BLS to Test Experimental CPI

The Bureau of Labor Statistics yesterday unveiled an experimental version of the consumer price index that rises more slowly than the official version because it takes partial account of how people switch what they buy when the price of one item goes up more than another.

The new experimental CPI rose about 0.25 percentage points less in both 1995 and 1996 than the official version, which is based on a fixed market basket of goods and services and makes no allowance for such substitutions. The official CPI increased 2.5 percent in 1995 and 3.3 percent last year.

BLS officials said that using a different formula for one step in calculation of the CPI could move the important index closer to a true measure of changes in the cost of living because consumers respond to changes in relative prices when a close substitute is available. For example, a consumer choosing between romaine and iceberg lettuce may buy more iceberg and less romaine if romaine is more expensive.

From now to the end of the year, BLS statisticians plan to study the experimental index to determine how to adjust the official CPI. Any change would likely be made in January 1999.

Altering the CPI would affect the millions of Americans who receive checks for Social Security, veterans' pensions and other federal benefits that are annually adjusted for inflation according to changes in the index. A change also would affect features of the income tax code, such as the size of the standard deduction.

Many economists say the official CPI overstates yearly increases in the true cost of living by 0.5 percentage points to 1.5 percentage points. A small portion of that overstatement stems from failing to take substitutions into account.

The Clinton administration and some congressional leaders have been trying to find some way to reduce the annual cost-of-living adjustments in federal benefits programs as part of their drive to balance the federal budget.

The experimental index, which is based on a formula called a geometric mean, would be a small step in that direction. However, BLS officials stressed that their concern is producing a CPI that is as accurate as possible, regardless of its impact on the budget.

To explain the experimental index, BLS used the example of a shopper buying two heads of lettuce, one of romaine and one of iceberg, both selling for $1. If the price of the romaine rose to $1.50 and the other price was unchanged, the customer would be paying $2.50 instead of $2, and the CPI would show a 25 percent increase in the price of lettuce.

But suppose the shopper noted the higher price for romaine and decided not to keep buying the same quantities as before but to continue to spend similar amounts of money on each type. In this case, the geometric mean formula would show the consumer buying 1.225 heads of iceberg and 0.816 heads of romaine for $2.45, with the price of lettuce going up 22.5 percent in the experimental CPI.

Of course, buying partial heads of lettuce isn't usually possible, but it is possible for millions of consumers to shift their purchases so that the quantities substituted make statistical sense.

In the experimental index, the geometric mean formula is applied when the prices of goods and services are brought together into 207 categories, such as apples, laundry equipment, prescription drugs and gasoline.

The impact on the CPI of adopting a geometric mean approach will depend on how many and which of the 207 categories BLS determines should be handled that way. The areas in which it is likely to be used are food, clothing and entertainment, officials said.

Consumers also may substitute very dissimilar goods and services when prices change—perhaps deciding to eat out rather than go to a movie. The geometric mean will not be used to handle that type of substitution.