Improved estimates of hours worked in Productivity: Implementation of hours worked from the Labour Account

Updates estimates of industry multifactor productivity (MFP) and experimental state productivity. For further details, refer to the Explanatory notes.

Cite
Print
Released
13/12/2021
Source
Estimates of Industry Multifactor Productivity, 2020-21 financial year
Released
13/12/2021 11:30am AEDT

Background

Historically, productivity estimates have been compiled using hours worked estimates taken from the Labour Force Survey (LFS). This release sees the implementation of hours worked estimates from the Australian Labour Accounts used in the compilation of productivity.

Presented in this article are revised productivity estimates from 1994-95 to 2018-19[1] for the market sector and selected industries. The revisions are a result of implementing the Labour Account hours worked estimates, as well as the national accounts historical revisions, into productivity estimates[2]. Estimates of hours worked for the period prior to 1994-95 were backcast using movements from the LFS and hence did not impact productivity estimates.

Labour Accounts hours worked estimates in the national accounts

The Australian Labour Account provides improved coherence between employment information collected from household surveys and business surveys, increased coverage and better alignment with the 2008 System of National Accounts production boundary compared to the LFS. Specifically:

Coherence: balancing LFS data with business survey and administrative data sources resulting in improved industry classification

Coverage: secondary job holders are accounted for and hours worked are redistributed to industries of employment appropriately

Alignment: out of scope units from the LFS that are in scope of the national accounts production boundary are accounted for, including:

  • Short term non-residents
  • Residents employed by overseas organisations
  • Defence employees
  • Employed children aged 5-14 years

This has resulted in improvements to industry level hours worked, thereby improving the quality of the industry productivity measures.

For more information about the use of Labour Account estimates compared to the LFS, see Technical note - Labour Account: the best source for industry employment and jobs data.

Revisions to market sector productivity

The Labour Account hours worked, and associated measures of multifactor productivity (MFP) for the market sector, show similar trends to the previously published series that was based on LFS hours worked estimates. Figure 1 and Figure 2 provide a comparison of hours worked and MFP measures.

Hours worked grew slightly faster on a Labour Account basis compared to LFS hours worked from 1994-95 to 2018-19 (Figure 1). Over the 24-year period, hours worked recorded a 38.4% rise on a Labour Account basis and a 34.5% increase on a LFS basis.

The increase in hours worked growth on a Labour Account basis resulted in slower MFP growth by almost three percentage points between 1994-95 to 2018-19 (Figure 2).

The majority of market sector productivity growth cycles remained robust to the changes in hours worked, with three of the four productivity peaks (in 1998-99; 2003-04; and 2017-18) remaining unchanged. A new peak emerged in 2009-10 replacing the previous 2011-12 peak[3] as a consequence of the changes to hours worked. Productivity growth cycle peaks for the market sector are detailed in Table 3 of the Data downloads.

Industry revisions

The implementation of Labour Account hours worked estimates and updated GVA estimates have resulted in revisions to industry level MFP. This is due to improvements in the quality of industry level labour market statistics attributable to the Labour Account. For example, the LFS, while appropriately measuring the labour force status of the population, relies on households self-reporting their industries of employment. Additionally, the Labour Account redistributes the LFS hours worked across industries to account for secondary job holding. This adjustment overcomes a limitation with industry-based measures of hours worked in the LFS, whereby all reported hours worked are allocated to a person's industry of main job. The productivity revisions resulting from changes in hours worked can be sizeable for some industries.

Among market sector industries, the largest downward revisions in MFP were seen in Administrative and support services and Wholesale trade, while the largest upward MFP revision was in Arts and recreation services. Minimal revisions were seen in some market sector industries including Agriculture, forestry and fishing and Information, media and telecommunications (Figure 3).

(a) For more information on revisions to GVA for the Financial and insurance services division, see Improved estimates of the annual national accounts: Results of the 2021 historical revisions.

Industry based growth cycles were impacted to varying degrees due to the revisions (see Table 1). The Administrative and support services division experienced the most significant impacts to its growth cycles with no original peaks retained following the implementation of the Labour Account. The Wholesale trade and Retail trade divisions were similarly affected, retaining only one original peak in 1994-95 and 2003-04 respectively. Only four industries retained all original peaks, Agriculture, forestry and fishing, Mining, Rental, hiring and real estate services and Other services. 

Table 1: Revised industry based growth cycle peaks, 1994-95 to 2018-19
       1995199619971998199920002001200220032004200520062007200820092010201120122013201420152016201720182019
Agriculture, forestry and fishing  ✔   ✔    ✔  ✔       ✔  
Mining ✔    ✔     ✔  ✔     ✔   
Manufacturing    ❌    ✔   ✔   ✔   ❌  ❌
Electricity, gas, water, and waste services   ✔  ❌     ❌ 🟡   ❌   ❌🟡 
Construction    ✔   ✔         ❌🟡   🟡 
Wholesale trade✔     ❌  🟡 ❌   🟡 ❌🟡   ❌🟡 
Retail Trade  ❌      ✔   🟡 ❌      ❌  
Accommodation and food services    ✔   ❌   ✔    ✔   ❌  🟡
Transport, postal, and warehousing  ✔     ✔   ❌🟡    ✔   ❌ 🟡
Information, media, and technology✔   ✔    ✔   ✔  ✔   🟡❌   
Financial and insurance services     ✔       ✔    🟡      
Rental, hiring, and real estate services ✔      ✔  ✔   ✔   ✔  ✔  
Professional, scientific, and technical services✔   ✔  ❌ 🟡      ✔   🟡  ❌ 
Administrative and support services❌  🟡   ❌🟡    ❌🟡    ❌    🟡
Arts and recreation  🟡   ✔   ✔   🟡❌   ❌  ❌ 🟡
Other services  ✔   ✔     ✔    ✔   ✔   
16 industries market sector    ✔    ✔     🟡 ❌     ✔ 

✔: Growth cycle peak retained

❌: Growth cycle peak disappeared

🟡: New growth cycle peak

The industries with the largest revisions are discussed in more detail below.

Administrative and support services

Administrative and support services recorded the largest revisions in hours worked of all the sixteen market sector industries. Figure 4 shows a divergence of the two hours worked series, notably between 1998-99 and 2007-08. This coincided with the expansion of Employment Recruitment and Labour Supply service firms in the late 1990s which was enabled by an advancement in information technology[4]. The faster growth of hours worked on a Labour Account basis reflects the improvements made to the measure of industry labour input via better aligned industry classifications[5].

Labour Account hours worked grew strongly over the last two decades resulting in comparatively weaker MFP performance in this labour-intensive industry when compared to the LFS-based MFP estimates (Figure 5). Moreover, the labour income share in this industry has not dropped below 90.0% since 1994-95, making it the most labour intensive of the 16 market sector industries. As a result, the industry MFP measure is very sensitive to revisions in hours worked.

Due to the large revisions in the hours worked estimates, all productivity growth cycles peaks for the industry were revised. New peaks are observed in 1997-98, 2002-03, 2008-09 and 2018-19 replacing peaks in 1994-95, 2001-02, 2007-08 and 2013-14 (see Table 1) (also see Table 26 in the Data downloads).

Wholesale trade

Wholesale trade was also significantly impacted by changes to hours worked. On a Labour Account basis, the industry hours worked rose steadily, reversing the decline previously observed in the LFS hours worked series. The two series show a 41.1 percentage point difference over the comparison period (Figure 6). The revisions in hours worked are attributable to improved coverage through industry allocations across main and secondary jobs, in addition to the improved coherence and alignment of the Labour Account industry statistics.

Revisions to hours worked resulted in a significant revisions to MFP for Wholesale trade throughout the period (Figure 7). Similar to Administrative and support services, changes to hours worked impacted the growth cycles for Wholesale trade. Only one growth cycle peak remained unchanged, in 1994-95. Previous peaks in 2000-01, 2005-06, 2011-12, and 2016-17 have been replaced with new peaks in 2003-04, 2009-10, 2012-13, and 2017-18 (see Table 1).

Arts and recreation services

Arts and recreation services was the third most impacted industry, with hours worked growing at a slower rate on a Labour Account basis than a LFS basis (see Figure 8). Two subdivisions, Heritage activities and Creative and performing arts activities, drove the difference between the two hours worked series from 2003-04 with both subdivisions recording much lower growth in hours worked on a Labour Account basis. Also contributing to the hours worked revisions was Sports and recreation activities, the largest contributor to employment for this industry, which recorded slower growth in hours worked on a Labour Account basis over the time series.

Hours worked revisions resulted in an upward revision to MFP, which grew almost 19.0 percentage points over the time series on a Labour Account basis compared with a decline of 3.3 percentage points on a LFS basis (Figure 9).

Growth cycle peaks were impacted due to the sizeable revisions to MFP. Three peaks disappeared in 2009-10, 2013-14 and 2016-17, while three new peaks emerged in 1996-97, 2008-09 and 2018-19. Two peaks in 2000-01 and 2004-05 remained unchanged.

Endnote

[1] In 2019-20, the national accounts moved to hours worked movements from the Labour Account to more accurately capture the pandemic’s impacts to hours worked.

[2] Analysis of impacts to the market sector and the industries selected in this article, will also include revisions to GVA from the historical revisions. However, these revisions are small and do not drive the overall revisions to MFP – see Figure 3 for more detail on GVA revisions.

[3] As first published in Table 14 of Australian System of National Accounts, 2020-21.

[4] See The Australian Recruitment Industry (The Australian Department of Education, Skills and Employment, 2016).

[5] For example, households employed by Labour Supply Services firms for large scale infrastructure projects in Mining and Construction, were incorrectly reported as working in one of these industries in the LFS but are now correctly attributed to Administration and support services under the Labour Account.