Thursday, September 25, 2014

The preference whisperer

I had a chat today with a freshly minted PhD from an elite US economic school about rationality and utility maximisation. As you do.

There was simply no phenomena he could not explain with his beloved utility theory. But if the theory explains everything imaginable, then it predicts nothing. I resolved that the theory as it stands, in the revealed preference form, was not falsifiable, though he didn’t seem to understand why that would matter.

I’ve since dubbed him the preference whisperer - someone who can observe any apparently illogical combination of behaviour and reverse out a utility function that is consistent with that behaviour. Insurance, check, Gambling, check. Addiction, check, Suicide, check. 

Of course I could develop a theory of space monkeys controlling the universe that fits will all observable phenomena. I could add a sense of mathematical rigour to it by demonstrating that our space monkey is optimising confusion in human society to amuse himself through the infinity of time his poor immortal soul is cursed with. 

You would laugh it off of course. But for some reason you are not laughing off utility theory just yet. Over the decades an air of credibility has sprung up around the idea. Economists who believe it to represent reality sometimes wear suits, have fancy degrees, and jobs in the respected halls of academia.

Yet the theory remains as unfalsifiable as my theory of space monkeys pulling the strings of the universe. And it makes the same predictions. In my view, a theory without falsifiability or predictability is no theory at all.

Wednesday, September 24, 2014

Seeking housing supply logic

A four year long debate over the potential role of town planning regulations in determining the price of housing has occurred at MacroBusiness. The latest instalments are here, here, and here, and my model of understanding price effects of locational distributions of housing is here.

I want to briefly look at the logic of those who argue that planning controls strongly influence housing prices, and ultimately the affordability of housing in all forms (including rental affordability).

So far it seems there are multiple conflicting views being simultaneously held by the shortage protagonists, which I call the response dynamic, the expectation dynamic, corner the market, and the squeeze. Each of these suffers from the problem of invoking an implicit baseline for comparison which simply can not exist.

Response dynamic

In this argument there exist regulatory barriers, such as the need for a development approval, that delay housing supply in the short run while prices are rising, amplifying any resulting cycle.

While intuitively appealing, this argument suffers from a similar logical problem that most supply arguments do - the implicit invoking of a baseline that doesn’t exist.

You see, implicit in this argument is the idea that there exists a world of no constraints in which an unlimited supply of some asset could exist (as asset, by the way, is by definition a share of a finite bundle of property rights, which if could be infinitely supplied, would have a zero price). This baseline arises in economic theory of perfect competition under the implicit, but poorly understood, assumption of free entry. Yet free entry invalidates positive prices in general. So the logic of this baseline relies on the same faulty logic hidden in the economic model of perfect competition.

So if this fantasy world is not a relevant baseline, then what is? At best it is a world of monopolistic competition, where all land owners compete in slightly different markets, as there are no perfect substitutes to each location. Here, each land owner faces an independent choice of when to develop for housing given current market conditions. Essentially, is it better to develop today, or will I make more money developing later?

In this case a land owner, or even a developer who has approvals and is ready to build, will delay sales if the present value of the revenue from delaying sales and achieving a higher price later exceeds the cost of delaying. In practice this means that when prices are rising rapidly and sales are occurring quickly, it usually pays to delay sales by increasing prices. This is a simple matter of logic, and is supported by a large literature on real options theory that takes into consideration the one-shot nature of development, and the optimal timing of investment and sales decisions.

What this means for the response dynamic argument is that regardless of the nature of the town planning system, and even if the development approvals process was removed entirely, land owners and developers would still delay sales and construction of new dwellings during the price cycle. They would not, as is argued, somehow ignore their independently available option to delay sales for higher prices.

Expectation dynamic

Another argument goes as follows - a housing bubble can occur in an oversupplied or undersupplied market, but what determines the long run average price level is the potential abundance of future supply.

This ignores a number of things, the main one being that current supply should, and does, matter for housing affordability because rental prices arise from competition for the current stock of dwellings. What we are discussing in this argument is how the expected growth rate of rents and prices is capitalised into current prices. If potential future supply enters the price equation through an expectation of future growth in prices and rents, then this expectation can of course increase prices.

The flaw in this argument is that expectations of growth in prices can, and do, arise in markets where future supply is essentially infinite, both in experimental settings, and in many real life examples, such as the public land auctions of the 1830s. So invoking expectations of future supply as the specific mechanism for the formation of price growth expectations is completely arbitrary.

By the logic of this argument I could promise to flood a city’s housing market with new dwellings in 50 years time, and by backwards induction, current home buyers would reduce their willingness to pay and overall prices would fall. If you believe this would work, I’m happy for you to believe the expectation dynamic argument on housing supply.

Corner the market

This argument says that because there are finite locations for future housing supply, large developers are able to corner the market by acquiring sites and drip-feeding new supply in order to maximise their prices.

Holding such a view relies on the same implied baseline as the response dynamic argument, in that a world of free entry and unlimited supply is assumed to be a possible alternative but for town planning controls. Of course a world that has no town planning control, but merely divides the land in bundles of private ownership, will also have limited supply, and will suffer the same problem.

What this argument boils down to is a claim that even with many thousands of developers operating in our housing markets, and even with many millions of potential development sites whose owners will seek development in the future, somehow adding additional potential development sites will increase competition.

It is fantasy to imagine that if competition between 1,000 developers does not exist, that when there are 2,000 some kind of magical competitive pressure arises.


Lastly, there is the basic squeeze argument that says that because planning constraints are binding, in terms of setting an effective quota system (which they don’t, more on that in a moment), that we can think of their effect in terms of basic constraints in supply and demand equilibrium approach. This is the approach the RBA has used before to attempt to understand housing supply.

But controlling what is allowed to be built where does not also control that rate at which new development of any type is constructed. Local councils approve developments at the rate they are sought, and in my home town the rate of approvals increased by a factor of five during the early 2000s property price boom. Clearly there was no quota system on quantity, only location restrictions on where different type of development would be allowed.

The separation of location controls and quotas can be clearly seen with an example. Think of road space. We have a ‘road planning system’ that says cars driving east can only use half the road space, and cars riving west can only use half. Therefore this should create, by the same logic that location regulations restrict housing supply, a shortage of commuting.

But it doesn’t. It facilitates better commuting because conflicting uses are segregated. Imagine then a city that says one side of the river can only be used for residential development, while the other side can only be used for non-residential uses. Does that automatically create a shortage and a housing squeeze? Only if you assume it does.

Taken to the extreme, this argument again invokes the free entry assumption and the backwards induction logic of the expectation dynamic. If segregating housing to one half of the city creates a squeeze, so do does ANY geographic restriction, including the mere finite size of the world. If only the world was bigger, housing would be cheaper!


Even you edge to the middle ground and argue that it is some combination of these arguments, this doesn't address the fact that they independently ignore the basic realities of property markets. It is not to say that at some point there could be a supply side effects on prices, such as following natural disasters and so forth. But we must realise just how small the new supply of housing is compared to the whole housing stock to understand that any effect of the rate of new supply on prices can only be miniscule. The real problem with these arguments (apart from their inability to make sense of rental prices) is that invoking a baseline of perfect competition is absolutely a wrong assumption.

Even if you still want to ignore the logic of the arguments at hand, and insist that the evidence shows that, for example, houses are cheap in Houston, Texas, therefore their planning system is the reason why, you need to a) ignore that they have a strict planning system, which just does not have zoning as it is usually used, b) ignore that at the fringe of all Australian cities there are also cheap houses (eg. I found 121 houses for sale under $300,000 with over 3 bed, 2 bath, 600sqm of land, in Ipswich, west of Brisbane),  c) ignore the price effect of state property taxes in Texas, and d) ignore commuting costs from sprawling cities, which are avoided in more compact cities and get capitalised into prices.

Tuesday, September 9, 2014

Australia outbuilds Texas, but will the shortage myth die?

The claim of an Australian land and housing shortage is a myth of epic proportions, perpetuated by vested interests across the media landscape. An Internet search for the phrase ‘Australian land shortage’ produces a deluge of pronouncements confirming this view.

Those with an intimate knowledge of property-related data, or with experience in the development industry, can tell you that the alleged housing shortage is fiction. It is in the developers’ best interests, however, to remain quiet and reap windfall gains from beneficial rezoning (surely betterment taxes should be imposed on that?). The imagined scarcity of housing is promoted by think-tanks and other vested interests, and follows a familiar script:

Australia is not producing enough new land for housing due to policies pursued by state and local governments that prevent land supply and land use from responding to price signals. In fact, the supply of new land for housing has declined over the last decade, with the average number of lots produced in the five largest capital cities declining by 21%. The decrease in the supply of new land has not surprisingly seen an increase in land prices.

Everyone is familiar with varying forms of this argument that are ceaselessly promoted by think-tanks and housing lobby organisations. Unfortunately for the vested interests, it is nonsense. For an abnormal surge in housing prices to be attributed to supply constraints that arise from town planning regulations, the following events must have occurred:
  • Rents must rise relative to household income (the rent to income ratio), for if fewer homes are built, tenant bidding wars for scarce shelter must result in a higher ratio. 
  • Real rents must increase (the nominal rent to inflation ratio), as supply is squeezed by regulations that restrict the number of new dwellings. 
  • The occupancy rate must rise, as a lower dwelling construction rate results in more people per property. 
  • The council approval rate for new dwellings must fall. 
  • The rate of new construction must decrease. If government meddling prevents developers from building, the ratio of new housing construction to recent arrivals in a town, city, state, region and/or country must rise. 
The following data tests these assumptions taken for granted by most economists. In the figure below, the top panel shows the mean housing rents and prices in five of the larger Australian states, while the bottom panel displays mean housing rents and prices relative to incomes. In these states, the rent to income ratio during the 2000s is approximately the same as throughout the 1990s, but the price to income ratio has risen.

RBA statistics reveal the rent to income ratio actually fell during the late 1990s through to the mid-2000s, despite the unprecedented inflation in housing prices. After 2007, the ratio increased, but has remained under the peak established in 1999. Furthermore, the rent to income ratios for all income quintiles has remained steady during the housing price boom, except for a slight increase in the lowest income quintile.

Regarding the second criterion, real rents were trending close to zero during the late 1990s, and even turned negative during the early to mid-2000s as housing prices boomed, invalidating the assertion that supply shortages were squeezing the number of new dwellings.

The third criterion of the mainstream account also withers under scrutiny, for the occupancy rate (persons per household) has continued to fall during the boom, hitting a record low in 2007 at 2.72 before flattening out. This trend is in line with similar developed nations and has persisted, despite the average size of residential dwellings increasing to a record high.

The ratio can only fall so far, constrained by the social dynamics that affect family households and shared housing arrangements. The ratio of dwellings to people will never approach a one to one ratio, even when considering the surplus housing stock comprised of long-term vacant properties, second homes and holiday houses, and the extremely large size of the average Australian home.

Available data to assess the fourth criterion (housing approvals) also paints an interesting picture. For instance, South East Queensland provides an excellent measure – ‘Stock of lots approved by Council’ – that covers the early 2000s boom. Between 2000 and 2004, annual approvals increased from around 10,000 to 24,000, while the stock of approved but undeveloped lots surged from 25,000 to 46,000. If councils and planning controls really constrained development, the pattern should in fact be inverted.

Finally, in relation to the fifth criterion, the methodology often employed by economists when examining dwelling construction is faulty. New dwellings are sometimes compared to the existing population base, resulting in a downwards sloping trend. One could just as easily generate a rising trend by comparing the flow of new population against the existing dwelling stock (see here for an example). Further, measurement of the absolute and relative change in the number of dwellings constructed is pointless, for it says nothing about the change in population.

Flow to stock comparisons are occasionally valid, but it is nonsensical in the case of new supply. The existing (housed) population cares little about the availability of new dwellings, whereas it matters a great deal to the new net inflow of persons migrating into a city, town, region, state, or country. Australia has a relatively high home ownership rate of 68 per cent, with owners moving every decade on average.

Contrary to common perception, Australia has experienced a blistering rate of dwelling construction. From 1981 to 2013, Australia produced an average of one new dwelling per 1.76 new persons. Over the course of the housing price boom (1996 to 2013), the ratio is still a remarkable 1.93 - still way below the average 2.7 persons per existing household.

Due to moderate volatility a 2-period moving average trend line has been applied to smooth the series. The ratio is actually overstated, because it compares the new estimated residential population to new construction. A more appropriate measure would only include adults because children don’t purchase homes, make mortgage payments or pay rent, and it ignores increasing size of existing housing true to renovations and extensions.

Over the long-term, the post-WW2 ratio exhibits less volatility as a consequence of implementing stringent town planning regulations, perhaps by reducing uncertainty (a ratio of 2.0 between 1946 and 2012 compared to 3.8 between 1881 and 1945). This era also has a lower average ratio than pre-WW2, though this may be partially explained by more accurate data post-WW2. The supply of new housing has improved dramatically and remained responsive in the post-WW2 era in defiance of stricter planning regulations.

In contrast to the conventional supply argument, Australia has even out-built Texas, typically advocated as the ideal of supply-side efficiency. Between 1990 and 2012, where comparable data are available, the Australian and Texan ratios averaged 1.85 and 2.86, respectively. Over the course of both housing booms between 1996 and 2006 (the peak of the US bubble), the Australian and Texan ratios averaged 1.88 and 2.20, respectively.

When Australia and Texas experienced downturns during the GFC, the ratios in both countries increased, though more significantly in Australia’s case, leading to a sharp peak in 2009. Australia’s housing bubble failed to burst, following banking and housing interventions by the Rudd government. Consequently, the rate of construction has resumed its long-term average.

The comparison of the ratio of new population arrivals to new dwellings has some interesting implications for those blaming developer land-banking and other non-competitive practices on restrictive supply-side regulations. If the supply of new housing is being artificially constricted in Australia, how many new dwellings should developers be constructing? Should it be one new dwelling per new person? Ten new dwellings?

Fortunately, developers are run by businesspersons who are finely attuned to the real world functioning of property markets. Quaint notions of neoclassical equilibrium theory are discarded; such as the assumption the real estate market should operate in a perfectly competitive manner, with plots of land and dwellings modeled as if they come off a factory assembly line like candy.

In reality, developers carefully assess market conditions and make future trend estimates in an environment of uncertainty. They will certainly not imperil their future profits by constructing an absurd amount of dwellings in response to high prices. Rather, land banks are used strategically, in spite of the clamor against these practices.

The totality of the data discredits the mainstream housing shortage argument. In particular, the rent to inflation and rent to income ratios provide compelling evidence, alongside the occupancy rate. Apart from a few years during the GFC, both ratios have remained steady and/or even fallen for the most part during the housing price boom between 1996 and 2013. The data reveals the supply of dwellings has remained responsive and generally fulfilled the needs of a growing population, even as regulatory burdens have increased.

Conventional economic theory makes no allowances for private debt, land rent, speculation, bubbles, irrationality, instability and uncertainty – the characteristics of real world property markets. Yet, this has not prevented a number of economists from claiming that accelerating private debt results in greater volatility in restricted land markets.

This schizophrenic approach is inherently contradictory: on the one hand accepting that markets operate (correctly) according to disequilibrium price dynamics, but simultaneously modelling real estate markets as if they operate according to equilibrium price statics. Markets must be modelled with one or the other, not mixed and matched when convenient.

As equilibrium econometric modelling does not account for the role of accelerating private debt on asset (land) prices, a causative relationship cannot be established between supply factors and prices, only a tentative correlation. Even then, a thorough analysis of the data leads to misgivings about the role of restricted supply, with a strong rate of dwelling construction over the long-term in the midst of a supposedly strict town planning system.

The same cyclical pattern is found in the stock market, evidenced by the recent Dot-Com bubble which was the largest of its kind. What supply-side factors generated that enormous bubble in the virtual world? For centuries, developed nations have experienced recurring stock market bubbles irrespective of the regulatory environment. Thus, in a manner analogous to the real estate market, it would be absurd to blame high share prices on regulation, for instance, the government curbing the supply of new stock via IPOs.

Australia suffers not from a lack of supply, but high housing prices – seemingly contradictory if one abides by conventional economy theory and the standard account of town planning. Housing prices are abnormally high due to a bubble generated by financial deregulation and generous tax expenditures.

Even with plentiful supply, concerns over affordable housing, however, are still valid. The bottom 40 per cent of households by income has always struggled; it is not a recent phenomenon. Higher social welfare payments, rent controls and a greater supply of public housing would help to ameliorate affordability issues for the poor.

In conclusion, Australia has built a persistently responsive supply of new dwellings, relative to the flow of new net population. There has neither been a housing construction boom nor crash, but a healthy rate of supply at around one new dwelling for every 1.8 persons in recent decades. Assertions of a housing shortage and/or restricted supply are not supported by long-term data and a host of metrics.

Comments and data provided by Paul D. Egan and Philip Soos

Tuesday, September 2, 2014

Land taxes, rent-seeking, boom and bust

I came across this video today. It's a fairly concise argument about the rent-seeking activities and incentives that perpetuate bad economic policy. Some great comments in there by Mason Gaffney, Ed Dodson and of all people Milton Freidman.

Friedman explains how a tax on the unimproved value of land is economically the best tax, but politically the worst, and he also takes some blame for personal income taxes which he helped introduce during his time in the US treasury in 1941. He says
The question is, which are the least bad taxes. And in my opinion, and this may come as a shock to some of you, the least bad tax is the property tax on the unimproved value of land - the Henry George argument from many many years ago. 
It's an interesting thing if you're talking about taxes about why it is that the property tax is so unpopular. It's not unpopular for good economics reasons. It's unpopular in my opinion for one good reason.  It's the only tax left on the books for which people have to write a big cheque. 
The income tax is a far worse tax. But, and I have to admit that I have some part of the guilt in this process, because during World War Two I worked at the Treasury and helped to design the withholding system (my wife has never forgiven me for it).  But with respect to the income tax we've arranged it so it's taken off bit by bit and it's almost painless. With respect to the sales tax we pay a little bit of it each time. With respect to the corporate tax and excise taxes, they're hidden in the price of things we buy - we don't even know we're paying them. 
But with respect to property tax that remains a tax that we as individuals have to pay, and then we have to write a big cheque for it. That's the fundamental reason in my opinion why it's so unpopular. 
But if you wanted to reduce the unpopularity of the property tax the way to do it would simply be to provide for an effective method whereby it could be withheld at source in small payments. 
For me Friedman's last comment could be the key to unlocking land taxes as the primary source of government revenue in many jurisdictions. People in general have intense loss-aversion - they hate getting something and having it taken away much more than never having it in the first place. The practicalities of implementing a withholding system for land taxes should be seriously considered by land tax advocates.

I could suggest that for residential property, each State's rental authority (the authority that holds rental bonds in trust) also operate as a rent intermediary, deducting estimated land taxes from periodic rental payments in the same way that income taxes are withheld. This would allow landlords to see only rents net of land tax, and make investment decisions on the basis of this new net rental amount.

But rentals are only about 30% of the housing stock. For owner occupiers, vacant and non-residential land owners, I currently see no practical alternative but collect State-level land taxes in the same manner as the already are.

In truth, this loss-aversion problem with land taxes might reflect Friedman's world view more than the real political problem. After all, in Australia most State's already have land taxes, but with extensive exceptions, particular for owner-occupied housing. Incrementing the tax rate is a costless exercise politically, as only a tiny fraction of voters pay it, compared with alternative sources of taxation. It seems to me the hinderance here is more a case of the political power of the wealthy, who would be on the hook for most land taxes, regardless of how they are disguised.

In any case, a good video.