4th Pacific Rim Real Estate Society Conference
Perth, 19-21 January 1997
WHY THE HYPOTHETICAL DEVELOPMENT METHOD IS FLAWED.
R.T.M. Whipple, PhD发传单
Emeritus Professor
Curtin University
e-mail: au
Keywords: Development Projects, Valuation, Valuation Standards
Abstract:
荠荠菜The method of hypothetical subdivision has been favourably viewed by the courts and is endord by the relevant professional body.
It is one of three variants that need to be carefully defined in order to asss reliability. The are the method of hypothetical subdivision in asssing the value of in globo land and the method of hypothetical development in asssing the value of a single site. The third variant is a combination of the other two.
All three are flawed to varying degrees. Two flaws are common to all: first, the u of magnitudes that are comparable only if the real rate of return is zero; cond, the importation of cost estimates produced in markets only tenuously related to the real estate market and which cannot be expected to predict the lling price of real property.
The third variant introduces unnecessary complications, magnifies the possibility of rious error and fails the test of Occam’s razor. Furthermore, its u assumes the margin for risk in both land subdivision and building operations is known. As this is frequently not the ca, rious distortion results – even if the previously noted objections are without force.
Concomitantly, u of the methods to analy a profit and risk factor from sales of land that has subquently been developed and sold will produce a result that is difficult, if not impossible, to interpret. This is so for all variants. Further, the traditional profit and risk factor is unrelated to the business decision process and is a concept without meaning.
WHY THE METHOD OF HYPOTHETICAL DEVELOPMENT IS FLAWED To study without thinking is futile
To think without studying is dangerous.
Confucius, Analects, 2.15
Introduction
The proposition I wish to investigate in this paper is that the method of hypothetical development is flawed and should not be ud as a real estate pricing model. I acknowledge at the outt that it does have a part to play in feasibility analys and in the economic ranking of alternative land u proposals as part of the valuation process, but I am not concerned with tho applications here. My concern is with its u as a means of predicting the price likely to be fetched in the real estate market. This is very much the concern of the valuer becau the objective of valuation is price prediction.
Some of the problems detailed below I first raid in 1987 (published in Whipple, 1998) and repeated them in Whipple (1995, pp. 442-443). Becau they are so obvious I did not give them a d
etailed treatment but, as the errors persist in valuation practice, a more comprehensive treatment is called for.
Variants
The general method has a number of variants that need to be carefully distinguished. Variant 1 – Predicting the Selling Price of Land In Globo
This variant generally goes under the name of “the method of hypothetical subdivision”. It is accepted as a valid method of price prediction by the courts – most recently in Joondalup Gate Pty Ltd v The Minister for Lands, 1996 (unreported). It is also listed as a method of valuation in the Australian Institute of Valuers and Land Economists’ Due Diligence Guidelines, page 4, issued 1/96. The variant is therefore well known and will be illustrated below.
Variant 2 – Predicting the Selling Price of a CBD or Shopping Centre Site传统童谣
In this application a site is hypothetically developed to its “highest and best u”(more accurately, its “most probable u”). The building is notionally rented and outgoings computed so as to calculate its net annual income. This net income is then capitalid at a lected rate to give an estimate of capital value. From this the cost of the building is deducted to yield an estimate of land value.
It is not uncommon to encounter this method in the context of predicting the price of a CBD site. The same approach is adopted in the feasibility analysis of a drive-in shopping centre site after the results of market analys are to hand. While there are some technical differences between the two applications, they are similar in kind.
Variant 3 – Predicting the Selling Price of a Site for an Integrated Development
This is a combination of the preceding two Variants. The site is subdivided, buildings notionally erected on each site, the buildings are lead, the net income is capitalid and, from this capital sum, all development costs are deducted to result in an estimate of the raw land value.
We now consider each variant in turn.
Variant 1 – Hypothetical Subdivision
To illustrate the discussion, refer to Table 1, taken from the source indicated. The layout is typical of the method adopted by most practitioners. By way of background, it was assumed the project would span 19 months from the date of acquisition – the reader is referred to Whipple (1995) for other assumptions underlying the magnitudes in Table 1.
It is important to note that the figures entered into the model are costs and returns using current money values.
The model predicts a price of (say) $680,000 whereas the site sold for $870,000. A model which mis-prices so widely in a particular ca cannot be a general model –
<, ud at large. The situation is akin to the well known problem regarding the argument from induction. One obrvation inconsistent with the conclusion invalidates the generalisation.
Such underestimation is typical of the writer’s experience when employed in the real estate development industry. Having been outbid for development sites on a number of occasions, my colleagues and I probed the model and quickly came to the realisation that it is fundamentally flawed. In a growth situation we found it usually underestimated raw land value; converly, in a period of economic contraction, it usually overestimated raw land value. We stopped using it some 40 years ago and it is rather surprising that it is still in vogue among valuers and accepted by the courts. Well, perhaps one shouldn’t be too surprid. Nevertheless, as a price prediction model, it is unstable.折字开头的成语
There are three flaws in the u of the model – each is now reviewed.
Flaw 1:
The magnitudes recorded in Table 1 are not comparable. Recall they are all expresd in to-day’s dollar values and purport to be prent values of their associated cash flows. The only condition when they will be prent values of tho cash flows is when the real rate of return is zero.
Table 10.20 in Whipple (1995) is a cash flow analysis of the same project. That is not reproduced here but readers who refer to it will verify that gross realisations occurred as follows:
Month Lots Sold Prices PV @ 5%
$ $
128472,000449,027
136354,000335,373
144236,000222,654穿越言情
153177,000166,298
163177,000165,608
173177,000164,921
182118,000109,491
----------------------------------------------------------------
Total: 29 1,711,000 1,613,372
The third column shows the pattern of lot sales using to-day’s dollar values. At the time this estate was developed, the real rate of return would have been around 5%. The last column shows the prent value of the relevant cash flows using that rate. Obviously, a nominal rate should not be ud becau the cash flows have not been inflated.
Note the difference between totals of the last two columns: $97,628. In prent value terms, this is equivalent to the lling price of 1.7 lots.
The figure for gross realisations that should be entered in Table 1 is $1,613,372 (shown in italics in Table 1) – not $1,711,000.
Similar comments apply to the other entries. It would be tedious here to discount all of them for prent purpos but, just to reinforce the point, take the ca of earthworks, drainage and road construction.
In to-day’s dollars, that item costs $193,000. Reference to Table 10.20 already cited shows the following actual timings (using to-day’s dollars):
Month Cost PV @ 5%
$ $
1156,21353,700
1257,15154,369
1379,63675,446
--------------------------------------------
Total: 193,000 183,515
Prent values are shown in the last column (e also the italicid entries in the last column of Table 1). The difference is $9,485 – the per lot cost of this item is equivalent to 1.4 lots.
Whereas in Table 1 the sum of $1,711,000 is treated as being in the same currency units as the $193,000, the correct comparison is between $1,613,372 and $183,515.
When repeated over almost all the items, it may be en that the table is misleading. The manipulation of numbers that are not comparable can only yield meaningless results. In fact, Table 1 is without content.
The only time when this will not be so is if the real rate of return is zero. That is easy to demonstrate:
PV = FV(1 + i) -n
from which it may be en that PV = FV when i = zero.
Hence, on this score, it follows that the method of hypothetical subdivision should be ud only if the real rate of return is zero. There are, however, other problems with it. Flaw 2:
The method of hypothetical subdivision is really another manifestation of the cost approach to price prediction and it therefore suffers from the same illogicalities.
The approach may be characterid in the form of what is referred to as “the developers’ equation”:
Land “Value” = Net Realisations – Profit – Development Costs – Finance Costs
For the equation to hold, we require
考研录取分数线1.the dollar equivalents to be comparable which, as shown above, is so only if the
real rate of return is zero AND
2. that the items on the right hand side of the equation to be in terms of values
determined in the real estate market – more correctly, in units of land value.
小酸菜We shall refer to the “profit” factor below. Net realisations, if correctly estimated, are reasonably assumed to be determined in the real estate market. This is not quite so, however, with development and finance costs.
Finance costs are t in the finance market. Now it is true that the demand for finance employed in the real estate market is determined partly by that market – but by other markets as well.
Similarly, the level of development costs is a product of forces within and beyond the real estate market. One simply has to think of the materials supply market, the market for transport rvices and such like. The level of professional fees is at least
recommended by professional bodies as well as influenced by the demand from clients operating beyond the real estate market.
It may be argued that costs so determined are paid by developers who are, of cour, an integral part of the real estate market and the resulting level, when paid, is real estate market determined. This overlooks the fact, however, that such estimation is no guarantee that the developed sites will fetch a price in the real estate market sufficient to cover tho costs and to provide an adequate profit and risk margin.
It follows, therefore, that an equation which has elements which are only partly determined in the real estate market and which (on its right hand side) are not in units of land value cannot coalesce into a prediction of the price the raw land may fetch in the real estate market – unless the whole eco
nomy is in a state of stationary equilibrium. That ems never to happen. Therefore the cost approach, while it has an important role to play in feasibility analys, should not be ud as a real estate price predictor – which is to say valuers should not u it as a valuation method. It is, however, listed in the Institute’s publication cited above.
Furthermore, the way in which items of capital expenditure interact directly with each other and indirectly with others in the list is little understood. To asrt that the effects are additive (as the model does) is to simplify that which is a very complex t of interrelationships.
In short, the model is illogical – but so, too, is the cost (or summation) approach. I have canvasd the arguments for this at some length in Whipple (1995, p. 488-492). Flaw 3:
We turn now to the profit factor. There are some real problems with this item – both conceptual and mechanical. Let us examine the mechanical aspects first. There are two sides to this coin.
First, it will be en from Table 1 that this is computed by proportion. The rate of 20% is ud in the example and this ems to be the generally adopted rate in contemporary valuations of subdivisional land.
Whereas the calculation is bad on net realisations of $1,621,250, rather it should be bad on the prent value of gross realisations of $1,613,372 (as shown above) less the prent value of lling costs. The latter is $84,627. Hence the ba figure should be $1,538,745 (shown also in the last column of Table 1).
The profit and risk factor, computed from this discounted sum, is $254,790 –compared with $270,208. The difference is $15,418. By virtue of the method of constructing Table 1, that error is carried down to the final value estimate. In the process, other errors are picked up as well for the reasons just given.
But is this quite right? No, it isn’t. This is becau the actual margin over costs is itlf a t of cash flows. They occur as net cash (hopefully) emerges from the project. The correct measure of the profit and risk factor is the prent value of its component of the net cash flows at the real rate of return.
一分一档
