The following was first published in the November 2004 Edition of the SGS Newsletter

The markets generally follow two measures of consumer confidence the Consumer Confidence Index as published monthly by the Conference Board and the Consumer Sentiment Index as published twice a month by the University of Michigan’s Institute for Social Research. Where the Confidence and Sentiment series date back to the 1960s and 1950s, ABC News publishes a weekly Consumer Comfort Index, which dates from the 1990s and is not widely followed. A history that covers multiple business cycles is useful in establishing the relationship of a given index to other economic series.

While the two major indices have significant quality issues, many of the problems can be overcome by viewing the data in terms of the year-to-year change in a three-month moving average. On that basis, the series are useful as lagging indicators, signaling where the economy has been in recent months, not as the leading indicators hyped in the financial media. Lagging indicators are useful, particularly considering that the economic consensus can be three to four quarters or more behind recognizing shifts in economic activity.

A major problem facing both indices is that consumers are, in effect, asked to be economists, to assess how the economy will be performing at some time in the future. Since most consumers are not economists, they tend to parrot what they hear and read in the financial media. In the 1990s, Geoffrey Chow of the University of Minnesota published the Media Climate Index, which measured the relative positive or negative tone of newspaper stories on the economy, as published around the country. His index served as a highly correlated leading indicator of the Consumer Confidence and Sentiment Indices.

The Conference Board’s Consumer Confidence purportedly is based on a survey of 5,000 consumers per month. While 5,000 post cards are mailed out, reportedly to paid participants, typically about 3,500 cards are returned. What is extraordinarily questionable about the series is that it is "seasonally adjusted," but the unadjusted series is never made available. The Conference Board’s purpose is to promote business, and it is not unusual to see regular reporting of heavily positive Consumer Confidence numbers at the onset of the holiday shopping season.

Al Sindlinger, who published his own consumer confidence measure for nearly fifty years–until his death in 2000–ran the initial survey for the Conference Board. Al always chuckled when he described the early efforts. It seems that when he reported positive results to the Conference Board, he’d hear the details an hour later on the radio. When the results were negative, he’d always get a call asking him to recheck his results.

The issues with the University of Michigan’s Consumer Sentiment Survey include sampling problems. The telephone survey of 500 consumers each month is too small to have tight statistical significance. Even worse, the early-month results usually are based on something less than half that amount, and are virtually worthless in terms of having meaningful statistical significance. Sampling problems are compounded by the increasing use of voice mail and cell phones, where certain consumers never are surveyed.

Beyond suspicions that reporting of Consumer Sentiment occasionally has been massaged as a favor to Alan Greenspan, there also is the questionable use of the series as a market-trading tool. Subscribers pay heavily to get the first release of the data. Once they’ve taken their market positions, they "leak" the data to the press, and the markets respond. The University of Michigan does not release the data to the public until well after the insiders work their trading games.Consumer Confidence Measures Financial  Media Sentiment