…it appears to me that the design of markets…presents both the biggest challenge and the biggest opportunity facing applied game theory at the brink of the twenty first century.1

For the past few years, among other activities, I’ve been working on creating three markets:

Outgrowths of the Buller Centre for Business

The first two–Find Good People and BiomassBrokerage.com–were developed as part of my role as Director of Providence’s Buller Centre for Business. There is still a lot of work to do on each of them, but they are both online and functioning well.

BiomassBrokerage.com has been spun off from the Centre and operates as its own business. Technically, it is two markets–a market for buying and selling biomass material, and a second market (the trade show) for buying and selling biomass-related equipment and expertise. However, because the same participants will most likely engage in both markets, and they are housed at the same web site, participants will experience them as a one market, so that’s how I’m treating them here.

Neither Find Good People nor BiomassBrokerage.com breaks new ground in market design, but each uses the database capacity, reach, and flexibility of the internet to create a niche market.

And each demonstrates a little-noticed shift that is happening in the internet. For most of its existence, the internet facilitated two functions–exchanging information, and exchanging goods and services. From fairly early on, both large and small participants could create forums for exchanging information, but only large participants could create mechanisms for exchanging goods and services. So, for example, anyone could buy and sell on eBay, but creating and maintaining eBay–or creating its competitor–was beyond the reach of most of us.

DIY Market Creation

This confinement of market creation to larger participants is beginning to evaporate. Both Find Good People and BiomassBrokerage.com were created using inexpensive, easily-available software and platforms (WordPress, Formidable Pro, MySQL) and simple programming languages like html and php. Although the sweat equity has been quite large, the cash outlay is tiny (about $200), and the programming knowledge required has been basic.

We are now seeing a shift to where individuals and very small businesses can create and manage markets. Find Good People and BiomassBrokerage.com are both examples of this shift.

There has been a lot of commentary on other shifts in the internet–like the maturing of the maker movement and the potentials of 3D printing when integrated with the internet. But the emerging ability of small participants to create and manage markets is happening with little fanfare. And, I think, it could be at least as important as these other trends.

Innovating in Market Design

The third—TradeMyPlayers.com—is being developed out of a personal interest in market design in general, and in the design and practicality of multi-participant trading in particular.

What is Market Design?

A good working definition of market design is the “use of economic theory, experiments, and empirical analysis to design market rules and institutions“, which is how Stanford’s Department of Economics describes market design in its Explore Course, ECON 136: Market Design2.
Market design is both a very old field and a very new one. Plato (ca. 360 BCE) devotes part of Book II of his Republic to an exercise very close to market design, positing the ideal market for the ideal state. Yet, today, it is still possible for one of its leading practitioners, Alvin Roth, to consider it an emerging discipline.3

Currently, market design is usually treated as a branch of game theory, although some consider game theory a branch of market design. I haven’t given this question enough thought to have an opinion.

Alvin Roth and Lloyd Shapley, another pioneer of market design, jointly received the 2012 Nobel prize in economics for their work in a part of market design–matching agents like students and schools, kidneys and patients, or husbands and wives. (For those wanting a fuller overview, Niederle et al.4 is a good place to start.)

Market design can be observational, attempting to understand the rules by which a market operates. However, it is more usually an applied discipline–creating rules for a market which could function in the real world. And it is usually prescriptive–proposing rules which are intending to have the market “succeed” in some sense.

Redefining “Free” Markets

A “free” market is not a chaotic market. To be durable, a market needs to have rules. These can be formal or informal, acknowledged or un-recognized, but they need to be agreed to and followed by the overwhelming majority of market participants, if participants are to be eager to return for multiple encounters.

So what rules sustain markets? Who decides on those rules? How do they decide? How are the rules changed? Why are some markets fragile or erratic, while others appear durable and stable? For me, these basic questions are at the core of market design.

Multilateral Trading

Roth and Shapley won their Nobel prize for work on two-participant matching. (Even if you have a dozen kidneys and a hundred potential recipients, in the matching process, you always ends up with pairs—each kidney goes to only one patient.)

This pair-matching—bi-lateral trading—is far and away the most common type of market interaction. There may be thousands of sellers and millions of buyers in a market, but the actual exchanges are almost always bilateral—agreements between a single buyer and a single seller.

The practical limitations that have constrained almost all trades to two participants include:

  • complexity and trust constraints
  • research costs in finding willing participants to a trade
  • transaction costs in completing trades involved more than two participants.

However, not all market interactions are bilateral. Sports teams, for example, sometimes make three-way or even four-way trades, but these sorts of trades are ad hoc and rare.

There could be a number of interesting, real-world uses for multilateral trading—people wanting to trade houses for their holidays, for example. However, at least so far, very few functioning multilateral markets exist.

Perhaps only one model for multilateral trading exists in the academic literature—the Top Trading Cycle algorithm. It was invented by David Gale, and first reported by Lloyd Shapley and Herbert Scarf5. One of the most interesting real-world applications is probably in New Orleans where, apparently, they are intending to use the Top Trading Cycle to facilitate school choice.

Of course, the Top Trading Cycle is not the only potential algorithm for multilateral trading, but it is the one that has been given the most thought.

Writing About Multilateral Trading

Like bi-lateral trading, multi-lateral trading can come in a variety of flavours:

  • Are the trades single-encounter or multiple-encounter?
    • Do the buyers and sellers typically trade only once, or is the practice in the market to have multiple encounters?
    • While, in the real world, a single market can facilitate both single and multiple encounters, in game and market modelling, it is recognized that the incentives for participants in single encounters is significantly different than for participants in multiple encounters.
  • Are the goods homogenous or heterogeneous?
    • This is a misleading question, in that it is asking whether the participants assign the same utility to the same good (homogenous) or whether they assign different utilities to the same good (heterogenous).
  • Are the goods divisible or indivisible?
    • In other words: Can you exchange a part of a good (divisible), or does the whole good need to be exchanged (indivisible)?

A few years ago I developed a possible approach to multiple-encounter, multilateral trading in heterogenous goods, and wrote about a draft paper on it called “A model for n-trading in ongoing markets in heterogeneous indivisible goods”.

Trade My Players – a Multilateral-Trading Demonstration Project

TradeMyPlayers.com, which is in development, is a project I’ve been working on through Boke Consulting to create a real-world application of multi-lateral trading. I initially created it off-line, using Excel (for data entry by participants and results display) and Visual Basic to run its algorithm.

For those who are curious, I’m happy to email you a copy of the Excel document, if you’re happy to sign a non-disclosure agreement.

The step I’m working on now is to put it online.

TradeMyPlayers enables participants in a sports fantasy league to make multilateral trades quickly and easily. Although it is intended as a demonstration project, it could easily be integrated into one or more of the current online fantasy leagues, or it could operate as its own standalone site. The application could, potentially be quite large, as millions of people participate in fantasy sports leagues (both on and off-line) every year.

Other potential applications for multilateral trading beyond that of TradeMyPlayers and those mentioned above include:

  • real sports teams trading real players (and other trading in services)
  • nurses in a hospital system or firefighters in a department trading shifts
  • wine collectors trading bottles (and other collectors trading objects)
  • airlines trading airport terminal docking rights
  • time-share trading
  • trading in fishing permits and some other resources in ocean zoning negotiations

It seems unlikely that expect that every application would use the same algorithm. We could expect, for example, that the “best” algorithm for single-encounter multi-participant trading would differ significantly from multiple-encounter multi-participant trading.

My intention is to brand applications of multi-lateral trading developed by Boke Consulting as BokeTrading, regardless of the algorithm used or the field in which it’s applied.

1. Roth, A. E. 2000. Game theory as a tool for market design. Game Practice: Contributions from Applied Game Theory (F. Patrone, I. García-Jurado, S. Tijs, eds.) 7-18. Dordrecht: Kluwer Academic Publishers.
2. . From the short summary of the course here. A slightly longer summary is found here.
3. Roth, A. E. 2002. The economist as engineer: Game theory, experimentation, and computation as tools for design economics. Econometrica 70 (4) 1341-1378.
4. Niederle, M., A. E. Roth, T. Sonmez. 2007. Matching. The New Palgrave Dictionary of Economics, 2nd ed. Palgrave Macmillan.
5. Shapley, L. S., H. Scarf. 1974. On cores and indivisibility. Journal of Mathematical Economics 1 23-37.