Content uploaded by Garrick Small
Author content
All content in this area was uploaded by Garrick Small on Sep 11, 2017
Content may be subject to copyright.
12/10/02 22:26 PM Auction Market Behaviour page 1 1
A paper for Presentation at the PRRES annual conference Christchurch, New Zealand,
Jan 2002
An Experimental Study of Auction Behaviour
Dr. Garrick Small
UTS, Australia, Email: garrick.small@uts.edu.au
Abstract
A laboratory simulation of property sales revealed that in a mildly
optimistic market, auction prices returned premiums to the rational
capitalised values, whereas tender sale did not. Results suggest that auction
premiums may tend to peak near the middle of the auction, however, the
properties were sold in order of diminishing value, so further research may
be necessary in order to distinguish between these two factors.
The paper relates the experiment to the developing literature on the
behavioural study of property auctions that tends to use analysis of actual
auctions. One advantage of the experimental environment is that unlike
actual auctions, the rational value of the property can be perfectly visible
to all subjects. The shortcoming of experimental simulations appears to be
that subjects may be less risk conscious.
The discussion raises several issues for future experimental design,
including the manipulation of market sentiment and the sensitivity of the
experimental situation to rewards and punishments in order to achieve
external validity. The paper concludes with a summary of the usefulness of
this type of property research and makes suggestions for its future
direction.
KEY WORDS
Price formation; behavioural property; property cycles; experimental property research; auction
behaviour
INTRODUCTION
Prices obtained at auctions have been thought to be the best indication of value since they result
from the free interaction of market participants. Lusht (1994a) raised some queries concerning
this conventional wisdom, citing the problem of what he referred to as the winner’s curse. He
12/10/02 22:26 PM Auction Market Behaviour page 2 2
contended that in analysing the value of a property, some prospective purchasers would make
errors in their appraisal, either through mis-allocation of risk or simple arithmetic error. He
contended that the person who erred most optimistically might win the property, but eventually
suffer financial hardship. It is also possible that the second last bid could be the rational price,
on the basis that the final bid clears competition by exceeding the financial value of the asset.
Lusht (1994b) found that bids at auctions were at discounts compared to rational valuations, as
did Allen and Swisher (2000) and Mayer (1998).
These authors were also interested in the patterns of prices within an auction. Lusht (1994b)
found that prices declined through the auction, while Allen and Swisher (2000) found they
improved and Mayer (1998) found no significant price trend.
The peculiarity of auction price formation suggests extra-economic behaviours that produced
unexpected pricing results. Their results indicate that real estate markets may be better
understood using behavioural methods following the general suggestion of Earl (1983) and its
property specific expression by Diaz (1999) or Hardin (1999). This study is an attempt to
understand market behaviour by studying market participant behaviour experimentally. It seeks
to better understand biases evident in the behaviour of actual bidders compared to its rational
financial value computed by sober analysis of the asset’s earning potential.
A shortcoming of this approach is that it leaves open important questions regarding the bidders’
analysis. In actual property markets, potential purchasers have differing perspectives on a
property’s investment potential, as well as behavioural inclinations in response to the auction
situation itself. The present work is aimed at controlling buyer analysis of potential properties in
order to observe behavioural biases that emerge from the auction situation itself. It achieves this
by using a simulated market environment to provide uniform market intelligence and valuation
methodology. In this laboratory environment, behavioural tendencies that distinguish bidder’s
behaviour from purely rational bidding behaviour can be studied.
AIM:
To experimentally examine behavioural biases resulting from the property auction environment.
OBJECTIVES
1) To identify pricing differences between auctions and tenders in a controlled market.
2) To examine behavioural tendencies within an auction.
3) To test if the auction prices conform to rational expectations.
HYPOTHESES
1) True value should equal the rationally derived financial valuation.
2) Auction prices differ from tender prices.
12/10/02 22:26 PM Auction Market Behaviour page 3 3
3) The auction situation can influence bidding behaviour.
4) Earlier sales within an auction can influence bidding behaviour.
PROCEDURE
An experimental methodology was employed involving a simulated property market following
generally accepted experimental design (Sarantakos, 1993). The experiment involved groups of
subjects playing a game that simulated a property market. The simulation began with the
competitive sale of a limited number of investment properties followed by simulated annual
rental negotiations between property owners and tenants. The general description of the game
and its rules is found in (Small and Oluwoye 1999).
Subjects were given an outline of a market where the currency unit was the bag of gold (B) and
the highest and best use of the property assets were publicly known. Subjects aimed to achieve
the greatest wealth by either occupying property as tenants, or earning rent from it as landlords
over a number of simulated years of negotiating annual rents. All subjects could bid to purchase
property parcels at the beginning of the game/simulation. Each parcel consisted of five identical
units, although the productivity of the units varied by parcel as shown in Exhibit 1. Property
purchases had to be paid for out of the rent earned over four years. Since the productivity of
the units and the cost structure of tenants was public knowledge, this meant that it was possible
to compute the value of the land very easily using a simple capitalisation calculation.
Land Market
Grade Parcels Lots
Product Notional
Rent
1 1 5 150 100
2 1 5 140 90
3 1 5 130 80
4 1 5 120 70
5 1 5 110 60
6 1 5 100 50
7 1 5 90 40
8 1 5 80 30
9 1 5 70 20
10 1 5 60 10
11 state 5 50 commons
12 state 5 45 commons
13 state 5 40 commons
12/10/02 22:26 PM Auction Market Behaviour page 4 4
Exhibit 1
The game has been found to be an effective teaching device as well as an experimental tool
(Small, 1999). The running of the simulation as a teaching exercise had the additional advantage
that rewards and penalties could be easily set as a consistent part of the simulation. The
exercise was allocated 5% towards the final subject grade, with bonuses for outstanding
performance enabling the winners to score up to 9 marks towards their final grade. As well as
proving to be a successful behavioural experiment, it has proven to be very popular with
students who tend to rate it as one of the outstanding exercises in the subjects in which it is run.
The simulation is structured to include observation of the formation of both rents and sale
prices. Small and Oluwoye (1999) found that it was a consistent and valid experimental tool
that yielded useful insights into the operation of rental theory. Those studies focused on the
rental market. That study found that the simulation returned results consistent with rental theory
operating within a near-perfect market. On that basis, the rental estimation and capitalisation
valuation can be adopted as reliable for this study.
The ownership and property value formation were a necessary part of the total simulation. By
allowing subjects to competitively bid for land, the prices were expected to absorb the marginal
value of the different parcels and therefore place property owners on an equal footing. In that
way it was expected that they would all be similarly motivated to seek optimum rents, which
was what was found in Small and Oluwoye (1999).
Property sale was competitive so that the subjects themselves formed the market. In each run
of the simulation either auction of tender bidding was used. The properties were all sold in
order of diminishing value. Future trials may consider reversing this order. Each subject could
enter the bidding, though ownership was limited to one parcel per subject. Subjects were not
obligated to bid if they considered the risks too great.
The experiment was run a total of eight times between March 1995 and March 2001, usually
on different groups of subjects. Runs 3, 5, & 7 were second year students who had
experienced the game also in their first years as runs 1, 2, & 4 respectively. The effect of
market learning was considered in Small and Oluwoye (1999) who found that learning had little
influence on rental market formation, as both were highly efficient. The impact of learning on
price formation will be considered in this study. Runs 1 & 2 used auctions, while the remainder
used closed tenders.
Subjects in simulation 3 were told that "productivities would increase sometime after year
one" before they computed their tenders. This was to create optimism by introducing an
unspecified expectation of future opportunity. The simulation typically ran for five or six
simulated annual rental cycles, of which the first four were critical for landlords since they were
placed in receivership if they failed to pay for their property in that time and were punished with
a zero grade for the exercise. Since the productivity increase was not specified, and could
happen in years five or six, a prudent bidder was not expected to place great value on it,
especially in view of the penalty that over-ambitious expectation would have on final subject
12/10/02 22:26 PM Auction Market Behaviour page 5 5
grades. Simulation 5 was used as a control for this run as both it and simulation 3 were second
year students with experience of the exercise.
SUBJECTS
Subjects were first and second year undergraduate real estate students. They were considered
an appropriate sample because of the following:
1) Their interest in real estate could be expected to be greater than the average and
representative of lay investors/tenants.
2) Their knowledge of real estate could be expected to be reasonably uniform, though not
well developed. In this they could be expected to reasonably mirror the profile of the
majority of small real estate investors and tenants. Their behaviour could be expected to
follow rational economic utility optimisation.
3) By incorporating performance in the exercise into subject assessment, meaningful
rewards and punishments could be incorporated into the game.
4) Their attention and continuous availability was reasonably assured.
5) Their motivation could be confidently expected on the basis of the game's learning
potential and the competitive spirit encouraged within the programme.
Ethical aspects of the simulation were considered due to the use of human subjects with limited
experience. The risk of severe penalty for poor performance was identified as a possible issue,
especially as it contributed to final subject grades in an undergraduate subject. Subjects were
permitted two options in participating in the exercise. One option was to be eligible for the
bonuses and penalties that could result in marks between minus 2 and plus nine out of a nominal
5 mark allocation. The second option was risk adverse, and involved being graded in a
conventional manner out of five, with no rewards and penalties. Subjects taking the second
option were not permitted to bid for property and on average earned about 3/5. Since students
had the choice to take risks or not, and even the risk-takers did not have to bid for property, it
was considered that the simulation met ethical requirements. The high regard held by students
for the simulation is an additional support for its claim to be ethical in its treatment of subjects.
RESULTS
The prices paid for the parcels sold in each simulation are shown in Exhibit 2:
12/10/02 22:26 PM Auction Market Behaviour page 6 6
Simulation
number
1 2 3 4 5 6 7 8
Sale type
Auction
Auction
Tender Tender Tender Tender Tender Tender
Market sentiment
Optimistic
Student year
1 1 2 1 2 1 2 1
Grade Rational 1997 1998 1998 1999 1999 2000 2000 2001
1 1880 1950 1920 2800 1880 1880 1800 1880 1880
2 1680 1815 1790 1680 1680 1590 1680 1710
3 1480 1700 1720 2080 1480 1480 1480
4 1280 1626 1655 1280 1140 1280 1280
5 1080 1900 1200 1120 1060
6 880 1420 1586 880 880 780 880 1080
7 680 1323 1215 1080 680 500 880
8 480 1040 855 460 420 480 480
9 280 610 510 780 280 220 120 240 300
10 80 290 370 80 60 800 40 87
Average premium: 339 322 648 0 -11 3 -5 44
Exhibit 2: Parcel Sale Prices
DISCUSSION
For the purposes of comparison, in this experiment there are four bases for price formation.
The first is the rational determination, the second is the result from auction simulations, the third
is the result from tender simulations and the fourth is the result from tenders in an unduly
optimistic market.
The rational prices are based on zero vacancy and rents that normalise returns to tenants. The
simulations were all run to ensure a slight under-supply of rental property that caused some
tenants to be forced onto the commons. The commons were necessary in order to provide an
alternative to the property market and provide the equivalent of public welfare into the game.
The commons provided a floor to wage expectations and a mechanism for allowing uniform
under-supply. For these reasons the assumption in the financial valuation would appear
reasonable, at least in setting a hard upper limit to value. Shrewd bidders were expected to
apply a discount to this value to cater for rents that may be struck under the optimum, though
such discount bids were relatively rare in the winning bids. The financial values and have been
used as the reference and premiums above rational values have been charted for each of the
simulations, as shown in Exhibit 3.
The auction simulations have a net average premium of B331, whereas the five tender
simulations have an average premium only B7.8. Given that the standard deviation of the
sample mean for the tender simulations is B20.3, the tender prices are not significantly different
to the rational values. Using the 42 observations that compose the normal tender simulations,
the standard deviation for a single price within the sample is B131.6 Using them as an estimate
of the population of rational market pricings, returns an estimated population standard deviation
of B133.2. The small difference between these two statistics suggests that B133.2 may be
12/10/02 22:26 PM Auction Market Behaviour page 7 7
confidently adopted as a conservative estimate of the population standard deviation for further
analysis. This facilitates the analysis of the auction and optimistic tender simulations using the
simpler Z test rather than the more common T test. Since T testing uses particular sample
means to estimate population parameters, it has the weakness of being forced to rely on small
sample sizes to estimate parameters that are consequently more likely to be less reliable.
Exhibit 3
AUCTION BEHAVIOUR
Exhibit 3: Premiums
On this basis, the standard deviation of the sample mean of the combined auction results is
B31.4, which produces an observed Z score of 4.42 for the auction mean compared to the
tender mean. For a 0.01 two tailed test, the critical Z score is 2.57, which means that the
results support the hypothesis that the auction results do not come from the same population as
the tenders. That means that the auction prices are statistically different to the tender prices, in
this case exceeding them. Similar analysis of the optimistic tender simulation reveals that it has
an observed Z score of 10.8, again indicating the optimistic tender simulation is statistically
different to the normal tender simulations.
The preliminary conclusion from these results is that in a sober, well-informed market, tenders
appear to return rational prices, whereas auctions under the same conditions tend to be more
Bid Premium by grade
-400
-200
0
200
400
600
800
1000
12345678910
Property Grade
Winning bid Premium (B)
1 Auction 1
2 Auction 1
3 Tender 2
4 Tender 1
5 Tender 2
8 Tender 1
6 Tender 1
7 Tender 2
12/10/02 22:26 PM Auction Market Behaviour page 8 8
bullish. The simulation that contained expectations of unspecified future growth appeared to be
the least restrained by rational computations. Each of these three situations will be further
considered in detail.
AUCTION BEHAVIOUR
Inspection of Exhibit 3 reveals that auction premiums grew through the first half of each auction,
them subsided. The parcels were sold in order of diminishing value, so the growth in premiums
may have been related in some way to the values of the properties. This would appear to have
been contradicted by the fall off in premiums towards the end of each auction, however, it may
have been the case that bidders were also wary of attaching high relative premiums to low value
properties. This would explain the later restraint.
Exhibit 4: Relative Premiums
Relative Auction Bid Premium by
Grade
0%
50%
100%
150%
200%
250%
300%
350%
400%
12345678910
Property Grade
Winning Bid Relative
to Rational Value (%)
1 Auction 1
2 Auction 1
Exhibit 4 charts the auction premiums as percentages of the rational prices. It is apparent from
this chart that the premiums did grow relatively throughout the auction in what appears to
approximate an exponential trend. A behavioural explanation for this may be that the early
premiums may have come from bullish bidders who were keen to become landlords at any
cost. The successive prices may have resulted from the effect of a psychological reassurance
provided by each previous sale. This is illustrated in Exhibit 5 that compares the average
auction premium by property grade against a notional initial 6% premium grown at 55%.
The tendency for premiums to diminish from the mid-grades to the least productive grade
(grade 10) may also have been influenced by growing realisation that the poorer parcels could
not support the higher repayments from rentals sufficient for successful purchase over four
years. In this case the cause of the latter restraint does not have to be tied to some subtle
mathematical analysis on the part of the bidders. This is more satisfying because simulation three
12/10/02 22:26 PM Auction Market Behaviour page 9 9
suggests that subjects are not capable of analysis of this sort. The bulge towards the middle of
the auction may either be explained simply by optimism feeding on the encouragement of the
prior behaviour of others, or by some subtle evaluation linked to property value. Future
simulations may use different orders of property value to explore this further. It is clear that
prices were well in excess of the rational bids, reflecting an over-optimism that resulted in the
majority of landlords going into receivership in those simulations.
0%
50%
100%
150%
200%
250%
300%
350%
1
3
5
7
9
55%growth on
6% base
Averge
premium
Exhibit 5
TENDER BEHAVIOUR.
Simulations 4-8 returned very conservative bids that relate well to rational expectations. The
average premium across the five simulations was negligible at about B7.8. Also apparent were
several sales at discounts to the rational prices. These were probably prudent considering the
riskiness of achieving optimum rents through the whole of the simulation. On the whole, it would
appear that tendering returns prices that are closer to rational valuations. This would appear to
contradict the conventional wisdom that auction prices are the superior indicator of true value.
For the vendor, this would appear to be a shortcoming of tendering, though other factors
should also be taken into account before choosing to use auction. Experienced property
brokers do appear to believe that auctions will return the better sale result in particular cases.
The results here confirm this popular belief, contrary to the findings of the studies cited earlier
that suggest auctions always return discounts.
TENDER BEHAVIOUR UNDER UNCERTAINTY
Simulation 3 was executed using tender, but subjects were told that there would be a
productivity increase sometime after year one. The prices bid reflect an inability to price this
risky prospect of future benefit. There is no evidence of the hump favouring the middle grades,
which further suggests that the latter was an artefact of the auction situation.
12/10/02 22:26 PM Auction Market Behaviour page 10 10
The simulation involved landowners having to cover their purchases using their first four years of
rent or face severe penalties. Diaz et al. (1999) considered the importance of rewards in this
type of experiment and the design of this experiment is in good accord with their
recommendations. The penalties were very real to the subjects as the simulation carried a
nominal 5% weight towards their final assessment and bankrupt landlords faced grades
between minus two and zero. The game is constructed so that the long-term winner is more
likely to be a property owner than a tenant, which is in conformity with common experience.
The typical tenant scored between three and four marks out of five.
Subjects therefore had to balance the attraction of the bonus marks against the risk of a
substantial real penalty. To bid high meant to be in the running to win a property, but also to
raise the likelihood of bankruptcy. All subjects in simulation three were bankrupted, as were
most in the auction simulations.
It may be argued that the subjects in simulation three were too naïve to bid responsibly. To
answer this criticism, their behaviour may be compared to simulation run 5. Both simulations
were second year classes that had also done the exercise in their first years, as simulations 1 &
2 respectively. The experience of over-optimistic landlords in the early simulations should have
warned subjects against bullish bidding. Simulation 5 returned the most conservative prices in
the entire experiment, while simulation three was at the other extreme.
The pattern of premiums is hard to validate financially. Premiums ranged from 50% to over
100% above the certain rational values. Given that the increases may have occurred anywhere
in the five years of the simulation, any risk-adjusted valuation of them would infer an
expectation of productivity increases several times that level. It would appear that little attempt
was made to systematically quantify this, with bidders focused only on winning the properties.
In actual markets, knowledge is seldom as complete as in these simulations. If it is the case that
optimism can so bias bidding behaviour as to take it well out of the realms of what could be
considered rationally defensible on financial grounds, it may be a major factor in price
formation.
IMPLICATIONS FOR PROPERTY CYCLES
There have been many attempts to explain property cycles, and there appear to be several
possible explanations for them. However, the behavioural tendencies suggested by this
experiment may also be applied to general market behaviour in a way that results in a
mec hanism that is capable of producing regular market fluctuations.
The two behavioural tendencies found in the simulations appear to be consistent with wider
market behaviour. If bidders do take signals from recent sales, then the effect may extend
outside the auction room to other instances of sequential sales to suggest behavioural
mechanisms in the real estate market generally. The trend in recent sales may therefore be a
major factor in bidder expectations of the likely winning bid in a coming sale. This is common
experience; in a rising market, all participants tend to expect the trend to continue.
12/10/02 22:26 PM Auction Market Behaviour page 11 11
Likewise, if the irrational pricing of optimism is a widespread reality, then it may have an
inordinate influence on any market that adopts unduly optimistic expectations. A market that
somehow develops optimism that property worth will increase in the future may therefore
inordinately value this expectation.
These two mechanisms could combine under optimistic conditions to explain the boom phase of
the property cycle. If pessimism causes comparably inordinate undervaluing of assets with the
market taking cues from previous sales, then the negative correction phase of the cycle is also
explained
Small and Oluwoye (2000) suggested a conceptual model for the real estate cycle based on
these two mechanisms with turning points related to financing and yield triggers. Such a model
has the advantage that it results in simple mathematical relations that describe the oscillations of
the market and has the capacity to accommodate both deterministic and chaotic fluctuation
patterns. This research provides an important support for that conceptual approach. The key
variable then becomes the actual quantification of optimism and the key research issue for
understanding market fluctuations is the mechanism by which markets switch from pessimism,
through neutrality to optimism.
DATA ISSUES
The data may be slightly misleading due to the fact that only winning bids were recorded and
parcels were sold beginning with Grade 1. Lusht has suggested in describing the Bidder's
Curse that the bidder who errs most optimistically will win the property, but faces ruin (Lusht
1994). In these simulations, the winner of any particular grade may have been over optimistic
compared to the rest of the group. Hence, the group’s behaviour is not completed observed.
Likewise, the simulation procedure allowed any subject one property only, so any inordinate
bids by that person for lesser grades dropped from sight.
This could partly explain the occurrence of discount prices towards the end of simulation 5 and
the reining in of premiums towards the ends of each simulation. The explanation would be on
the basis that the earlier sales had flushed out the most bullish purchasers, leaving the more
conservative ones to win the latter properties.
On one hand a more complete analysis of bidder behaviour may be available from a more
detailed analysis of all bids, or some modification of the simulation rules to allow multiple
purchase. On the other, it may not be desirable to explore this issue too far, because it risks
focusing too much on individual behaviour, rather than the more important social process of the
market’s behaviour as a whole. Auction prices are only those that come from the highest
bidders, and in general, purchasers do drop out of the market once their needs for property
ownership are sated. The object of property economics is to understand property market
behaviour, and while this may be informed by an understanding of individual participant
behaviour, the latter is not its primary focus.
12/10/02 22:26 PM Auction Market Behaviour page 12 12
CONSISTENCY AND VALIDITY
Experiments are only useful for research if they produce consistent results when repeated and
have a valid relationship to the actual situation that they model. In this experiment simulations 1
& 2 and also 4-8 may be used to check for consistency for auctions and tenders respectively.
In both cases, the results indicate consistency.
Validity is more difficult to identify. The simulations are simplifications of conditions in actual
markets. For example, few market participants have the level of certain knowledge that was
available in the simulations. Conversely, brokers are aware that certain market conditions do
favour auctions over other methods of sale, and these insights appear to have been validated by
the experiment, contrary to the previous studies on actual auctions. Likewise the phenomenon
of market feedback, where early sales appear to be used to set bidder attitudes later in the
market, appears to be consistent with actual market behaviour.
The most important issue in assessing validity is assessing the degree to which the experimental
situation models the actual situation. Two aspects of this are especially important, one is the
degree to which the behavioural mechanisms in the experiment parallel those of the real world
and the other in the relative balance of incentives and deterrents between the artificial laboratory
situation and the real world.
The experiment appears to utilise the same mechanisms as the real world, albeit simplified.
Purchasers have to evaluate the prospective properties in terms of their rental potential within
their financial constraints. Purchase is only profitable if subsequent rentals are sufficient to cover
financial obligations and the cost of failure is high. Although the financial, operating and leasing
mechanisms are simple, they are essentially similar to real world situations.
The appropriateness of the rewards and punishments is more difficult to defend. Resource
constraints have limited the running of the simulations, so fine tuning of the parameters is
difficult. Moreover, it may be dangerous to adjust the reward structure too much for fear of
merely setting it so as to obtain expected results. As it stands, the author would like to use the
severity of rewards and punishments as an independent variable in future trials in the hope of
containing the apparent over-optimism of auction markets. There may be ethical constraints to
taking this too far considering that the rewards and penalties relate to grades in university
subjects. It may be that more severe penalties for bankruptcy may reduce both tender and
auction prices. This is clearly an aspect for further research. What can be concluded however,
is that the experiment does appear to display reasonable consistency and defensible validity.
CONCLUSION
This experimental approach appears to offer insights into the behaviour of property purchasers.
In a well-informed market, the tender process appears to return prices close to rational financial
evaluation. By contrast, auctions appear to encourage prices at a premium to the underlying
12/10/02 22:26 PM Auction Market Behaviour page 13 13
rational value. Uncertain prospects of future benefit appear to be priced with irrational
premiums.
At first sight, the positive internal auction price effect appears to only correspond with some of
the literature, and is not substantiated by the majority of real-world studies. This may be
because of other behavioural issues. Likewise, the auction premium effect is counter to most of
the literature; Lusht (1994), Mayer (1998) Allen and Swisher (2000), and Brennan (1971) all
found that auction prices were at discounts to expected values. However, the instances
examined may have carried other implicit expectations of lower prices. For example, Allen and
Swisher (2000) studied properties sold at an auction devoted to mortgage defaults. Buyers
could be expected to be cautious about the valuations and also they would expect that the
vendors would not be as discriminating in setting reserve prices. Similarly, Lusht (1994) studied
the sale of bank branches in Australia as they were released in bulk onto the market. Buyers
could likewise have expected discounts and could also have been suspicious of the long-term
prospects signalled by the bank's choice to liquidate its property. Brennan (1971) examined
early auctions in Canberra (Australia) finding them to return heavy discounts as evidenced by
later private re-sales. In that case, there was considerable uncertainty regarding the viability of
the Australian Capital Territory. All of these instances were blighted by either undue pessimism
or the self-fulfilled expectation that the sales would be at a discount.
One advantage of the experimental approach is that it can be designed to control for these
extraneous variables. Indeed, it would appear possible to design experiments to explore their
actual operation in a way that cannot be done by post-hoc study of actual auctions. The
simulation/game that forms the basis of this study would appear suitable for this purpose.
The inordinate pricing of uncertainty is a finding that deserves closer study. If optimism can
produce unduly bullish markets, then its absence may explain the discount pricing found in
previous studies. Auctions do appear to be well regarded as an effective marketing method,
which would not be the case if they always returned discount prices. This author noticed the
trend towards auctions during the strongly rising market 1988 in Sydney that appeared to return
premium prices, though a systematic study of this relative popularity would be necessary before
drawing firm conclusions.
The behaviours revealed in this experiment are sufficient to explain market behaviour in boom
conditions and perhaps in the pessimistic phase of the property cycle as well. This would
appear to provide sufficient behavioural mechanisms to construct a behavioural theory of
property cycles.
The experiment appears to produce consistent results and is arguably a valid tools for
understanding actual property market behaviour.
References
Allen, M. T. and J. Swisher (2000). "An Analysis of the Price Formation Process at a HUD
Auction." Journal of Real Estate Research 20(3): 279-298.
12/10/02 22:26 PM Auction Market Behaviour page 14 14
Brennan, F. (1971). Canberra in Crisis. Fyshwick, ACT, Dalton.
Diaz, J. (1999). "The first decade of behavioural research in the discipline of property." Journal
of Property Investment & Finance 17(4): 326-332.
Diaz, J., R. Zhao, et al. (1999). "Does Contingent Reward reduce negotiation anchoring?"
Journal of Property Investment & Finance 17(4): 374-79.
Earl, P. E. (1983). The Economic Imagination: towards a behavioural analysis of choice.
Armonk N.Y., Wheatsheaf Books.
Hardin, W. (1999). "Behavioural research into heuristics and bias as an academic pursuit for
real estate disciplines." Journal of Property Investment & Finance 17(4): 333-52.
Lusht, K. 1994. Forecasting and Investment Selection in Property Markets. In Forecasting
and Investment Selection. Sydney: UWS.
Lusht, K. (1994). “Order and Price in a Sequential Auction.” The Journal of Real Estate
Finance and Economics 8(3): 259-66.
Mayer, C. (1998). “Assessing the Performance of Real Estate Auctions.” Real Estate
Economics 17(Autumn): 41-66.
Sarantakos, Sotirios. 1993. Social Research. Melbourne: Macmillan.
Small, G.R. 1999. Simulations, Property Markets and Behavioural Research. Paper read at
ISGA conference, at Sydney, Australia.
Small, G. R. and J. Oluwoye (1999). An Experimental Study of Market Formation Behaviour.
RICS Cutting edge Conference, Cambridge, England, RICS.
Small, G. R. and J. Oluwoye (2000). The Significance of Debt, Human Nature and the Nature
of Land on Real Estate Cycles. Pacific Rim Real Estate Society International Conference,
Sydney, Australia.
