Maggie Humphrey, a price collector for the Bureau of Labor Statistics, visits the same grocery store every month in the Chicago suburbs to punch the cost of a pound of bananas into her Lenovo tablet computer.
“That price has not fluctuated since I’ve been here,” says Humphrey, who started gathering prices for the BLS in 2006 and has checked bananas at this particular establishment for about a year. She records it as 69 cents a pound and includes their country of origin, whether they’re on sale and any applicable sales tax.
Humphrey is among 400 price collectors who visit 23,000 locations in 87 cities every month to determine the cost of 80,000 products and services, from breakfast cereal to haircuts. She and her colleagues feed a database in Washington, where statisticians compile the monthly inflation report, used as benchmark for everything from Social Security payments to the value of Treasury’s inflation-indexed bonds.
The bureau’s price-gathering and statistical methods are standard practice from Japan to Switzerland. That hasn’t averted a lashing from critics who say the government is engaged in a campaign to hide inflation of 10 percent a year or more. Assurances by Federal Reserve policy makers that inflation remains “subdued” also haven’t deterred the skeptics.
“I’m as hawkish and worried about inflation as anybody,” said Stephen Stanley, chief economist at Pierpont Securities LLC in Stamford, Connecticut and one of the top forecasters of CPI over the last two years in Bloomberg News surveys. “But the idea that inflation is 10 percent is not a proper reading of the data.”
One such critic is John Williams, the author of Shadow Government Statistics, a newsletter that he has run since 2004. Williams says the federal government understates the level of inflation to keep increases in Social Security payments and other costs down.
“The reporting system increasingly succumbed to pressures from miscreant politicians, who were and are intent upon stealing income from Social Security recipients, without ever taking the issue of reduced entitlement payments before the public or Congress for approval,” Williams says on his website, shadowstats.com.
Williams’s alternate measure of inflation was 10.3 percent for the 12 months through March, compared with 2.7 percent for the Consumer Price Index. He calculates unemployment at more than 20 percent rather than the official 8.2 percent in March. His assessment of gross domestic product has clocked negative economic performance in every quarter since 2005. The Department of Commerce’s measure turned negative in 2008 and 2009, recording the worst recession since the Great Depression. The economy is nearing “hyperinflationary Great Depression,” he says on his web site.
Williams takes issue with statistical methodology adopted by BLS since the early 1980s.
The first is “substitutability,” which accounts for people buying cheaper goods as prices increase. If the cost of two types of chicken breast rises, the BLS assumes consumers buy more of the less expensive one, giving it somewhat greater weight in the index.
The second is “quality adjustment,” which seeks to measure how goods change over time. For example, as ever-more-powerful computers are available for the same price, the BLS records this as a type of deflation. As the amount of fabric in clothing shrinks (think skinny jeans) that is noted as a type of inflation.
The third is “owner’s equivalent rent” which replaces the cost of owning a home with what it would cost to rent it.
Had the BLS not altered its statistical practices over the years, Williams says, inflation would be reported about 7 percentage points higher each year, enough that the price level will quadruple in a little over a decade.
Williams sells subscriptions to his website for $175 a year. He declined to give the number of subscribers. “My business has been picking up over the years,” he said in a phone interview. “I don’t know I can attribute it to any one thing other than to say that most people do think that inflation is higher than the government is reporting and economic growth is weaker.”
Williams graduated from Dartmouth College in Hanover, New Hampshire in 1971 with a bachelor’s degree in economics and subsequently earned a master’s in business administration from Dartmouth. The background on his website says he worked as a consulting economist.
Some economists favor alternative inflation measures. The Federal Reserve aims for a 2 percent annual rate of inflation based on the Commerce Department’s personal consumption expenditures index, which averages about 0.5 percentage point a year less than the CPI. The Social Security Administration uses an index known as CPI-W, which rose 2.9 percent in March from a year earlier, compared with the main CPI’s 2.9 percent increase.
“These various measures, the CPI and others, PCE index and others, move very closely together,” Fed Chairman Ben S. Bernanke said at a Jan. 25 press conference in Washington. “And you are not going to have a situation where the CPI is 10 percent and the PCE is 2 percent. There may be a few tenths difference, but generally speaking they move very closely together.”
The BLS has responded to Williams’s claims in papers over the years. One such report in 2008 noted the degree of international acceptance of BLS methods, with 13 of the then-30 members of the Organization for Economic Co-Operation and Development using rental equivalence and 12 using the quality adjustment known as hedonics. Eurostat reports that 20 use the CPI’s method of substitution.
“There are places where you could quibble with what they’ve done to change the CPI,” said Pierpont’s Stanley. “But to say that everything that’s been done over the last 30 years to the CPI is illegitimate and what we should really do is look at the methodology from 30 years ago? Would we want to use the state of the art from 30 years ago for computers?”
The BLS provided details on its methodology because the “myths” about the CPI construction methods “continue to circulate,” BLS economists John Greenlees and Robert McClelland said in an August 2008 rebuttal.
The index has its roots in a World War I-era effort to adjust wages to rising prices. The first cost-of-living measures, published in 1919, concluded that not everyone could afford “all the necessaries, many of the comforts, and a goodly supply of the luxuries of life.”
In 1995, Congress created a commission to study the index, headed by Michael Boskin, a Stanford University economics professor and former head of the White House Council of Economic Advisers. Its review concluded that the index was overstating the inflation rate by 1.1 percentage points a year.
Congress authorized the commission during a period when members looked to cut government spending. The panel made 16 recommendations for changes to the index, estimating that they would shave $691 billion off the debt over the next 10 years.
Robert Gordon, an economist at Northwestern University and one of the Boskin commission members, estimated in a 2006 paper that the BLS changes removed only some of the bias, and the index remained about 0.8 percentage points too high per year, a sharp contrast to Williams’s claim that the level is 7 percentage points too low.
The CPI’s methodology drew attention again last summer when a group of senators looking to reach a deal to cut the deficit considered switching government indexing to a CPI measure that, because of a methodology change, would lower reported inflation by 0.3 percentage point per year. Ultimately it wasn’t adopted.
“At BLS, we’re just dealing with the data, not the policies behind the data,” said Gary Steinberg, a press officer and more than 20-year veteran of the bureau. “We don’t have an opinion one way or the other on how the data are used.”
Williams’s criticism of the CPI focuses on the effect of substitution and hedonics.
“I believe Williams is right directionally,” said James Bianco, president of Bianco Research LLC in Chicago. “The measures as they were constructed 30 years ago would show higher inflation if we were using them today.”
Bianco said that the U.S. government does have an incentive to favor lower reported inflation because then it saves money on cost of living adjustments, union contracts and inflation-adjusted bonds that are benchmarked to the index.
“The current measures might only be a few percentage points different but even a few percentage points is still pretty significant,” Bianco said. “Lots of money rides on these numbers.”
The BLS doesn’t dispute that its adjustment for substitution tends to lower the index. Yet the size of the BLS changes is nowhere near the 7 percentage point difference that Williams claims, BLS officials said.
On average, the hedonic quality adjustment has increased, not decreased, the reported rate of inflation, the BLS says.
Ken Stewart, a BLS economist, compared the old and new methodologies over a 21-year period and has said the current procedure produces an annual rate of inflation about 0.45 percentage points lower.
“Hedonic adjustment makes sense to me,” said Michael Pond, an expert in inflation-indexed bonds at Barclays Plc in New York. “If I’m paying the same price for a good that is better, then I’m getting a better deal, getting more value for my money.”
From 1977 to 1998, the CPI showed a 141 percent increase in the level of prices. A product or service costing $10 in 1977 ought to have cost $14.10 more in 1998. The old methods would have produced a 163.9 percent increase, according to a paper Stewart co-wrote with BLS economist Stephen B. Reed.
Williams draws his estimate of a 7 percentage-point bias per year in part from this paper, he said in a phone interview.
“It’s a simple approach to it,” he said. A complete reconstruction would be “the type of thing that takes extraordinary time and computing ability, which I don’t have the resources to support.”
Independently of Williams, two economists at the Massachusetts Institute of Technology developed their own methodology.
Roberto Rigobon and Alberto Cavallo launched the Billion Prices Project in 2010, which builds on analysis that found Argentina understated its inflation rate. In the U.S. alone the project tracks over 5 million prices, Rigobon said, on pace for a global goal of 1 billion.
Basket of Goods
Rigobon said the project includes about 60 percent of the goods used in the CPI. The study reported inflation about 1 percentage point higher than CPI for much of 2010 and slightly more than 3 percent inflation in 2011, in line with the CPI.
“In a three-to-four month window, the inflation rate we report is almost identical to the CPI,” Rigobon said. That suggests that the CPI measure is accurate, he said.
For her part, price collector Humphrey in Chicago says she never can guess what the total CPI will end up being even after logging hundreds of prices month after month.
“It’s hard to have an intuition for the overall number,” she said. “You see a lot of prices, but it’s such a small part of the nation as a whole.”