Positive Liberty

In which the armies of Isengard take a Hayekian battering ram to the citadel of electronic markets.

Tyler Cowen of Marginal Revolution notes that online betting sites are offering futures trading on the identity of the next pope. I am both intrigued and skeptical (as is the Mises Economics Blog). What follows is my own self-indulgent analysis of the phenomenon.

Political Futures Markets: As discussed in my review of Bruce Caldwell’s Hayek’s Challenge, markets act as analyzers of distributed information, effectively collecting, processing, and disseminating imperfect, speculative, and partial knowledge. The process maximizes overall utility better than any central planning agency could ever hope to do. It may even be likened to a vast computer program–or more properly to a human brain, where a massive number of interrelated connections, themselves largely meaningless, all combine to form what we experience as conscious, purposeful thought.

Perhaps drawing on this insight, the idea that futures markets shed light on upcoming events has been quite a fad in recent years. The Iowa Electronic Markets offer futures in a wide range of local and national elections, while other markets do likewise on the outcomes of sporting events, awards shows, and even the verdict in the Michael Jackson case.

The logic behind these markets is simple: Assume that distributed among the traders, there is enough information to solve the “problem” of how the event will turn out. If so, then the only remaining obstacle is that the knowledge is so finely dispersed: None of our actors have all the information they need to solve the problem. Instead, they possess only what they have been able to gather elsewhere–divided by a factor representing their biases, logical errors, and conceptual blind spots.

Biases aside, such markets deserve some benefit of the doubt, because the people who really do have some scrap of information about the identity of the next pope will be far more likely to participate in the papal futures market, to buy more shares, and to respond more intelligently to fluctuations in market prices. The prices of shares will thus tend to follow the smart money–wherever that money happens to be.

But there are at least two reasons that futures markets on the identity of the next pope, and futures markets on elections more generally, are not likely to collect much useful knowledge.

Bad Aggregators: First, a futures market in a political event does not serve as a good aggregator of information because it lacks the iterative quality of a genuine commodity market. In real markets, the same goods are traded again and again from day to day. Unlike wheat or oil, there is only one election for pope that is ever going to resemble the current one. The “futures” in the various candidates are unique goods in a one-time market.

Where commodities traders accumulate market knowledge continuously–and have to live with the consequences of their trades from day to day–futures markets in political events can seldom make use of information gathered through the successes or failures of past trades in the same market. Previous elections have only a distant relation to the one at hand, and it is a notoriously bad political strategy to re-fight the last election.

So while the prices of shares may follow the smart money in both commodities markets and political events futures, the “smart money” in the latter is never very far ahead of the rest. And why should we expect it to be? In a market that only iterates once, there is no way for traders to improve their performance.

An example of this comes from the Iowa Electronic Markets futures trading for the 2004 U.S. presidential election, an experiment which failed to predict anything at all before it became screamingly obvious. Worse, the IEM even seems to have had some very deep structural inefficiencies that a really smart trader could have taken advantage of.

Note that for an election that is not terribly close, we do not need the IEM or any other system of futures trading to tell us what the likely outcome will be. It is only in relatively thin, but not razor-thin elections (like that of 2004, but not so close as 2000) that we might expect to gather marginally useful information. Yet even in 2004, the IEM did little more than to follow the news reports; although I cannot find a graph of it now, I do recall that the final hours of trading produced few surprises. Minute by minute, the IEM followed the news releases, reacting to exit polls that favored Kerry, then to early returns suggesting that Bush was leading in the swing states, and finally to the news that he seemed likely to win both Ohio and Florida. The market gathered specific information of no better quality than what was also disseminated among bloggers, the mainstream media, and general word of mouth.

The papal futures market seems to suffer all the same difficulties as the 2004 election market–and maybe more, seeing how there are absolutely no agents who have ever traded in any sort of papal futures before: These markets were not invented until the 1990s, while John Paul II was elected pope in 1978. Any information peculiar to a papal election yet enduring from one papal election to the next will have to be aggregated from scratch.

Moreover, the dispersed information in the papal market is clearly of lower quality as well. In 2004, millions of people, including the mainstream media, bloggers, office know-it-alls, and even kindergarten classes, all speculated on the presidential election, gathering and creating more information as they went. The commentariat had never been larger, and arguably it had never been more well-informed. A large segment of this same commentariat also turned out personally at the polls, meaning that their pre-election data included the knowledge of how at least one actual elector would vote. In the aggregate, the information market really did possess a vast store of genuine election data to draw upon. Of all election markets, this one would surely be the one to find something useful–yet it failed.

Now consider the papal market, where few people know very much about the College of cardinals, and where fewer still will actually vote in that College. Cardinals are also sworn to secrecy about all that goes on in their proceedings, making them far less likely sources of information than the typical campaign staffer. If distributed information is out there, then I have no idea where it might be found.

Knowledge “How” and Knowledge “That:” The presence of real information, of iteration, and of at least roughly standardized conditions all seem crucial to how the market aggregates information. But even when everything works properly, the real knowledge product of a market is not a specific datum, like the name of the next pope. Instead, it is know-how: It is the elusive, hard-to-communicate property that separates expert traders from novices. Know-how is not necessarily going to produce any one specific insight on the market even when it exists abundantly.

As is my wont, I now propose a fanciful example.

Let’s consider real time strategy games (RTSes). In an RTS, players create buildings, gather resources, recruit fighting or laboring units–and attempt to destroy one another in combat. (Extra credit to those who immediately noticed the economic inputs of land, labor, and capital in the structure of the RTS. Interestingly, RTS land returns rent, in the form of resource points, much as land also does in the real world; likewise, soldiers return wages, in the form of the benefit I get by destroying enemy inputs. But there is no return on capital in an RTS; resource points that sit in one’s account cannot gather any interest. Practically speaking, all capital is kept under a mattress. This means that accumulating capital without a specific goal in mind is always a strategic mistake in an RTS, while planned capital expenditures should be made as soon as they can be afforded).

My current favorite RTS is Lord of the Rings: The Battle for Middle-Earth, a game drawing on the world of J. R. R. Tolkein. Players command the armies of Mordor, Isengard, Rohan, and Gondor, each with distinct units and structures. Besides knights and archers, Gondor can recruit Gandalf to its aid; Rohan has strong horsemen and fearsome Ents; Isengard boasts nasty siege machinery and quasi-industrial upgrades for its units.

As in all RTSes, the players in The Battle for Middle-Earth are economic decision-makers in the most classical sense of the term, for they face the challenge of how best to allocate limited resources to meet their goals. Crucially, we Middle-Earthlings are in direct competition with each other–over and over again. When a player wins, his economic choices are shown to have been more efficient than his opponent’s.

While the rules are clear, the winning or losing strategy is basically never apparent to a novice at the outset. As players first approach the game, all that they have are lists of data on the prices and capabilities of various units and structures. Few of these provide any real clue toward their economic utility; inert data proves a poor substitute for the know-how that one gathers in competition.

To give an example of the value of iteration and consistency, consider a series of games I have played within the past few weeks. In the first, I commanded the armies of Gondor. I built some farms, an archery range, a market (of course!), and I then began saving for Gandalf, a powerful wizard who can destroy a battalion of orcs with a single blast.

Before Gandalf arrived, though, the army of Isengard was already at the gates. Nor was it any ordinary army; it consisted solely of battering rams and uruk-hai foot soldiers. My opponent had purchased only the “forged blades” upgrade from the extensive menu of quasi-industrial choices at his disposal. Clearly he was following a nonintuitive strategy, one probably gleaned from prior competition–while I had relied on intuition alone.

“I think we’re in trouble,” I messaged my teammate.

Within moments the specialized army had decimated my archers, leveled my citadel, and won the game. I saved a recording of the debacle for later review, and it turned out that my opponent had pursued his strategy both fiercely and single-mindedly: Though many options were available, he had not made even a single choice that fell outside his clearly defined plan.

Immediately I adopted his strategy against a new opponent–and won an overwhelming victory. The next game I played, though, I used the same strategy and lost: Horsemen are quite effective against uruk-hai foot soldiers, even with the forged blades upgrade. What’s worse, battering rams are powerless against anything but walls; even a single horse can trample a battering ram ingloriously to death.

On my third game, I tried using pikemen instead of foot soldiers. I also made sure to protect each of the battering rams with blocks of pikemen until they reached the castle gates. Slightly modifying a strategy that I had acquired elsewhere allowed me to win another decisive victory.

These examples show that practical know-how, nonintuitive and rather difficult to communicate, is precisely the benefit that one reaps through competition and choice in the market. Specific predictions–like the name of an upcoming pope–have little to do with it, and may remain unknown, even to “expert” agents within the market: To this day, I cannot give accurate predictions of how other Middle-Earthlings will fare when they play; even when observing a game, I still sometimes can’t tell who is winning until the battle is concluded. All the same, I am a much stronger player than I used to be.

In short, practical know-how improves incrementally over time, often without our ever noticing it–but it depends absolutely on an iterated market, and its information is not always of the type we would wish. For all I know, my defeated opponents are reviewing their games even now, hoping to adopt “my” strategy in later encounters–or perhaps to improve upon it. If they do, I hope to learn from them.

This is why free markets, aided by the rule of law, do so remarkably well at supplying economic needs–while centralized planning tends to follow the planners’ intuition (or worse) into all manner of inefficient distractions. It is also why a futures market on the papacy bears only a surface resemblance to a genuine market, and why the chance of anything more than a trivial success is close to nil. Worse, the chance that only one iteration will allow our market to rise above the common perceptions is even closer to nil. Where the ordinary news reports fail, the markets in popes, basketball champions, and guilty former pop stars can scarcely hope to succeed: What has failed is not the market, but our understanding of what market aggregation is really able to do.

Trackback URI |