The consumer-price index and its companion economic indicator, the personal consumption expenditures price index, are in the spotlight as never before. Fed Chair Janet Yellen and her Federal Open Market Committee have made it known that unless the increase in these indexes runs at an annual rate of 2%, the FOMC will stay its hand in hiking the interest-rate target on federal funds.
 
In the 12 months through September, there was no change in the CPI and only a slight increase in the PCE price index (+0.2%) as both were hit hard by the plunge in energy prices. In the same period, the core CPI and core PCE, which exclude the subcomponents that track volatile food and energy prices, have had gains of 1.9% and 1.3%, respectively, as indicated on the charts nearby.
 
Illustration: Lars Leetaru for Barron's
 

The flat performance of the CPI brought unwelcome news last month to the nearly 60 million folks on the rolls of the Social Security Administration. The index is used to determine annual hikes in the SSA’s monthly payouts, and recipients were informed that they will get no cost-of-living adjustment in 2016. The CPI also governs, among other things, escalator clauses in divorce settlements, union contracts, and tax exemptions and brackets on the personal income tax.
 
These indexes are routinely used to convert nominal gains into real, or inflation-adjusted gains.

The PCE price index adjusts nominal consumer spending over time into real consumer spending. Similarly, a high-paying bank certificate of deposit bought five years ago had an annual yield of 2.8%, and the CPI rose 1.8% a year in the same period. Thus, the inflation-adjusted, or real return, was just 1%.
 
THESE INFLUENTIAL indexes constitute a kind of statistical sausage, filled with a blend of exotic quantities that are often laced with informed judgment calls.
 
Just for starters, nearly 25% of the consumer-price index incorporates not actual prices but an estimate of the rent homeowners would pay themselves if they were their own landlords. Other parts reflect estimates adjusted for changes in the quality of the goods being priced.
 
But imperfect as these indexes are, and as much as they might still be in need of improvement, they serve their purpose as very rough approximations of overall price change.
 
 
The Bureau of Labor Statistics became the keeper of the consumer-price index due to a quirk of history.
 
During the inflationary period of World War I, the government saw the need to keep labor peace in the shipbuilding yards, which meant that wages had to stay in line with the rising cost of living. With the help of the celebrated economist Irving Fisher, the BLS conducted surveys of family expenditures in 32 shipbuilding and industrial centers to determine the cost-of-living escalator.
 
The CPI was born, and the BLS has been generating it ever since. The PCE price index, maintained by the Bureau of Economic Analysis, uses most of the same prices generated by the BLS, but applies a somewhat different weighting scheme, along with an altered mix of goods and services.
 
No single individual contributed more to the math of price indexes than Fisher, author of two definitive works on the subject. In his 1911 treatise, The Purchasing Power of Money, Fisher comes off as two parts scientist to one part mad scientist, claiming, for example, that his method will “ultimately be recognized as an exact science, capable of precise formulation, demonstration, and statistical verification.”
 
TO ILLUSTRATE HOW index numbers work, Fisher used an example of a three-commodity basket consisting of sugar, coal, and “a given grade of cloth.” We can keep things even simpler by imagining a basket of just two commodities, Good A and Good B.
 
Say that, from the first period to the second period, the price of Good A has risen by 10% and the price of Good B by 1%. In order to calculate the percentage rise in the cost of living, we need to know how these commodities are weighted in the typical consumer basket. Based on our consumer survey, Good A accounts for 30% of all expenditures and Good B for the other 70%. We set the base period equal to 100, with the price per unit of Good A accounting for 30 points and B, for 70 points.
 
Formerly worth 30 points, A is now worth 33 points, since its price has risen by 10%; formerly worth 70 points, B is now worth 70.7 points, since its price has risen by 1%. Add the two components together (33 + 70.7), and we get 103.7. Against the base-period figure of 100, then, CPI inflation has run 3.7%.
 
Instead of tracking two commodities, however, the CPI is recorded from the prices of literally thousands of goods and services, classified under 200 broad categories. To update the index on a monthly basis, BLS data collectors call or visit retail stores, service establishments, rental units, and doctors’ offices to determine the price of things.
 
Since there are millions of these establishments, the economic assistants are sampling the universe of price data, which necessarily means that the numbers reported are vulnerable to sampling error. There is also nonsampling error, which can arise, for example, when the good or service being tracked is no longer available. Moreover, the data-gathering work is generally not done on weekends and holidays, times when shopping has become increasingly prevalent.
 
These approximations are then fed into the BLS weighting scheme, which is itself based on approximate weights derived from Consumer Expenditure Surveys for 2011 and 2012.
 
The weighting raises the vexing problem of substitution. Say that the price of blueberries goes up and the price of strawberries stays flat. Consumers naturally respond by purchasing more strawberries. Should the full effect of the blueberry price rise be recorded, or should strawberries receive a greater weight to reflect the substitution? The CPI adjusts the weighting to reflect the substitution. Within the steak category, for example, the formula permits substitution between flank steak and filet mignon. Contrary to myth, however, the CPI never allows hamburger to be substituted for steak.
 
The Bureau of Economic Analysis, keeper of the personal consumption expenditures price index, gives its PCE index more latitude with respect to substitution—one key reason the PCE index generally rises more slowly than the CPI. Through September, the five-year average annual increase in the PCE has been 1.5%, compared with 1.7% for the CPI.
 
ANOTHER PROBLEM to which there is no good solution: tracking the “price” of living in the home you own. While 36.5% of U.S. households are renters, and thus pay rental prices that can be sampled and compiled, the other 63.5% are their own landlords and therefore pay no recordable rent.
 
Until 1983, the homeownership component of the CPI tracked monthly mortgage payments, which in turn reflected house prices and mortgage interest rates. That was scrapped in favor of “owner’s equivalent rent,” which is supposed to estimate what homeowners would pay themselves if indeed they were their own landlords, renting their homes to themselves. The estimates are arrived at by finding comparable homes that are actually being rented, an imperfect process at best.
 
Another imperfect process, but likewise unavoidable, involves products whose quality changes noticeably.
 
A simple, if now outdated, example, is to consider two TV sets that are identical in every way, except the new model includes a remote control. If the new model sells for 5% more than the old, should we say that TV-set inflation is running 5%? Hardly. Since consumers are more than willing to pay the premium for the convenience of having a remote, a statistical method called “hedonics” is applied to the change, and no price hike is recorded.
 
Hedonics is routinely applied to items that include men’s and women’s apparel, major household appliances, and, of course, housing, which often gets renovated and enlarged.
 
Then there are new products, which can also disrupt the price trend. When economist Fisher declared in 1922 that his price indexes were precise to a tolerance of one part in 800, he neglected to mention that the automobile was not at that point in the CPI. It was introduced in 1935; air conditioners, in 1964; and cellphones, in 1998.
 
THE CLAIM MADE BY many skeptics that the current CPI and PCE grossly understate price inflation simply doesn’t hold up.
 
Their case is refuted by a statistical form of reductio ad absurdum. Recall that the PCE index is used to turn nominal consumer spending into real consumer spending, and that over the past five years, the PCE has risen at an average annual rate of 1.5%. Over the same period, nominal consumer spending has grown at an average annual rate of 3.8%. Back out PCE inflation of 1.5%, and real consumer spending has risen by 2.3%.
 
But what if the skeptics are right that the actual rate of inflation has been 4% or greater? If that were true, we would get an absurd result; it would mean there has been no growth at all in real consumer spending over the past five years, since 4% inflation turns nominal growth of 3.8% into a small negative. But no growth in real consumption defies belief, if only because wholesale and retail trade establishments have added 1.7 million jobs over the past five years, an employment increase of 8.6%.
 
These price indexes are rough approximations, but they are still useful. For example, when the Nasdaq Composite hit a nominal high of 5073 this past April, the media declared that its March 2000 peak of 5048 had finally been topped. But not in real terms. The CPI has risen 38.9% since March 2000. So to keep up with inflation, that peak of 5048 must be 38.9% higher, or more than 7000 in 2015 dollars. Give or take.