Friday, March 11, 2011

Do prediction markets work?

There are several advantages to prediction markets:

  • They introduce common sense to such basic management follies as beta testing taking more time than originally planned.
  • They can be used to get a general sense of any project that is devoid of hard management considerations such as financial viability, expenses etc. In other words, whether the project has enough firepower to create buzz. This can be a potentially rewarding input for new technology companies, smartphone apps, etc.
  • They can be used to check one’s intuition about a project: it can do wonders during those gruelling decision-making times.
  • Prediction markets allow top management to keep the ear to the ground and not get carried away with their management-speak.
  • Conditional prediction markets can be used to predict likelihood of events that are contingent on some other events. Often for financial managers, the choice is between two projects that look equally attractive from a cash flow perspective. Say, an FMCG company X wants to get into the biscuits space with choco-sandwich Y to take on entrenched rivals, or alternatively, expand its fast-growing potato chips business Z. Both projects look attractive from a purely financial perspective, and the marketing department has a mountain of data that is unable to help. In such a situation, a conditional prediction market can help with gauging the popularity of Y on purely consumer attractiveness measures contingent upon X taking up Y and giving up on Z. Or, consumers might like new variants of Z and that input will gather more traction on trades than the stock corresponding to launching Y.
  • Prediction markets have major advantages in the political domain. Consider the current unrest in Libya. One question before president Obama is whether or not to exercise the military option. The US administration is chary of a quick decision because of the morass it finds itself facing in both Iraq and Afghanistan. Besides, the issue of body bags (bodies of dead soldiers from frontline areas returning to the US) has gained an especial emotional currency in the US. If the President did not want sophisticated military models to drive his decision and wished to read the pulse of the nation, a national, real-time prediction market can help. How the prediction market can help in such cases is by nuancing the debate with several viable options.

1. Threat of return to recession in the form of an oil shock should conflict in Libya prolong.

2. Strategic rewards from establishing a no-fly zone over Libya

3. Arming the opposition rebels and defending them from aerial attack launched by Gaddafi’s forces

4. Perception of Americans on war, ranging from important (defeating communism in Vietnam, terror post-9/11) versus cautious (messy states with a history of perennial conflict)

Would managers want prediction markets?

At first glance, any manager would be skeptical of prediction markets since leaving out the decision to a diversified marketplace is fraught with danger. There is the chance that a rival manager in another department might strategise to “play” the market. But this argument is not right since it overplays the role of the individual investor (to borrow stock market terminology) in market trades.

There is another risk to managers from prediction markets. What if a prediction market is able to accurately and consistently predict information that companies pay to get from marketing research personnel? This would jeopardize a number of jobs and lead to much bad blood on the ground.

Most managers will likely not want a prediction market to suggest negative outcomes for their products/projects. The risk to management from a prediction market comes in the form of dilution of “information asymmetry”. We know that a prediction market, much like a stock market, endeavours to reduce the information asymmetry that surrounds any product/project. Managers often play upon this asymmetry to sell their projects. A prediction market will substantially reduce the agency cost associated with running an enterprise.

There is another risk that prediction markets pose to middle management. If top management begins using them to listen directly to the employee workforce, then the best ideas, the most exciting innovations will directly reach the top management without the filter of the middle management, and this may not go down well. Such “disruptive information flow” can engender new ways of intra-organisation communication at multiple levels.

However, for obvious reasons, prediction market trades will be but one factor of consideration in the final decision. So the input should be welcomed and not feared.

New technology companies use different ways of boosting innovation, for instance, Google allows employees 20% time to work on something they like and which can later develop into a sellable product. The real learning here is the focus on incremental innovation, as against hankering after a blockbuster product. Apart from the pharma space, where revenues are typically driven by blockbuster drugs, most sectors will do well to tap the productivity of their workforce. Prediction markets can help bridge the current gap in realizing incremental innovation.

How to get employees interested in prediction markets?

Between cash awards or other soft awards, there is another system called the lottery system that makes better sense since it would promote the kind of behavior (participants should bid based on true beliefs, and not financial incentives) that is desirable.

However, cash awards in themselves are not a bad idea. There is the fear, in my opinion unfounded, that if cash awards become the basis of reward, then some employees would bet on an extremely unlikely event and win big if that event actually happened, but incur no financial loss if it did not. But this is how any efficient market, in the sense of the term coined by economist Adam Smith, functions. Traders buy and sell stocks on any number of parameters ranging from personal bias to deliberate strategizing. To expect an intranet to not ultimately follow such behavior is to miss the point of a prediction market.

Another positive of prediction markets is greater cohesion within the company, since they induce people to show their intuitive superiority in a quantifiable way. In other words, it earns them “bragging rights”. In this regard, T-shirts or other soft gifts make little sense since they would not entice someone to actually log on and start trading. They may only act as a hygiene factor for people already interested in prediction markets. However, to really scale up the technology, hard cash/lottery needs to come in.

Prediction markets can only work if there is enough liquidity in the system, in other words, if more members login and join the trades (Metcalfe’s Law). After the initial euphoria of the early traders, hard incentives will need to be brought in to increase liquidity.

(Part of a writeup submitted for a case in the Management of Information Systems course, Term III, MBA)


John Delaney said...

Very solid post and article. Well done.

Would you agree to us posting it on Intrade as an addition to the "primer" articles for those interested in PM's generally and not just public prediction markets like Intrade.

Vikram Johri said...

Thanks, John! Glad to share my post with InTrade.