Last month, RBA Assistant Governor, Christopher Kent, gave a speech entitled Recent Developments in the Australian Housing Market, which contained a number of erroneous statements about land/housing supply (click to read my critique).
Yesterday, the RBA’s Head of Financial Stability, Luci Ellis followed up with a speech entitled Housing and Mortgage Markets: The Long Run, the Short Run and the Uncertainty in Between, which contains a number of factual errors and mis-diagnosis of land/housing supply constraints across Australia’s various housing markets. Lets take a look.
Australia had high population growth during the post-war era (Graph 5). Since then it has slowed, but with some bumps along the way. During much of that period, growth in the dwelling stock outpaced population growth. Household sizes were shrinking, partly because the population was ageing and partly because people were having fewer children than in the past.
Graph 5
Population growth surged in the years after 2004, largely resulting from immigration. It was not immediately recognised, because part of the increase came from students and former students, whose long-term residency wasn’t immediately obvious. It’s fair to say that the fact of this acceleration in population growth wasn’t fully appreciated at first. As such, it was therefore not planned for. So at least in some cities, we saw strong demand for housing and rising prices, even though the credit boom had ended, because construction did not rise to meet this surge in demand.
The housing and credit boom of 2002/03 was clearly demand driven, but even it sparked concerns about constraints on housing supply. Those concerns only intensified afterwards, when the pressure really was for more dwellings to accommodate the population, rather than just nicer ones to absorb the extra borrowing capacity. My colleague Christopher Kent spoke about this recently. He noted that if constraints on land supply were the most important factor explaining high housing prices, we would see prices rising fastest where those constraints were most binding – the greenfield sites on city fringes. But that is not what we see.
The RBA’s claim that land prices have not risen quickly in greenfield sites is not backed-up data. As shown in the below table from RP Data, land prices in each and every capital city surged in the decade to 2012:
In fact, the land cost per square metre more than tripled in all capital cities, with the exception of Sydney, where vacant land prices were already expensive (see next chart).
If the above data is not proof of rampant inflation in fringe land costs, I don’t know what is.
Ms Ellis’ suggestion that the pre-2003 price boom was solely demand driven is also questionable. While credit was undoubtedly the fuel behind the surge in house prices, the RBA’s own chart (above) shows that the rate of dwelling supply contracted sharply from the mid-1990s, just as Australian house prices began their decade long run. If housing supply was responsive (‘elastic’), such an increase in price would have elicited a supply response, which would have mitigated further price rises. This clearly didn’t happen, hence the conclusion that Australian housing supply has become increasingly unresponsive (‘inelastic’).
Back to Ms Ellis’ speech.
Part of the perception of supply constraints comes from our view of Australia as a big country. Yes, we have a lot of land. What we don’t have much of is land where people actually want to live. In general, desirable land is the land within or adjacent to existing settlements. Australia’s population is heavily urbanised compared with other developed countries, let alone most emerging countries.
It is also quite concentrated in a couple of large cities (Graph 6). So the range of acceptable locations is actually more limited than the map would lead you to believe.
Graph 6
Perhaps more to the point, part of the reason why land seems in such short supply is that we each consume so much of it. As this graph shows, Australian cities are the least dense, in terms of population per square kilometre, than the cities of any other sizeable country, developed or emerging (Graph 7). Only New Zealand comes close, and that is partly because its cities are smaller. (Smaller cities tend to be less dense on average.) This is not even a case of comparing Australia with crowded European cities, or the megacities of developing countries. At around 375 people per square kilometre, Sydney is only 40 per cent as dense as Toronto. Melbourne is a bit more than half as dense as Toronto. And of the 276 US cities and towns in these data, only eight have lower densities than Sydney; the largest of those eight cities has only a little more than half a million inhabitants.
Graph 7
I fail to see the point of Ms Ellis’ remarks. Australia has a relatively small population of only 23 million people, or roughly 1/14th that of the United States. Moreover, none of Australia’s cities are large by global standards. In fact, Greater Sydney (population 4.4 million) wouldn’t even rank in the top 10 of largest metropolitan areas in the United States. Yet our residential land prices are absurd by global standards, running well above most other countries where land supply is genuinely scarce (rather than made scarce through regulation and lack of infrastructure provision).
Back to Ms Ellis.
If land availability were the problem, we’d expect land costs to dominate the costs of producing a new home at the city fringe. But that’s not what we see (Graph 8). Neither do government charges dominate total costs. Rather, it turns out that construction costs are the largest contributor to the total costs of production, and they seem quite high compared with the total cost of a newly built home in some other developed countries.
Graph 8
This is nonsensical analysis by the RBA. The latest HIA-RP Data Residential Land Report showed that the weighted median fringe capital city land price was $223,000 as at December 2012, implying that land prices comprise the majority of the cost of newly constructed housing.
Further, according to the ABS, the cost of building new project homes (i.e. the cost of constructing the physical dwelling, excluding land costs) has been lacklustre when compared against the growth of house prices (which includes land costs), suggesting that the bulk of the cost escalation of new homes has been in land prices, not construction costs (see next chart).
The above charts are broadly supported by aggregate land and dwelling values data produced by the ABS and RBA, which shows land prices roughly doubling against GDP since the mid-1990s, whereas the value of physical structures has flat-lined (see below chart).
Back to Ms Ellis.
What are the long-run implications of our demography and geography? I am not a demographer, but it seems reasonable to me to conclude that long-run population growth will be slower in future than in the decades following the Second World War. It also seems reasonable to expect that the decline in household sizes has come to an end. The expansion in the housing stock needed to accommodate any given population increase will therefore be smaller than past data relationships implied.
Even with slower population growth, the price of our low-density life has become unaffordable for some. It therefore seems likely that our cities will become denser over time. If so, the mix of residential construction will be tilted more towards medium-density and high-density dwellings than in past decades. We have already been seeing this shift over the past decade or so (Graph 9).
Graph 9
What Ms Ellis fails to mention is that the recent uptick in average household size (see next chart) has likely been caused by the increasing cost of housing, rather than demographic change or consumer preference. That is, many who would like to enter the housing market simply cannot, due to prohibitively high housing costs caused, at least in part, by rising land costs from supply constraints.
Indeed, according to the 2011 census, the percentage of group households increased to 4.1% from 3.9% in 2006, whereas the percentage of single (lone person) households declined to 24.3% from 24.4%, suggesting that higher housing costs have worked to increase occupancy rates.
More broadly, Ms Ellis’ and Christopher Kent’s testimony on the supply-side contradicts a 2008 speech by the RBA’s Anthony Richards, which properly diagnosed the issues around Australian housing supply:
…supply-side factors should have a much greater influence on prices towards the fringes of cities, where land is less scarce and accounts for a smaller proportion of the total dwelling price. In principle, the price of housing there should be close to its marginal cost, determined as the sum of the cost of new housing construction, land development costs, and the cost of raw land. And in the absence of any restrictions on supply, the price of raw land on the fringes should be tied reasonably closely to its value in alternative uses, such as agriculture. So unless there has been a marked increase in the value of this land when used for other purposes, the availability of additional land towards the edges of our cities should have limited increases in the cost of housing there…
So if we are looking for explanations why housing is not as affordable as we might like, it may be necessary to look at factors on the supply side as well. One obvious place to start is the cost of land for building new houses near the edges of our cities…
There are no doubt a number of factors that could be contributing to the observed level of land prices… One factor that has been widely mentioned is the existence of various constraints on land development, including growth corridors and boundaries. Another factor that has been mentioned is the existence of a range of government charges, including developer levies or infrastructure charges. More broadly, concerns have also been expressed that zoning policies and building approval processes have hampered in-fill development closer to the city centres.
Both economic theory and international evidence suggest that housing prices can be boosted by land usage policies (which can create artificial scarcity of residential-zoned land), problems with the complexity of the development process (which creates rents), and the fees and charges imposed on development. Accordingly, the fact that higher prices for housing have not resulted in a more significant supply response could be a reflection of various supply-side costs that have represented a wedge in the cost of bringing new housing to market.
…the fact is that real price increases in the outer suburbs have been quite large as well.
Ms Ellis’ and Mr Kent also seem to contradict the Productivity Commission’s 2011 planning report, which found that the land supply system is slow and unresponsive, caused in part by slow approval and planning processes, as well as poor infrastructure delivery:
Based on a sample of 20 residential developments in greenfield areas across Australia’s five largest cities, it can be 10 years after the commencement of rezoning before a subdivision of that land is completed, infrastructure is installed, and building can commence. If processes outside of planning are included, it can take up to 15 years between site assembly and building construction.
Across all jurisdictions, the most common causes of delays and extended timeframes in land supply processes are associated with rezoning and planning scheme amendment; structure planning; and overcoming community concerns and objections. The substantial length of time associated with rezoning and structure planning processes (up to six years) is not surprising given the complexity and absence of statutory time limits in most jurisdictions.
Perhaps the underlying reason for the RBA’s new found supply-side obtuseness is because it has become the Australia’s key defender of high house prices. Not only has the RBA seemingly changed tack on the supply-side, but has offered cherry-picked analysis of other markets that concludes more responsive housing supply increases financial instability.
These arguments are not supported by the weight of local or international evidence or economic principle. We should expect better from our central bank.
Leith van Onselen is Chief Economist at the MB Fund and MB Super. He is also a co-founder of MacroBusiness.
Leith has previously worked at the Australian Treasury, Victorian Treasury and Goldman Sachs.
