Tuesday, April 17, 2018

Delay or Develop? What really determines the rate of new housing supply

Recent reports by Grattan Institute and the Reserve Bank of Australia have argued that zoning is a significant cause of Australia’s high home prices. Yet neither organisation has applied the appropriate economic theory to the property market, leading to conclusions that are almost the complete opposite of reality.

The main issue in property is that the static equilibrium assumptions of short-run supply and demand economics do not apply. If you try to apply these models you will interpret the market, and policy effects on it, incorrectly. Please read this article for some background on the gap between reality and what static equilibrium means when applied to land markets.

I’m not being radical. I’m not trying to earn street cred by being the anti-establishment economist. All I am doing is applying the totally standard, but correct, economic framework of real options.

It took me a long time to learn about property markets and how important real options are to understanding them. When I began studying economics the stories most economists were telling about property markets conflicted with my previous experience as a trained valuer working in the development industry. I had to really search to dig out this often-ignored but crucially important part of economics.

I want to use this post to explain why static-equilibrium analysis of the supply and demand type does not apply to property, provide a quick lesson on real options, and show how real development behaviour is best predicted by a real options approach.

First, the monopoly question
Land is a monopoly. This is fundamental to understanding property markets.

The reason is that there is no free entry — any potential market entrant must buy land from an existing monopoly owner. In practice, this means that property developers cannot be in the business of maximising turnover or undercutting each other on price since once they have sold all their new dwellings on one parcel of land they are out of business. They must buy back into the market from another land monopolist.

It is only because of this monopolistic power that land has a non-zero market value at all. Indeed, for a land market to be competitive we must be able to produce land (locations) with non-land (non-location) inputs.

In practice this means that each landowner is their own ‘little monopolist’ and their individual incentives are reflective of the incentives of the market as a whole.

The vacant land problem
The trick to understanding the dynamics of land development is to ask the question ‘why is there vacant or underdeveloped land’? In a short-run equilibrium model of supply and demand this can’t happen — all options to develop must be taken up (this underlying model gets a geographical twist in the Alonso-Muth-Mills model).

Let me quote David Pines on this.
The static approach in the Alonso-Mills-Muth model is useless in explaining many stylized facts regarding the urban structure and its evolution through time.

The reason for the failure of the static model in explaining these ‘irregularities’ is that the housing stock is assumed to be perfectly malleable, which, of course, is highly unrealistic.
What he means is that to ‘clear the market’ the model requires complete demolition and rebuilding of the city in response to any change in price, population, or preferences in order to ‘clear’ the market. This is obviously not how the dynamics of housing supply operate. In the model, there cannot be any vacant land nor opportunities to develop — all development has already taken place!

The reason there are still vacant plots of land able to be developed becomes clear only in a real options framework. A vacant plot, despite making no current income, contains options for future uses, such as to build a house, a new retail centre, or new commercial or industrial facility. It has a value because of the future options for income flows it represents.

Because development is a one-shot game, the decision for a landowner is a joint one of what to develop (residential or commercial, a 10 storey or 20 storey building, etc.), and most importantly, when to develop. Developing land now eliminates potentially valuable options to develop differently in the future.

A real options example
Let me try and convey the basic idea of real options as they apply to land development with the aid of the below diagram. Only through this lens can we consider the crucial question when development will take place, as well as how much will take place on a particular plot of land.

If you are worried about the total growth in the housing supply then the ’when to develop’ question is the much more important one.

The diagram shows on the left a ‘binomial options tree’ with the available future options for apartment development in two years time compared to the optimal (profit-maximising) development today for a hypothetical plot of land without zoning controls or limits on development density.

Two possible future states of the world are shown; one where price growth for apartments means a 20 storey building is optimal and profit-maximising at that point (providing a $15m profit), and one where prices rise only a little, and a 10 storey building is still best but at a lower total profit (of$12m). Each is judged to have a 50% chance of occurring.

Under this scenario, we can now consider the joint problem of the landowner — when to develop, and what to develop (10 or 20 storeys)?

The ‘when to develop’ question can be answered by comparing the present value of building now or waiting and having a 50% chance at each of the two future options. With a 10% per year discount rate we simply consider whether the present value of building today exceeds the expected present value of waiting.

Build now: PV = $10m Wait two years: PV = ($12m x 0.5 + $15m x 0.5) / 1.21 =$11.2m

In this case, the best thing to do is wait and keep the property vacant for two more years. Then, in two years, the same decision will again be made, and perhaps then it will also be optimal to delay.

On the right part of the diagram I have shown a scenario with zoning that applies a strict height limit of 10 storeys. Here there is no upside option from waiting. In the language of real options we have ‘reduced uncertainty’.

We can then rerun our calculations to see whether waiting or building is the profit-maximising choice.

Build now: PV = $10m Wait two years: PV =$12m x 1 / 1.21 = $9.9m Look at that! Now the profit-maximising decision is to develop a 10 storey apartment building today. By imposing zoning we can increase the supply of housing by a 10 storey building’s worth of apartments compared to the alternative no-zoning situation! Providing the option to build higher in the future increases the present value of the land, but also provides the incentive to delay development! The same logic applies to zoning rules that allow both commercial and residential uses. Removing commercial development options for landowners can bring forward residential housing supply. If that sounds a bit crazy and contrary to ‘economic intuition’, maybe you will take more seriously Sheridan Titman who made the exact same argument in the American Economic Review back in 1987. … if uncertainty is increased in a manner that keeps the state prices constant, prices of both land and building units as well as rental rates will increase, a larger portion of the land will remain vacant, but taller buildings will be constructed. Let me translate. If “uncertainty is increased” means that more future options are added to a landowners rights, like what happens when zoning controls are removed. Keeping the “state prices constant” means that the relative prices of different types of dwellings or commercial buildings are expected to be the same when the uncertainty change happens. The rest I hope is straightforward. In effect, this is the opposite of what anti-zoning economists have been saying. This logic applies to the land market as a whole, and to any individual parcel in it. There is no magic economic mechanism that means that removing zoning controls in one place increases those land values and can delay development there, but removing it everywhere decreases land values because somewhere else development has accelerated. Shouldn’t prices of zoned and un-zoned land equalise? No. Land is not a physical object. It is a set of legal rights that define the available real options. Property valuers and lawyers call this a ‘bundle of rights’ approach. Land is worth whatever the highest and best option is from the selection of legally defined rights. For example, you don’t own the minerals under your land, nor the airspace above it. If the law is changed to provide you the right to access and sell those minerals, your property will be worth more because of that option (if there is a positive probability of using the new minerals right). A new ‘property right’ that is separate and additional to the previous rights is now bundled together with those previous property rights. The same applies to zoning. There is no economic logic to the often-repeated argument that land with different zoning rights should equalise in value under market conditions. New zoning creates a different set of property rights. The value of two different set of property rights will be different if the highest and best option available in each is different. Actual development behaviour reflects real options This real options approach is the only way to make sense of the actual behaviour of landowners and developers in the market. There is no point arguing for removing of planning controls to ‘let the market work’ without understanding how land markets actually work. Here are some examples. 1. The Brisbane City Council has been repeatedly up-zoning an inner-city industrial site owned by Parmalat. The problem here is that this increases the value of waiting to develop, offering the global dairy company a free boost to the balance sheet by sitting on the inner-city site rather than selling and it moving to an industrial area. 2. When I was working for the property developer FKP we had a new building approved and ready to start off-the-plan sales at the Sunshine Coast during the early 2000s boom. There was a queue at the sales office on opening day, and by mid-morning dozens of sales were made. The prices were set months ago and market prices had unexpectedly increased since then. Continuing to sell quickly at these older prices and undercutting rivals was not the optimal thing to do in a real options world. So we closed the sales office early and put all the prices up. It then took nearly three years to sell the remaining apartments in that building. But that was profit-maximising in the sense of exercising our option to delay selling. There was only one chance to maximise profits from that site. 3. A recent paper on rent-control in San Francisco shows that when you eliminate the option to keep your current tenant at a higher rent next year, you are more likely to exercise your option to redevelop the site into apartments. This is a classic example of decreasing uncertainty in a real-options world and bringing forward in time execution of the remaining options. 4. Adding costs to development on a per dwelling basis can bring forward development because it reduces the payoff to waiting to develop to more dense uses. This pattern was seen in Queensland when developer charges were changed suddenly and those areas where charges increased saw faster new development compared to those areas where charges were decreased. 5. When Lend Lease had their site at Yarrabilba (in Queensland) rezoned from rural to residential they had argued that there was a housing shortage and that only if their site was rezoned could new homes be built. Once they got their approval they told their investors the project would take 'approx. 30 years’ to build those promised homes. Their optimal strategy is to delay and dribble out new homes, not to flood the market and undercut others on price. A similar situation has happened at Springfield, where the developer has had their own act of Queensland parliament granting them extensive freedom to develop as they choose since 1997 — you can't get more freedom to develop than that. A good summary of these planning gifts is as follows. The Queensland and federal governments have invested more than$1.2 billion in the region’s infrastructure and the Springfield rail station is state of the art. But the most extra­ordinary gift from the Queensland government was the Local Government (Springfield Zoning) Act 1997. This law puts all the planning and development powers for Greater Springfield in the hands of Sinnathamby’s Springfield Development Corporation.
And yet, 20 years later, the area is one-third developed. They are optimally delaying development.   Between Springfield, Yarrabilba, and the dozens of other similar developments in South East Queensland, there are hundreds of thousands of zoned plots of residential land waiting to be developed. The delay is because the landowners possess attractive future options.

A comment on the political economy of property
Developers hate zoning and planning rules because they want the flexibility to delay. If removing zoning controls did lead to a flood of new supply and lower prices, as the static supply and demand approach might suggest, then developers are the worst lobbyists you can imagine!

Is it plausible that they have been lobbying for years to drastically reduce their profit? Or more plausible that they use supply and demand economics as a cover story for what is really happening?

Only a real options view can make sense of this lobbying. Removing zoning gives current landowners, especially the large land-owning developers, more valuable future development options without requiring them to build them until they decide it maximises their profits!

In sum
It is time to start using the correct economic framework to analyse property markets. Only then can we make policies that deliver planning and housing outcomes we want. Otherwise, we will implement all the wrong policies, and in the process providing windfall gains to the development industry.

Friday, April 6, 2018

Queensland is giving its gas away

Australia's mining and energy industry often claims to be a significant source of public revenue. One case where that is not true is in Queensland's coal seam gas sector. Back in 2012, the sector gave the impression that it would provide a massive financial windfall for the State government via royalties. Even The Economist magazine noted the sector's "glittering promise of jobs and royalties for governments."

As you can see below, the forecast growth in royalties to over \$600 million per year did not happen, nor does it seem likely to in the near future.

Coal seam gas companies had an incentive to overstate their economic benefits when applying to the Queensland government to approve their operations in the face of substantial opposition from farmers and environmentalists. Unfortunately, these exaggerations were taken seriously in the budget and were a likely influence in the approvals process as well.

One way to get a more accurate estimate of expected royalty revenues royalties is to insist on upfront payments of the first 5 years forecast royalties at a discounted rate. Companies that finance this payment will demonstrate that their royalty forecasts are credible. The economic claims of those that can't (or won't) can be dismissed as not credible.

Forcing the mining and energy sector to put their money where their mouth is one way to stop exaggerated economic claims being made. If approvals for resource projects are going to hinge on economic outcomes, including future royalty incomes for the State, it seems important to get a forecast backed up by dollars rather than by economic modelling and wishful thinking.