Prediction Markets in the Firm

Tyler Cowen asks why prediction markets are not common in business. The obvious answer is that the concept is still too new for most organizations to really adopt it. It took decades for U.S. based businesses to catch on to quality programs.

Having answered the question, there are some mis-conceptions about prediction markets that I feel the need to dispel:

"Prediction markets threaten the hierarchical control of top managers."

Actually, the experience at Microsoft suggests that top managers love the prediction markets, it is the middle managers that feel threatened. Prediction markets allow for a much flatter organizational structure, but not a perfectly flat one.

"Prediction markets make a big chunk of the bettors into 'losers.' "

Why? Are we talking about a decision market, or a betting pool? If a top manager wants a check on his intuition and/or research on products or markets the organization serves, why shouldn't he pay for it? Those who want answers from the markets should set themselves up to be the "losers". They are not losers though, as they gain information that might otherwise be lost going from the line level up the chain through middle managers. Markets are win-win, make more pie. Betting pools are zero sum. If a properly set up and run prediction market had an index fund, the index fund should grow over time. This isn't to say there won't be 'losers', there will be some participants who will trade on imperfect information and/or who act irrationally.

"Most employees have no rational basis on which to bet."

In most, if not all, organizations the line level employee knows the products, skills, and capabilities of their team better than the management. That should be more than sufficient for rational trading in a decision market.

"If someone knows the truth, but is otherwise locked out from credibly signaling that knowledge to management, something is wrong with the organization of the company."

All centrally planned organizations suffer from knowledge and incentive problems. That we are talking about business firms means the quantity of the problem is much smaller than a socialized nation. There is not some fundamental difference in the essence of the problems. The problems still exist, prediction markets are a means of further reducing those problems.

Like any market, different people should be coming to a decision market with different desires and resources wishing to trade their resources for the things they desire and thereby improve their own wealth. In the case of decision markets, the underlying value being exchanged for money is accurate information.

Share this

It may be that prediction

It may be that prediction markets have a built in bias that makes their results unreliable when the market participants already have a financial interest in the events concerned. If an event would already be undesirable for someone, no matter how they view the likelihood, they have an incentive to bet that it will happen. The prediction market becomes an insurance market. This inflates the estimated likelihood of any events that bettors would prefer to avoid.

I was unaware that Microsoft

I was unaware that Microsoft used predictive markets internally - is there a source that describes their usage? A quick google did nothing more than return their troubles with the EU.

James, I believe that

James, I believe that scenario has been looked at, and the effect was negligible.

Chris check out this bit from bayesianinvestor

David, The effect probably


The effect probably is negligible in many cases. I would be adversely affected if the US economy were to tank, but I don't maintain a short position in the stock index futures market as insurance. On the other hand, I do buy auto insurance. What gives? Do I believe a car accident is more likely than an economic downturn? No. I'm just more likely to insure against events that would constitute a large adverse effect on me. The larger the adverse effect, the more I'll want to wager that it will happen. I suspect this is true of most people. In corporate prediction markets, many (not all) of the events would be so relevant to the participants, that the predictions would have bleakness bias that couldn't be easily remedied. That might leave many other events where prediction markets would work just fine, but I suspect the events that firms would most like to forecease are those which are negative and which would be harmed by this source of bias.

James, Perhaps this effect


Perhaps this effect counters the normal positive bias in healthy companies? The data so far (which is admittedly not at all significant) suggests that there is no effect.

In existing derivatives markets (where most hedging occurs) we see that those who use the markets to hedge must pay a premium. The prices in existing derivatives markets are not lower than would be expected - i.e. they reflect reality rather than the hedger's biases.