Understanding the CPI
The idea of the CPI is that it is an index number for the cost of everything American consumers buy. That idea needs lots of adjusting to be practical. Much of the fiddling uses the results of large sample surveys.
Who is covered? The official name for the common version of the CPI (there are others, but we will ignore them) is the Consumer Price Index for All Urban Consumers. The CPI market basket represents the purchases of people living in urban areas. The official definition of “urban” is broad so that about 80% of the U.S. population is covered. But if you live on a farm, the CPI doesn’t apply to you.
How is the market basket chosen? Different households buy different things, so how can we get a single market basket? From a sample survey. The Consumer Expenditure Survey gathers detailed data on the spending of more than 30,000 households. The Bureau of Labor Statistics (BLS) breaks spending into categories such as “fresh fruits and vegetables,” “new and used motor vehicles,” and “hospital and related services.” Then it chooses specific items, such as “fresh oranges,” to represent each group in the market basket. The items in the market basket get weights that represent their category’s proportion of all spending. The weights, and even the specific market basket items, are updated regularly to keep up with changing buying habits. So the market basket isn’t actually fixed.
375
How are the prices determined? From more sample surveys. The BLS must discover the price of “fresh oranges” every month. That price differs from city to city and from store to store in the same city. Each month, the BLS records 80,000 prices in 87 cities at a sample of stores. The Point of Purchase Survey of 16,800 households keeps the BLS up-to-date on where consumers shop for each category of goods and services (supermarkets, convenience stores, discount stores, and so on).
Does the CPI measure changes in the cost of living? A fixed market basket price index measures the cost of living the same over time, as Example 2 illustrated. In fact, we don’t keep buying the same market basket of goods and services over time. We switch from LP records to tapes and CDs and then to music downloads. We don’t buy new 1995 cars in 2005 or 2015. As prices change, we change what we buy—if beef becomes expensive, we buy less beef and more chicken or more tofu. A fixed market basket price index can’t accurately measure changes in the cost of living when the economy itself is changing.
The BLS tries hard to keep its market basket up-to-date and to compensate for changes in quality. Every year, for example, the BLS must decide how much of the increase in new-car prices is paying for better quality. Only what’s left counts as a genuine price increase in calculating the CPI. Between December 1967 and December 1994, actual car prices went up 313.4%, but the new-car price in the CPI went up only 172.1%. In 1995, adjustments for better quality reduced the overall rise in the prices of goods and services from 4.7% to only 2.2%. Prices of goods and services make up about 70% of the CPI. Most of the rest is the cost of shelter—renting an apartment or buying a house. House prices are another problem for the BLS. People buy houses partly to live in and partly because they think owning a house is a good investment. If we pay more for a house because we think it’s a good investment, the full price should not go into the CPI.
By now it is clear that the CPI is not a fixed market basket price index, though that is the best way to start thinking about it. The BLS must constantly change the market basket as new products appear and our buying habits change. It must adjust the prices its sample surveys record to take account of better quality and the investment component of house prices. Yet the CPI still does not measure all changes in our cost of living. It leaves out taxes, for example, which are certainly part of our cost of living.
376
STATISTICAL CONTROVERSIES
Does the CPI Overstate Inflation?
In 1995, Federal Reserve Chairman Alan Greenspan estimated that the CPI overstates inflation by somewhere between 0.5% and 1.5% per year. Mr. Greenspan was unhappy about this because increases in the CPI automatically drive up federal spending. At the end of 1996, a group of outside experts appointed by the Senate Finance Committee estimated that the CPI had, in the past, overstated the rate of inflation by about 1.1% per year. The Bureau of Labor Statistics (BLS) agreed that the CPI overstates inflation but thought that the experts’ guess of 1.1% per year was too high.
The reasons the CPI shows the value of a dollar falling faster than is true are due partly to the nature of the CPI and partly to limits on how quickly the BLS can adjust the details of the enormous machine that lies behind the CPI. Think first about the details. The prices of new products, such as digital cameras and flat-screen televisions, often start high and drop rapidly. The CPI market basket changes too slowly to capture the drop in price. Discount stores with lower prices also enter the CPI sample slowly. Although the BLS tries hard to adjust for better product quality, the outside experts thought these adjustments were often too little and too late. The BLS has made many improvements in these details (most recently, in February 2015). The improved CPI would have grown about 0.5% per year more slowly than the actual CPI between 1978 and 1998.
The wider issue is the nature of the CPI as essentially a fixed market basket index. What sort of bias does such an index create? Does it produce an upward bias; that is, does it overstate the cost of living? Or does it create a downward bias; that is, does it understate the cost of living? Why?
Even if we agree that the CPI should look only at the goods and services we buy, it doesn’t perfectly measure changes in our cost of living. In principle, a true “cost-of-living index” would measure the cost of the same standard of living over time. That’s why we start with a fixed market basket price index, which also measures the cost of living the same over time but takes the simple view that “the same” means buying exactly the same things. If we are just as satisfied after switching from beef to tofu to avoid paying more for beef, our standard of living hasn’t changed, and a cost-of-living index should ignore the higher price of beef. If we are willing to pay more for products that keep our environment clean, we are paying for a higher standard of living, and the index should treat this just like an improvement in the quality of a new car. The BLS says that it would like the CPI to track changes in the cost of living but that a true cost-of-living index isn’t possible in the real world.