Problems Measuring Inflation

Just as there are problems measuring unemployment with a single number, there are problems in trying to capture with a single number what is happening to millions of prices. First, the Consumer Price Index is based only on a small fraction of the many goods and services that are available. If a person's buying habits differ substantially from the market basket on which the index is based, that person may experience a very different change in the cost of living than what the CPI shows. To represent this problem in terms of the example of the previous section, consider a person who never eats hamburgers, drinks beer rather than soft drinks, and does not own a car. A price index built on the buying pattern of this person might be totally different. The Labor Department addresses this problem to some extent by constructing two price indexes, one for all urban consumers (the CPI-U), and another for urban wages earners and clerical workers (the CPI-W)

This question of buying patterns has practical importance when wages and government benefits are indexed (tied to) the rate of inflation that the CPI measures. For example, Social Security payments are indexed to the CPI-W from the end of one third quarter to the next. When the CPI-W rises, so do Social Security checks (though the adjustment is made only once a year). Many retired people have considerably different buying patterns than urban workers, and thus their rate of inflation may be greater or less than what the CPI-W measures. If it is less than what the CPI-W measures, retired people on average benefit from inflation, and if it is greater than what the CPI-W measures, they suffer.

Changes in the quality of products are difficult to incorporate into the CPI. If a product becomes better with time and the price also rises, how much of the change in price is due to the improved quality? Changes in quality are rather common over long periods of time. The automobile of the 1990s was very different from that of the 1970s, which in turn was very different from that of the 1950s. Sometimes changes in quality are evident in short periods. The tremendous improvement of electronics products—televisions, audio equipment, and computers—has changed the way we live and work. The Department of Labor tries to make adjustments for quality changes, but by their very nature such adjustments are in part subjective.

The table below illustrates a final problem with the CPI. In this market basket of only two items, apples and oranges, the purchase of apples drops from period 1 to period 2 because their price rises and the purchase of oranges rises because their price drops. We have a choice of two different market baskets when constructing a price index to measure inflation. We can choose the market basket based on the early weights (10 apples and 9 oranges) or the market basket based on the late weights (8 apples and 11 oranges). We will get different answers depending on the choice we make.

WHICH MARKET BASKET?

PERIOD 1 PERIOD 2

Items
AMOUNT YEAR 1 PRICE YEAR 1 AMOUNT YEAR 2 PRICE YEAR 2

Apples
10 $.20 8 $.25

Oranges
9 $.25 11 $.21

If one constructs a price index using the early quantities, one finds the market basket increases in price from $4.25 to $4.39, or a 3.3% increase. If one uses the market basket based on the consumption pattern of the second period, one finds the cost of the market basket drops in price from $4.35 to $4.31, a decrease of .9%. This pattern is normal; a price index based on early quantities will usually show a larger rise in prices than one based on late quantities. The reason is that neither basket takes into account that people substitute as the result of changes in price structure.

Beginning in January of 2002, the Labor Department began to update the consumption-expenditure weights every two years. Prior to the 1990s, the Department had updated the weights in ten-year intervals. The CPI for 1990 was based on a market basket formed from consumer buying patterns that existed in 1982. If the Department of Labor had done another survey of consumer buying patterns in 1990, it would have found that consumers were buying more of those items that had dropped in price or that had risen in price by small amounts. People would have been buying less of those items that had risen in price much more than average. If these new weights were used to recompute the price index, the large price rises would have a smaller impact on the total. The smaller price rises would have more of an impact on the total because they would be more heavily weighted. Therefore the resulting price rises would look smaller, which is what the example above explained. With more frequent updating of weights, the CPI should be a more accurate measure of inflation.

Another important economic measurement is that of production, done with Gross Domestic Product.

¹ If you want a technical discussion of this problem, check more advanced books under the topics Paasche and Laspeyres indexes.
Copyright Robert Schenk