to the subjects, it was rigged to land on only two different numbers. The
subjects were then asked if the percentage of African countries in the UN
was higher or lower than their wheel number, and by how much. The
results were dramatic. For the subjects who saw a “ 10” on the wheel, the
median estimate for the countries was 25. For those who saw a “65” come
up on their spin, the median number was 45. Subsequent experiments
done by numerous other researchers using different prompts have yielded
the same conclusions and the effect is now well-established. What’s truly
surprising is that the anchor might be obviously unrelated to the question.
Effects on Appraisers and Brokers
In 1987, Gregory Northcraft and Margaret Neale at the University of
Arizona showed a marked anchoring bias in valuation of real estate. The
researchers carried out two identical experiments—one on a group of
students and the other on a group of real estate agents. The teams were
given a single-family home to appraise, were sent to the site to inspect
the home and were given information packets that included the subject’s
list price, the MLS sheet for the subject, a summary of sales and other
industry-typical data. The subjects were then asked to estimate the
following:
1) The appraised value of the property.
2) An appropriate listing price.
3) A reasonable price to pay for the house.
4) The lowest offer they would accept for the house if they
were the seller.
The information packets were identical with the exception of the subject
property’s listing price, which was set by the researchers to one of four
separate values. All of the test subjects’ value estimates showed significant
evidence of anchoring in the results, as seen in one summary chart
below:
Based on the agents’ responses, the chart
shows a significant variance in appraised
value of $14,550 or 12 percent. These
subjects were experienced real estate agents
who had been practicing for several years in
the field. Because the only difference in the
information was the listing price, the only
possible explanation for the variance is the
information about the list price. Moreover,
when polled about their decision-making
processes, very few agents identified the
listing price as an important consideration
in their deliberations.
Similar results have been found using
professional appraisers as subjects. The
subject’s list price serves as an anchor to
the value of the property and is frequently
not adjusted adequately to account for the
bias. Given the potential for the subject
listing to improperly influence opinion, it
is interesting to consider the utility of the
requirement under USPAP to analyze the
current subject listing.
Effects on Property Buyers and
Negotiators
Buyers may also be subject to anchoring on
numbers that are not relevant. As shown
above, if a list price is too high, it can
influence the offers, the appraisal and the
final sale price, leading a buyer to end up
paying too much for a property. In typical
market transactions, a potential buyer’s
agent may have a tough time convincing
their client that a price is irrational if the
buyer has anchored onto an inappropriate
number. More likely is the challenge facing
a buyer or a buyer’s agent if the seller has
anchored on a bad number. Sellers are very
susceptible to anchoring on non-market
indicators such as previous price paid, a
bad appraisal, or a nearby property they
perceive as comparable but in fact is not a
good comparison.
The “loss aversion” phenomenon plays a
large role in inducing anchoring behavior.
This is a concept whose key idea is that
people react differently to losses than to
gains. More specifically, losses are shown
to be felt twice as powerfully as equivalent
gains. For example, it hurts twice as much
to lose a $20 bill than it feels good to
Pre-Set
Listing Price
$119,900
$129,900
$139,900
$149,900
Appraisal
Value
$114,204
$126,772
$125,041
$128,754
Purchase
Price
$111,454
$123,209
$124,653
$127,318
Lowest Offer
$111,136
$122,254
$121,884
$123,818
Listing Price
$117,745
$127,836
$128,530
$130,981
Results for Experiment 2: Mean Estimates of Expert Subjects
