Sunday, January 11, 2015

Survey disagreement is not the same as uncertainty




Often analysts will associate differences of opinion with market uncertainty. I have made this shorthand mistake even when I know better. Saying there is a dispersion of opinions is not the same as saying that market views are uncertain. Forecast uncertainty is the measure of dispersion surrounding the mean opinion of any forecaster.

Opinions can be very different but at the same time very confident and certain. We often refer to consensus as a lack of differences in opinion of survey participants but that does not tell us anything about the uncertainty of those opinions.  See Zarnowitz. What research has found is that the standard deviation of point estimates may underestimate the true uncertainty with some forecast. 

Recent research by the NY Fed pushes this issue even further. It finds that there is not a strong link between the uncertainty and disagreement in forecasts as measured by inflation forecasts. We have seen EU inflation disagreement go down but uncertainty go up since 2010. There was an opposite effect in 2003-2007. Data in surveys are complex and nuanced and requires deeper analysis than we often employ. 




So what doe this mean for the macro investors? Uncertainty is more subtle than just looking for dispersion of opinions. When there there is more uncertainty, less emphasis on survey consensus make sense. Yet, we may not have the tools to measure this uncertainty since there are limited surveys. 

Market data, on the other hand, may give us a better view on market uncertainty. It could be something like the VIXX index or opinions based on option prices, but market votes through actually buying and selling in markets may provide an effective guide for market uncertainty. The crowd information in market prices especially at extremes is useful. 

No comments: