GENERAL EXPLANATION ON HOUSEHOLD LABOUR FORCE SURVEY

History

Since the first five-year development plan, information on the economically active population and its components, although on a limited basis, had been periodically collected in Turkey through population census, which were conducted every five years. Since 1966, additional data had been obtained on the basis of labour force survey between the census years. However, data gathered with the labour force survey, remained fragmentary and could not be easily compared across time because of differences in questionnaire, geographical coverage, definitions and concepts until the end of year 1985.

In 1988, the survey was re-designed to measure labour force improvement through time and to supply international comparability of data, using the latest international standards (ILO, Thirteenth International Conference of Labour Statisticians, 1982) concerning statistics on the economically active population, employment, unemployment and underemployment.

The aims of Household Labour Force Surveys are preparing the questionnaire and the tabulation plan to view labour market in different aspects, ending the concepts chaos in labour market and constructing the definitions to collect figures, which are internationally comparable.

For this reason the basic concepts and definitions accepted in the Thirteenth International Conference of Labour Statisticians have been used in the Household Labour Force Surveys which had been conducted since October 1988. In this series, the last week of April and October starting with Monday and ending with Sunday was taken as the reference week and survey was applied in the fifteen days following the reference week. The results of the surveys have been published by Turkey, urban and rural areas.

While a total of 11,160 sample households were interviewed with the method of substitution in each application of HLFS from October 1988 to April 1994, the sample design was changed in the October 1994 HLFS application. Beginning with this application, substitution method was given up and the number of sample household was increased to about 15 000.

In 1999, the survey was evaluated and re-designed in the meeting with the participation of national and international experts.

In this direction, in the year 2000, application frequency, sample size, estimation dimension, questionnaire and some other aspects of HLFS were changed. These changes were made taking into account the need of making comparisons with the existent series.

It is aimed to increase the sensibility of estimations of the survey with the new design. The sampling size was increased to 23 000 household and was used firstly in October 1999 HLFS application. Since the year 2000, the sampling size has been about 23 000 household in quarter.

Selected households are visited four times within the sixteen months with the new sampling design. Consequently, it would be possible to observe activities of household members in time and to measure effects of economic developments on household members. For this aim, some questions were added to the HLFS questionnaire. Another change in the questionnaire was about informal sector employment. The share of employed persons in informal sector to total employment is increasing in Turkey. Questions about domestic workers, legal status of workplace, type of taxation were added to the HLFS questionnaire in order to determine regular information on informal sector. In addition, a question about "time for seeking job by using any active method" which is one of the unemployment criteria was added to the questionnaire in order to calculate the number of alternative unemployed.

Since the year 2000, the fixed reference period was given up and the seven days preceding the application day of the HLFS have been used as the reference period. In the new series, the field application of the survey is initiated on the 8th day of each month.

Another changes in the new series started in 2000 were about the age criterion used in the survey results. From the beginning of the year 2000, the results of the survey have been given for the population 15 years old and over instead of 12 years old and over. In addition, since 2000, employed and unemployed who had worked before, have been classified according to the International Standard Industry Classification 1990. Rev.3 and since 2001 they have been classified according to the International Standard Classification of Occupation 1988.

Weighting

Up to 1990, 1985-1990 General Population Census results, for 1991 and after 1990 General Population Census and 1997 Population Count results, were used for weighting of HLFS results. The necessity of the revision of HLFS results according to the 1990-2000 population projection has been inevitable because of the detailed results of 2000 General Population Census have been available and also there are important differences between the results of 2000 General Population Census and 1997 Population Count by the detail of age and rural-urban distribution. Another reason of differences in total population is being used the different hypothesis in population projections. The most important difference emerged in the rural population when used projections based on 2000 General Population Census results. In this press release, 2002 revised results are given.

It has been considered that the revised results of HLFS should be used after now on. Because 1997 Population Count was not a census and just applied for the purpose of election, it has some disadvantages. Moreover 2000 General Population Census gives the most current results and reflects the demographic structure of Turkey in a better way. The revised results of HLFS belong to 1991-2000 period will not be published, but it would be possible to reach these results through “Household Labour Force Database” existing in the SIS web site.

The new sample design of HLFS has been designed to give estimations on the level of Turkey, urban, rural, seven geographic regions (total, urban, rural) and selected nine province centers (Adana, Ankara, Antalya, Bursa, Erzurum, Gaziantep, Istanbul, İzmir, Samsun). According to this design, the annual data set was consisted by combining the data set of four rounds. The yearly estimation results of HLFS were re-weighted by using mid-year population projections.

Sample size and non-response rate

According to the press release of 2002 HLFS, the estimation results were given for Turkey, urban rural, seven geographic regions and selected nine province center because of the sufficient sample size.

A total of 96 802 sample households, 75 495 of which from about 163 urban areas and 21 307 of which from about 289 rural areas were used in the data set of 2003. 78 174 households were interviewed and for 18 628 households non-response forms were filled in due to the reasons like rejection, not being at home, lack of communication etc.

In the year 2003, the annual non-response rate was 7.2% in rural areas, 12.4% in urban areas, and 11.2% for whole Turkey.