How much do you spend a week on alcohol? Whatever it is, you're probably underestimating. By a lot. And that’s a problem for the statisticians who compile the nation's main inflation gauge.
Despite the flurry of attention paid earlier this month to the surprise drop in unemployment, there’s another indicator that some economists say could be improved: the closely watched Consumer Price Index, a key measure of inflation. Businesses and the Federal Reserve Bank follow it closely, and economists use it to analyze the effects of policies on price trends.
But part of the CPI is due for a reboot, experts say. And a panel of statisticians and social scientists is meeting throughout the day on Tuesday to share the conclusions of a two-year review of how to do that. The problem lies with a key input in calculating the CPI—interview responses on how Americans spend their money.
“People are being asked to do a lot, and recalling back everything that you’ve spent in a great deal of detail over three months is a difficult task,” said Carol House, who works for the National Academies' Committee on National Statistics and serves as the study director for the panel on redesigning the way the surveys are conducted.
The review comes as the Labor Department on Tuesday released its latest CPI reading, showing the index jumped 0.6 percent in September, driven by higher gasoline costs. But the "core" CPI, which excludes volatile food and energy costs, rose a tame 0.1 percent for the third month in a row.
The CPI is maintained by Labor's Bureau of Labor Statistics, which compiles a number of economic indicators, including the all-important monthly employment report. Two things generally go into calculating the indicator: the changing costs of more than 200 items, and the share of spending that Americans devote to those items, as reported through BLS’s Consumer Expenditure Survey.
The survey responses dictate how much weight to assign each item within the inflation index. The cost of dried beans, peas, and lentils could jump up, as it did from August 2011 to August 2012, but if no one’s buying them the impact on inflation would be negligible. The problem is that, with the increased use of credit cards and online shopping, people just aren’t that good at remembering how much they spend on what and where.
“Basically, if you ask people how they spend their money, they can’t tell you,” said Barry Bosworth, a Brookings Institution expert on fiscal and monetary policy. “And this problem has been getting progressively worse over time.”
People have a tendency to underreport how much they spend on some items, according to an August study in which researchers compared the 2010 Consumer Expenditures Survey responses to a similar, but broader, measure of spending--the Bureau for Economic Analysis’ Personal Consumption Expenditures.
Rent and spending on cable and satellite TV services tracked closely between the two measures, they found. But gambling reported in the CPI surveys came in at a mere one-tenth of what was reported in the other measure, signaling that consumers greatly underestimated their expenditures in that category. Alcohol bought in stores was one-fourth. Neither measure perfectly tracks spending, but the discrepancy underscores the challenge of asking Americans to report things they may not accurately remember.
But it’s not just the accuracy of the survey reports that’s a concern. Some argue that they don’t reflect variety among different groups of people. The standard CPI measure is accurate for the nation, but it’s a “one-size fits nobody” indicator, said Keith Hall, who in January completed a four-year term as the head of the Bureau of Labor Statistics.
Seniors, especially, feel the effects of that, said Ross Eisenbrey, a labor- and employment-law expert and a vice president of the Economic Policy Institute. The CPI is used to calculate cost-of-living adjustments to Social Security benefits, but it fails to account for the fact that a big portion of senior spending goes to health care, he said.
“Anyone who has the goal of having a better measure—a more technically accurate measure—of what the effects of inflation are on the Social Security population would want to much more heavily weigh health than the CPI-U does,” he said.
Identifying how the CPI is used and how BLS would like it to be used is key, the review panel found. In an advance report released earlier this month, they recommended a “major redesign” of the Consumer Expenditure surveys, but not until BLS has prioritized “the many uses of the CE data.”
BLS has “to think very carefully about which priority needs are most important to be met,” said Don Dillman, a Washington State University sociology professor, a survey methodology expert, and the chairman of the panel. “It’s no small set of issues to think about making changes to it right now.”