Thursday, January 2, 2014

Consumer Confidence Index: What does it mean?


The  headline is:  "Consumer Confidence Blows Past Expectations."

Finance.Yahoo.com reported that the University of Michigan's Consumer Confidence Index "surged" to 78.1 in December (but still below its level before the 2008 financial meltdown).

Is this good news?

Well, of course it's good news—"consumer sentiment" got better.

But what does it mean, y'know, economically, consumer behavior-wise, what "surged? Who knows….

First, keep in mind that the Consumer Confidence Index  is based on survey data, so, technically, we really don't know what it means because no survey produces valid, precisely accurate data. Every survey produces some kind of data, and repeated surveys (done by professionals) can nominally illuminate trend changes, but I don't give surveys much more credit than that.

When is the last time you sat still to answer a telephone interviewer's questions? Besides the decline of public willingness to participate, all publicly reported surveys are cooked by arbitrary respondent selection/screening procedures, sloppy question design (both wording and sequence) and “statistical weighting” of the data, which is the wonk way of correcting for all the other failures to reach a “true random sample.”

By the way, a "true random sample" means, technically, “every member of the population which is to be represented by the survey results must have an equal opportunity to be explicitly questioned and must actually participate if asked.” No public survey achieves this basic requirement, so we don't really know whether most survey data really reflects the opinions and behavior of the people who are supposed to be represented.


And here's another thing: statisticians have confirmed that the Consumer Confidence Index in the last 35 years has had only a 4.5% correlation with the government's monthly index of Personal Consumption Expenditures—the PCE measures actual consumer spending, and you'd guess that more consumer confidence  would match up pretty well with more consumer spending. Well, it doesn't.



A correlation of 100% would mean that increases or decreases in confidence are consistently, proportionally and directionally related to increases or decreases in spending—if one goes up, the other goes up, and vice versa. A correlation of  zero would mean there is absolutely no relationship between changes in one index and changes in the other one. A correlation of 4.5 is pretty close to zero….

So, changes in the Consumer Confidence Index do not indicate that similar changes are occurring in the Personal Consumption index. So, there is no economically useful interpretation of what those changes in consumer confidence really mean.

What does the December CCI report really tell us?  Something "surged." What was it? Your guess is as good as mine.








No comments:

Post a Comment