How to run the Kiviq Double Auction experiment
- Divide students into an even number of teams. Two or three per team is the ideal. Kiviq will assign half the teams to be buyers, half sellers. It is strongly advised to promise a few dollars as a reward for the most profitable buyers and sellers to motivate a degree of competition. Teams will be assigned their roles as buyers or sellers, they do not choose. There is no reason to tell students that the roles will be evenly divided. The point of the experiment is to demonstrate that teams ("agents'") do not need full information for the market to work.
- Each team will share one device (smartphone, tablet or laptop) to do the experiment.
- Everybody including the instructor goes to the Kiviq home page: www.kiviq.us
- Ideally the instructor should project on a screen from their own computer to assist with instruction, trading and debriefing.
- Before anybody (students or instructor) log in, both students and the instructor cilck through the "Double Auction Experiment Tour" slide show at the bottom of the Kiviq home page. (Students can also watch it on their own devices by clicking the "tutorial" icon after login, should projecting the experiment not be possible, but watching with the rest of the class is preferable.)
- The instructor explains how the experiment will work.
- The instructor answers any questions the students may have on how the experiment works.
- Students then log in to Kiviq via the "join experiment" icon, adding the names of each team member, and waiting to be given the experiment "code word. " Students do not have to enter an email address.
- The instructor logs in and sets the parameters for the experiment session, creating a unique name and date for this experiment "group." The "group" name (which is really just a label for the specific class and date doing the experiment) will be important for locating and projecting the total earnings for each individual team after the trading session ends.
- When setting parameters the default is four tradeable units per team (not all will be profitable to trade) so for a small class (twelve teams or less) increase units per team to six or more to generate more trading action. A very large class (over 30 teams) should be given a long trading period (ten minutes.) Five minutes is enough for a medium size class.
- The instructor clicks "launch session." Kiviq generates a "code word" which is announced to the class. Students enter the code word and join the experiment automatically. There is a counter that shows the number of teams that have joined the experiment.
- The instructor presses "start" once all the teams have joined the session. Trading begins automatically on student screens. The instructor screen shows the experiment progress and timer. The instructor by clicking on the lower axis legend can cause the instructor screen to show current bids, current asks, and the history of completed trades. Demand and supply curves can also be selected but that obviously should only be done during debriefing after the timer stops and trading has ended.
- For debriefing it is preferable to first show and discuss the pattern of bids, asks, and trades. Then the demand and supply curves can be revealed, layered over the trading data. If all socially desirable trades are executed (and no trades beyond the equilibrium quantity are made) total consumer plus producer surplus will be maximized and "efficiency" is 100 per cent. Distribution effects do not affect total surplus so even if a team mistakenly trades at a loss that by itself does not decrease total efficiency.
- At the conclusion of trading the experiment green "statistics" box icon allows the instructor to project the earnings results for teams on a screen. Each team member has their whole team earnings next to their name.
- Results appear in "statistics" according to the alphabetical position of the group name that was chosen, not by the time of the experiment.
- The JPE 1962 paper by Vernon Smith describing the original double auction classroom experiment conducted at Purdue University in the 1950s is freely available from the academic paper repository Researchgate. The first diagram in the paper shows Purdue students from the 1950s trading in a pattern usually similar to how students in the 21st century do - i.e. there is a period of "price discovery" and trades approach the predicted equlibrium.