Academic journal article The Economic and Labour Relations Review : ELRR

Australian Labour Force Data: How Representative Is the 'Population Represented by the Matched Sample'?

Academic journal article The Economic and Labour Relations Review : ELRR

Australian Labour Force Data: How Representative Is the 'Population Represented by the Matched Sample'?

Article excerpt

1. Introduction

Data on gross flows between various labour market states in Australia has been available since early 1980. From time to time researchers (e.g. Foster 1981, Foster and Gregory 1984, Fahrer and Heath 1992, Borland 1996a, Leeves 1997 and Leeves 2000) visit this data with a view to gaining extra insight into issues related to the determinants of movements in the level of unemployment over time and/or the equilibrium or natural rate of unemployment. Little attention however has been given to the implications of the survey methodology and related ABS statistical procedures for the representativeness of the data derived from the matched sample.

This paper aims to address two related sets of questions. The first set concerns the proportion of the population represented by the matched sample (and thus the gross flows data). What proportion of the total population is in fact represented by the matched sample, why is this proportion what it is and how does it vary over time? Second, given that slightly over 20% of the population are not represented in the matched sample, it-is sensible to ask how representative are labour market indices derived from the matched sample data and, if bias is present, what can we say about the direction and source of the bias? The structure of this paper is as follows. Section 2 details the way in which the Labour Force Survey is undertaken and the method by which gross flows data is derived from successive surveys. Section 3 examines the behaviour of the proportion of the population represented by the matched sample over time. Section 4 compares the time series properties of the unemployment rate for those persons represented in the matched sample as against the rate for the population as a whole. (1) Section 5 looks at the behaviour over time of the matched sample's unemployment rate and the unemployment rate for the groups not represented in the matched sample. Section 6 presents a framework which enables us to decompose the bias in the matched sample into its constituent parts and to evaluate their numerical importance. The final section considers the representativeness of the matched sample in capturing flows and transition rates and concludes with a discussion of the implications for future research and policy.

2. The LFS and the Matched Sample

The Labour Force Survey (LFS) has been undertaken on a monthly basis since February 1978. (2) Households selected for the LFS are interviewed each month for eight months, with one-eighth of the sample being replaced each month. Prior to August 1996, all interviews were conducted face-to-face at the homes of respondents. Over the period August 1996 to February 1997, the ABS introduced the use of telephone interviewing to collect LFS data. (3) In the interviews an attempt is made (inter alia) to establish whether each person is in or out of the labour force, if in whether employed or unemployed and, if employed, whether the employment is full-time or part-time.

To derive labour force estimates for the relevant component in the Australian population, expansion factors (weights) are applied to the sample responses. Weighting ensures that LFS estimates conform to the benchmark distribution of the population by age, gender and geographic area. A weight is allocated to each sample respondent according to his/her State/Territory of usual residence, region (capital city or other), age and gender. The weights are computed in such a way so as to also adjust for any under-enumeration and non-response.

For the LFS, private dwellings (such as houses and flats) and nonprivate dwellings (non-private dwellings are those that provide a communal or transitory type of accommodation (4)--such as hotels and motels, boarding houses, short-stay caravan parks, hospitals, nursing homes and homes for the aged, educational colleges and boarding schools, and Aboriginal and Torres Strait Islander communities) are separately identified and sampled. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.