Wednesday, February 15, 2012

Dispersion in Confidence as a separate indicator

The economic data of the past week were not decisive in any way. It was a week of mostly minor reports compared to the week before that put so much at stake and showed such unexpected clarity about the economic condition in the US. Still the U of M Consumer sentiment index for Feb (prelim) did move lower in both its outlook and current components last week. But the weekly Bloomberg Consumer comfort survey continues to press to higher readings. The U of M set back simply took some of the punch away from the previous month’s large gain. The Feb sentiment reading is still the highest U of M sentiment reading since May 2011, excepting of course the January level from which it just dropped. But, obviously, the gain on the books is smaller now with this month’s drop in the index to 72.5 from 75.0. The weekly reading on comfort suggests that we may be seeing some uncertainty about the pace of the improvement of consumer attitudes more than a new unexpected set back.

The U of M report also tells us that by region some large gaps have developed in sentiment. The Midwest showed a pop in sentiment in February and finds sentiment is still low in the 29th percentile of its historic queue (weaker than this only 29% of the time). And that is similar to the North East which stands in the 26th percentile. But the South and the West stand in their 11th percentile and 13th percentile, respectively.  This is a new development and is worth watching. The Midwest has tended toward the strongest relative readings and the West and South have tended toward the weakest as the expansion has progressed. Still, this month’s gap is extreme. Interestingly if we go back to the 2007 rankings the West and South were typically as much stronger than the Midwest at that time as much as they are laggards now.  Regional disparities are not new but this one is interesting as two parts of the country do seem to be advancing faster than two others.  

There is one other aspect of consumer sentiment that is very interesting. That is that in recessions all age cohorts seem to feel the same relative pain but in expansion periods the confidence as experienced by the three age cohorts in the U of M survey varies quite widely as a rule. Used as a diagnostic, this economic regularity suggests that the recovery period is truly transitioning into something more normal. (See graph below)

The chart above clearly identifies by its low readings periods of recession. And used as a diagnostic the increased variability among age cohorts as this recovery ages is suggesting that the economy is getting back toward normal. We are not there yet but we are making strides.

Let me describe this measure in more detail. The U of M sentiment report offers consumer sentiment readings for three age cohorts 18-34, 35-54 and 55+. Call them A, B and C. we take data back to 1978 and rank each cohort as a percentile standing in its rank expressed in February as the 26th percentile for 18-34, the 22.8th percentile for 35-54 and the 24.1 percentile for 55+ cohort.  To create our measure of confidence disparity/similarity we calculate the three squares of the ranking differences (a-b)^2, (b-c)^2, and (a-c)^2 and sum them. The chart plots the four-month moving average of this measure over time. The chart is not about the rankings being high or low but about the being similar or dissimilar.

In non-recessions and as recoveries transit to a period of normalcy, various age groups experience different relative confidence levels. But in recessions and in early recoveries or poor recoveries, age groups tend to see their confidence levels relatively the same- that means as poor. Because of this regularity this measure of confidence dispersion can serve as an indicator of economic normalcy. Right now the measure seems to imply that the economy is getting to a more normal period of activity as age cohorts are beginning to experience different relative levels of confidence.