A primer on labour productivity

There’s been a lot of discussion on recent subdued productivity growth in Australia, particularly in the non-market sector which has also been the source of strong employment growth. In this primer on labour productivity, we contribute to this discussion by explaining how the ABS measures labour productivity and what this means for interpreting productivity statistics.

ABS labour productivity statistics are produced using Gross Value Added (GVA) and hours worked from two key datasets – the national accounts and the labour account. 

GVA is a measure of output published in the national accounts. It is equal to total output produced, minus the inputs used up in creating that output such as raw materials, energy and contractor services[1]. GVA is published for each of the 19 industries in the economy. 

To measure the quantity of labour used to produce output, the ABS could use the number of people employed in each industry, but a better measure is to use the number of hours people work. The number of hours worked in each of the 19 industries is published in the labour accounts. 

Labour productivity is calculated as the quantity of GVA divided by the number of hours worked to produce it. Ideally, the ABS would adjust outputs and inputs for changes in quality to provide a more accurate measure of productivity. Quantities that have been adjusted for changes in quality are called volume measures[2].

The ABS also produces annual multifactor productivity (MFP) and KLEMS multifactor productivity statistics to help users analyse productivity and inform policy decisions (Box 1). 

Productivity growth is the most important determinant of economic growth and living standards in the long run. When labour productivity grows, it means our workforce is producing more output for each hour spent at work. This makes it more likely our real wages will grow and our living standards will rise.

Australia’s recent subdued productivity growth is well documented[3]. In the June quarter 2025, the level of Australia’s labour productivity was at pre-pandemic levels, roughly equivalent to where it was in December quarter 2019 (Chart 1).

There is considerable variation in labour productivity across industries, in level and growth terms. This is to be expected given very different levels of capital intensity, use of technology, market structures and competitive pressures.

Low productivity growth is evident in the market sector. In the past 3 years, average annual productivity growth in 9 of the 16 market industries has been lower than its 20-year average (Chart 2). 

The ABS defines the non-market sector as those parts of the economy that produce goods and services not sold at economically significant prices. The ABS classifies 3 of the 19 industries as being in the non-market sector[4]. These are:

• Education and training (e.g., schools, universities),
• Health care and social assistance (e.g., hospitals, aged care and disability services), and
• Public administration and safety (e.g., government departments, policing and defence). 

A relatively low level of measured productivity in the non-market sector is to be expected given low levels of capital, lack of competitive pressure and the pursuit of other social goals over maximising profit.

However, it is striking that in these industries average productivity growth in the past three years has been negative, and well below their 20-year averages (Charts 3 and 4). This has occurred alongside very strong employment growth (Chart 5). 

The rest of the article focuses on whether measurement limitations are contributing to the observed negative productivity growth in the non-market sector. 

There are several challenges in measuring and interpreting labour productivity growth, which are particularly pertinent in the non-market industries. These include:

• Measuring the quantity and quality of output (particularly for services),
• Measuring the quantity and quality of labour inputs,
• Differences in the drivers of employment for non-market activity,
• Challenges in measuring quality improvements, which is particularly difficult in the non-market sector, and
• Differences between the price of outputs produced and the outcomes desired by consumers.

These challenges are discussed in greater detail in the following sections of the article. 

For most businesses in the market sector, the ABS estimates the dollar value of output using business surveys and non-survey data, such as scanner data from supermarkets. The ABS assumes that improvements in the quality of a product are embedded in the market price. ‘Volume’ estimates of output are derived by deflating dollar estimates with an index of ‘pure’ prices that abstracts from any quality changes[5]. This means volumes derived by removing pure price changes include quality changes. This is important because an increase in the quality of output has the same benefit as an increase in the quantity.

Producers in the non-market sector typically provide their services free of charge, or at very low prices. Because non-market prices aren’t the product of supply and demand, they don’t reflect quality changes or the true economic value of what has been produced. This means for non-market producers it isn’t possible to estimate the dollar value of outputs and use price changes to derive volumes, as is done for market producers. It means we need to develop a different mechanism to ensure changes in quality are included in volume measures[6]. 

Instead, the ABS directly measures the growth in the quantity of non-market services by using an indicator series, though as discussed later, it is often difficult to measure improvements in the quality of those services over time. Where quality improvements aren’t captured, output and labour productivity are underestimated. 

Labour inputs are measured as the number of hours worked by employed people who are paid for their time. Volunteer labour is excluded. 

In the Labour Force Survey (LFS), the ABS asks respondents how many hours they worked in a particular week, and how many hours they work in a normal week. This information, combined with data from other surveys, feeds into the labour account to create an estimate of total hours worked in each industry. Adjustments are made to account for employees not covered by the LFS, including defence force personnel and short-term international visitors to Australia who are in the labour force[8].

ABS estimates assume all hours worked are equal. However, it would be more useful to adjust hours worked for the quality of labour input to get a more accurate measure of productivity gains from using the same quantity and quality of inputs. One simple reason the quality of labour might change over time is due to fatigue. For example, a tradie who works 10 hours of overtime to finish a job that is behind schedule might produce work of lower quality (i.e. the worker would be less productive) in those 10 hours than they were producing during standard hours. The ABS does not make these types of adjustments. 

Another structural reason the quality of labour might change is due to changes in the average level of education and work experience. The ABS estimates this change in quality through ‘quality adjusted labour input’ (QALI) indexes. Education and work experience data is sourced from the five-yearly Census of Population and Housing, with estimates interpolated in the intervening years. Since QALI captures improvements in labour quality, when the average level of education and work experience increases over time, the growth of labour input as measured by QALI is higher than the growth measured by hours worked. Consequently, labour productivity will grow at a slower rate. The Reserve Bank of Australia has recently explored the use of microdata to adjust the quality of labour inputs more frequently[9].

In the market sector, firms will increase employment to the point where profits are maximised. Non-market producers operate differently. Their aim is not to maximise profits but to provide a service to the community, and/or to increase the quality of the service. Consequently, hours worked in the non-market sector tend to reflect demographic trends and government policy rather than profitability trends in the private economy.

The decline in measured labour productivity in the non-market sector could in part reflect the way the ABS measures outputs. For example, some of the additional hours might improve the quality of services, such as increasing the ratio of nurses to patients or the ratio of teaching staff to students. However, this could cause measured productivity to fall in the short run, because (a) it can take time for the additional labour inputs to translate into more outputs and (b) it can be difficult to capture the lift in service quality in statistical terms[10]. This challenge is described in the next section.

Another possible explanation for subdued productivity growth is that the increase in hours worked is being undertaken by less experienced people. This may have required firms to spend extra time skilling up their new workforce. If so, it’s likely that time spent at work by less experienced workers would have been less productive. If true, this effect should only be temporary. Additionally, with record levels of labour force participation and a historically low rate of unemployment, some of the people who are currently employed may be intrinsically less productive than the rest of the workforce.

As mentioned earlier, because prices can’t be used to deflate the dollar value of non-market output, we need to find tools and data that we can use to measure quantities directly. A basic count of the quantity of services (e.g. number of children enrolled at school, number of operations performed by hospitals) is a starting point, but it can be improved to account for changes in quality.

A measure of quality can be introduced by weighting the quantities depending on how costly they are to produce. For schools, universities and hospitals, this is what the ABS aims to achieve[11]. In this sense, costs of production are used as a proxy for missing prices. It means, for example, that each open-heart surgical procedure carries more weight than each broken bone treated in the emergency room. The heart surgery costs more to perform, so it is assumed to be more valuable. Aged and disability care services can be weighted by the intensity of care provided in each package – more intensive packages cost more to deliver, hence carry more weight. It costs more on average to educate a secondary school student than a primary school student, hence more weight is given to each secondary school enrolment. In the absence of meaningful prices, weighting different services by their relative cost is an attempt to introduce a measure of quality.

There are further challenges around non-market services that are consumed collectively, and not individually. Examples include defence, national security and public policymaking. It isn’t possible for individuals to opt in or out in terms of consuming these services. This means there is no way to measure the ‘quantities’ of these services being produced and consumed, and the ABS assumes volumes of output grow at the same rate as inputs (hours worked). Consequently, measured growth in labour productivity is zero for these activities. 

National accounts measure outputs. Outputs and outcomes are not the same thing. Outputs refer to the quantity of goods and services produced, whereas outcomes are the results or impacts of the goods and services provided. 

For most market-based activity, outputs and outcomes are similar. However, this isn’t the case for the non-market sector. In terms of health care, outputs consist of the provision of medical care through processes such as hospital treatments and medical consultations, whereas desired outcomes might include longer life expectancy, faster recovery from surgery, and shorter waiting times for admission. Similarly, for education, output might be the number of students taught, whereas outcomes might be examination scores or lifetime earning capacity. 

To give a concrete example relating to health, technological progress often means it becomes possible to conduct procedures on patients that were not previously thought feasible. An example is the introduction of hip replacements in the 1980s when earlier techniques made such operations too risky. Another example of technological progress is when existing procedures such as knee reconstructions can take place via keyhole surgery on day admission, when previously, such procedures would’ve been more invasive and would’ve required lengthier stays and longer rehabilitation time for the patient. These are cases where improvements in outcomes might not be captured as a rise in output, and hence productivity.

It is difficult to measure the outcomes in the non-market sector because prices don’t reflect the underlying value of services provided. Possible metrics in health care include changes in waiting times for surgery, patient recovery rates, life expectancy, patient quality of life, and the rate of surgical failures (readmissions and deaths). It is not clear which of these attributes should be included in an index and how to weight them. Measurement would also be an issue. Given these challenges, the ABS has not gone down this path. 

It is also possible that outcome indicators might not be correlated with output. For instance, standardised test scores to measure the quality of school education might improve due to the innate ability of cohorts, rather than teaching methods. Similarly, in the case of health care, improvements in rates of surgical success and recovery times might reflect a community becoming healthier over time, which is unrelated to the quality of health care services.  

Some countries have experimented with ‘decoupling’ productivity statistics from the national accounts by introducing explicit outcomes-based measures[12]. The Productivity Commission recently published an alternative approach to measuring productivity in health care, which takes an outcomes-based focus from the perspective of the patient[13].

Labour productivity isn’t easy to measure, particularly for service sectors that make up most of the economy. We infer labour productivity from estimates of output and hours worked, which in turn depend on the accuracy of numerous surveys and non-survey datasets. Estimates will embody an element of ‘noise’ from measurement error, particularly for quarterly data, which is why the ABS has for many years recommended users examine longer-term trends in productivity. 

Measuring labour productivity in the non-market sector is even more difficult. The absence of meaningful prices means it is difficult for the ABS to estimate quality-adjusted output from the dollar value of goods and services provided. Furthermore, it is the outcome for consumers that matters, yet outcomes are not measured in the national accounts. 

One implication of these difficulties is that it is likely that attempts to improve the quality of non-market services will show up, at least in the short term, as an increase in hours worked but not an increase in output, thus biasing productivity estimates downward. Comparing market sector productivity to its non-market counterpart is therefore not an exercise in comparing like with like.

Productivity experts around the world have grappled with these kinds of questions for decades. Difficulties in measuring quality change accurately directly affects our measures of labour productivity, and this should be kept in mind when analysing productivity statistics.

1. Labour and capital depreciation are not subtracted when calculating GVA. They both contribute to gross value added but are not consumed like raw materials or services bought from other firms.
2. Imagine a phone company introduces a new phone that represents an improvement on the previous model. If the company sells the same number of phones this year compared to last, but the new phone is ‘better’ in some way, volumes have increased.
3. An example of the Productivity Commission’s research can be found here: Productivity before and after COVID-19 - Commission Research Paper - Productivity Commission
4. A fourth ‘notional’ industry, Ownership of Dwellings, is conceptually part of the non-market sector. For the purposes of compiling labour productivity estimates, this industry is excluded from the scope of the non-market sector because there is no labour input.
5. The aim is to measure ‘pure’ price change. This is the change in the price of a product where the characteristics of the product have been held fixed, where any change in quality has been removed from the price. More information can be found here: Quality change in the Australian CPI. It is usually easier to account for quality change in goods than it is for services.
6. A detailed paper on the measurement challenges for non-market activity can be found here: Non-market output measures in the Australian National Accounts: a conceptual framework for enhancements, 2020.
7. It also goes the other way. The ABS, a non-market producer, is classified as part of the professional, scientific and technical industry, which is part of the market sector.
8. More information can be found here Hours worked guide.
9. Measuring Labour Quality in (Closer to) Real Time Using Emerging Microdata Sources
10. As discussed by Dean Parham in Productivity in Welfare Service Industry—Measuring Productivity in Health and Education: An Exploratory Study of Selected APO Countries - APO.

11. This paper explains how the ABS measures the output of hospitals: Hospital output measures in the Australian National Accounts: experimental estimates, 2004-05 to 2017-18.

    This paper explains how the ABS measures the output of primary and secondary schools: School output measures in the Australian National Accounts: experimental estimates, 2007-08 to 2017-18.

    This paper explains how the ABS measures the output of universities: University output measures in the Australian National Accounts: experimental estimates, 2008 to 2017.

12. The UK Office for National Statistics is an example. Sources and methods for public service productivity estimates - Office for National Statistics
13. Advances in measuring healthcare productivity - Commission Research Paper - Productivity Commission