6311.0 - Information Paper: Construction of Experimental Statistics on Employee Earnings and Jobs from Administrative Data, Australia, 2011-12  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 11/12/2015  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All

APPENDIX 2: COHERENCE OF EXPERIMENTAL STATISTICS WITH ABS SURVEY COLLECTIONS

The data used in the following comparisons is compiled from a variety of ABS survey collections. The Labour Force Survey (LFS) data is based on an average of the quarterly original data (see Labour Force, Australia, Detailed, Quarterly (cat. no. 6291.0.55.001)) for the 2011-12 financial year. The Average Weekly Earnings (AWE) data is based on averages of quarterly trend data (see Average Weekly Earnings (cat. no. 6302.0)) for all four quarters during the 2011-12 financial year. The Employee Earnings and Hours (EEH), Australia, (cat. no. 6306.0) data is for the May 2012 reference period, and the Employee Earnings, Benefits and Trade Union Membership (EEBTUM), Australia (cat. no. 6310.0) is for the August 2011 reference period. The experimental statistics on mean weekly earnings are calculated by dividing the annual earnings by 52.29 weeks (this differs slightly from the usual 52.14 weeks as 2012 was a leap year).

GENERAL DIFFERENCES

The general differences in the number of employees and earnings were due to:

    • differences in the concepts, scope and methodology used in the LEED Foundation Projects and those used in household and business surveys;
    • the LEED Foundation Projects containing a combination administrative data collected for taxation purposes from both individuals and businesses, and ABS Business Register data collected for statistical purposes, whereas other ABS data sources are compiled for the explicit purpose of creating statistics;
    • unreported cash in hand payments which are excluded from the experimental statistics but may be included in household and business surveys if reported in the reference period; and
    • the experimental statistics categorising individuals as employees if they had worked at any point during the 2011-12 financial year, whereas any point in time measure of employees includes only those who were employed during reference period (often the last week/fortnight or last pay period on or before a specified cut-off date).

NUMBER OF EMPLOYEES

The following graphs show comparisons between the aggregate experimental statistics and estimates of the number of employees from selected ABS business and household surveys.

Age

There were minimal differences found in the number of employees by age group between the experimental statistics and LFS, especially for employees aged between 35 and 59 years. The experimental statistics are expected to be higher for younger and older employees due to more frequent periods without employment during the financial year (observed in the 20 to 34 year age groups in particular, as well as the 60 and over age groups due to bridge retirement practices). The experimental statistics are expected to be lower for younger employees due to unreported cash in hand work (observed in the 15 to 19 age group).

Graph 2.1: Number of employees, by age, 2011-12

Graph 2.1 compares the number of employees from the experimental statistics with LFS estimates, by age group


By sex

Due to the way employees are defined and measured in the LEED Foundation Projects, it was expected that the experimental statistics would be more coherent with estimates of employment from household rather than business surveys. However, the LEED Foundation Projects measure the gross volume of employees in the financial year, which would inflate the experimental statistics against all sources, while employment not reported to the ATO will deflate them. For male employees, there is no statistically significant difference between the experimental statistics and the LFS estimates, whereas for females there is a 6% difference. This may be due to greater numbers of women transitioning between labour force status during the financial year. The difference between the experimental statistics and EEH is much higher for males (12%) and for females (3%). The differences are likely caused by general differences in concepts, scope and methodology employed in EEH. The exclusion of the Agriculture industry (which employs approximately twice as many males as females) from EEH may deflate the number of male employees in EEH compared to the experimental statistics.

Graph 2.2: Number of employees, by sex, 2011-12

Graph 2.2 compares the number of employees from the experimental statistics with LFS and EEH estimates, by sex


EMPLOYEE EARNINGS

The following graphs show comparisons of average weekly earnings between the experimental statistics and the estimates from selected ABS business and household surveys.

Age

It is expected that average weekly earnings from the LEED Foundation Projects will be broadly coherent with EEBTUM. Periods without employment during the financial year (more prominent for younger or older employees) as well as unreported cash in hand work (also more common in younger age groups) are expected to deflate the experimental statistics (as observed in the 15 to 34 year age groups, as well as the 65 and over age group). The differences in the 65 and over age group were potentially influenced by the tendency of some older employees not to report their earnings on an ITR. The experimental statistics include the gross value of fringe benefits, while EEBTUM includes only the salary sacrificed component, which will inflate the experimental statistics, particularly in the older age groups where fringe benefits are more prominent. There is no statistically significant difference in the 60 to 64 year age group, which is potentially due to a balancing of these effects.

Graph 2.3: Mean weekly earnings, by age, 2011-12

Graph 2.3 compares the mean weekly earnings from the experimental statistics with estimates from EEBTUM, by age group



Sex

There is no statistically significant difference between mean weekly earnings from the experimental statistics and estimates from AWE. There are minimal differences between the experimental statistics and EEBTUM, which are likely due to the inclusion of fringe benefits (inflating the experimental statistics) and unreported cash in hand work (deflating the experimental statistics), as well as seasonal variability in the EEBTUM estimates. The EEH estimates are higher (4.7% for males and 8.6% for females) than the experimental statistics (as well as AWE and EEBTUM). This is likely due to the general differences in concepts, scope and methodology employed in EEH, including the exclusion of Agriculture (which has lower mean earnings).

Graph 2.4: Mean weekly earnings, by sex, 2011-12

Graph 2.4 compares the mean weekly earnings from the experimental statistics with estimates from AWE, EEH and EEBTUM, by sex