Official name of the survey/data source:
Income and Property Distribution Survey

LIS name:
Norwegian Income Distribution Survey - 1991

Administrative unit responsible for the survey:
Central Bureau of Statistics.

The first income distribution survey in Norway was conducted in 1958. The scope of the survey was, however, rather limited. It did for instance only include taxable income from the tax return, and did not include any tax-free transfers. The household definition was restricted to only cover married couples and singles. Adult children still residing at home where, furthermore, considered to be separate households. The same scope and definitions where utilised in later surveys in the 60s (1962, 1967). The surveys of the 60s only exist as written reports, and not as micro-data.

In the 70s the Income Distribution Surveys was conducted every three years (1970, 1973, 1976, 1979). The household definition was still restricted to only cover the family unit (i.e. persons having common address and common family name). Persons living in a consensual union where, for instance, considered to be two households. Up until the 1979 survey, only taxable income was included in the survey. For all the surveys of the 60s and 70s the household definition was strictly based on information from administrative records and no household interview was ever conducted.

The Income Distribution Survey was reorganised in 1982. The household definition was changed to include all persons having common board and living in the same dwelling. The household composition was, furthermore, established after a household interview. The income concept now included both taxable income and social transfers not liable to taxation (for instance, family allowance and scholarships). These changes made it possible to construct disposable household income. The Income Distribution Survey has been conducted annually since 1984.

B POPULATION AND SAMPLE SIZE, SAMPLING METHODS  back

The Income Distribution Survey 1991 covers all persons except people living in institutions. People with foreign citizenship are included if they are registered in the Population Statistics System. They are normally registered here if they have a working and residence permit. Children born during the year are included, so also people that died during the year. The sampling frame is the Central Population Register.

The sampling procedure for the 1991 survey was the same as for the 1986 survey.

The IDS has information about persons and households. Household information are gathered through interview. If household information is missing, for instance if a household decline to participate in the survey, they never the less are included in the survey, but household information is then substituted with information on family composition from administrative registers (The Population Statistics system). This means that there is no non-response in the IDS.

It should also be noted that in almost 90 percent of all cases there is full overlap between actual household composition and family composition from the register. The response rate of the household interview was ca. 70 percent.

The sample of the 1991 IDS consists of the following 4 subsamples:

(1) The Household Panel of the IDS (2512 households)
(2) The rotating panel '91 of the HBS (482 households)
(3) The rotating panel '92 of the HBS (483 households)
(4) A special sample of self-employed (4595 households)

Both the Household Panel (1) and the rotating panel from the Household Budget Survey (2, 3) was drawn from the Central Population Register. The sample of self-employed (4) was selected on the basis of information from the Census 1990 and Tax Register of 1990 (persons with SE-income).

C. MEASURES OF DATA QUALITY   back

Comparison between estimates from the Norwegian Income Distribution Survey 1991 and external sources

The main strength of the Norwegian Income surveys is the use of administrative registers in the reporting of income. One obvious advantage is, for instance, that there is no non-reporting of income in the survey. Any deviance between survey estimates and administrative registers is therefore caused either by sampling errors or difference in population (e.g., the exclusion of the institutionalised in the survey but not in the registers).

As a measure of data quality the survey estimates have been compared with the National Accounts and with administrative registers.

Table 1 compares survey estimates for some main income components with the adjusted household sector of the National Accounts for 1991 (R›stadsand & Heldal, 1994). In order to compare with the IDS, the NA have be adjusted by deducting income from Consumer organisations. Further adjustments have been made by deducting income items where the NA definitions and book-keeping differ significantly from the income survey. This concerns, for instance, the treatment of pension and life insurance, health benefits and certain intra-household transactions (e.g., alimonies etc.). The figures from the NA after these adjustments are presented in column A in the table.

However, there are still differences between the NA and the IDS that can be explained due to different book-keeping and definitions. These explained differences are reported in column B in the table. In the treatment of wages and salaries, for instance, one difference that can be explained is that in the IDS these income items are classified as wages, whereas the NA (more correctly) consider them to be transfers. Similarly there exist differences that can be explained and accounted for in the treatment of other income components as well.

After adjusting for explained differences there are only some minor differences between the NA and the IDS (the difference between columns D and C).

(a) The wage amount is somewhat higher in the IDS, compared to NA, but a reason for this is that the sum of wages is underestimated in the NA.

(b) For self-employment income the amount in the IDS is 82 percent of the level in the NA. Part of the difference may be explained by different ways of calculating consumption of fixed capital.

(c) The coverage of property income is 93 percent of the level of the adjusted NA. Some of the remaining difference may be explained by the fact that the IDS only covers private households, and not people living in institutions.

(d) The IDS covers 89 percent of the level of transfers of the adjusted NA. Some of the differences can again be explained by the fact that the IDS only covers private households, and not people who live in institutions.

(e) In respect to taxes, total income and income after tax there is almost full overlap between the IDS and the adjusted NA.

In Table 2 we present the estimated figures for some social transfers from the survey, and the corresponding figures from administrative sources.

The survey estimates of family allowance is very close to the administrative register. For the other transfers (unemployment benefits and social assistance) the survey estimates slightly exceeds the amounths from the administrative registers. There are, however, very few observations in the sample receiving these benefits.

References

Heldal, J. (1992): A Method for Calibration of weights in sample surveys. Working Paper No. 3. Divison for Population and Living Condition. Statistics Norway.

Røstadsand, J. I. & Heldal, J. (1994): «Hva kan inntektsundersøkelsene for husholdningene fortelle oss? Om bruk av utvalgsundersøkelser i nasjonalregnskapet». Økonomiske analyser. nr. 2. Statistisk sentralbyrå.

Table 1.
Comparison of survey estimates from the Norwegian Income Distribution Survey (IDS) to National Accounts Aggregates. Million kroner. 1991 

Source: Røstadsand and Heldal (1994)

Table 2.
Comparison of some survey estimates with administrative registers. Million kroner. 1991

D. DATA COLLECTION AND ACQUISITION  back

The IDS collects income and property data from the Tax Return and from administrative registers. Additional administrative registers provide biographical information such as level of education and marital status. An obvious advantage of using administrative registers is that there will be no item-nonresponse.

Information on taxable income and property is collected from the local Tax Authorities. For each person in the IDS sample Statistics Norway received a duplicate of his or her Tax Return form. These forms were later processed within Statistics Norway. In addition several other administrative registers provided information on tax free social transfers.

In 1991 the IDS collected the following information from the following sources:

1. The Tax Return :
The Tax Return provides information on all kinds of taxable income(s). The IDS collect all information available from the tax return form. Among these are :

• Wages and salaries
• Entrepreneurial income
• Property income
• Taxable income from public and private pensions
• Alimonies
• Interest paid
• Deposits in domestic banks
• Shares in domestic and foreign companies
• Real property
• Debt

2. The Population Statistics System:
This register provides important background information on all persons residing in Norway. The most important variables extracted from the PSS to be included in the IDS are:

• Family number
• Municipality
• Sex
• Age
• Marital status 

3. The Tax register for personal tax payers:
This register covers all individuals above the age of 12, irrespective of whether they have any taxable income or not. Some of the basic variables included in the IDS are:

• Assessed income and property taxes
• Occupational income
• Parent's tax deductions 

4. The State Educational Loan Fund:
This register covers all current students receiving support from the State Educational Loan Fund. In addition, former students paying interest and repaying debt are included. The following variables are included in the IDS:

• Scholarship (annual amount received)
• Loan (annual amount received)
• Debt by the end of the year

5. The State Housing Bank
This register provides information on:

• Annual amount of dwelling support received.

6. Social assistance register
This register gives information on annual amount received of:

• Social assistance payment

7. The National Insurance Administration
All persons receiving some kind of benefit from the National Insurance System are registered in the National Insurance data base. This register makes it possible to distinguish different types social security recipients. The following variables are included the IDS:

• Pension type (i.e. old-age pension, disability pension, survivor's pension, etc.)
• Disability status
• Basic grants
• Assistance grants
• Family allowance

8. The Register for End of the year Certificates
This is a register that provides detailed information on different kind of wage and salary income, as well as taxable social security benefits. This register provides more detailed information on both wages and social security benefits than the Tax return register. In the IDS the following variables have been included:

• Unemployment benefit
• State subsidised pension pursuant to agreement
• Rehabilitation allowance

9. Statistics Norway's Educational Register
This register supplies information on the highest level of completed education for all individuals, as well as education that is still going on. The following variables are included in the IDS:

• Highest Level of completed education
• Type of education (main subject, etc.)

10. Statistics Norway's Central Register of Companies
This register provides information on industry.

11. The register of Employers and Employees
This register gives information on name, addresses and municipality of the employer.

E. WEIGHTING PROCEDURES  back

The Income Distribution Survey of 1991 introduced a new method of weighting. After estimating weights on the basis of the traditional Horwitz-Thompson method, the weights in the IDS was adjusted using a method of calibration. This method, in brief, constructs new weights by the use of regression technique in order to produce estimations that for some variables are identical to known totals (from registers). The method is described in the paper by Heldal (1992).

The income variables used for calibration in the 1991 survey (i.e. those variables where the estimations are identical to known totals) are:

• Employment income, wages and salaries
• Employment income, self-employed in primary sector
• Employment income, self-employed in other industries
• Pensionable income, basis for future social security benefits
• Property tax

All the known register totals was collected from the Tax register for personal tax payers.

In addition the sample was post-stratified with respect to age and according to Population statistics.

F. DETERMINATION OF SURVEY UNIT MEMBERSHIP  back

G. CHILDREN AND SPOUSES  back

H. AVAILABILITY OF BASIC SOCIAL AND DEMOGRAPHIC INFORMATION  back

I. AVAILABILITY OF LABOUR MARKET INFORMATION  back 

J. AVAILABILITY OF GEOGRAPHIC INFORMATION  back

K. SOURCES AND AMOUNTS OF CASH INCOME  back

L. TAXES  back

M. BIBLIOGRAPHY OF MAIN PUBLICATIONS back