Let's try prediction markets for contract cost and schedule goals
A group of researchers -- my friend Matt Potoski at Iowa State (one of the brightest young public administration scholars working on contracting issues), along with Ricardo Valeri and Taroon Aggarwal of MIT -- are launching a pilot research project to examine whether prediction markets can help improve cost and schedule forecasts in federal acquisition programs. It is
one of several such projects that Nancy Spruill in the Office of the Secretary of Defense is funding at the Naval Postgraduate School in Monterey.
In a prediction market, buyers and sellers trade contracts.The payoff depends on future events, such as the cost or delivery date of a product. The results of prediction markets can be informative because market traders have incentives to discover and reveal information through their trades, and the market weighs traders’ beliefs based on how they spend money on contracts in their market.
Pilot efforts in large companies using this approach to predict completion times for new product development and for IT projects have generally produced more accurate forecasts than those using the more traditional method of having project leaders at the top of the hierarchy develop the projections. These better predictions have highlighted potential problems of which management may not be aware, and thus allow corrective action earlier.
The aim of the research is to initiate several prediction markets in federal acquisition programs. The researchers will compare the prediction market forecasts with traditional estimation techniques, such as parametric, analogy, activity-based approaches, to see if prediction market forecasts are more accurate, and to see how the prediction markets responds to new hard and soft information.
Ideal candidates for the study would be acquisitions programs with delivery dates in the spring to early summer of 2010 and with a sufficiently broad following to attract engaged traders from inside and outside the government. The research team will supply easy-to-use prediction market software, technical assistance, and will manage the market. If you are interested in participating in this study and have a program that will be approaching a milestone in the next 3-8 months, or have questions about confidentiality, participation terms, etc., please contact the research team at firstname.lastname@example.org.
This is an example of the kind of innovative approaches we should be examining in order to improve the performance of the procurement system. Dan Mintz, former CIO of the Transportation Department, has publicized this on his Facebook page -- and maybe Mary Davie can post this in her GovLoop acquisition forum. Let's try to find some volunteers.
Posted by Steve Kelman on Oct 22, 2009 at 12:08 PM