02 CLASSIFICATIONS

Doc. ISCO/08/07/FI 3November 2008

______

Collection of data on occupations for Finnish Employment Statistics

Document prepared by Airi Pajunen, Statistics Finland, Population Statistics

______

Workshop on the data collection of occupational data

28 November 2008

Agenda item 3.2

STATISTICAL OFFICE OF THE EUROPEAN COMMUNITIES

02 Statistical governance, quality and evaluation

Tel. (352) 4301-33622, Fax (352) 4301-33899

Collection of data on occupations for Finnish Employment Statistics

Airi Pajunen

Statistics Finland, Population Statistics

FI-00580 Helsinki, Finland

E-mail:

1 Introduction

Statistics Finland collects data on the population’s occupations with several annual surveys, mainly to serve as background data with which the respondents can be classified into socio-economic groups. Data on the numbers in each occupational category are produced by two sets of statistics, the Labour Force Survey and Employment Statistics.

The Labour Force Survey is a sample survey from which occupational data have been published since 1976. The survey is today conducted monthly with telephone interviews. Its data on occupations are based on the respondents’ own reporting and the data obtained from the sample are raised to concern the whole employed population. Employment Statistics, in turn, represent total statistics for which data are mainly collected from administrative registers. Annual data on occupations have been published from the Employment Statistics since 2004. Prior to this the data were collected in population census years.

This presentation first describes the formation and processing of data on occupations in the Employment Statistics. It then goes on to compare how the occupational data produced in the Employment Statistics and in the Labour Force Survey differ from each other. The presentation concludes by discussing the factors that cause these differences.

2 Why are data on occupations collected annually for Employment Statistics?

According to recommendations on population censuses, data on occupations are among the basic data of population censuses. Up to the year 2004, data on the population’s occupations were collected in Finland at five-year intervals in connection with the population census. The latest collection was made in 2000. However, the importance of data on occupations has grown in recent years. Especially ministries and administrative bodies need them. Combined data on the population’s occupations and education have been regarded as important for future scaling of educational supply and related anticipatory activity. There is also growing need for occupational data by region. The data on occupations obtained with the relatively small samples of the Labour Force Survey have not been able to satisfy these information needs.

This is why Statistics Finland decided to start producing occupational data for all employed persons annually. Employment Statistics have produced most of the population census data in Finland annually since 1987. Part of the reason for the late start with annual occupational data was that their production requires combining of data from a variety of different sources.

3 Sources for occupational data

The process of producing annual data on occupations for all employed persons is comprised of many phases.It rests on the Finnish annual register-based population census system in which data on basic population arederived from the registers maintained by the Population Register Centre. These registers cover all people residents in Finland. All statistical units can be linked in them by means of identification systems. A general code system, such as the personal identity code or the business identity code, makes it possible to combine directly collected data and register data.

Most of the data on occupations for the Employment Statistics are collected from diverse administrative registers. For some employees information about occupation has to be inquired direct from their employer enterprises.In addition, for some employees, especially those working in small enterprises, occupation has to be inferred from background information.

The sources used in the production of occupational data in the Employment Statistics can be roughly divided into four categories.

First, they include primary registers comprised of diverse registers of employment relationships. These are maintained by employment pension schemes.The most important of them are the Finnish Centre for Pensions, the State Treasury, the Local Government Pensions Institution and some other small information producers.The data of the Finnish Centre for Pensions cover the private sector employees who are employed by organised employers (employer organisations’ data on occupations).The data are collected jointly with employer organisations so that they collect data from their own member enterprises. These data become available for the Employment Statistics via Statistics Finland’s Statistics on Wages and Salaries. These primary sourcesaccount for roughly 60 per cent of the data on occupations.

The second type of data source is the Employment Statistics’ own inquiry on occupations which is a direct inquiry conducted annually among around 10,000 enterprises. The respondent enterprises can answer it either with a paper questionnaire or electronically via the web. The Employment Statistics’ inquiry on occupations covers the non-organised employers who employ at least 3.5 persons. The largest enterprises are covered by the inquiry every year while the smallest are covered every three years.

The third type of data source are secondary administrative registers from which data cannot be combined direct with those on a person’s employment relationship or any enterprise but which contain occupational data. These include e.g. the registers of job seekers maintained by the Ministry of Labour. The data on occupations in the Population Information System, which are updated from notices of change of address of individual persons, are also regarded as a secondary source.Thus, data on occupation can be obtained via the Population Register Centre for persons who have moved in the course of the year, otherwise the Population Register Centre’s data on occupations are deficient.

Finally, for those employed persons for whom data on occupation cannot be obtained from any administrative register or inquiry, occupational code is estimated from background information. Such background information includes data on professional status, industry, education and income. For the majority of all entrepreneurs the occupational code is estimated from data on industry and size of company.

The most important data sources are shown in the following table where the percentages refer to the share of the data source in the production of data for 2005.

Table 1. Data sources in the production of occupational data for 2005, %

Primary registers: / 62
Employer organisations’ data (via Wage and Salary Statistics) / 34
Local government sector data on employment relationships / 20
Central government sector data on employment relationships / 6
Other data on employment relationships / 2
Inquiry on occupations / 9
Secondary registers / 6
Estimation (industry, education, income, former occupation) / 20
Unknown / 3
Total / 100

4 Processing of occupational data in the Employment Statistics

Occupations are coded according to thenational 2001 Classification of Occupations, which corresponds to the ISCO-88 (COM) up to the 4-digit level. To accommodate national needs, a 5-digit level has been created for some occupational groups.

A large proportion (around 56 per cent) of the data on the occupations of employed persons in the Employment Statistics are obtained ready-coded from the Wage and Salary Statistics or by using the conversion keys of the Wage and Salary Statistics with which the code of an occupational title in the source data is converted into an occupational code. Ready codes are mainly available for the data of the employer organisations and conversion keys are used for the data of the local and central government sectors.

Occupational titles obtained from other register data and inquiries are principally coded with an automatic software application. The application’s dictionary of occupations is supplemented with new search words and conditions as new occupations surface in sufficient numbers from various sets of data. In the 2005 statistics, 20 per cent of occupations had to be coded automatically.

Even though the objective is to obtain occupational codes automatically, either as ready-made or with conversion keys or automatic coding, there are always some occupational titles which do not pass the automatic coding. Such titles are dealt with case-by-case by manual coding. Manual coding is very time-consuming and the aim is to decrease its share to the minimum by developing automatic coding procedures further. A person’s background information on sex, age, education, income, employer and former occupation is used in the manual coding. In the 2005 statistics, four per cent of occupations had to be manually coded and occupations for the remaining employed persons had to be inferred from background information. Occupation remained unknown for three per cent of employed persons due to insufficient information on background or occupation.

The registers may include overlapping data, as well as contradictory data. Most importantly, the registers complement one another. Several occupations are found for some employed persons, so that data from different sources must be put in order of priority.As a rule, efforts are made to link the occupational data in the Employment Statistics to the employment relationship in the last week of the year and describe a person’s occupation during it.The main principle is that wherever possible the data on occupation should be obtained from the same source as the data on employment relationship.

5 Comparing occupational data in the Employment Statistics and in the Labour Force Survey

Data on occupations have been published from the Employment Statistics since 2004.Apart from the Employment Statistics, the Labour Force Survey also publishes occupational data in Finland.The data of the Labour Force Survey are based on a sample and the occupational data are collected with computer-assisted telephone interviews of the sampled persons.

In general, with the exception of major group 1, Legislators, senior officials and manager, differences between the statistics were not large at the major occupational group level in 2005.Nearly 10 per cent of the employed persons of the Labour Force Survey belonged to major group 1 whereas in the Employment Statistics their share was under five per cent. The differences were smaller than this in the other major groups but considerable at the 2 and 3-digit levels even within the major groups.For example, there were (over 30 per cent)more persons in the Labour Force Survey than in the Employment Statistics in the following 3-digit level occupational categories:Writers and creative or performing artists (Code 245), Computing professionals (Code 213) and Artistic, entertainment and sports associate professionals (Code 347).By contrast, there were 50 per cent more persons in the Employment Statistics than in the Labour Force Survey in the category of Finance and sales associate professionals not elsewhere classified (Code 3419).

Figure 1. Employed persons by occupation major group according to LFS and Employment Statistics, 2005

The occupational groups for which best matched numbers were obtained includedSenior central government officials (C111), Dentists (C2222), Medical doctors (C2221), Dental assistants (C3225), Optometrists and opticians (C3224), Accountants (C2411), Lawyers (C2421), Psychologists (C2445) and Cleaners (C91322). The occupational titles in these groups are fairly unambiguous.

6 Reasons for differences between the Employment Statistics and the Labour Force Survey

The differences in occupational data between the Employment Statistics and the Labour force Survey largely arise from differences in data collection methods, reference points of time and concepts but there are also different coding practices.

The reference points of time for occupational data deviate in the two sets of statistics.In the Labour Force Survey, data on occupations are collected monthly and because the survey samples are small the published data represent means for the whole year.In the Employment Statistics, occupational data describe the last week of the year.In addition, due to differences in the definition of an employed person, the Labour Force Survey has over five per cent more employed persons than the Employment Statistics.In the Labour Force Survey, persons who during the survey week have been gainfully employed for at least one hour for pay or benefit in kind or for gaining profit, or who have been temporarily absent from work are counted as employed persons. In the Employment Statistics the employed labour force comprises all persons aged between 18 and 74 who were employed during the reference week. The information on employment is based on data from employment pension scheme and tax authorities.

The statistics also use different coding practices.Differences in the numbers of major group 1, Legislators, senior officials and managers, arise especially because occupations for entrepreneurs are coded differently in the Labour Force Survey and in the Employment Statistics. In the Labour Force Survey, entrepreneurs are generally coded as managers (main group 1) if they have hired employees. In the Employment Statistics only entrepreneurs with at least five employees are classified as managers.The rest of entrepreneurs are coded into the major occupational groups 2 to 9 according to the industry of their enterprise.

Differences in the data also cause discrepancies in the numbers of managers. In the Labour Force Survey, occupational data are coded on the basis of occupational titles. This also largely happens in the Employment Statistics. An exception from this is the data of employer organisations that come to the Employment Statistics via the Wage and Salary Statistics. Employer organisations have instructed enterprises to consider the real job tasks of their employees when they code the data (the data only contain codes for tasks and not for occupations). An employee is coded as a manager and professional on the basis of job tasks clearly more seldom in the Employment Statistics than in the Labour Force Survey in which the coding is based on occupation only. Diverse occupational titles ending in manager or director but involving hardly any or no leadership tasks have recently grown popular in Finland. Coding of occupations that is based on occupational titles only often gives a person an incorrect occupational category. Unfortunately, information about the real tasks of individual persons is available from very few sources.

In future Statistics Finland intends to harmonise the occupational data of the Labour Force Survey and the Employment Statistics.Due to the differences in the specifications and the data the discrepancies cannot be totally eliminate but can be diminished, for instance, by unifying the practices on the coding of entrepreneurs and by also otherwise increasing co-operation between the two sets of statistics.