Thursday, October 05, 2006

A bettor way to forecast 

The National Association of Business Economists has a new means of generating forecasts. By using an auction mechanism, they are estimating private employment to get a better feel for the dispersion of opinion around a forecast.
The experiment involved a type of �pari-mutuel� betting process in which participants spend a limited budget of �house money� to purchase and place tickets on a range of possible outcomes�in this case, job gains in September 2006. In describing the experiment, former NABE board member and president of Macroeconomic Advisers Chris Varvares said that the �overall objective is to get a consensus view from the NABE Outlook panel of the entire probability distribution of possible outcomes rather than just an average of forecast modes.� The chart on the next page shows the distribution of �bets� generated by today�s exercise.

�We believe that this experimental approach to forecasting is potentially a useful and valuable addition to the NABE Outlook Survey, a quarterly consensus macroeconomic forecast of 50 NABE members,� said Ellen Hughes-Cromwick, NABE vice president and chief economist at Ford Motor Company.

Participants in today�s experiment revealed that they believe that a private payroll gain in September of 100,000 to 124,000 is the most likely outcome, with roughly a 25% probability. The average expected gain was also within this range at 121,000. This compared with an average gain so far this year of 127,000, and an average increase over the last four months of just 102,000. The distribution of expected outcomes implies that the participants believe there is roughly a 75% chance that the employment change in September will be between 75,000 and 174,000. By way of contrast, consensus estimates available through Bloomberg project total job gains in the range of 90,000 to 125,000. Over the past year, governments have added an average of 11,000 jobs per month. Subtracting this from the Bloomberg consensus range would put it at roughly 80,000 to 115,000 for private payroll employment gains.
Winners get real cash, which means NABE will have to find some money to fund this further. Such markets have occurred in many other places (Robin Hanson at George Mason has a good description of these), and in other places like elections these have been pretty good predictors.


Filed in: