A natural experiment in excess housing supply

Why did Melbourne rents bounce so quickly after a massive 5% supply shock?

Thank you to the 2,900 subscribers who read Fresh Economic Thinking.

I want to ask a small favour. If you enjoy these articles, the best thing you can do to support me is to share them on your social media and forward the emails to your friends.

And don’t forget to subscribe. It’s free after all.

---

Here are the opening paragraphs from a new paper by Tim Helm¹ at Prosper Australia.

You’d never get ethics permission for an experiment like this. Take a city of 5 million, growing at 100,000 per year, construction rates steady – a city on a stable trajectory of rapid growth. Now freeze time, drop in 130,000 extra dwellings, and resume.

What happens to housing costs?

That experiment was, broadly speaking, the experience of Melbourne over the course of 2020 and 2021. With one of the world’s longest and harshest lockdowns, borders more or less closed to internal and international migration, and a one-way escape valve for an exodus of foreign citizens, the city’s population shrank in absolute terms by 80,000 people, while construction continued more or less unabated.

The result? An urban population by mid-2021 some 340,000 residents short of the level expected in projections from just two years prior – or, expressed another way, a city with 130,000 dwellings more than previously thought to be needed. Pandemic policies engineered a massive over-supply of dwellings.

What were the consequences for housing costs?

Here’s the key chart. The bars show the cumulative imbalances between the population and the housing stock since the start of the series in 2013. The imbalance becomes more negative if the population grows at a faster rate than the housing stock, and more positive when the housing stock grows faster than the population.

What you notice is that between 2013 and 2018 both the population and the housing stock grew in tandem (on average, with population more so in 2013-16 and housing stock more so in 2016-19).

But in 2020, 2021, and even 2022, there was a rapid and historically large increase in the relative abundance of homes. By 2022 there were about 6% more homes than people compared to 2018.

This is an enormous figure. It’s like getting three years of new housing construction dumped into the city.

You’d expect rents to be about 10-12% lower if the standard view is correct.

Advertised asking rents did fall by this amount.

But it didn’t last. By 2022 rents were above 2019 levels and rising fast (a 20% per annum rate). This is despite the fact there remained a 6% excess of dwellings compared to the population, and the excess was still rising during 2022.

If you prefer the rental measure for all dwellings in the CPI rather than advertised asking rents, then all rents in Melbourne fell 2.5% over an 18-month period, which reversed during the next 6 months.

So what?

Responses to this report over the past week or so have fallen into two main categories.

1. Rents falling 12% is a bad thing. Ha ha ha. This shows how good more supply is.

2. Supply didn’t cause a sustained decline in rents because of demand-side factors that counteracted it.

Regarding the first response, no one argues that a sudden and unexpected supply shock won’t have a short-run price effect or that lower rents are bad.

The housing supply and price question is always about the long run. After all, the story about a lack of supply causing high rents and prices is a long-term argument—new housing supply was a little too low for many decades and this accumulated slowly over time into a big price effect. If a 5% shock to the stock (not a shock to the rate of growth of the stock) can’t sustain a price effect for more than two years, then how is slowly reversing the apparent supply deficit meant to add up to a big price effect?

The second response is extensively dealt with in the report. In fact, understanding adaptive responses to a sudden change in the housing stock is one of the main points of the report (covered on pp25-31).

Adaptive responses

Here’s how the important question of adaptive responses is described in the report.

The message is that when we predict the consequences of economic or policy change we too often focus on the initial impact, and pay inadequate attention to subsequent adaptive responses. We worry about or elevate first-round effects, while downplaying second-round reactions.

In the real world, whenever there’s a change, there’s a countervailing adaptation. That means neither doomsday scenarios nor promised policy utopias tend to pan out that way.

It also means that all we can ever exploit through policy are the relative timeframes and scales of the initial shock and the subsequent reaction.

Property owners leaving homes vacant was the first adaptive response considered. Using analysis of water meter data the author was able to show that there

…was a significant increase: in 2021 around 35,000 more homes were left empty or under-used for the entire year than in 2019 (a 51% increase).

This increase represents around 1.8% of the housing stock, or around one-third of the pandemic excess supply shock, indicating that leaving property empty was an important margin of adjustment.

There was also a slow reversal of this unoccupied dwelling adjustment when rents began to rise.

In terms of household size, this also adjusted to the lower rents. The report quotes the RBA Assistant Governor Luci Ellis about this economic adjustment.

"If average household size hadn't declined, there would have been fewer households to fill the new homes being built. Instead of the result being a swathe of empty homes, though, prices would have adjusted – in this case, rents... The subdued level of rents was likely one of the factors that induced the decline in average household size I mentioned earlier... The lesson here is that we shouldn't focus only on the original shock – but rather how people will react to that shock (RBA 2022).

Note, however, that fewer people in each home as an adjustment to lower rents doesn’t need to imply rents rise again. Just like when clothes fall in price I buy more clothes, this doesn’t automatically imply the price of each clothing item goes back up.

The report considers the possibility of lower rents attracting in-migration, but couldn’t find evidence of that happening.

To be clear, people were returning to Melbourne in 2022, but into a housing market that was still building new homes faster than the rate at which people returned. The excess of dwellings to people was still rising in 2022.

The final possibility is that this enormous and unusual supply shock was accompanied by an enormous and unusual shift in preferences such that households now wanted to spend more on housing and occupy the exact number of homes left empty by the supply shock.

No doubt there was a change in housing preferences. But why wouldn’t those also occur if the market suddenly increased supply in the absence of the COVID experiment? After all, the number of people per household has declined enormously over the past decades, so we seem to quite naturally demand all the homes that exist.

Spatial and rental equilibrium rule of thumb

A rule of thumb that is extremely good at predicting housing markets is that households spend roughly 20% of their gross income on rent and rearrange their household composition to match the number of physical homes. This is a rental and spatial equilibrium process.

Why should we expect this to be any different?

We can look at an example of a massive supply shock in the reverse direction during the 2011 floods in Brisbane and more recent floods in northern New South Wales. In these areas, rents increased for about a year and then returned to their previous level within two years, despite a large and sustained negative shock to the stock of physical dwellings.

Property markets always adjust this way. There are equilibrium forces that make it so.

What did developers say?

When asked about this report, the UDIA developer lobby group simply chose not to offer an explanation of how a sudden 6% increase in the housing stock compared to the population had no sustained effect on rents.

In fact, they had nothing to say at all except that the planning system is approving lots of homes and the property owners of those sites are choosing not to build them.

Max Shifman, president of developers peak body the Urban Development Institute of Australia, said the report contained “a lot of conflated issues” and it wasn’t entirely correct to say there was a major oversupply.

“There were so many unusual things that happened concurrently during the pandemic period,” he said.

“They’ve pulled a lot of interesting data together, but I don’t think they’ve assessed it and come to the right conclusion from that data.

“The simple fact is, we have consistently under-delivered the number of dwellings we need, and it’s only getting worse.

“Even approved stuff is not getting built anymore. And it is absolutely the number-one thing we need to fix. We need to get more capacity in the system where its needed and the actual type of housing that people want.”

1. I know Tim well and have worked with him on other projects. I read an early draft of this report but made no contributions to it.