Official name of the survey/data source:
Income distribution survey 1991

Administrative unit responsible for survey:

Copies of the codebook and other documentation can be obtained from at the above address.

The Income Distribution Survey is conducted once a year.

The first year in which the survey was conducted was 1977, and was first made available as microdata in 1986 (complete and comparable data).

The main purpose of the survey is to measure household and individual income and its distribution. The survey has been funded by the budget of Statistics Finland.

The main users of the survey have been Statistics Finland, various researchers, e.g. at the Government Institute for Economic Research, The Ministry of Finance, the Ministry of Social Affairs and Health.

The process of data collection (i.e. field work) LIS data was begun in January 1992 and completed in May 1992.Most of register sources have however been merged later to the interview data (late 1992, early 1993).

The public has access to the data only under the condition that the data are used for research or statistical purposes only.

A bibliography of the most important publications based on the survey is provided in Section M.

B. POPULATION AND SAMPLE SIZE, SAMPLING METHODS back

The sampling frame for the survey consists of the Finnish taxation register for 1989. The sampling frame includes the total population of household heads.

However, persons (most often not heads of household) who became 16 after 1989 and people living permanently in institutions or abroad (prison, long-term hopsital care or at leat 6 months abroad) have been excluded from the sampling frame.

Once the sample was selected from the sampling frame, all sampling units were eligible for interview.

1. The Sample Design

In principle a rotating two year panel design. The sampling procedure in each panel is following: At first a stratified list sample of individuals 12 strat were formed according to crude socioeconomic status and income level. The household memebers were subsequently gathered round this "sample persons" by using a unique housing code. That resulted in

a. A stratified PPS sample of households
b. A cluster design when considering the individual household members.

Some populations groups have been oversampled, such as farmers, other entrepreneurs and other high-income groups.

A total number of 11754 (unweighted, weighted= 2 200 200) repondents were actually interviewed from the 14477 sampled. A total number of 997 households were not included because of oversampling.

C. MEASURES OF DATA QUALITY back

1. Item and Group Nonresponse Rates

The survey design is two year rotating panel design. Therefore the response and non-response rates must be calculated separately. All figures are unweighted:

2. Reporting and Under-Reporting and Income Data Quality

a) Due to sampling procedures applied and due to non-response, families with children have been overepresented. Furthermore, the age distribution of individuals was slightly skew to working age people.

b) Households receiving living allowances are badly underrepresented

c) Undercoverage in income from interests and dividends (about 33% of the national accounts figure).

3. Data Corrections and Adjustments

Some corrections have been used to improve interest income, income in kind (grown vegetables and other food) and housing benefits. Mainly regression imputation has been used (close to "cold deck")

4. General Assessment of Income Data Quality

Good, since 90% of income items were obtained from registers. Some coverage error in the income from interests.

D. DATA COLLECTION AND ACQUISITION back  

1. Method of Data Collection

First an letter was sent to all respondents, followed by a call a few days later by the interviewer. Participation in the survey was voluntary. Privacy of the respondents was assured.

2. Structure of the Survey Instrument

Sections: Household data; data on main activities; some income items, housing (an additional questionnaire on the effects of economic recession).

3. Survey Respondent Rules

In principle, all members should have been present during the interview, however the data still contain a lot of proxy information by other household members. Respondents were asked to consult pay records, tax returns, etc. to help provide the most accurate income information possible. However, we should stress that more than 90% of income data is received directly from register sources.

E. WEIGHTING PROCEDURES back

A survey weight has been assigned to each sample case in order to represent the true population/household structure and post-stratification in order to fight non-response bias.

The sum of the survey weights do not equal the total number of units in the sampling frame, because of the age of the frame (1989).

F. DETERMINATION OF SURVEY UNIT MEMBERSHIP back

A household, the basic unit of aggregation of the survey, consists of persons either earning for the household or using their income together or abroad. People permanenetly in institutions are excluded, where as those attending military service, temporary students, etc. are excluded if they are not regarded to be members of some other household.

The head of the household was determined after the data collection and defined as the person with the highest income.

G. CHILDREN AND SPOUSES back

Children are defined as those under 18, unmarried and living with their parents. Spouses are defined as the person with the lowest income among cohabitating couples.

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 BASED ON THE CSP PANEL back

Tulonjakotilasto 1991. Income Distribution 1991 Satistics Finland (in Finnish with English Summary)

Lehtinen, Teemu 1994. Valittoman verotuksen tulonjakovaikutukset (Effects of Direct taxation on Income Distribution) (in Finnish)

Government Institute for Economic Research, Research Reports 23, Helsinki.