Commercial, not residential, property in a price bubble

Commercial, not residential, property in a price bubble
Christopher JoyeDecember 8, 2020

As we look to what's ahead for 2012, Property Observer is republishing some of our most noteworthy stories of 2011.

 

It never ceases to amaze how distorted investment thinking can get among even seasoned professionals. A classic example is the very different approaches taken to Australian commercial and residential property in the superannuation fund sector. 

Of the nearly $1 trillion in non-self-managed superannuation money (i.e., held by retail, public, government and industry funds), the average portfolio weight to “real estate” is between 10% and 15% of all fund assets. 

Now the rational observer might reasonably assume that there would in turn be a say, 90:10 split in favour of residential over commercial given the relative asset class sizes. The total market capitalisation of private residential property is about $3.6 trillion based on Rismark’s best estimates. In contrast, the total value of all core commercial, industrial, retail and wholesale property is less than $300 billion, according to RMIT. That gives us a rough 90:10 split. 

Another way of building portfolios is via much more sophisticated “mean-variance optimisations”. We have applied this technique over the last 30 years of data and the results are even more damning: you don’t allocate any money at all to commercial property because of its lower total returns, higher volatility, and strong correlation with the business cycle. 

One could also fairly argue that since more than 30% of contributing superannuants have zero exposure to residential property, and those who do have access own one very risky individual asset and presumably possess a powerful motive to get diversification to a truly national portfolio (much like if you only owned Fortescue shares and wanted to diversify across the whole market), super funds have a logical incentive to deliver these imperatives. 

The final nail in the asset-allocation coffin is the risk, return and correlation characteristics of Australian housing. The table below summarises all the major asset-classes’ performance over the last three decades, which is the longest period over which can objectively evaluate the data.

On a risk-adjusted basis, a national portfolio of Australian housing has consistently outperformed Australian shares, global shares, Australian bonds, and Listed Property Trusts (LPTs). The “uncorrelated” feature of residential property returns vis-a-vis equities, whereby housing yielded strikingly superior performance during the 1987 stock market crash, when shares fell 44%, the 2001 tech wreck, when global equities collapsed by around 50%, the 2002-03 period, when Australian shares were down 10% to 15%, and 2007-09, the GFC, when Aussie and global equities once again halved in value, implies that housing is a valuable portfolio “diversifier”. 

The most remarkable thing is that notwithstanding the compelling evidence above, your typical Australian super fund has no allocation at all to housing. That is, the 10% to 15% investment in real estate is dedicated almost exclusively to commercial property. There seem to be few sensible reasons for these decisions. 

One controversial explanation that has been proposed to me is that super funds suffer from an insidious form of “industry capture”. In short, most of the executives responsible for managing real estate investments within super funds previously worked in the commercial property sector in one capacity or another. The same is true of the super funds’ advisors: almost all “asset consultants” that focus on real estate come from commercial property backgrounds. 

This possibly gives rise to an instinctive bias to pursue commercial property strategies. At the margin, there are a few funds that do have some residential development holdings. But in an investment sense these are far removed from the traditional, established housing stock. 

There are good reasons why banks happily lend to established housing but avoid developers like the plague: the probabilities of loss are far greater with the latter, as they are with commercial property. Development assets tend to be located on the fringes of cities with lower quality public infrastructure and amenities. The demand for these homes is often less certain while the buyers frequently have lower incomes. This is why the “risk-weighting” APRA forces the banks to apply to commercial property and developer loans is multiples the risk-weighting attached to established residential loans. Through the business cycle, established housing credits have proven to be much safer (as we shall see shortly). 

It is easy to highlight the comparative risks of commercial and residential property. In the first chart below we contrast the performance of the ASX LPT Accumulation Index (reinvesting dividends) with the total net returns generated by Australian housing in the period since 1982. To proxy for housing’s performance, we have used the capital gains plus only half of the gross rents to control for transaction costs. 

 


 

Over the past 29 years, a national portfolio of Australian residential property has produced superior total returns (10.6% per annum) with, crucially, nearly one-third the risk. This was borne out most graphically during the GFC. The peak-to-trough fall in the LPT index was a stunning 69% even allowing for dividends. On the other hand, the peak-to-trough fall in boring old bricks and mortar was less than one-10th this amount. Click to enlarge

 

Those in the commercial property industry would argue that the ASX-listed returns tell an artificially negative story apropos their asset-class. They instead prefer to use “unlisted returns” measured by a group called IPD. The problem with these unlisted performance estimates is that they are based on human valuations of capital growth over time. They are not based on actual sales data, or the judgments of thousands of investors participating in an actively traded market.

Many of these unlisted valuations are also commissioned by the fund managers that own the underlying assets, which presents a well-known, albeit rarely acknowledged, conflict of interest. In brief, it has been repeatedly documented that human valuations of commercial properties significantly lag real-world events and materially underestimate true losses. This is why we prefer to rely upon the day-to-day pricing afforded by a market such as the ASX. For the record, the house price data we use here is based on around 40,000 home sales a month, and does not include any human valuations.

On the basis of the analysis presented herein, the really big “bubble” was in commercial real estate, not residential property. 

A similar observation applies to Australian shares. In the next chart I have added the All Ordinaries Accumulation Index to our analysis (grey line), and telescoped in on the past 10 years of data (although note this stops in March 2011, and does not account for the recent share market correction).

Once again, the take-away is that by far the two biggest bubbles were in domestic shares and commercial property. This was illustrated by the tremendous price appreciation generated in the years before the GFC (observe the stunning rise in the blue and grey lines) followed by precipitous 50% plus declines. Australian housing just did what it always does: produced low volatility (viz., risk) total returns that typically track changes in disposable household incomes and borrowing capacity. Click to enlarge

 

 


 

In recent years, the RBA has gone to great lengths to highlight the fact that commercial property investments are more hazardous than established housing exposures (see here). It is widely accepted that bad commercial property loans in 1991 almost bankrupted ANZ and Westpac. 

In this context, the chart below depicts the losses realised by CBA in different parties of its credit portfolio over time. Observe that during the 1991 recession, when unemployment peaked at around 11%, CBA, like the other major banks, suffered an enormous increase in bad debts. These were, however, almost exclusively attributable to CBA’s “business lending” activities, which includes commercial property. Despite a huge increase in unemployment and double-digit mortgage rates, CBA’s losses on its home loans (lower red line) hardly budged. Importantly, a similar pattern reasserted itself during the GFC.

The RBA’s bi-annual Financial Stability Review (FSR) is the single most authoritive Australian study on the banking system’s risks. In examining these issues, the RBA offers some fascinating comparisons of the commercial and residential property asset-classes.

The first RBA chart I want to highlight is effectively two diagrams compressed into one. The left-hand-side panel shows commercial property prices in Australia, the US, UK, France, Ireland and Spain over the period 2003 through to 2011. What is especially interesting is that the RBA’s analysis demonstrates that Australian commercial property prices (blue line) surged way past peer country rates of growth in the years preceding the GFC (note the large protrusion above the other lines). And while Australian commercial property prices naturally fell by a very substantial margin, they remain materially above any comparable country today.

In the right-hand-side panel the RBA conducts the same analysis for residential property. Here the results are very different. You can see that in the period 2003 to 2007 Australian house price growth was substantially less than that witnessed in the US, UK, France, Ireland and Spain. In contrast to commercial property, the relative “outperformance” only emerged in the years following the GFC, which makes sense given Australia’s underlying economic outcomes. Indeed, even accounting for the deep European recessions, Australian house price growth since 2003 has been substantially less than its French equivalent.

Rismark recently undertook a similar study comparing Australian and UK house price growth over a longer period. We used the UK Academetrics house price index because it employs the UK Land Registry’s information, which includes 100% of all UK home sales. Other UK house price measures, such as the Halifax or Nationwide benchmarks, cover less than 15% of all sales.

Notwithstanding the implosion and ensuing nationalisation of much of the UK banking system, and the exceptionally acute recession experienced in that country, house price growth in Australia and the UK has been nearly identical over the last two decades. In fact, what is most noticeable from this analysis is the much more rapid run-up in UK house prices (red line) in the years before the crisis, which significantly exceeded Australian gains. Click to enlarge

 

 


 

I want to conclude today’s column with some more charts from the RBA’s Financial Stability Review. The first examines “non-performing” home loans (ie, mortgages in arrears) in Australia, the US, UK, Canada and Spain. What is important to remember when studying this chart is that mortgage rates in Australia are much higher than they are in other countries. All things being equal, that would suggest that the probability of default should also be higher in Australia. Yet according to the RBA’s latest information, only 0.8% of all Australian home loans are currently three months or more behind.

If we assume that there are about 5 million households with mortgages, that implies that only 40,000 are materially behind on their repayments. That is slightly less than the monthly number of sales transacted in Australia’s housing market. While the default rate has understandably trended up as the RBA has lifted interest rates to above-average levels, it remains very low by international standards (compare the lower blue line with the US, UK, Spanish and Irish rates). 

The next chart I want to reflect on quantifies the size of “impairments” in Australia’s banking system by loan type. As is the RBA’s wont, we are looking at three different charts compressed into one. The first left-hand-side panel shows impairments of residential home loans. Focus on the red line. We can infer here that there are about $2 billion worth of “bad” home loans in Australia. In total, there are more than $1.1 trillion home loans outstanding, so only 0.2% are truly impaired.

The blue “past due” line above the red impairments looks to be around $8 billion. That makes intuitive sense: if we multiple the 0.8% national mortgage default rate by the $1.1 trillion of loans outstanding we get $8 billion to $9 billion. If we multiply the 40,000 borrowers who are three months behind on their repayments by the average loan balance of $200,000, we also get about $8 billion. 

Now move to the next, middle panel in the chart above. This shows the same analysis for all the banks’ “business” lending activities. Observe how total impairments are around $20 billion. That is, the size of business impairments in Australia is about 10 times larger than residential home loan impairments. Recall that this is similar to what happened in the 1991 recession: most bad debts in the banking system originated from the business, not residential, lending portfolios. 

In the next chart, the RBA splits out commercial property loans from total business lending. You can see in the left-hand-side panel that commercial property impairments are much higher than the overall business lending average, which gels with the regulator’s view of commercial property’s more elevated risks.

 


 

My final two charts illustrate RBA analysis of changes in Australian banks’ lending standards, and Australia’s house price-to-income ratio.  In the first we can see that loan-to-value ratios (LVRs) have fallen significantly in the period since the GFC. Indeed, elsewhere in the Financial Stability Review the RBA explains that the recent increase in mortgage arrears has been largely driven by loans that were extended before the crisis hit, not subsequent to it, as is often claimed.

The last chart reinforces Rismark’s analysis of changes in dwelling price-to-income ratios over time. The RBA finds that Australia’s “all regions” dwelling price-to-disposable household income ratio is about four times (the actual levels are less useful than the change in the ratio over time). Importantly, this ratio has been falling for a while now; we have had no house price growth in about one and a half years while ABS-reported disposable household incomes have been expanding at a circa 8% per annum pace based on the latest June quarter National Accounts. Like us, the RBA finds that with the exception of a brief dip during the GFC, Australia’s house price-to-income ratio has fallen to its lowest level since about 2003.

Christopher Joye is a leading financial economist and works with Rismark International. Rismark and RP Data provide house price analytics products, and solutions that enable investors to go long and/or short the housing market. The above article is not investment advice.

Editor's Picks