We DON'T KNOW if housing construction productivity is RISING or FALLING (Part I)

What we got from Australia's Productivity Commission recently were silly methods without sound economic logic

Since we know little about the causes of productivity increase, the indicated importance of this element may be taken to be some sort of measure of our ignorance about the causes of economic growth…
- Moses Abramovitz, 1956

Last month, the Australia Productivity Commission (PC) released a report that began

Improving productivity can help fix housing affordability

Australian housing is increasingly unaffordable. Decades of inadequate supply coupled with high demand has driven this outcome.

In response, Australian governments have committed to build 1.2 million homes over 5 years – 240,000 homes each year. In the 12 months to June 2024, just 176,000 homes were built.

Governments have focused on alleviating constraints to new supply via changes to planning regimes. These are important reforms. But the speed and cost of new building is also a constraint on new housing supply.

Increasing the productivity of the construction process would lower construction costs – meaning more approved projects would be viable, increasing housing supply – even with no change in the size of the workforce, interest rates or the cost of materials.

But dwelling construction productivity has been stagnant at least for 30 years.

Fixing productivity will require a genuine focus on prioritising housing supply and affordability. This report provides policy directions for improving housing construction productivity by reducing the regulatory burden, streamlining and speeding up approval processes, supporting innovation and improving workforce flexibility to help turn the dial on this persistent policy challenge.

Download a copy here.

After you’ve read that, please take a look at a previous article of mine on the subject.

William Baumol's insights into the construction productivity puzzle explain why my old house would not be built the same way today

As I said in that article:

The puzzle for many is that labour productivity in manufacturing and many other sectors keeps increasing—we get more cars per car factory worker, more bottles of soft drink per Coca-Cola worker, and more agricultural produce per farm worker—but in construction, we get the same output value per on-site construction worker.

Why?

I want to focus on two things in this article—1) measurement issues and 2) the economics that helps us interpret those measurements.

That article was a deep dive.

This one dives even deeper. You’ve been warned!

I’ve broken it into two parts—Part I now and Part II in a few weeks time. I hope to explain to someone who isn’t an expert but is willing to follow the details of the logic why the analysis in the PC report doesn’t lead to the conclusions it makes.

There is a lot to get through to bring an intelligent reader who is not familiar with the issues along for the ride.

Background

Declining labour productivity in housing construction is a widespread economic concern in many nations. In February 2025, Australia’s Productivity Commission (PC) released the latest in a line of reports making a sequence of claims that:

1. construction labour productivity is declining,

2. causing higher construction costs, which in turn

3. increase the market price of housing.

There is no doubt that improvements in labour productivity across the economy improve living standards. It is also desirable for our regulatory environment to allow for, even encourage, new production methods and capital investment to facilitate this.

For example, the PC notes, as many have done before, that prefabrication in housing construction is a possible avenue for improving labour productivity.

Unfortunately, the sequence of three claims made is not supported by the data. We don’t know if housing construction labour productivity is increasing. Nor do we know if it is decreasing. Moreover, the link between productivity and house prices isn’t clear cut.

Concerns about the inability of construction productivity to keep pace with the overall economy are at least half a century old—as old, in fact, as the economic statistics used to measure industry-level productivity.

For example, Allen sought to understand in 1985 “[w]hy construction productivity is declining”. Further back in 1967, we learn from Cassimatis that “[a] persistent view in the economic literature is that productivity growth in construction is far below the national average.”

These days, you can find many attempts to measure and muse over the common pattern of construction productivity decline that is observed globally, either in the academic literature or economic magazines and websites.

The common cry is: “What is wrong with construction?”

But the economist, who realises that this sector is subject to market forces and pressures to innovate and economise as much as any sector, might want to be more circumspect.

Perhaps a better starting place is to ask: “What are we measuring and what would we expect that measure to show based on economic logic?”

That is the purpose of this article.

Rather than jumping to conclusions about the causes of changes in construction productivity, it is worth first reflecting on the puzzle of declining measured productivity during a period of sustained productivity gains across the economy. Is there some economic logic being missed that could explain the common trend across countries – one that each country, looking at its own experience, takes to be the result of local factors?

This “explainer” walks through the economic logic and measurement issues at the heart of the construction productivity debate and the recent PC report on the topic.

PART I - Labour productivity and its economic and measurement issues

What is labour productivity?

Productivity is the general name for a ratio of a given economic output to a certain input of interest.

Labour productivity is a class of productivity ratios where some form of labour input is used, whether in the form of work hours, total workers, or some other measurement.

It is an intuitive macroeconomic concept—with a given population and workforce input, an economy with higher productivity is an economy that can produce more valued outputs (numerator) for each person (denominator).

The way that labour productivity ratio’s outputs and inputs are measured is quite important.

Labour inputs can be work hours, the number of jobs, or some quality-adjusted metric that captures the various skills in the workforce. Which labour to include is not clear. Is it only on-site labour? What about upstream materials manufacturers? And what about the labour used upstream to build manufacturing facilities?

How much labour is used to make a product is not an easy question. This is why similar questions, like how much energy is used to make a product, are hard to grapple with for concerned environmentalists. The same economic and statistical limits mean we don’t ever really know the answer to either question.

Housing construction outputs are also tricky to measure.

Often, the number of dwellings is used (as the PC did in part of their analysis). But the quality and size of dwellings vary a great deal over time and space. It must require more people to make bigger and better houses. To deal with this, quality adjustments can be made to estimate a real measure of housing (adjusting for quality change, just like real prices adjust for inflation).

Another way to measure housing construction output is by using the total value of housing construction contracts. This overcomes some, but not all, quality adjustment issues. A problem here is that about 80% of the cost of construction comes from off-site inputs like materials. Why include only the labour involved with the final step in production, which accounts for only 20% of the total cost, but include in the output measure all the inputs like materials and contracted services, which make up 80% of the cost? It is like including only the labour at the car dealership, the last stage of production, but ignoring the labour at the car manufacturer when looking at the labour input needed to create the total car value.

To overcome this issue, it is possible to measure only the Gross Value Added (GVA) in the housing construction sector, not the total construction cost. GVA is the construction contract price minus non-labour inputs. This is the output measure the PC uses in their main analysis.

In the end, most labour productivity measures are ad hoc and often not comparable. Even classifying businesses into specific industries to know what industry is being measured is tough to do. Yes, there are groups of specialist economists working on all these measurement issues. But they’ve never found a practical way to measure the concept they have in their mind.

What the Productivity Commission did

Australia’s Productivity Commission (PC) looked at the labour productivity in the residential construction sector and concluded that their “new estimates of housing construction productivity tell a story of decades of poor performance.”

They claim that improved labour productivity in housing construction “can help fix housing affordability” and direct their policy recommendations towards “alleviating constraints to new supply via changes to planning regimes” and reducing “regulatory impediments” that get in the way of “prefabricated and modular construction.”

The mechanism by which this is thought to occur is that “increasing the productivity of the construction process would lower construction costs – meaning more approved projects would be viable, increasing housing supply – even with no change in the size of the workforce, interest rates or the cost of materials.”

They explain that in the past three decades that

• the number of dwellings completed per hour worked by housing construction workers has declined by 53% (physical productivity)

• gross value added per hour worked – a more comprehensive measure that controls for quality improvements and increases in the size of housing – has declined by 12% (labour productivity)

Like many previous authors from Australia and around the world, they have jumped to the conclusion that what they have measured is what they understand the concept of productivity to be.

What did the PC measure?

First, they counted the number of new dwellings produced as the output (the numerator) and the number of on-site work hours as the input (denominator) and called this Dwelling construction physical productivity.

We know that changes to the quality and size of dwellings make this metric economically rather meaningless. For example, it is possible to increase this measurement of labour productivity by building much smaller dwellings than desired by homebuyers. Because of this serious and well-known issue, I ignore this metric in the following critique.

Second, they added up the Gross Value Added (GVA) of residential builders and on-site contractors (which are part of an industry called Construction Services) as the output (the numerator) and the number of on-site work hours as the input (denominator) and called this Dwelling construction labour productivity.

The diagram below helps to explain how this is done (it is a good one, helps to show the main point, and is worth the paid subscription to drop the paywall).