The future inflation rate is one of today’s great investing imponderables. It’s never been as important as it is now to forecast it accurately, and opinion has never been so divided. At one extreme sit those who believe that huge fiscal deficits and the Federal Reserve’s reckless monetization of debt will lead to currency debasement and rampant inflation. The bull market in gold is one manifestation of such fears. At the other extreme are those who note persistently high unemployment as evidence that ample spare capacity exists in the economy. Today’s very low bond yields suggest that inflation fears are not dominating investors’ decisions.
While future inflation will affect investors’ returns, less attention is paid to how inflation is calculated. Correct measurement is of no small concern to just about everybody. Savers ultimately need returns net of inflation, and many commercial contracts as well as substantial payouts from Federal and state governments (including Social Security) are indexed to inflation. The Bureau of Labor Statistics (BLS) calculates multiple inflation measures to reflect different spending patterns and to smooth out volatile items (such as food and energy). The most widely used measure is the CPI-U, U.S. City Average, All Items (CPI for short). Over the past twelve months it has risen at 1.2%. In 2009 it was -0.4%
Individuals routinely complain that prices are rising faster than the BLS calculates. Education (3.7%), Medical Care (3.4%) and Fuel (13.4%) are all causes of this opinion. The statisticians have an answer, which is that frequent, small purchases (school books, prescription medicine, gas) affect perceptions disproportionately compared to infrequent, large ones (new cars, housing, computers).
Few people outside government statisticians and obsessive fixed income investors spend the time to analyze the construction of the CPI. It’s built by statisticians and while the subject of criticism the BLS maintains that it does its job. However, the BLS notes that it’s not a cost of living index and it excludes most taxes which obviously have an impact. But most crucially, the CPI measures a basket of goods and services of constant utility (a theoretical concept related to how much use an individual gets from each purchase). Measuring the cost of maintaining a constant standard of living would be much more useful, but the CPI isn’t designed to do that. Because of this and other quirks, the CPI does what statisticians want but fails to fully measure what’s important to most people. One of the most bizarre components of CPI is Owners’ Equivalent Rent (OER), which is intended to capture the cost of shelter. Rather than measure actual rents, or the costs of owning a home (mortgage, taxes, maintenance), OER is derived from a homeowners survey that asks what rental income owners think they could earn on their home. This is currently 24% of the entire index. How many people stay current on what they could rent their home for? And given the high incidence of home ownership in the U.S., how relevant is OER to the actual cost of owning a home experienced by most people? The statisticians respond that houses are an asset, like stocks and bonds, while they’re trying to measure the cost of shelter, a service. But surely a more meaningful statistic would be to survey data on the direct costs of housing which would include (for owners) mortgage rates, taxes and maintenance.
Another feature that only a statistician could love is the BLS’s treatment of quality improvements. Companies improve their products all the time, but in CPI-land, if a product improves in quality with no price increase, that is the same as a price decrease. This works for a box of breakfast cereal increased in size from 12 oz to 15 oz sold at the same price. But how should this be applied to computers, whose speeds and capacity are increasing regularly? If a laptop cost $500 last year and this year’s current version with Microsoft’s latest operating system costs $500, the BLS will register a price decrease because of the quality improvement. This is known as a “hedonic quality adjustment”, and uses statistic modeling to estimate the value of the quality improvement. However, the laptop hasn’t really fallen in price (and even if you could buy last year’s model for $400, who would?). The BLS’s aim for constant utility turns unchanged prices into deflation and in the case of computers drives down the overall CPI. The same applies to cars which now carry airbags. Since an airbag adds utility, the price of a car with an airbag is adjusted down so as to compare with the old price of a car without an airbag. But you can’t buy a new car without an airbag, just as you don’t want to buy an old computer. The result is that the inflation statistics are not presenting information that reflects how most people experience price changes. While it works for the BLS, this may explain why many individuals believe they experience higher inflation. It’s the difference between constant utility and a constant standard of living. If the CPI wasn’t so widely used as a price index its quirky construction would relegate it to obscurity.
Conspiracy theorists have speculated that the government deliberately understates inflation to reduce its expenditure since substantial transfer payments are indexed to the CPI. However, that would presume implausible statistical sophistication on the part of our elected leaders. Nobody watching the legislative process of the past few years could seriously believe this same group is secretly poring over subtleties in how inflation is calculated.