MBA 211 Game Theory

Class #3 Highlights

2012-01-30T16:13:00.000-08:00

In class # 3, we talked about rationality. While caricatures of economics and game theory posit that rationality is the same thing as maximizing money, the reality is more nuanced. One needs to pay attention to non-pecuniary factors like fairness, identity, pride, altruism, spite, and so forth. The tools of game theory are flexible enough to handle these sorts of additions. Indeed, the art of successful "mind reading" is getting the game right in the sense of including all relevant factors determining a rival's payoffs.

We also talked about the fundamental rule of game theory: look forward, reason back. Through this rule, we discovered that the competition among teams for a player in the presence of a right of first refusal option was largely illusory. We also discovered that a press release designed to deter entry can only be successful to the extent that the threats and promises offered are credible.

Side notes: Results from McCain are due by midnight of 1/30/12. Also, I've fixed the calendar entries in the syllabus.

Class #1 and 2 Highlights

2012-01-25T08:39:00.000-08:00

As I mentioned at the outset: The goal of this course is to change your mindset, to provide you with a new and useful lens through which to view the world. The shorthand for this is inside versus outside thinking. Inside thinking focuses on how to make optimizing decisions when faced with either a static setting or environmental risks. For instance, in marketing you identify the willingness to pay of key customer segments and then choose optimal prices for each.

Outside thinking deals with strategic risk--there are other players in the game besides yourself. These players have goals, ambitions, and strategies, and it is important to factor these in when making decisions. The metaphor for outside thinking is a chessboard. The "best" space on a chessboard depends on where all the other pieces are and, perhaps more importantly, depends on the anticipated moves by the other player.

We illustrated the difference through the Race for the GOP Nomination game. In that game, inside thinking assumed that all other players simply chose the candidate corresponding to the signal received and then optimized. Viewed in this light, later voters are very likely to make the correct decision since they have lots of data on which to base that decision. Outside thinking paid attention to the incentives of earlier voters to vote strategically, i.e. to vote against their own signal. Using outside thinking, we discovered that the primary race is very likely to produce a "cascade"---a run of decisions all favoring a single candidate. Indeed, we saw this behavior in the experiment as well. The key implication is that later voters do not have much data on which to base their decision and, in fact, information is not collected by voting in this fashion.

Why is this important? Under inside thinking, you would (correctly) conclude that voting sequentially is likely to produce the correct choice given enough voters. With outside thinking, you would realize that there is a serious strategic problem with sequential voting. It is quite likely to produce the wrong answer. In our setting, there was a 20% chance that voting would produce the wrong answer even if we had an infinite number of voters.

We might then compare this to another common system--everyone voting at the same time. In our game, this would produce the correct answer always. Indeed, using game theory, we can discover that there is a general result that if all voters have reasonably similar incentives (like choosing the more electable candidate in the primary), then a national primary will choose the correct voter with certainty as the number of voters gets large. Thus, by using outside thinking, we learn that there is a right way and a wrong way to use voting to make good decisions. Since voting is a common way to make decisions in business settings as well as political settings, this is quite important in developing effective ways of having your company make decisions.

Game Theory Teams

2012-01-25T08:09:00.001-08:00

Building a Team

Here is the incomplete list of team members from emails received to date. If you are not on the list, please send me an email with your team members and team name.

Game Theory Teams Spring2012

Gameblers: Maria Alejandra Gonzalez, Alper Batur, ismaelghozael, Fabio Povoa, Mohamed Shommo

Rafa de la Guia: Bradley Okamoto, Iris Shim, Juan Sanchezand Dave Lewis

Latin Gambits: Paul Kisiliuk, Rodrigo Donoso, Andres Pachano,Matias Bebin, Inaki (Jose Ignacio) Ruiz

Good Swimmers: Alden Woodrow, Patrick Flemming, Laura Bentzien,Kyle Bentzien, Dana Ledyard.

Pokerface: Gustavo Ribeiro, Jasmine Hellings, JavierFigueroa, Juan Manuel de los Rios Wakeham, Shannon Riley

Nash-ty equilibrium: Gene Boyle, Eamonn Courtney, Billy Hwan,Ella Yanai, Janice Yuen

4 Young Men: Aditya, Archit, German, Sue, Darren

SwimSocks: Caroline Mock, Oscar Salinas, Sam Mathias, SamWiggin, Patrick Mar
Beautiful Minds: James Cook, Pei-fu Hsieh, Patrick Schneider, Derek Simmons and Jenni Tonti.
YMCA²: Yoichi Katayama, Matthias Egler, Carlos Facanha, Alexandre Montoro, Aaron Tang

4 Men and a Baby Mama: RahulBijor, BrandonYahn, LauraAndron, JoshMogabgab, AndrewHamilton

The Race for the GOP Nomination

2012-01-25T07:53:00.000-08:00

Choosing the Right Move

From the Game Theory in the Wild files comes this piece on the Colbert Report spotted by your colleague Dave Lewis. The piece is about the book The Dictator's Handbook, written by the NYU game theorist Bruce Bueno de Mesquita. The piece talks about how the professor uses game theory in a computer model to predict geopolitical events. Here's a useful tool for applying these ideas: The Predictioneer's Game.

Key highlight from the interview: A prediction on the GOP nomination. Enjoy!

http://www.colbertnation.com/the-colbert-report-videos/406665/january-23-2012/bruce-bueno-de-mesquita

Welcome to Game Theory

2012-01-17T10:29:00.000-08:00

Welcome to the Spring 2012 edition of game theory. All of the assignments are available from the main game theory website. I'll use this blog to recap the key highlights from each class. It will also be used to make announcements about assignments and to offer takes on whatever I happen to be thinking of at the moment. You should check the blog regularly for announcements and results from the various games.

The blog is also an excuse to show off my photos. Unless otherwise credited, all the photos on the blog are done by me. Photography is a real passion of mine, and I especially enjoy sharing the images I create. Creativity is a major subtheme of the class.

I hope you enjoy the semester we'll have together!

Final Project Deadline

2011-05-04T14:22:00.000-07:00

Final projects are due on May 16 at 1159pm Hawaii time. Please send as a PDF.

Mixed Strategies Highlights

2011-04-29T08:41:00.000-07:00

Choosing the Right Strategy?

In our section on mixed strategies, we studied games that had no set of deterministic best responses. For instance, in soccer penalty kicks, the best response to a keeper who defends left is to kick right, but then the keeper's best response is to defend to the right, which leads to kicking to the left as being optimal and we end up going in a circle. The solution to the "circle" problem is to study mixed strategy equilibria. These strategies can be understood literally as using randomization schemes to determine which strategy to play.

An aside: The can also be understood as representing population measures of play. For instance, suppose that the keeper's strategy is to defend left when he wakes up on the left side of the bed and to defend right otherwise. If half of the keeper's wake up on the left and half wake up on the right, then it is as if each is randomizing his strategy even though all are playing pure strategies.

The goal of randomization is not pure payoff maximization. Indeed, in equilibrium, all strategies yield the same (expected) payoff. Rather it is to keep the rival off balance and thereby avoid being exploited. In strategy terms, one can view mixed strategies as providing a rationale for "strategic ambiguity"---remaining intentionally vague about one's strategy. In terms of inward/outward thinking, mixed strategies represent the culmination of outward thinking since the mixture is done purely for the sake of the rival.

Mixed strategies can sometimes lead to surprising insights as we saw in the Volunteer's Dilemma. In this situation, a group of n people can decide to help or not. Helping provides a public benefit B but requires a private cost c. If a person knew that she was pivotal in providing effort (i.e. no effort would be undertaken without her), then since B > c, she would volunteer. In that sense, there is no apparent free rider problem. But what if there is uncertainty about whether you are pivotal? Here, we solved the game using mixed strategies. Not surprisingly, volunteerism declines as n increases. But more strikingly, the chance that anyone helps also declines with n. That is, the phenomenon that social psychologists refer to as "diffusion of responsibility" can be understood as a rational response to the volunteer's dilemma game using a mixed strategy equilibrium.

Signaling Highlights

2011-04-27T15:43:00.000-07:00

Please find the results of the value of education experiment here.

We studied signaling through the value of education game. In signaling games, the key parameter in the mental model are the beliefs of employers about a worker's type following each signal. In a separating equilibrium, different signals produce different beliefs. In our game, we saw that a separating equilibrium consisted of believing that grade school (or BS) were associated with low types while MBA was associated with being a high type. In that case, a low type could not gain by pretending to be a high type and, therefore, these beliefs are confirmed by actions. We also saw that one could not sustain a separating equilibrium by believing that low types choose grade school while high types choose BS or MBA. The reason is that the marginal benefit to a low type from imitating a high type and getting a BS was 100k while the marginal cost was only 80k; therefore, low types will "invade" the BS population and thereby destroy the signal.

Let us reflect on this for a moment. 20-30 years ago, an MBA was unnecessary to climb to the highest levels in business whereas now it more or less is required. What has changed over that time is undergraduate enrollments. 20-30 years ago, many people ended their education with a high school degree. Now, there are strong incentives to proceed to get an undergraduate degree and many subsidies and loans to make this affordable. In a sense, an undergraduate education is now too cheap to offer a useful signal. In our game, this signal inflation was altogether a bad thing since there were no gains to human capital from signaling. Of course, the reality is that there are some human capital gains. Nonetheless, the results of the game indicate that there is an incentive to overinvest in education purely for the signaling aspect. In other words, the problem might be too much education rather than too little.

Besides the separating equilibrium, there was also a pooling equilibrium in the game. In a pooling equilibrium, everyone is supposed to get a grade school degree. So long as employers are sufficiently skeptical about the signal value of a BS, then no type will wish to deviate. Obviously, this equilibrium is better for low types since their quality is no longer being revealed. What is more surprising is that, for the parameters of our game, it is also good for high types. While high types earn more than low types when they separate, these gains are more than burned up in the cost of the signal.

Again, it is useful to reflect on this. In effect, the signaling game is a type of coordination game based on employer beliefs. If employers believe that education contains little signal value, then this is self-reinforcing and workers are happy to go along. If, however, they believe strongly in the power of the signal, this too is self-reinforcing, and types divide by signals. In our game, the payoffs were higher in the pooling setting than in the separating setting. In effect, this is like choosing the worse outcome in a coordination game, but in a more subtle way. This suggests that policies like restricting advertising (as was the case for many years in the distilled spirits industry) can actually enhance the profits of all brands, both low-end and high-end.

Preso Order

2011-04-27T15:11:00.000-07:00

Here is my best reconstruction of the preso order on the basis of your cards from last class. Please alert me if any of this is screwed up.
Class #28 (Weds)
1. YNK
2. MAX
3. AGT
4. WSID
5. NAS
6. F
7. HAZ

Class #29 (Mon)
1. REPS
2. H8
3. MOAB
4. RDR
5. MOTN
6. BNT
7. SEA
8. PLAYA

Aside: Photos, creativity, and conversations

2011-04-19T08:51:00.000-07:00

I was delighted to get a comment after class on the photos on the blog. When I'm not professoring, I'm an amateur photographer. Indeed, other than working on game theory, photography is my other great passion. I mentioned creativity in an earlier aside and I do think this is perhaps the most important thing to gain from an MBA education. To me, both game theory and photography offer a chance to "see the world differently," which is the essence of creativity in my view.

For those interested, you can check out my Flickr stream here. Flickr is far and away the best online photo site for the aspiring amateur. It combines the cool social aspects of Facebook with the ability to upload photos at full resolution. If you're shooting party photos with your phone of point and shoot camera, this is not a big deal, but if you're as picky about quality as I am, full resolution really matters.

Social also matters since it's nice to get feedback about your work. In that regard, the blog has been less successful than I had hoped. I get very few comments from you guys on it. I'd be happy for any thoughts about how to make this more "conversational" than it currently is.

Class # 23 Highlights

2011-04-19T08:38:00.000-07:00

Surviving the War of Attrition

We closed the auction unit with a cautionary tale about modeling behavior. The key to fighting a war of attrition is to determine the hazard rate of concessions. Now, there is a tendency toward overoptimism in estimating this hazard rate and, as a consequence, the war tends to lost longer than expected. We saw this vividly in the $20 bill auction where the total spend exceeded the value of the bill by $7. This also illustrates that the revenue equivalence theorem is a starting point rather than an ending point. Accounting for behavioral factors, risk preferences, framing and so on is essential in developing good mental models.

The second half of the class was about pricing. While most of pricing is concerned with identifying and segmenting consumer WTP in various ways, in our setting pricing strategies were strongly driven by outward thinking. The "right" price is the one slightly lower than what is being offered by rivals in the market. Thus, the mental model is to guess what this price might be based on past actions, market dynamics, and so on. This interdependence means that a predictable pricing strategy is vulnerable to exploitation. Thus, the right strategy is an unpredictable strategy. In class #24, we will discuss how to operationalize this idea through mixed strategy equilibria.

In the meantime, check out the results of the experiment here. On the first page, I list all of the prices charged by each team. Look for patterns. Are there tendencies you could exploit? Is your strategy exploitable. On page 2, the market interactions are shown. Notice that period 4 is very different from the other periods. Here, cooperation is almost achieved. Thinking back to OPEC, is it possible to maintain high prices in this setting?

Mid-Semester Projects

2011-04-18T10:01:00.000-07:00

I've now added links for all the mid-semester projects, so you can read what your colleagues have written. I would especially call your attention to MOAB's mid-semester project. This is by far the nicest mid-semester project I've seen since running game theory. They study the NYSE/Euronext acquisition. Notice how they pin down 4 key bidding points in the negotiation based on the inside versus outside option equalization in second price auctions. Notice also the care in devising what these valuations are. They are facing a potentially super-complicated game, but used the tools of game theory to isolate and analyze the key aspects.

Check it out here.

Class #22 Highlights

2011-04-13T22:07:00.000-07:00

A War of Attrition

This class completed our auction unit. There were two key highlights:

1. N is a big deal. More than anything else, the number of bidders is the key variable in constructing a successful auction. We illustrated this in two ways. First, a quick case about the failure of the German 3g auctions owing to the fact that the number of licenses equaled the number of large players in the industry. Second, we showed that the worst auction beats the best negotiation if it attracts one additional bidder. This latter result is important in thinking about M&A.

2. How to think about wars of attrition. Wars of attrition arise in many business situations. In strategy, both the bitter competition and Ryanair cases are about anticipating wars of attrition.
The key problem in wars of attrition is that it is often difficult to forecast the costs of the war. Auction theory can help a lot in this regard.

Key points: You should drop out of the war of attrition whenever the value of winning times the hazard rate--the chance your rival will concede in the next period conditional on how long the war has lasted to date, exceeds the cost of fighting for one more period. Thus, the central piece in the mental model is to forecast the hazard rate of concessions. One needs to be mindful of two things: The current duration of the war is informative. One needs to account for this in the forecast. Second, there is a tendency toward overoptimism. Auction theory provides a tool for calculating the hazard rate and thereby avoiding this bias.

The second key point is forecasting the cost of the war. So long as you and your rival are similar, the revenue equivalence principle applies and this can be used to forecast costs, both overall and at each point in time during the war. Forecasting costs is obviously critical in making good judgements about continuing to fight or not.

Finally, knowing the beliefs about the hazard rate of concession are the key driver, one can try to influence this. One way to do this is to try to signal that you have a very high value for the prize. A credible signal will lead the rival to conclude that the war is not worth fighting, thus shortening its duration. A second way to do this is to commit to fight a long war. This may be done by creating a war chest or hiring an agent with a reputation for not conceding to fight on your behalf. Again, if you can credibly influence the rival's perceptions of your concession probability, she will give up quickly allowing you to win the war on the cheap.

An Update from the Penny Auction Front

2011-04-12T23:11:00.000-07:00

The auction unit seemed to inspire a burst of interactivity on your part. I got a number of comments about penny auctions and slot auctions. Here are some highlights:

Leigh Fiske writes:

Quibids (http://www.quibids.com/) is a new "penny auction" site in the model of swoopo.com. Each bid costs $0.60 to make, and in the last few minutes of the auction, each new bid extends the remaining time by a certain amount. It's pretty interesting. They must think that their business model has a higher likelihood of success than their predecessors'!

Rosie Wang notes the demise of Swoopo and finds the effective markups on penny auctions rather yucky:

Thought you'd be interested to note that Swoopo has recently filed for bankruptcy. (http://technologizer.com/2011/03/25/swoopo-quietly-files-for-bankruptcy/) I was originally going to write you and let you know about another site called "oohilove" that had a very similar penny auction but for luxury handbags and accessories, only to find the site shuttered and that it's owned by the same company as Swoopo. I once did a calculation on oohilove though to see how much money they were making (since each bid is also $1 and bids go up in $0.02 increments), and was a bit disgusted by the multiple they were getting on the actual value of the product.

Gwyn Jones notes the burgeoning academic literature on position auctions. Here is a piece by Google's Chief Economist (and Berkeley Professor) Hal Varian title "Position Auctions" Here is a broad survey on Google style auctions from the source of all knowledge, Wikipedia.

Finally, a piece about waiting in line to see Star Wars.

Happy reading!

The Two Principles

2011-04-12T08:57:00.000-07:00

Our auctions unit highlighted two principles: The Revenue Equivalence Principle and The Linkage Principle. The Revenue Equivalence Principle is a handy result allowing us to compare all sorts of auctions. It roughly says this:

Any auction where: (1) High bidder wins; (2) Free opt-out; (3) Bidders are similar; and (4) Bids are for cash yields the same expected revenue as a Vickrey auction. In fact, it says something a little stronger. It says that the expected payment for a bidder with value v is the same across auction forms. This auction pulls together a lot of non-auction settings such as price wars, R&D races, waiting times in line, and so on. It should be viewed as a first step and not a final answer. For instance, risk aversion will "break" the result. Bidders who differ a lot (i.e. one bidder known to be high value with a competitive fringe surrounding) will break the result. Multiple units that differ from one another (i.e. the slot auction) will break the result. But it's an essential starting point.

Our second principle was linkage. This roughly states that the more linked a bidder's payment is to value, the more revenues the auctioneer earns. We saw this sharply in comparing debt to equity. Since equity has linkage, it yields more money. It can also be used for disclosure strategies when values are correlated. For instance, suppose that an auctioneer is selling rights to an oil tract. Here, bidders have a common value for the tract (equal to the amount of oil under it). The linkage principle says that the auctioneer is better off, on average, if it reveals information about the value of the tract (i.e. seismological studies) than when it does not. In other words, disclosure is the best policy even if it means sometimes disclosing unfavorable information.

Class #21 Highlights

2011-04-12T08:36:00.000-07:00

This class featured guest speaker Elizabeth Churchill talking about the evolving experience of shopping. Her slides are now posted up on the course website. More importantly, here is Elizabeth's light-hearted blog post about the topic:

Rat, Rational or Seething Cauldron of Desire: Designing the Shopper

Class # 19 Highlights

2011-04-04T22:08:00.000-07:00

Structuring a deal

This is the first part of a two part class on structuring deals through auctions. In the experiment, we considered two dimensions of a deal: the degree of transparency and debt versus equity. Deals made via auction can be transparent, as in an English auction. Here, the auctioneer discloses the leading bid along the path to getting to a final deal. Deals can also be secret. Here, the auctioneer asks for sealed bids and the results are revealed only at the end. Each type of deal has advantages, which we will flesh out in class #20.

The other dimension concerns the structure of a deal, typically debt versus equity. In a debt deal, the winning bidder promises to pay back a fixed amount. This leads to less uncertainty in the cash received by the auctioneer, but less upside if the deal proves to be really successful. An equity deal exposes the auctioneer to more risk but also more upside. In class #20, we will talk about the pros and cons of structuring deals.

Class # 18 highlights

2011-04-04T22:00:00.000-07:00

The Road not Taken

The highlight of this class was to introduce the Nash Bargaining Solution. This suggests that surplus should be allocated to maximize the product of the happiness of the two sides. The solution follows as a consequence of 4 principles: symmetry, no money left on the tables, frames don't matter, and the road not taken. It is the only rule satisfying these principles.

The Nash solution can be used many ways. In preparation for negotiation, the solution offers a guide about the likely outcome. In the negotiation, the principles can be used to move conflict from the personal to the intellectual and thereby defuse problematic emotional responses. The solution also explains how improvements in the outside option affect the inside option. For instance, when 2 risk neutral parties compete, a $1 improvement in the value of one side's outside option raises their allocation in the deal by 50 cents.Overall, the framework provides a powerful shortcut for predicting outcomes in "messy" bargaining situations where drawing game trees is impractical.

Class #18 Highlights

2011-03-30T22:01:00.000-07:00

A Cake Cutting Problem?

In this class, we developed a framework for assessing bargaining situations. Bargaining is complex and messy. Moreover, the answers provided by game theory are often frustratingly sensitive to the exact form of the game tree. What is needed is a toolkit for assessing outcomes without having to write down the game. Here, we took a page from cooperative game theory.
Cooperative game theory is an older line of game theory that is less used. It mainly involves developing principles (axioms in formal language) and then applying these to predict payoffs in bargaining and coalition building situations. While this branch of game theory is interesting in its own right, scarce course time limits our coverage.

Our main tool for assessing bargaining is the Nash Bargaining Solution. This says that, given a payoff frontier and outside options for each party, choose the outcome that maximizes the product of the payoffs.

This odd suggestion derives from 4 simple principles: 1. Symmetry, 2. Frames don't matter, 3. No money left on the table, and 4. The road not taken. Agreeing to these 4 principles is the same as agreeing to the Nash Bargaining Solution.

The solution has several sensible predictions. First, better outside options produce more payoffs in a settlement. Second, more risk averse parties do worse than less risk averse ones.

Class #17 Highlights

2011-03-30T21:49:00.000-07:00

The Weakest Link in Cooperation

In this class, we reviewed the results of the OPEC game. The main lesson of this game is to provide practice in achieving cooperation in difficult environments. The two main difficulties in this setting are: (1) lack of observability; and (2) limited tools for punishment and reward. As we saw in GE v Westinghouse, these are common obstacles in many settings.

How do we solve this problem?

1. Limit the temptation to cheat. While both markets sought to hit the joint profit maximizing output quotas, this may be less effective than a less ambitious strategy. By raising the quota, the temptation to cheat is reduced. Ratcheting the quota in steps then creates the possibility of building trust over time.

2. Limit the size of the problem With so many parties to coordinate, the problem is daunting. Yet about 80% of the gains from cooperation can be realized with a 4 party alliance consisting of the big producers. This smaller group has two advantages: coordination is easier and detection is easier. A smaller group means that the behavior of the 3 smaller countries is entirely predictable; thus, the detection problem is made much more manageable.

3. Have a clear target It is fine to agree in principle on production quotas, but there needs to be a clear target for what counts as good behavior and what triggers punishment. In the game, price offers an economically important and publicly observable coordination role.

Backing up a step, many of you stressed the need to "do your homework" before coming to the auction. This mostly meant doing a careful valuation analysis before bidding. While this is undoubtedly important, at least as critical is doing your homework in terms of figuring out a workable plan for cooperation. Effective leadership involves preparing a detailed roadmap for coordination and supporting this with careful economic arguments that can be shared across the parties seeking to cooperate. Too often this aspect--the outward thinking aspect--is neglected in determining valuations. This is true not just in OPEC but in M&A more generally.

Judo Highlights

2011-03-30T21:35:00.000-07:00

An Unpromising Landscape?

The judo class was a capstone for the ideas of the first half of the course. In strategy, the easy thing is to identify promising business landscapes. The problem is that others can identify these same situations, and the space quickly becomes crowded. In strategy, being successful when you have a competitive advantage, while not easy, is much more promising than situations where you have no such advantage. So how do you find opportunities in unpromising settings where you have no advantage? (One good thing about such settings is that others are probably not looking there.)

Perhaps the most important contribution of game theory is to see the possibility in such unpromising settings. Our unit on judo strategies illustrated one avenue for using game theory to this end. Judo strategies are useful in settings where there is an opportunity for a niche player to compete in such a fashion that it provokes no competitive response to rivals. Avoiding a competitive response amounts to finding circumstances where the rival is sufficiently inflexible that aggressiveness does not pay. In our experiments, entrants pulled this off by staying small and pricing aggressively in the face of an incumbent that could not price discriminate over the targeted customers of the small firm.

But this idea is much broader and more powerful than merely that setting. Inflexibility can be used against incumbents with brand advantage. By choosing a niche where the appeal would undercut the brand associations of the incumbent, the entrant can survive and prosper. Red Bull is a nice example of this strategy.

Another type of exploitable inflexibility is the worry about cannibalization of existing product lines. A niche player can operate by selling a (possibly inferior) substitute to the main product line of the incumbent. Since retaliation drives down the value of the main line, incumbents are often reluctant. Connor Peripherals executed this type of strategy in hard disk drives. Similarly, Compaq pursued this strategy in "portable" computers against IBM.

Inflexibility might be tied up in channel relationships. For instance, the strength of IBM and HP's relationships with retailers allowed Dell to use the direct channel without swift retaliation. IBM did not pursue this strategy since it cost too much in terms of damaging the relationship with the main channel.

(An aside: In strategy, we often talk about the difficulty of undertaking a straddle strategy. Straddles are a classic type of inflexibility that can be exploited by entrants.)

Game theory provides a powerful analytic framework for identifying and exploiting these opportunities. The lesson for the first half of the class: Outward thinking opens up whole new avenues for value creation.

Final Projects Clearinghouse

2011-03-21T09:41:00.000-07:00

A Potato Clearinghouse?
To help mitigate the coordination problem in project selection, please use the comment area below to report your final project. The rules for this are as follows: (1) If you chose topic X for the mid-semester project, then you own that topic and are free to pursue it for the final. (2) Any new topic Y is owned by the first team to make a comment on the website. No other team can claim this topic.
My goal is to avoid the level of topic duplication we saw in the mid-semester projects.

OPEC Round 9/Cold Fusion Round 10

2011-03-21T09:27:00.000-07:00

Solar Energy
Round 9 Results:
Market A: Price $26.20, Q = 68,259
Market B: Price $26.97, Q = 68,153
Cold Fusion Coin Flip = Heads. So cold fusion is invented prior to round 10. This means that Round 9 was the last round of the game.
The P&L results of the OPEC are posted on the P&L spreadsheet. I've also posted the results for the Judo game. I notice that some spectrum results are still not posted. Please add these unless the payoff was zero for this game.

Cold Fusion Round 9

2011-03-18T19:43:00.001-07:00

OPEC Round 8

2011-03-17T15:44:00.000-07:00

Results:
A: Price 26.23 Q = 68,182
B: Price 26.00 Q = 69,356

Please note that the results for the first round of cold fusion will be reported on Friday. There is a 10% chance of invention. Following this, I will sim out all the rest of the results of the game assuming no cooperation for the remaining periods and report final results.

Mid-Term Project Due Date

2011-03-16T23:03:00.000-07:00

Mid-term projects are due by 1159pm Hawaii time on Weds March 30, 2011. Please submit these either as hardcopies (recommended if you use color) or pdf. Please note, if submitted pdf, I only have a BW printer, so any color stuff will be lost. Make sure that graphs etc look good in BW.

Upgrades

2011-03-15T18:12:00.000-07:00

In my efforts at continuous quality improvement, I have gone over the mid-semester evaluation forms with the class rep. Overall, you seem to like the class (with one exception) but there are always ways to make it better. Here is a list of the upgrades so far:

1. Change the channel. Some of you complained that the run time for the videos was too long for the information provided. Therefore, all new videos now come with exec summaries for busy people who do not have time to watch.

2. Dates not class numbers: Some of you complained that using class numbers to refer to classes in the syllabus and in my discussion of past concepts is confusing. Therefore, I've now retrofitted the syllabus with dates. I've also tied the take aways on the blog (which are listed by class number) to the syllabus, so that you can more readily correlate the class number with the date. If you turn to the syllabus, you'll now see hyperlinks to the takeways to make this connection easier.

3. Too many problem sets: Several of you complained about the workload of the course, particularly the problem sets. Apparently the three problem sets required over the first 8 weeks of class were unduly burdensome. You also wondered what the point of the problem sets was. The point of the problem sets is to get you to see the math behind the games (something that several of you wanted more of). Nonetheless, I aim to please. Since problem sets are a pain for me to grade and you don't want them...they're gone. PS3 is the last problem set for the course.

4. Why do I have to post the results? This is the first year that I had you post the results of experiments. This change in the course, which was suggested by last year's MBAs, has received mixed reviews. Several of you thought that you really shouldn't have to post your own results and that I should take care of all the logistics. When I handled all scoring, I would send numerous nag emails to remind you to send me the results of experiments so I could input them. This process just ends up wasting (mostly my) time though some teams complained about the nag emails. You still have to send me an email with the results (which is about the same as inputting them to the online spreadsheet), while I have to nag you and then transcribe the results from the email. I'm not changing this since it really won't save you any time.

Now for the one exception: One of you wrote that the games I designed for the class were "stupid and useless." You were clearly very unhappy in the course. Since the games are the main things I have to offer, I'm happy to give you a refund as best I can. I don't want dissatisfied customers. Therefore, I'll sign whatever forms are needed to allow you to exit the course. Please come see me offline.

For the record, I believe that the games are some of the most effective tools for learning about strategy and leadership you'll encounter as an MBA.

Please note that if any of these upgrades are, in fact, perceived as downgrades, if I hear enough dissension, I'll switch back.

I'll post more as I come up with additional upgrades. I'll also discuss the results of the evals in more detail in class #17 (3/28/11)

OPEC Period 7

2011-03-15T17:15:00.000-07:00

Market A:
Price: $25.19 Q = 68,734
Market B:
Price: $42.33 Q = 64,075

OPEC Period 6

2011-03-13T15:01:00.000-07:00

And for those too busy for the video, here's the executive summary:

A: Price = 26.34, Q = 69,647
B: Price = 38.22, Q = 65,091

OPEC Round 5

2011-03-11T08:31:00.001-08:00

No time for video on this go. Aidan is too busy with his new Pokemon Black game. Here are the results:

Market A: $25.33 with Q = 68,269
Market B: 45.31 with Q = 61,204

The moderated market seems to have settled into a happy equilibrium while market A remains as dysfunctional as always.

Feedback

2011-03-09T22:20:00.000-08:00

Feedback?

As it is halfway through the course, now is a great time to get feedback. While the mid-semester evals provide useful feedback for the course itself, I wanted to get a little feedback on the blog and other electronic aspects of the course. What do you like? What would you like to see more of? Less of? Leave me some comments to let me know. If you think everything is great, feel free to leave me a comment saying that. If not, tell me what to improve.

Halftime

2011-03-09T21:58:00.000-08:00

Getting Ready for the Second Half

We're at the halfway mark in game theory. It's time to take stock of the main lessons learned.

1. Outward thinking

The goal of the class is to change how you think, to get you to view the world through an interactive/game theory lens. Outward thinking is the shorthand used to describe this process. It means analyzing situations by anticipating the rival's response and then "changing the game" with this in mind to find a successful outcome. Outward thinking sheds light on new opportunities (like how to make money when you have no form of competitive advantage whatsoever) as well as new ways to use the tools you've already developed (like using an option to stifle competition).

2. Mental Models

At the heart of outward thinking is the process of building mental models. To forecast a rival's moves correctly requires that you get inside the head of the rival. This means determining pecuniary aspects of various strategies as well as non-pecuniary aspects such as pride, perceived fairness, reference points and so on. Good mental modeling is a mix of financial forecasting and psychology. Good mental modeling also factors in the mix of possible moves--it takes the strategy idea of core capabilities and asks how these capabilities might be used in new and unexpected ways to create opportunities.

3. Commitment

Many times, there are opportunities to shape the strategic response of a rival. Typically, these shaping opportunities are referred to as commitment. By committing to an action or strategy, the rival is forced to react and this can be advantageous. Commitment strategies are not always obvious. They may involve divulging information that one might think should best be kept secret. For instance, in the McCain-Schumer exercise, Pharma was able to gain an advantage by disclosing its lobbying strategy even though this was a key input for the rival to develop its most effective counter strategy. Commitment also takes the forms of threats or promises. The key here is that these threats or promises must be credible to be effective. Sometimes, the best strategy to ensure credibility is to destroy the flexibility to choose other options. For instance, in the Coors case, it was better for Coors to have financially committed to its aggressive marketing strategy for deterrence to be effective.

4. Leverage the future

The fourth key idea is to leverage the future in order to promote good behavior today. This is an intuitive idea--if reputation matters, then I may not act opportunistically today since the reputation loss will be costly tomorrow. Most informal contracts rely on this idea. Game theory helps in two ways. First, it identifies situations where informal contracts will not be sufficient. For instance, in declining industries, reputation is worth much less. Second, it identifies key aspects for making informal contracts work: clarity, detection, rewards/punishments, and forgiveness. Often, one can change the game to strengthen reputational aspects and obtain good performance, as in GE v Westinghouse.

5. Change the game

While many analyses consist merely of assessing a given strategic situation and determining the best path to follow, in the long-run, there is often scope for changing the game. This idea is sometimes called market design or mechanism design. Game theory provides a powerful set of tools for determining the right design to solve a problem. For instance, in the teams problem, we saw how game theory could be used to shed light on finding the right types of incentives to generate employee effort. In the auctions unit, we saw how different designs affected the surplus created as well as value captured by the auctioneer. The second half of the course dives much deeper into the design problem. Here, we will not simply use the tools of game theory to analyze situations as we find them. Rather, we will use the tools to reshape these situations into the outcomes we want.

Class 14 Take Aways

2011-03-09T21:33:00.000-08:00

Small is Beautiful

This class completed our segment on game theory in practice. It also marks the end of the first half of the course. In this unit, we learned how a "small is beautiful" strategy can allow an entrant to prosper even if it lacks any form of competitive advantage whatsoever. The key is the correct marketing mix---targeting only a subsegment of consumers and charging prices much lower than the incumbent. The strategy works because the incumbent lacks the flexibility to price discriminate between targeted and untargeted customers. It's another example of how outward thinking can create new uses for standard tools (like targeting).

Formally, the entrant looks forward and reasons back. It's targeting and pricing strategy determine the "game" played by the incumbent. In this game, the entrant wants the incumbent to accommodate (so it needs to both make fighting costly by setting a low price and accommodating relatively painless by not targeting too many customers. Here, the marketing mix is for the benefit of the rival, not as a means to capture value from consumers.

When does this strategy fail? It obviously fails if the incumbent can price discriminate. It also fails if the entrant cannot commit to stay small. Imagine a longer game where the entrant can increase its targeting in each period. In that game, the incumbent might well decide to crush the entrant early, even though this fight is more costly than accommodating in the short run, simply to head off the entrant growing larger in the future.

The small is beautiful strategy is closely tied to the puppy dog ploy we saw earlier. In both cases, the idea was to be as non-threatening as possible so as to avoid provoking an aggressive response.

Class Presentation Times

2011-03-09T17:39:00.001-08:00

Class 15 (March 14th)
1. MOAB
2. Max Payoff
3. Getting’ Nashty
4. Moral Hazards
5. F is a Big Deal
6. Class Reps
7. Yankees 27
Class 16 (March 16th)
1. Seagull
2. Mrs. Bento
3. Don’t Hate the Playa
4. Playadize
5. Free Rider
6. Money on the Night Stand
7. What Should I Do
8. Free Agents

Judo

2011-03-09T16:39:00.000-08:00

The judo spreadsheet is located at

https://spreadsheets.google.com/ccc?key=0AtUueq554t8-dEw2anh2cWVtY0F2SDNHaTd3VENRUmc&authkey=COjgnZwF&hl=en#gid=2

Please input your P&L for this game on the P&L spreadsheet. All P&L numbers for this game should be your dollar profits divided by 10.

Syllabus - Revised Topic Order

2011-03-09T09:34:00.000-08:00

Timing is Crucial

As I mentioned, the syllabus is slightly rejiggered for the timing of OPEC. This class we will cover Judo (formerly class 19). I've also moved the timing and identity of the guest speaker. My friend from Deloitte is unable to make it, so I've added a guest speaker from Yahoo, an ethnographer named Elizabeth Churchill. She will be speaking during class #19 on how social motivations affect market design. Social is obviously a huge deal in online these days, so her talk should be really useful for those of you thinking about startups with a social dimension (i.e. all of you contemplating tech startups).

Class 13 Take Aways

2011-03-08T23:18:00.000-08:00

The Practice of Game Theory

This class was all about translating the ideas of game theory into practice. In GE v Westinghouse, we saw how GE was able to use simple measures to create price discipline in the market. Specifically, it made prices transparent both in its price book and in disclosing prices and orders. The idea here is to make cheating easy to detect.
It used a price multiplier to make punishment and reward easy to implement. It offered its customers price protection supported by an outside auditor to reduce the temptation to engage in price cutting. Finally, it forgave price transgressions from Westinghouse following a short punishment period. Westinghouse, seeing the virtues of this approach, embraced it by adopting the same "best practices."

This is a textbook example of practically solving a social dilemma.

In the browser wars case, MSFT faces a seemingly insurmountable problem--how to tip a market with network effects away from an incumbent with market power. The key for MSFT was to make its browser sufficiently attractive that new customers in this growing market would adopt it. To achieve this, Microsoft made a credible commitment to offer, in effect, a zero price for its browser and to eliminate hassle costs. It did this through bundling. In effect, MSFT leveraged its market power in operating systems to credibly commit to an aggressive strategy in browsers. This strategy was enormously successful. In just 4 years, MSFT tipped the browser market in its favor and more or less put Netscape out of business.

The broader lesson here is that outward thinking creates new uses for standard strategies like bundling. We've seen this theme several times already: the finance versus game theory value of an option, the accounting versus game theory value of an asset, and now the market versus game theory use of bundling. The ability to think strategically creates all sorts of new possibilities for the managerial tools introduced throughout your tenure as MBA students.

OPEC Period 4 Results

2011-03-08T23:03:00.000-08:00

Too busy for a video today, so the results are in boring format:

Market A:
Price: 25.76
Quantity: 70,646

The punishment period apparently continues for Market A

Market B:
Price: 43.13
Quantity: 62,660

Market B seems to have sorted things out and is back to cooperating (to some extent)

OPEC Period 3

2011-03-06T14:46:00.001-08:00

Apparently another punishment period.

OPEC Period 2

2011-03-03T19:42:00.001-08:00

Mid Semester Topics Clearinghouse

2011-03-02T22:53:00.000-08:00

A Coordination Game?

Coordination is a pretty good thing. In an effort to facilitate coordination, this blog entry represents a clearinghouse for mid-semester project topics. I'll list topics for a couple of teams. Please add your team's topic in the comment section below.

Money on the Nightstand: NFL lockout
Mrs. Bento: Amazon's entry into streaming video

Class 12 Take Aways

2011-03-02T22:40:00.000-08:00

The Better Angels of Our Nature?

Why do organizations grow bureaucratic and rule-bound as they scale up? In this class, we saw how game theory can explain this phenomenon. This class was all about impediments to achieving cooperation in the face of social dilemmas. Here, we highlighted the social aspect of solving social dilemmas. In a small enough society, punishment by ostracism can deter bad behavior. The idea is that, while interactions with any particular individual in a society might be rare, if individuals talk, then a society level punishment might be sustained in the face of bad behavior.

But when societies grow large, this glue dissolves. Put differently, social solutions to social dilemmas are not scalable. The trouble occurs when the interactions become too infrequent and information too diffuse for a social deterrent to be effective. This can take the form of a tipping point where I society functions well and then, suddenly, good behavior falls apart in a significant way.

In a corporate setting, this implies limits to the size of effective organizations and an additional cost to outsourcing. Much of good behavior in organizations is enforced through social contracts. In Japan in particular, these contracts are often quite explicitly understood. Elsewhere they are more subtle. As long as the organization is sufficiently cohesive--interactions occur frequently enough and news spreads widely enough, then informal social contracts can work well. But, as the organization becomes more spread out and information spreads more slowly, a well functioning organization can suddenly lose its effectiveness. Even if other aspects of the business scale well, the social aspect need not.

The solution, of course, is to replace implicit social contracts with explicit contracts with monitoring. Of course, this might be very costly. Still, the prediction of game theory is that organizations will become more bureaucratic with scale simply as a means of solving social dilemmas within the organization.

OPEC Period 1

2011-03-01T20:17:00.001-08:00

Samples

2011-02-28T21:22:00.000-08:00

Free Samples
Many of you have asked about ideas for the mid semester project. There are many samples on the course website under the handouts tab of the course website. There you will find about 12 mid semester projects and 14 final projects from last year's class. This should give you some ideas about the scope, depth, and breadth of these projects.

Class 11 Take Aways

2011-02-28T21:19:00.000-08:00

Thinking of Tomorrow

In this class, we learned about how to leverage the future to overcome social dilemmas. To be successful, a strategy must have: (1) rewards for good behavior, (2) punishment for bad behavior, and (3) forgiveness or proportionality of punishment in case of mistakes. The strength of the punishment needed depends on the size of the temptation in the present as well as the value of future rewards. Greater temptation or less future reward require greater punishment.

There are limits to sustaining cooperation and trust. When discounting of future cash flows is great or the chance that the game will continue is small then it may not be possible. The limit of when cooperation is sustainable is made by comparing rewards and punishments under a grim trigger strategy--the threat to be mean forever following any transgression.

Final Farmville Update

2011-02-27T21:32:00.000-08:00

The Sun Sets on Farmville

Here's the final update: For both markets, verdant sold for 3, meadow for 2, and rocky for 1 in round 9. The market collapsed in round 10 with verdant selling for 1 and the others for zero. Hyperfarming was invented in period 11, so 10 was the last period. I've filled in the P&L for each team on the P&L spreadsheet at the site. In doing so, I noticed only a few teams put up P&L numbers for spectrum. Please put these up on the site.

Happy Farming.

Apple and Farmville

2011-02-23T22:29:00.000-08:00

We typically think of Apple as using a classic differentiation strategy. They make elegant products that appeal to customers with a taste for high quality user experiences and charge a premium price for this. While that strategy has been very effective for Apple, it suffers from the imitability problem. As rivals produce products that are cheaper than Apple and reasonable substitutes in quality, Apple tends to lose share. The classic example of this was the development of the Window OS, which massively undercut Apple's advantage in UI quality.

Apple now seems to use game theory to anticipate this. When it rolled out the iPad, it adopted a Farmville type strategy. A key component to imitate the iPad is the capacitive glass used in the screens of these devices. This input is difficult to produce with only limited fabs currently manufacturing it. Apple prepaid for about 60% of world output of this glass. Obviously, this helps Apple to avoid shortages and production delays but, more importantly, it ensures that imitators will face high prices for this input as they come to the market. The result is that Apple now also enjoys a cost advantage as well as a differentiation advantage. This will obviously slow market share gains from rivals and allow Apple to sell in quantity thereby further lowering input costs.

The combination of offering a vertically differentiated product while at the same time having lower costs makes Apple a fearsome competitor.

Class #10 Take Aways

2011-02-23T22:17:00.000-08:00

A Pyramid Framework

In Class #10, we worked out a framework for analyzing games. At the bottom of this framework is data, the building blocks for doing interactive strategy. The level above this is competitor analysis. Here we take our first cut at working out the details of payoffs and strategies available in the game as well as starting to make a mental model of our rival.
Above this is behavioral analysis. The goal here is to flesh out the mental model. Does the rival suffer from certain tendencies that will drive her strategy? Is she analogizing incorrectly? Is there a way to reframe the situation to gain an advantage? Finally, game theory is at the top of the pyramid. Having laid out the game, the strategies, and the payoffs and developed a "theory" of our rival's moves, we can use equilibrium ideas like look forward reason back, dominance, and Nash equilibrium to determine our best course of action.

This framework can be helpful in taking complex situations and reducing them to their key strategic elements so that sensible analysis is possible. Part of the art of game theory is knowing what not to include in the analysis so as not to get bogged down in detail and lose the key strategic factors of the situation.

Class # 9 Take Aways

2011-02-23T22:01:00.000-08:00

A Teams Problem

In class #9, we talked about the teams or partners problem. The problem goes like this: two partners work on a project. Each unit of effort returns 1+x dollars and imposes a personal cost of $1. Assume x < 1.

The problem is that individual effort cannot be directly observed.
Working alone and receiving the full value of output, each partner is willing to work all out. But working as a team, then any profit sharing scheme will lead the partners to (optimally) free ride. It is a dominant strategy not to work at all.

This is a version of the holdup problem. The holdup problem occurs when another party can capture part of value created. Consider a 50-50 profit share. Now, each unit of effort produces 1+x dollars for the partnership, but the individual partner only gets half of this, which is less than the personal cost of effort.

One solution to the partners problem is to set threshold type incentives. Here, the idea is that pay is based on a sliding scale of total output. In the simplest version, partners only get paid if company profits exceed some target and not otherwise. This transforms the game into a stag hunt. Now cooperation is an equilibrium, but so is shirking. The culture of the organization determines which regime will prevail.

An aside: If the partners work for an outside manager, this scheme works fine, but it won't work so well if the partners own the company. The reason is that, for the scheme to work, the partners have to commit not to be paid in the event the target is not reached. But if they own the company's profits, they cannot really do that. The situation here is even worse, requiring some sort of repeated game incentives to sustain cooperation.

Class #8 Take Aways

2011-02-23T08:20:00.000-08:00

An Allocation Problem

In class #8, we discussed market design--what is the right way to allocate spectrum licenses? One way of allocating the licenses is to use a Vickrey auction.. This auction has the "pay your harm" property. You pay an amount equal to the harm you cause to the surplus of other bidders. With one unit for sale, harm is done to the second highest bidder in the amount of her bid, so the high bidder pays second highest bid.
This type of design has the nice property that it is a dominant strategy to reveal truthfully. (It also has some not so nice properties like complexity and lack of equity).

We compared this to the design actually used, the simultaneous ascending auction. This design is simpler and fairer, but leaves bidders to grapple with the exposure problem---when should you stop bidding on items in the presence of the synergy. The aspect of the design where none of the items close until all of them close is an attempt to deal with exposure.

Farmville Round 8

2011-02-22T22:07:00.001-08:00

No property transactions or upgrade. Prices hold steady at 3, 2, 1 in both markets.

Farmville Period 7

2011-02-19T11:23:00.000-08:00

Same results for both markets, so only one video.

Farmville Period 6

2011-02-17T21:33:00.000-08:00

Things were the same in both markets, so only one Farmville report.

Upgrades: Don't H8 the Playa upgrades 3 meadow to verdant. Class Reps upgrades 3 meadows to verdant, Playadize upgrades 1 meadow to verdant.

FarmVille Period 5

2011-02-15T22:42:00.000-08:00

Market A

Market B

Upgrade Decisions:
Key: Team, Meadow to Verdant, Rocky to Meadow
F is a Big Deal, 3, 3
Mrs. Bento 15, 1
Free Riders 9, 0
Money on the Nightstand 3, 0
Gettin' Nashty 3, 0
Free Agents 3, 1
MOAB 3, 1
Class Reps 3, 3
Dont H8 the Playa 0, 3
Playadize 3, 1

Class # 7: Key Take Aways

2011-02-15T15:36:00.000-08:00

In this class, we studied the game theoretic valuation of an asset. While the usual role for holding an asset in a company is for the cash flows it throws off, there can also be strategic reasons. The presence (or absence) of an asset can provoke certain types of strategic responses from rivals. In the case we studied, the presence of excess capacity created incentives for the rival to respond forcefully to a price cut.
The reason is that failing to respond resulted in sufficient lost market share that retaliation was cost effective. In contrast, holding smaller excess capacity weakens the threat of a price cut and changes the strategic response of the rival. Now a price cut leads to smaller market share loss. The cost of retaliating in that circumstance is lower than the benefit. This type of "weakness" strategy is sometimes called the "puppy dog ploy."

Puppy dog ploys appear in a number of practical contexts. One notable context is for startup airlines. Imagine a cut rate carrier decides to add a route already served by an incumbent. (Kiwi Airlines did exactly this against Continental in Newark some years ago). If the startup is small, then the lost share is such that it's not worth fighting a price war. If, however, the rival's capacity grows (or threatens to grow), then a different and altogether more aggressive strategic response is provoked. Ultimately, it was the growth of Kiwi that doomed its business model.

Spectrum Auctions: Key Takeaways

2011-02-15T15:24:00.000-08:00

Link: Results of Spectrum Experiment

Our first variant in the spectrum auctions experiment was the Vickrey auction. William Vickrey first designed this auction in a single unit context in order to obtain truthful revelation of values as a dominant strategy. The key point of the auction is that your bid only determines the threshold price for winning and losing. It does not determine the price you pay. When there's only a single object, it is easy to see how this works. Since I pay the amount of the second highest bid, my bid merely determines the maximum price I could possibly pay. As such, I can do no better than to set this threshold equal to my value.

Things are more complicated when there are many objects and synergies. The Vickrey principle is that you pay the amount of the harm you do to the values of other. Again, with one object the "harm" done to bidder 2 is equal to her valuation (which is her bid amount as well). Thus, paying the second highest price compensates for the harm. In the multi-object case, the harm consists of computing the allocation of the goods absent bidder 1 (say) and then the allocation with bidder 1 present. The difference in the valuations (expressed in bid amounts) of the others is the amount of the harm.

Most teams had an intuitive grasp that bidding truthfully was the best strategy.

Vickrey auctions are unpopular in practice for several reasons. First, they are hard to explain. Second, the high bidder does not always pay the highest amount. Third, when there are a large number of items, the auction becomes unmanageable. Finally, and we failed to mention this in class, for bidders with budgets, determining the right bid is difficult. A conservative approach is simply to bid up to your budget, but this merely ensures you'll have money left over and won't net you the objects necessarily. On the other hand, bidding your value when it's above your budget exposes you to the risk of going over budget.

Next, we studied the simultaneous ascending auction. This auction is simple to manage and easier to understand. The key difficulty is working out the right bidding strategy with synergies. Bidding with the synergy in mind creates the "exposure problem"---bidders are exposed to the risk that they won't get one of the objects and the anticipated synergies will then not be realized. This can create cautious bidding yielding low revenues and low value created.

Please input your profits from this experiment to the P&L spreadsheet.

FarmVille Period 4

2011-02-12T20:19:00.001-08:00

Market A:

Market B:

FarmVille Period 3

2011-02-10T22:49:00.000-08:00

Class #6: Take Aways

2011-02-10T18:42:00.000-08:00

A Hawk-Dove Game in Action

In class #6, we formally introduced the notion of Nash Equilibrium. A Nash equilibrium is a set of strategies that are mutual best responses. We also talked about archetypal games, games representing a wide array of real world business situations. These included: pure coordination games, battles of the sexes (coordination with a distributive element), hawk-dove (coordination with distributive and risky elements), stag hunt (simple model of trust), and prisoners dilemma. A key observation is that rationality is not sufficient to avoid bad outcomes. While individuals might be perfectly rational, strategic interaction can produce collective irrationality.

FarmVille Period 2 Results

2011-02-08T17:19:00.000-08:00

Market A

Market B

FarmVille Period 1 Results

2011-02-05T21:53:00.000-08:00

Market B Results

Market A Results

Please note, the video reverses markets A and B owing to a transcription error on my part. All references to market B are for market A and vice-versa. Apologies for the glitch.
FarmVille FAQ

Q: Is upgraded land available for rent in the same period it is upgraded?
A: Yes.
Q: Can I perform multiple upgrades (i.e. rocky to verdant or scrub to meadow) in the same period?
A: No. You can only upgrade a single level in a period.
Q: Are there more farmers in the market with 8 firms than the one with 7?
A: Yes. There are 4 farmers per landlord on average. There is the same randomness in the number of farmers in both markets.
Q: Can I specify the rental price of land?
A: No. You can only determine the amount of land you wish to make available. The rental price is determined by market clearing. If supply exceeds demand, the price will fall to zero.
Q: Can I use acquired property in the same period I acquire it?
A: Yes. Property transactions occur "between" periods and the acquired land is immediately transferred and available.
Q: Will I learn the land amounts that other teams put on the rental market?
A: No. I will only reveal the clearing prices. The total available land and who offered it is secret.
Q: What is the terminal value of land?
A: It is worthless once hyperfarming is invented.
Q: If I rent land in a period, can I sell it before the next period?
A: Yes. Rental contracts are for one period only.
Q: Who am I competing with?
A: You are competing with a simulation of last year's MBAs. Since this game did not exist last year, I will use other game results to simulate how much would have been made in this game. To win money at the end of the year, you need to accumulate as many experimental dollars as possible from all of the games.

Contact Information

2011-02-03T21:59:00.001-08:00

Apparently some of you are very shy since there are only 3 teams with contact info on the board at present. Team Don't H8 the Playa kindly asked me to prod all of you to provide contact information. Apparently, they're ready to wheel and deal and looking for folks to trade with. To provide your info, go to the game theory P&L and add it to the "contact info" column on the spreadsheet. The direct link is:

Link: https://spreadsheets.google.com/ccc?key=tJkfldzw-qQHHuX3GLk_4uw#gid=0

McCain-Schumer Debrief

2011-02-02T18:52:00.000-08:00

The key variation in the experiment turned on the timing of the game or, equivalently, the disclosure policy of Pharma. When a game is sequential, there is a chance to influence the rival's move with your move. In a simultaneous game, this is not possible. While it is normally to your disadvantage to reveal private information, when that information is your action choice, disclosure can be advantageous owing to this strategic influence.
In these situations, disclosure can never be harmful---simply choose the same action as in the simultaneous game and the payoffs will be the same. Moreover, disclosure can be positively helpful. In the lobbying game, Pharma can increase its bid over the simultaneous equilibrium levels and get its rival to lower its bid. The improvements in the chance of winning more than compensate pharma for the increased expense.

More broadly, timing and disclosure decisions are often under the control of a firm. Game theory provides tools to study how these tools can be incorporated into a firm's strategy to secure additional advantage.

View the results of the experiment here.

Class #4: Key Takeways

2011-02-02T18:40:00.000-08:00

Class #4 ended the unit on Look Forward, Reason Back. Here, we saw how the tools of game theory could be used to determine the appropriate course of action in the face of the threat of entry. In the case, Coors could opt for a proactive strategy, advertising in advance of the entrant, or a reactive strategy, advertising only if the entrant came into the market. Instinct would suggest that the proactive strategy is better, but game theory illustrates the possibility that the reactive strategy is the right one. The key condition is whether Coors will carry through with its threat to advertise if AB enters. If this is not credible, then Coors needs to take steps to make it credible.

The broader point is that simple tools of game theory can be extremely helpful in systematically analyzing strategic situations and determining the right course of action rather than relying purely on instinct.

FarmVille

2011-02-02T09:09:00.000-08:00

For the next couple of weeks, we will be running the game Farmville outside of class. This game illustrates the connection between game theory and competitive advantage in strategy. There will be two markets, one consisting of 8 teams and the other of seven teams. Please read the instructions in advance of class and I'll explain in class.

There will be two markets playing the game. You will only be competing with teams in the same market. They are:

Market A:

Seagull Managers

Don’t H8 the Playa

Getting' Nashty

What Should I Do?

MOAB

Mrs. Bento's Box

Yankees 27

Market B:

Money on the Nightstand
Moral Hazards
Class Reps
F is a Big Deal

Max Payoff

Free Riders

Playadize

Free Agents

Road Map: Classes 6 and 7

2011-02-02T08:52:00.000-08:00

In class #6, we will discuss how to analyze games with simultaneous moves and highlight a number of "building blocks" for studying strategic situation. Please read the class notes in advance. Class #7 introduces the idea of dominant strategies. Here, we will be doing an experiment simulating the radio spectrum auctions.

Class # 3: Key Takeaways

2011-01-31T18:49:00.000-08:00

The Long shadow of the Endgame?

Class #3 concluded our unit on look forward, reason back. There were three key observations to come out of this unit.

1. The long shadow of the endgame--the end of the game casts a "shadow" over the entire strategic interaction.
In the NBA case, that shadow was the competitive threat of the rival team. The right of first refusal nullified this competition and let the incumbent capture the bulk of the surplus. Absent this clause, competition allowed the player to capture the bulk of the surplus.

2. The value of an option - the game theory value of an option depends on how it affects others behavior. It is often most valuable when it is not exercised since this means the competitive threat of the rival has been thwarted.

3. Irrationality - We typically use the term rational to be synonymous with profit maximizing. But non-pecuniary factors drive payoffs too. Game theory is flexible enough to handle these aspects in the analysis. A key part of developing a good model of the rival is assessing payoffs accurately--fairness, reciprocity, and other ego factors are important to account for.

NBA Free Agency Key Takeaways

2011-01-26T20:15:00.000-08:00

The central point of the experiment was to illustrate the difference between the game theoretic value of an option compared to the valuation of options in finance. In the NBA free agency game, the option provokes a strategic response from the rival team. Specifically, the presence of the option destroys the possibility of bidding successfully for the player and hence nullifies competition. The option is only valuable when it is not exercised. Exercising the option means that it has failed to stave off competition.

The broader point is to illustrate the difference between inward and outward thinking. Options are normally thought of as providing value to the holder through the action of exercise. In this experiment, the option derives value from changing the action of the other players in the game.

The value of the option is simply the difference in payoffs to the incumbent firm with and without it. With the option, the incumbent captures the bulk of the surplus, typically around $7mm. Without it, the incumbent captures very little, only about $500k using its best strategy. Thus, the option is worth about $6.5mm even though it is never exercised.

Such options appear in a number of contexts. In core strategy, this same idea appeared in the form of "meet or release" clauses secured by Nutrasweet in advance of entry by Holland Sweetener Company in the US. In venture capital, such clauses are often put into place by early round investors to help damp down competition for later round financing leading to IPO.

The other key point illustrated in the experiment is that pecuniary motives are not the only things determining payoffs in a game. Non-pecuniary factors such as reputation, fairness, and regard for others often play a role. Even when the stakes are large, these factors still matter. A great many professional athletes have chosen to sit out for a season because they perceived being treated unfairly, despite leaving millions of dollars on the table. While the caricature view of game theory/economics has in mind that, in the words of Randy Newman, "it's money that matters," the tools of game theory are flexible enough to incorporate these less easily measured incentives as well. Part of the art of good mental modeling is determining how the non-pecuniary aspects of a strategic situation are valued and incorporating them into the analysis.

Class #2 Key Takeaways

2011-01-26T20:01:00.000-08:00

In class #2, we started talking about the fundamental rule of thinking strategically--look forward and reason back. We illustrated this in two ways. First, in durable goods industries, often competition comes from future "selves" of the firm. For instance, a firm selling a technology product can look forward and anticipate that it will want to cut prices in the future.
Reasoning back, this will affect the purchase decisions of today's consumers and hence optimal pricing policy. The result of this future competition is to erode today's margins. One solution to this problem is price protection. By "changing the game" for its future selves, it reduces the temptation to cut price in the future thereby maintaining margins today.

A second version of this idea occurred in the NBA free agency game. Absent the right of first refusal clause, the incumbent can look forward and anticipate strong competition from a rival. Reasoning back, this means that the incumbent must make an aggressive offer today. The competition between the incumbent and the rival allows the player to capture most of the value. The right of first refusal fundamentally changes this. Now, if the rival makes a counter-offer, the incumbent knows it can match this deal. Reasoning back, the rival now has little interest in making offers and hence the incumbent enjoys, effectively, no competition.

Road Map: Classes 4 and 5

2011-01-26T19:51:00.000-08:00

In class #4, we will be completing the Coors case and then playing the McCain-Schumer game. Please read the instructions carefully before class. In class #5, we will debrief McCain-Schumer and then talk about the case Game Theory and Business Strategy.

Experiment 1 Results

2011-01-24T17:37:00.001-08:00

Please send me the results from the NBA game by Tues at 1159pm. Also, if you have not already done so, please send me the name of your team and its members. Finally, we currently have 15 teams, which is one too many. If you are team consisting of 3 members, please consider joining a larger team. All teams should have 4-5 members

Road Map: Classes 2 and 3

2011-01-19T17:41:00.000-08:00

For class 2, we'll do the NBA free agency experiment and debrief. Class 3 will cover the Coors case (a follow on from core Strategy) as well as the analytics of look forward, reason back.

Before class 2, please remember to form your team and come up with a colorful nickname.

Welcome to Game Theory

2011-01-14T13:21:00.000-08:00

Photo Credit: John Morgan

Welcome to the Game Theory Blog. Here you will find updates and roadmaps for upcoming classes, results from experiments, videos, as well as key takeaways from each class. Please check back here for the latest. Please become a follower of this blog.

Also, it is ideal if the blog is interactive. Please submit comments if you have a take about what's covered in the blog or have suggestions for topics.

In preparation for Class 1, please check out the course overview lecture notes. You can find them here and here. The first of these links is the "live" syllabus. As new content is added, the links to that content will go live. The second is the handouts tab of the course website, for later reference.

For transparency reasons, the course website is not on bspace. You can find it here The reason is that bspace is walled off from Game Theory alumni. I want those who took the course in the past to be able to access the materials. As a result, I make everything available without walls.

See you in class!