Basic markets – how do they
Does the simplistic demand-supply
model make sense?
Is there actually a game behind it?
Will real people behave according to
the theoretical predictions?
Introductory example: the
first market experiment
Chamberlin (JPE, 1948) conducted bilateral trading
experiments with his graduate students at Harvard
to illustrate how perfectly competitive equilibrium might
Subjects could each trade one unit of a good
Price higher than equilibrium 7 times, lower 39 times
“… economists may have been led unconsciously to share
their unique knowledge of the equilibrium point with their
theoretical creatures. The buyers and sellers, who, of
course, in real life have no knowledge of it whatever.” (p.
Response by Vernon Smith
Vernon Smith, a former Harvard student (and Nobel Prize
laureate in 2002), changed Chamberlin’s trading institution in
the following way:
Instead of having subjects circulate and make bilateral
deals he used the oral double auction procedure.
He also implemented the method of “stationary replication”,
which is a sequence of trading days with stationary demand
and supply schedules.
“These two changes seemed to me the appropriate
modifications to do a more credible job of rejecting
competitive price theory, which after all, was for teaching,
not believing...” (Smith 1991, p. 155).
Details of the double auction
Each buyer i is paid according to Bi(xi)-∑pi where xi denotes the
number of goods bought and Bi denotes the buyers’ utility from
consuming xi goods.
Each seller is paid according to ∑pi-Si(xi).
There is a limited time for trading per “market day”. If trading
ceases before the time limit is reached the “day” ends.
Within a market period a buyer can make price bids to the group
of sellers for a specified quantity and/or accept a seller’s price
offer for a specified quantity at any point in time.
Within a market period a seller can make price offers to the
group of buyers for a specified quantity and/or accept a buyer’s
price bid for a specified quantity at any point in time.
Improvement rule: A new bid must be better
(higher) than the highest standing bid. A new
offer must be better (lower) than the lowest
If a bid (offer) is accepted a binding contract
In general, individuals only know their own
Bi(xi) or Si(xi) values.
Is the outcome in the DA
Demand and supply change during a trading period.
Nothing ensures that trade will take place at the
CE. Notice that the number of CE-trades is in
general smaller than the number of economically
feasible trades. In principle it might be possible that
all feasible trades take place.
There is no rigorous game theoretic prediction.
No well defined game!
Intro example 2: Cournot Mergers
N<5 firms in market - merger between two
makes them suffer and outsiders gain!
Merger beneficial if outsiders did not change production
Optimal response of merged firm to unchanged
behaviour of others : lower output
But optimal output of others rises => further drop in
optimal output of the new firm!
In new equilibrium industry profits rise. But market
share of merged firm drops from 2/n to 1/(n-1)
Huck et al (2007) lab study of mergers
Total output close to theory
However, individual outputs different
Merged firms produce more than unmerged
Unmerged firms best respond to that
Mergers in larger market weakly profitable!
Now what? The state of
Over time we have gathered a wealth of evidence
regarding deviations from Nash equilibria in many
The number of experimental papers has grown
Recording deviations is not enough, we have to
Modern economic theory is interested in
incorporating the insights gained from that process
to improve models
All pay auction
You pay your bid independently of winning or
Let’s try it!
How did you play? Is that an equilibrium?
Insight actually used to make real money (swoopo
and other penny auctions sites)
How is this similar to a patent race?
The owner of a company
A manager that wants to buy it
Value to owner V is unknown to manager,
but distribution uniform in the set
Value to manager is 1.5*V
How much do you offer?
What is behavioral IO
Traditional IO is based on Nash equilibrium
Rational firms, rational consumers
Consistent beliefs - best responses
Rational firms best responding to irrational
Irrational firms - no best responses?
Rational firms with wrong beliefs? Rational
consumers with wrong beliefs?
Common deviations from
“rationality” - bounded
Right beliefs - no best response
Levels of reasoning
Best response - wrong beliefs
N players, can say a number [0,100], winner is
the person closest to ½ times the average
Undercutting games Guessing games
Level k model also
behaviour well in 2x2
games, 3x3 games,
Using models of bounded
QRE can explain behavior in all pay auctions
Overbidding as we saw in lecture
Level k can explain the experience of the
ECB with liquidity auctions
Banks demand liquidity – ECB supplies, if total
demand lower than supply there is a proportional
Nash equilibrium if supply<demand: demand
What actually happened?
Common deviations from
“rationality” - psychological
A poker player does not remember his big gains, but
always remembers his big losses (Matt Damon in Rounders)
Discounting day 2 payoffs in day 1 stronger than day 12
payoffs in day 11
Inertia - Nudges
Sunk cost- and other fallacies
Example: all pay auction
Are these behavioral traits
useful in the real world?
replications in poor countries with payoffs comparable
to monthly wages
… repetition, experience
But many games played by non experts!
– Interaction with rational agents (product markets) …
Behavioral IO: firms may exacerbate rather than
– Sample too narrow
replications with many populations (professionals,
game theorists, chess players…)
Using psychological traits:
Kahnemann, Knetsch, and Thaler (1986)
proposed that sales might have an irrational
Survey: many subjects say that it is unfair for
firms to raise prices when demand goes up
Firms have incentive to hold sales rather than
reducing regular prices
if firms lower regular prices when demand is low,
they will be branded as unfair if they raise prices
back to normal when demand returns to normal
Why sales? pt2
Rotemberg (2005) - more complex fairness-
based model to account for firms’ occasional
use of sales and for the stickiness of prices
consumers have reciprocal preferences and
punish firms discontinuously if their estimate
of the firm’s altruism crosses a threshold
model also relies on the firms’ objective
function being a concave function of profits
and on consumers feeling regret
This leads to sales and sticky prices
Information and obfuscation
Milgrom (1981), Grossman (1981) –
Should you reveal relevant information to
E.g. about the quality of your product to your
Best firm should reveal, to attract clients
But then, the next best firm should also reveal,
else buyers would assume it is just an average
Unravelling- all others have to reveal!
Information and obfuscation
Ellison and Ellison (2005) find differences
btw theory and practice
mattress manufacturers put different model names
on products sold through different stores and
provide few technical specs so as to make it very
difficult to compare prices
Credit cards: hard to imagine that the complex fee
schedules in small print on the back of credit card
offers could not be made simpler.
Ιnformation and obfuscation:
Gabaix and Laibson (2004) suggest a very simple
Consumers have noisy estimates of utility they will receive
from consuming a product: they think they will get utility u+ε
from consuming product i when they actually get utility u.
Obfuscation increases the variance of the random evaluation
error ε in a model in which consumers have noisy estimates
of their utility
Such a model is formally equivalent (from the firm’s
perspective) to a model in which firms can invest in product
Firms will invest in obfuscation just as they invest in
differentation to raise markups.
Information and obfuscation:
Spiegler (2006) discusses another rule-of-thumb model
Products inherently have a large number of dimensions.
Boundedly rational consumers evaluate products on one
randomly chosen dimension and buy the product that
scores most highly on this dimension.
In this model, consumers would evaluate the products
correctly if products were designed to be equally good on
all dimensions. Spiegler shows that this will not happen,
Essentially, firms randomize across dimensions making the product
very good on some dimensions and not so good on others.
Non-linear pricing: Paying not
to go to the gym
DellaVigna, Malmendier (2005)
Survey of all health clubs in Boston area (100
Most common contract design:
monthly and annual fee & initiation fee
no per-visit fee
Estimated marginal cost: $3-$6 + congestion
Below-marginal-cost pricing of visit, p<C’
Paying not to go to the gym
Consumers are initially offered a two-part tariff
upfront payment L and additional per visit charge p.
If consumers accept this offer, they learn the disutility d that
they will incur if they visit the club, and then decide whether
costs p and gives a delayed benefit b.
two reasons why a health club will want to distort p away
from marginal cost
sophisticated rational consumers would like to commit themselves
to go to the health club more often
naive rational consumers overestimate the number of times that
they will go to the club
Las Vegas hotels and restaurants:
Price rooms and meals below cost, at
High price on gambling (the house always
Above marginal cost pricing of
addictive leisure goods, p>C’
Anchoring: Do workers know
their disutility of effort?
Ariely, Loewenstein and Prelec (2004)
Asked subjects whether they would pay $2 to attend
a 15-minute poetry reading
Asked other subjects whether they would attend if
they were paid $2.
Later, asked whether they would attend for free.
Among those who were anchored on paying: 33%
Among those who were anchored on being paid:
Shipping charges on ebay
Hossain and Morgan (2006) conduct
field experiments on eBay
find that auctioning goods with a high
(but not too high) shipping charge raises
more revenue than using an equivalent
minimum bid and making shipping free.
Do people perceive inflation
Georganas, Healy and Li (2010)
Present subjects with a basket of goods and instruct them to
buy a designated one each period
Manipulate the speed and frequency of price changes
People underestimate inflation if cheap, frequently bought goods
have low inflation
People overestimate inflation if such goods have high inflation
This influences their consumption, saving, investment
Real life example: The introduction of the euro led to many
complaints of high inflation because of rounding etc
Actually rounding hardly mattered, since 0.95 was rounded to
1 (5% difference) but 999.5 just to 1000 (just 0.05% difference)
What did we learn- Future?
IO models can be applied to the real world
Sometimes they need some adjustments to account for the
way reality is different from our assumptions
Consumers deviate from rationality in several different,
sometimes predictable ways
Companies can best respond to that and/or also deviate
from fully rational behaviour
We have just begun scratching the surface
There is a looooot of work to be done and many
things to find out…
But you will only find out in grad school :)
U.Malmendier Lecure notes
Excellent survey: G.Ellison http://econ-www.mit.edu/files/904
D.Ariely: Predictably irrational