. Where the relationship is good (about 85 percent of cases) we use the reported value. Where there is no reasonable relationship, a new job percentage is calculated using the number of paid hours or salary. Experience after five years with the a-ordning suggests that where there are deviations, the reporting of salary is usually correct. More about the method is available here (in Norwegian).

A revision is a planned change to figures that have already been published, for example when releasing final figures as a follow-up to published preliminary figures. See also Statistics Norway’s principles for revisions.

Revisions in previously published seasonally adjusted figures can take place when new observations (or revised previous observations) are included in the basis of calculation. The scope of the revision is usually greatest in the most relevant part (last 1–2 years) of seasonally adjusted time series. A corresponding revision in trends is also typical, particularly at the end of the time series. The extent of the revision of trends and seasonally adjusted figures is partly determined by the revision policy, see Section 4 of the European Statistical System (ESS) Guidelines on Seasonal Adjustment on the Eurostat website.

For more information on the revision of seasonally adjusted figures, see the About seasonal adjustment section in the relevant statistics.

Monthly and quarterly time series are often characterized by considerable seasonal variations, which might complicate their interpretation of change from one period to the next. Once data have been adjusted for seasonal effects by X-13ARIMA-SEATS or some other seasonal adjustment tool, they are far easier to compare between periods.

For more information on seasonal adjustment: metadata on methods: seasonal adjustment.

The quarterly statistics for both self- and doctor-reported sickness absence show seasonal variation. Lost man-days due to both self- and doctor-reported sick leave are significantly higher in the winter half of the year (quarters 1 and 4), partly due to more frequent respiratory infections.

Both lost and contractual man-days are strongly affected by the number of man-days in the quarter, which is the number of weekdays in the quarter minus public holidays.

Series with seasonal adjustment

The following quarterly time series are seasonally adjusted separately for men and for women:

• Sickness absence man-days documented with a doctor’s certificate.
• Self-certified sickness absence man-days
• Contractual man-days

We publish seasonally adjusted figures for lost man-days due to sick leave in relation to contractual, which is what we call the sick leave rate. This is divided into self-certified, doctor-certified, and total sickness absence.

In 2023 and 2024, several seasonally adjusted figures were published:

• Industry (SN07) and gender
• County of workplace (counties applicable from 2020) and gender
• Doctor-certified sickness absence rate, gender, and immigration category
• In addition to the sickness absence rate, we also expanded with seasonally adjusted figures for sickness absence man-days, broken down by gender and type of sickness absence.

Prior to conducting seasonal adjustments, it is crucial to pre-correct the raw data for calendar effects and outliers. This entails implementing a detailed process that uses customized models, which are not accessible as standard options within the seasonal adjustment tool.

Time series for doctor-certified sickness absence man-days with an influenza diagnosis by gender are seasonally adjusted. The irregular component from these time series is then used as an explanatory variable in the pre-correction. The purpose of this is to ensure that irregular fluctuations in influenza related absence are not misinterpreted as systematic seasonal variation in the seasonal adjustment. We apply the irregularity in influenza-related absence only by gender as an explanatory variable in the modelling of sickness absence, both broken down by different industries and counties.

Self-certified sickness absence is also pre-adjusted using an influenza variable based on doctor-certified sickness absence man-days with an influenza diagnosis, even though there is no direct connection. Statistical tests show that the pre-adjustment variable has a strong effect on doctor-certified sickness absence and a weak but significant effect on self-certified sickness absence.

Calendar adjustment

Calendar adjustment is performed on all series showing a significant and plausible calendar effect using a robust statistical approach, such as regression or the RegARIMA procedure (a regression model wherein the noise term is modelled by an ARIMA model). Calendar regression variables are processed following Norwegian public holidays.

Methods for trading/man-day adjustment

Average man-days per quarter are calculated based on the period from the 1st quarter of 2015 and one year ahead. Then, the difference between the number of man-days in the quarter and the average number of man-days in the respective quarter of the year is calculated. This is one of the explanatory variables in the modelling of sickness absence.

Adjustment for movable public holidays

The time series are pre-adjusted for the number of man-days in the quarter. The number of man-days in the quarter refer to the number of weekdays (Monday to Friday) in the quarter minus Norwegian public holidays and official holidays in the quarter that fall on weekdays. This includes the following movable public holidays: Maundy Thursday, Good Friday, Easter Monday, Ascension Day, and Whit Monday.

A separate Easter variable is not employed due to its lack of significance when data are also pre-corrected for the number of man-days in the quarter. Pre-correcting for Easter instead of the number of man-days yields poorer results, as determined by conducted tests. A national calendar is utilized in this process.

The Norwegian national calendar is used.

Treatment of outliers

The series are examined for extreme values, and any identified extremes are explained or modelled using all available information. When a clear understanding of the cause of the extreme values is established (e.g., strikes or consequences of policy changes), they are incorporated as regressors within the model.

Model selection

To pre-correct, it is essential to select an ARIMA model and determining whether the data should undergo log transformation is essential. The model is automatically chosen based on established routines within the seasonal adjustment tool.

Selection of Seasonal Adjustment Method

X-13ARIMA-SEATS

Consistency Between Raw Data and Seasonally Adjusted Figures

In certain cases, it may be desirable for the sum (average) of quarterly seasonally adjusted figures for a year to match the sum (average) of quarterly figures from the original raw series. However, no consistency conditions are imposed in this context.

Consistency Between Aggregates/Definitions for Seasonally Adjusted Figures

For some series, consistency is enforced between seasonally adjusted totals and sub-aggregates. Furthermore, there may be a relationship between different series, such as gross value-added equating to output minus intermediate consumption.

Equality between seasonally adjusted sub-aggregates and over-aggregates is enforced, with this equality automatically achieved through the indirect method.

Direct or Indirect Method

A direct method is employed when the time series for a total and its sub-aggregates are all seasonally adjusted separately. Conversely, an indirect method is used for the total if the corresponding sub-aggregates' time series are directly seasonally adjusted and then aggregated to the total level.

In this case, the indirect method is used, with the components being directly seasonally adjusted using the same approach and software. The totals are then calculated by aggregating the seasonally adjusted components.

We use direct seasonal adjustment for time series broken down by gender and type of sickness absence, since some subcategories (such as “unspecified”) in industry and county are not seasonally adjusted.

We use the indirect method for time series for “Both genders” and “Both self- and doctor certified sickness absence.” Here, we sum the seasonally adjusted total time series for men and women, and for self-certified and doctor-certified sickness absence.

Time Horizon for Model Estimation and Calculation of Correction Factors

We use the entire time series for calculating the model and correction factors for the whole country as well as self-certified and doctor-certified sickness absence.

For industry and county, the time series from 2015 onwards is employed for model estimation.

General revision policy

In the context of seasonal adjustment, revisions may occur due to the introduction of new observations or alterations in raw data. Various methods exist for addressing these revisions during the release of statistical information.

The figures up to 2015 is not revised. Seasonally adjusted data from 2015 undergo revisions in compliance with well-defined, publicly accessible revision procedures and release calendars.

Concurrent versus current adjustment

Partial continuous adjustment is employed, where models, including any log transformations, seasonal and trend filters, calendar variables, and outliers, are identified on an annual basis. Meanwhile, respective regression parameters and factors are continuously re-estimated whenever new or revised raw data become available.

Time Horizon for Publication of Revised Figures

The series from the 1st quarter of 2015 is revised when seasonal factors are re-estimated. Consequently, users may observe that published figures appear to change; however, this is an inherent characteristic of an estimated series.

Continuous/periodical evaluation using standard measures proposed by different seasonal adjustment tools.

For most of the series, the quality is evaluated trough iterative processes using a combination of graphical tools and statistical tests.

The table below provides some indicators of the quality of the seasonal adjustment:

Quality indicators (Excel)

For more information on the quality indicators in the table see: metadata on methods: seasonal adjustment

Impact of the a-orningen on Sickness Absence Rate

The a-ordning replaced the Aa-register starting from the 1st quarter of 2015. This has resulted in different figures related to employees. This is due to a slightly broader coverage of the group, as well as improved data quality. The position percentage for each job from the a-ordning was also used directly in the calculation of contractual man-days. Previously, we only had agreed working hours from the Aa-register and made a rough assumption about what constituted full-time (37.5 hours per week for everyone) to construct man-days. Nevertheless, the transition has had a small but noticeable effect on the sickness absence rate even at the aggregated level. This has been modelled and pre-adjusted in the seasonal adjustment tool.

Addressing the Coronavirus Crisis in 2020

The coronavirus crisis affects the sickness absence statistics in several ways that are important to consider when interpreting the figures.

In the seasonally adjusted data, we have followed Eurostat’s guidelines which state that the effects of the coronavirus crisis should not be included in the basis for the seasonal pattern.

This means we assume that the seasonal pattern remained unchanged during the coronavirus period (from the 1st quarter of 2020 to the 1st quarter of 2023), and we correct for systematic seasonal variation based on pre-pandemic data.

After the 2nd quarter of 2023, we ended the exception period for coronavirus handling in the sickness absence statistics and allowed new observations to once again influence the calculation of seasonal factors in our time series.

If the seasonal pattern after the coronavirus period is different from before, as the sickness absence figures may suggest, this will lead to changes in the seasonal factors. This entails increased revisions of the seasonally adjusted figures during a transition period.

Data availability

Raw and seasonally adjusted data are available.

Historical are available to enable revision analysis. Both seasonally adjusted data and seasonal, calendar, influenza and a-ordning break adjusted data are available in our StatBank.

Press releases

In addition to raw data, at least one of the following series is released: pre-treated, seasonally adjusted, seasonally plus man-day adjusted, trend-cycle series.

For each series, some quality measures of the seasonal adjustment are presented.