Employment in the low income group
An evaluation of the concordance between register and interview data
Previous analyses of employment in the low income group have shown large differences in the level of employment between interview data and register data. In this project we compare information from interviews and registers at an individual level in order to gain an insight into these differences.
The source of register information on employment is the Central Register of Employers and Employees for the years 2006-2008. Information from this register is included in the event database “FD-trygd”. The EU survey on living conditions EU-SILC is the source of interview information on employment.
Excluding the self-employed, who are not covered by the register, and full-year students, for whom survey information on employment is of low quality, we find that the employment level in the low income group is substantially higher when using the interview data compared to the register data. According to the interview data, about two thirds are employed, compared to about a half according to the register data. The difference in the number of working hours per year is even higher. The number of hours according to register data is about 60 per cent of the figure for interview data.
About half of the low income group (excluding the self-employed and full-year students) are employed according to both interview and register data. Nearly 20 per cent are employed only according to interview data and very few, 2 per cent, are employed only according to register information.
The group making the largest contribution to the difference in the number of working hours calculated from interview and from register is those employed according to interview data but inactive according to register data. Approximately a third of the difference is due to higher estimates of working hours from interview data than register data, for persons employed according to both interview and register data. For these, the number of working hours per year is more than 25 per cent higher according to interview data than according to register data.
Accordance between interview and register data is lowest among persons who are working few hours according to register data.
Those in the low income group who are employed during the year according to interview data but not register data are working two thirds of a man-year. Another source of information on employment is wage income according to the tax assessment. More than three thirds of the low income group are either employed according to interview data or have a wage income. Those not employed according to interview data only have a small wage income. Nineteen per cent in the low income group are employed according to interview data but not register data, and these have a very low wage income despite the relatively high number of working hours. An earlier analysis showed that total consumption expenditure in the low income group was considerably higher than the after-tax income (Andersen 2007). The combined results of these analyses indicate uncertainty about the completeness of the register data, and consequently the measurement of low income.
Are there certain groups in which the difference between interview and register data is especially large? The percentage employed according to interview data but not register data is almost twice as high among men than among women. Differences according to age, immigration background, education and type of household are rather small. The percentage is highest among persons aged 25-34 years and lowest among young adults below 25 and 45-66 year-olds. The percentage is highest among persons with only compulsory schooling and those without a registered education.