Learning but not always doing: Sources of multifactor productivity growth in Australia

26 February 2025 | Lawson Ashburner, Senior Research Economist, and Vincent Wong, Assistant Research Officer

Recent issues of the Productivity Commission’s (PC’s) Quarterly productivity bulletin have highlighted that the productivity bubble has now deflated, meaning that productivity rose during the COVID-19 pandemic but has since collapsed back to its pre‑COVID levels.

The productivity in this story is labour productivity – output per hour worked. Growth in labour productivity comes from two sources: more investment (capital deepening), and better ways of combining capital and labour to produce outputs – also known as multifactor productivity (MFP) growth.

MFP growth has been slower over the past two decades. This suggests that our economy has improved at using labour and capital, but the rate of improvement has been slower than it used to be. Why is this happening? Are we not learning new ways of organising labour and capital as rapidly as we once were? Or are we just applying fewer of the skills, knowledge and innovations that we have?

Both are true, according to the PC’s new decomposition of changes in MFP, based on a method devised by Diewert and Fox (2018) and previously applied to Australian data by Zeng et al. (2018).

The method decomposes MFP into the MFP frontier and inefficiency.

• The MFP frontier is the highest level of MFP achievable with contemporary knowledge and mix of capital and labour inputs. We never actually witness the true frontier, but the decomposition takes a conservative approach and estimates it to be roughly the highest level of MFP previously achieved in each industry. It is assumed that industries never ‘forget’ the best way to combine labour and capital, and so the frontier can only decline because of changes in the availability of labour and capital. In practice, changes in the availability of labour and capital have quite a small impact on the frontier. ¹
• Inefficiency captures how far actual MFP is from the MFP frontier.

The decomposition suggests two reasons for the slowdown (figure 1). ²

First, we seem to be finding fewer new and better ways to organise inputs. Growth in the MFP frontier has slowed since the mid-2000s (the transparent blue line grows more slowly since then). ³

Second, we are not using our resources as we best know how. Inefficiency crept into the economy in the 2000s and has persisted since (the light blue bars grow over the 2000s and do not subsequently decline). This inefficiency is now substantial – MFP would have been about 7% higher in 2023-24 without it. It is also somewhat cyclical, temporarily rising around the global financial crisis in the late 2000s and the recent COVID-19 crisis, which is probably because more capital sits idle in economic downturns than during better times.

Figure 1 – MFP growth slowed because growth in the frontier slowed and inefficiency crept in

The drivers of MFP differ by industry

MFP growth has varied greatly between industries, and the decomposition suggests the drivers have varied too.

Business services and distribution services have both seen continual growth in their MFP frontiers, but actual business services MFP was stagnant between the mid‑1990s and late 2000s, while actual distribution services MFP grew steadily from the mid‑1990s (figure 2, panels a and b). ⁴

Inefficiency explains the difference – business services became progressively less efficient between the mid-1990s and late 2000s and has only partly recovered since, while distribution services remained efficient throughout.

Personal services and the goods sector (excluding mining) have followed similar trajectories to one another (figure 2, panels c and d). Both saw rapid MFP growth until the late 2000s, driven by growth in their MFP frontiers, and stagnation in the 2010s due to a lack of frontier movement and creeping inefficiency. Both, especially the goods sector, also saw frontier expansion in recent times, boosting MFP.

The utilities and construction sectors stand out as poor performers. Their MFP was stagnant from the mid-1990s to early 2010s, as there was growth in their MFP frontiers and growing inefficiency in roughly equal measure (figure 2, panel e). Since then, MFP has continually fallen as there has been no movement in the frontier and rising inefficiency.

Figure 2 – Multifactor productivity growth, and its drivers, varies by industry group

What might explain these different movements in the MFP frontier and efficiency?

The differences in the growth of the MFP frontier are likely because some industries are better suited to exploiting new technology for MFP growth. For example, the extent to which service industries can utilise work from home, be offshored or delivered remotely is correlated with growth in their MFP (PC 2021), probably because the improvements in information and communications technologies and growing globalisation since the 1990s boosted competition in industries with these inherent features, incentivising innovation and leading to growth in their MFP frontiers. Many business and distribution services industries meet these criteria, such as professional and technical services and information, media and telecommunications services. In contrast, many personal services and utilities and construction industries require face‑to‑face contact or physical delivery of the service, which can make them more difficult to offshore or deliver remotely.

Education is another plausible driver of the differences in MFP frontier growth. Since the mid-1990s, the quality of the labour force (based on education and experience) in business and distribution services grew by more than in other industries (figure 3). More educated and experienced workers can, in theory, produce more per hour worked, even with the same amount of capital per worker.

The inter‑industry differences in efficiency are harder to explain. Inefficiency can occur because new technology makes old capital obsolete, but firms do not dispose of the capital and so it goes underutilised (Fox 2018). More stringent regulation is another possible driver, as regulation can prevent firms adopting new technologies and using resources efficiently. Indeed, even well‑designed regulation might reduce MFP in pursuit of other social, environmental or economic goals that are not captured by output. Measurement error may also be a contributor. Both regulation and measurement error reduced measured MFP in the utilities industry in the 2000s (Topp and Kulys 2012).

Figure 3 – The MFP frontier grew faster in industries in which workers upskilled

Australia’s MFP could be higher

The decomposition is not perfect. Most notably, it attributes the reductions to mining MFP to inefficiency, which is debatable and why we excluded the mining industry from the whole‑of‑market‑sector measure in figure 1 (box 1).

But even with these imperfections, the decomposition encourages us to broaden how we think about MFP. Discussions about boosting MFP growth often focus on boosting the rate of technological innovation – in other words, they focus on the MFP frontier. The frontier is important, but the decomposition shows that efficiency matters too. The PC’s 2023 Productivity Inquiry examined this in the volume Innovation for the 98%, and highlighted that only about 2% of Australian firms innovate in ways that are new to the world (PC 2023). Its central theme was that much productivity growth will come from the wider adoption of established, even dated, technologies and practices by the other 98% of firms.

The PC is revisiting this theme in the five inquiries it is currently undertaking under each pillar of the Australian Government’s productivity growth agenda.

An appendix to this article is available in the Productivity Bulletin.

References