BushTender

1
Experimental Markets

• Double Auctions

2
Set-up

• The double auction is a market institution that
  has been widely tested in experiments. It is also
  widely used in practice, e.g., financial markets.
• In each period, buyers are given values for units
  of a commodity, and sellers are given costs for
  producing units.
• Buyers make bids to purchase units. They compete
  to make the highest bid.
• Sellers make offers to sell. They compete to make
  the lowest offer.

3
Set-up

• Trades take place when a buyer accepts an offer
  or a seller accepts a bid.
• All bids and offers are centrally and publicly
  recorded.
• Buyers and sellers typically see nothing else
  other than
• their own values or costs, and
• a list of all bids, offers, and transactions.

4
Theoretical Predictions

• If a buyers valuation of a good exceeds the cost
  of producing the good, then it is efficient for
  trade to take place.
• Economic theory predicts that a competitive
  market will create an equilibrium price such that
  all efficient trades occur, and no inefficient
  trades occur.
• This means that buyers and sellers will
  collectively earn the maximum amount possible.

5
Predicted Results
Davis and Holt (1993)
6
What do We Observe in the Lab?

• Intense trading which results in competitive
  prices and quantities.
• A very large number of sellers is not necessary
  to generate competitive outcomes.
• In fact, markets organized under double-auction
  trading rules appear to generate competitive
  outcomes more quickly and reliably than markets
  organized under any alternative set of trading
  rules.
• For this reason, they have been frequently
  investigated as a standard against which the
  performance of other institutions is evaluated.

7
Experimental Results
Davis and Holt (1993)
8
Experiment Results
Davis and Holt (1993)
9
Evaluating Market Performance

• A measure commonly used is efficiency. It is the
  proportion of the maximum surplus extracted.
• To calculate the maximum surplus, consider the
  difference between the valuations and the costs
  of production for the units traded.
• Note Distribution is a different matter.
• Another possible measure of performance is the
  coefficient of convergence.

10
Evaluating Market Performance

• Double auction experiments nearly always perform
  according to theoretical predictions.
• Prices converge quickly to within 5 cents of the
  predicted price.
• Quantity traded is generally within 5 of the
  predicted quantity.
• The total earnings of participants is generally
  more than 95 of the predicted earnings.

11
Stress-testing Market Power

• What is the minimum number of sellers sufficient
  for competitive outcomes?
• Are there any conditions under which
  double-auction markets do not generate
  competitive outcomes?
• Competitive predictions are somewhat weakened
  when the market is reduced to only two sellers,
  but competitive price, quantity, and efficiency
  levels are often observed, even in monopolies.

12
Stress-testing Market Power

• Monopolists in double auctions often fail to earn
  high profits.
• They sell at high prices to high-value buyers and
  low prices to low-value buyers (price
  discrimination).
• In subsequent rounds, buyers know the seller has
  low costs, and refuse to pay high prices (Smith
  and Williams, 1989).

13
Stress-testing Collusion

• What if the sellers are allowed to collude?
• Suppose they are put in a separate room.
• Can discuss prices and quantities.
• Cannot exchange cost information or make side
  payments.
• Finding Collusion does not work. The sellers
  fail to earn high profits.

14
Stress-testing Collusion

• Sellers are tempted to cheat on their agreement
  to sell low-price units to low-value buyers.
  Buyers, upon seeing that sellers can afford to
  sell at lower prices, refuse to accept higher
  prices in subsequent periods (Isaac and Plott
  (1981).
• Hence, markets organized under double-auction
  rules also appear to generate competitive
  outcomes in the face of opportunities for even
  explicit conspiracy.

15
Conclusions

• Experiments can be cheap and effective tools for
  testing the predictions of economic theory, and
  for evaluating proposed market institutions.
• The double auction has been tested in experiments
  and shown to be a reliable, efficient market
  institution.