Economic statistics often undergo changes to underlying calculation methods, data collection, and other adjustments over time. Even today’s widely used GDP is a shift from the previous standard measure, GNP. Arguably, these changes are designed to improve the metrics’ quality and usefulness. However, some believe such changes either unwittingly or intentionally affect these statistics’ accuracy—one such example is the Consumer Price Index (CPI).
In CPI’s case, most critics feel methodology changes artificially reduce the level of the index, thus understating inflation. But does the new calculation method actually understate inflation or would the old method overstate it? In large part, the correct answer is largely contingent on what one seeks to measure with inflation.
CPI has shifted over time from less of a pure price index to more of a cost of living index—no longer measuring so much the price change for a set basket of goods than the cost of a changing basket of goods designed to reflect maintenance of a constant living standard. If strictly trying to measure price change, then current CPI may understate inflation. If trying to measure how changing prices impact consumers, then the old measure may overstate inflation.
The major changes often cited as distorting CPI were made in the early 1980s and mid-1990s. In the early 1980s, housing prices were replaced with owners’ equivalent rent, a measure designed to approximate what a homeowner would pay for equivalent housing in rent. Rather than including the direct purchase price of houses (a very rare expenditure), the index included an estimate of ongoing housing costs. Thus, if home prices were inflated by factors making homeownership easier (like lower mortgage rates), looking at a measure of the day-to-day cost of a home rather than the raw price may better reflect real housing costs. Meanwhile, a person buying a house with a fixed-rate mortgage could live there forever without the monthly financing cost ever going up. (Even if home prices rose 500% and the homeowner changed homes, the first home’s appreciation would roughly offset the increased cost of a similar second home.) While owners’ equivalent rent may seem convoluted, it attempts to provide a more realistic day-to-day measure of housing costs for homeowners.
Then, in the mid-1990s, the Senate Finance Committee appointed the Boskin Commission to study CPI. The study led to various methodological changes designed to make the CPI better reflect the cost of actual consumption patterns. This included increasing the frequency of updates for the basket of goods underlying CPI (the weights used in calculating CPI) to account for substitution and new items available to consumers (like computers and cell phones).
Some critics of today’s CPI measure even attempt to undo these changes to track what they believe undistorted inflation really is. Using this methodology, some skeptics suggest CPI inflation really averaged somewhere around 8%-10% annually throughout the 2000s, as opposed to the roughly 2.5% officially reported. If these skeptics were right, then on average, a consumer’s day-to-day expenses should have more than doubled over the past decade (assuming no change in the volume of consumption). While some prices, like gasoline, have doubled in the last decade, prices for many other commonly consumed goods have not, including some major purchase items. For example, the Apple iBook laptop was a US bestseller in 2000 and cost about $1,500.[i] Today, a 13” Macbook Air, which is more powerful by multiple orders of magnitude, has more memory, a much wider range of functionality, and is much more portable, costs only $1,299. Clearly, even at the same price, one would be getting more for their money, yet the price of this major purchase item has actually fallen significantly!
Other CPI critics suggest quantifying the impact of long-term product quality changes is impossible. Thus, they argue using raw commodity prices (largely the same goods today as decades ago) to measure inflation is more accurate. At first blush, this may seem logically appealing. But it ignores productivity gains in manufacturing final goods and the impact of substitution effects.
For example, in early 1999, Maytag introduced its ultra-high end Neptune front-loading washing machine for $1,099[ii]. Today, the MSRP of their highest-end washer is $1,399. Ignoring likely improvements to the product’s functionality, the increased price represents a mere 2.0% inflation compounded over the last 12 years. But the costs of major commodities used to build a washer (like steel and copper wire) certainly rose faster than CPI—and well above the finished washer’s price change. More efficient production and substitutions, like using plastic where steel wasn’t really needed (or where plastic would be better suited due to rust concerns), could easily account for keeping the price lower. Pure commodity prices would suggest a much higher price for the washer.
Overall, essentially all government and economic statistics are flawed somehow, whether it’s due to data gathering, measurement or adjustment, or how the calculations may have been manipulated over time. While it’s possible politics motivated some CPI changes (with economic thoughts secondary), that’s mostly irrelevant to analysis. Many economists and analysts have their preferred ways of measuring key metrics that differ from headline reports and can adjust data in a wide range of ways to suit their needs. And there’s no perfectly precise way to measure a subject as broad as price changes across an entire economy. However flawed headline government statistics might be, they do provide a common reference for discussion and a baseline on which to layer various views and analyses. Thus, regardless of whether the current CPI calculation overstates or understates inflation, markets are typically more concerned with how the reported number relates to expectations than with some theoretically more accurate measure.
[i]Source: Mac Observer, January 7, 2000.
[ii]Source: The New York Times, March 22, 1999.