Page 615 - Microeconomics, Fourth Edition
P. 615

c14gametheoryandstrategicbehavior.qxd  8/6/10  8:22 AM  Page 589







                                                       14.2 THE REPEATED PRISONERS’ DILEMMA                     589

                      8 wins (the number needed to ensure a winning   They find, for example, that the unexpectedly large
                      record) than would be expected by chance. Further,  number of wins by a wrestler on the bubble was
                      they find that winning percentages for wrestlers who  increased if the wrestler was engaged in a match with
                      are on the bubble are particularly elevated on the last  another wrestler against whom he had wrestled fre-
                      day of the tournament as compared to other days.  quently in the previous year. Further, they discovered
                         The natural alternative hypothesis that would ex-  that a wrestler who is in the last year of his career (and
                      plain these findings is that sumo wrestlers who are on  who therefore cannot participate in repeated play in
                      the bubble try especially hard to win the eighth match  the future) is less likely to win an unexpectedly large
                      so as to guarantee a winning record; that is, they “step  number of matches when he is on the bubble. These
                      it up a notch” and find a way to win. One way to dis-  patterns are consistent with the collusion hypothesis,
                      criminate between this hypothesis and the collusion  but there is no reason to expect to observe them if
                      hypothesis is to use insights from the repeated prison-  wrestlers who were on the bubble were simply exert-
                      ers’ dilemma model. That model tells us that the like-  ing extra special effort. Though  Duggan and Levitt
                      lihood of wrestlers being able to sustain a collusive  have not uncovered a “smoking gun” showing that
                      deal should be positively related to the frequency with  collusion in sumo matches occurred, their indirect evi-
                      which the wrestlers interact and the likelihood that  dence is very powerful and suggests that the authori-
                      they will be paired again in the future. Duggan and  ties who control sumo wrestling in Japan should be
                      Levitt’s findings are  consistent with this  prediction.  alert to any signs that matches are being fixed.




                      APPLICA TION  14.6

                      The Cost of War     17
                                                                       of inflation. However, in the late 1980s, the market
                                                                       began to change. Health concerns slowed the de-
                      An excellent illustration of what can happen when  mand for cigarettes in Costa Rica, a trend that hit the
                      one firm miscalculates competitor responses occurred  premium and mid-priced segments much harder than
                      in the cigarette industry in Costa Rica in 1993. The  it did the VFM segment. In 1992, B.A.T. gained market
                      most famous cigarette price war of 1993 occurred in  share from Philip Morris for the first time since the
                      the United States, when Philip Morris initiated its  early 1980s. Philip Morris faced the prospect of slow
                      “Marlboro Friday” price cuts. The lesser-known Costa  demand growth and a declining market share.
                      Rican price war, also initiated by Philip Morris, began  On Saturday, January 16, 1993, Philip Morris re-
                      several months before and lasted a full year longer.  duced the prices of Marlboro and Derby cigarettes by
                         At the beginning of the 1990s, two firms domi-  40 percent. The timing of the price reduction was not
                      nated the Costa Rican cigarette market: Philip Morris,  by chance. Philip Morris reasoned that B.A.T.’s invento-
                      with 30 percent of the market, and B.A.T., with 70  ries would be low following the year-end holidays and
                      percent of the market. The market consisted of three  that B.A.T. would not have sufficient product to satisfy
                      segments: premium, mid-priced, and value-for-money  an immediate increase in demand should it match or
                      (VFM). Philip Morris had the leading brands in the  undercut Philip Morris’s price cut. Philip Morris also ini-
                      premium and mid-priced segments (Marlboro and    tiated its price cut on a Saturday morning, expecting
                      Derby, respectively). B.A.T., by contrast, dominated  that B.A.T.’s local management would be unable to re-
                      the VFM segment with its Delta brand.            spond without first undertaking lengthy consultations
                         Throughout the 1980s, a prosperous Costa Rican  with the home office in London.
                      economy fueled steady growth in the demand for cig-  However, B.A.T. surprised Philip Morris with the
                      arettes. As a result, both B.A.T. and Philip Morris were  speed of its response. Within hours, B.A.T. cut the
                      able to sustain price increases that exceeded the rate  price of its Delta brand by 50 percent, a price that

                      17 We would like to thank Andrew Cherry (MBA 1998 Kellogg School of Management) for developing
                      this example.
   610   611   612   613   614   615   616   617   618   619   620