The size of the female labour force is often thought to stretch and contract in response to changes in the pressure of demand over the economic cycle. If so, estimates and projections of the labour force and of unemployment might make an allowance for the phenomenon of "discouraged workers'. But is the number of women in the labour force much smaller in recessions than it otherwise would have been? The answer to this question requires a consistently defined measure of the female labour force, as well as estimates of its size, holding demand constant. In Discussion Paper No. 31, Susan Owen and CEPR Research Fellow Heather Joshi are concerned with the first problem - how to measure the "actual' labour force consistently over time.

Labour statistics regularly report the working population, which includes only the employed plus the registered unemployed. A more desirable measure of the labour force would include all those seeking (paid) work - whether or not they are registered as unemployed. In Britain, such a measure of "economically active' persons has only been recorded intermittently. The decennial Censuses of Population (1951-81) seem to offer a possibility of a consistent measure of this preferred definition for the post war period. Joshi and Owen find that the construction of a series of female activity rates from these data is not at all straightforward.

The way in which economic activity was recorded altered at each Census. In 1951, for example, the questions referred to a person's "usual' gainful occupation, whereas subsequent censuses adopted the concept of a "reference week'. The censuses also use a "head-count' measure of labour supply. Joshi and Owen argue that this creates difficulties because it requires that people be placed in one of a series of mutually exclusive categories of "economic activity'. Women who divide their time between work in the home and the labour market therefore pose a classification problem. Those performing paid work for only a few hours a week may well be counted as "economically inactive' by the household member who fills out the census form - as comparisons with in- depth interview surveys reveal. Joshi and Owen found that 1971 was the only year in which the census did not specifically mention "housewife' as a reason for "economic inactivity'. They found that indeed the female activity rates derived from the census for that year match measures derived from other sources more closely than in any other census year!

Joshi and Owen document these and other changes in census procedure. They also use other sources such as National Insurance records, the Census Post-Enumeration Survey and the General Household Survey, to assess the reliability and likely direction of errors in the census evidence.
Joshi and Owen adjust the census information in the light of these findings to construct a series of decennial female activity rates by age group. They argue that their series is "less inconsistent' with the definition of economic activity adopted by the EEC Labour Force Survey, which is the most regular source likely to be available in the future. They find that the long- run trend towards a higher female participation rate seems smoother in their revised series than in the crude data.


How Long is a Piece of Elastic? The Measurement of Female Activity Rates in British Censuses 1951-1981
H Joshi and S Owen

Discussion Paper No. 31, September 1984 (HR)