As the agency that compiles the Consumer Price Index, the Bureau of Labor Statistics is aware of the CPI's limitations. But we also are aware of, and concerned about, criticism that is based on lack of understanding of indexation arrangements. Our role at BLS is to find the CPI's limitations and try to overcome them. We leave to others the solution of policy issues related to indexation.

The Consumer Price Index measures the change in price of a fixed market basket of goods and services. The CPI is a good measure of price change, but, like any other statistical measure, it is not perfect. The imperfections have received much criticism in recent years because the index has been widely used in income escalation. It is important to note, however, that the critics often disagree among themselves. Those who receive income tied to the index frequently feel that prices are rising faster than the CPI. They argue that the CPI is too low. Those responsible for funding indexed payments find that the cost of indexation is large. They argue that the CPI is too high. And, government officials fear that the indexed payments will raise costs and help to fuel inflation.

The most common technical criticisms of the CPI are (1) the items priced for the index are those purchased in the base year, not those currently bought and (2) the CPI overstates the price of homeownership.

The fixed market basket. The CPI is based upon a market basket of goods and services purchased in a base period. The basket now in use is based upon a survey of what consumers purchased in 1972 and 1973. BLS practice has been to stick with a market basket until a major revision of the index occurs--about every 10 to 12 years. The market basket is kept constant deliberately because we measure price changes, not changes that may occur in living standards.

In recent years, as prices have risen, some have argued that the CPI market basket has become outdated. They contend that, because rational consumers shift the quantities of the things they buy when the relative prices of them change, the CPI might overestimate the cost of maintaining current living standards.

Most research indicates that these shifts in consumer purchases have not usually created large differences in price indexes in the past. To evaluate this question more fully, however, we have begun field collection of a continuing consumer expenditure survey. In a few years, when this survey is fully in place, BLS will be able to monitor continuously the degree to which consumers change their consumption patterns. Then, we will be able to determine when future revisions of the CPI weights are needed and will have at hand the data needed to revise the market basket.

Owner-occupied housing in the CPI. The measurement of owner-occupied housing has been troublesome for many years. The basic problem is to determine what the index should measure.

The present CPI views a house both as an asset that can be resold and as a home in which the owner lives and consumes housing services. The index measures month-to-month changes in prices of five expenditures of owning a home--house prices, contracted mortgage interest, property taxes, insurance and maintenance and repairs. The total weight for home ownership under this approach is large, about 23 percent of the entire index.

More than a year ago, BLS began publishing five experimental indexes reflecting different concepts and measurement approaches. All exclude the asset value of an owned home and give a much smaller weight to the home ownership component.

The most widely discussed of these experimental alternatives is the "rental equivalence" (CPI-X1) index. CPI-X1 excludes the investment aspects of home ownership and includes only the cost of consuming the shelter services provided by a house. Over the last decade, CPI-X1 has risen less than the CPI; in recent months, however, the changes in CPI-X1 have been closer to the official index.

CPI-X1 uses rent as a proxy for home ownership costs. Since a true rental equivalence sample--one made up of housing units of the same types and in the same locations as owned units--is not currently available, BLS wants to develop and has requested funds in the FY 1982 budget to develop a rental equivalence measure.

So we believe that the asset value of an owned home should be excluded from the CPI. We are publishing -- for public discussion and possible use --five experimental approaches that exclude the asset value. To permit more frequent evaluation of the CPI market basket and weights, we are collecting consumer expenditures data on a recurring basis.

We also are evaluating the effect on the CPI of variable rate mortgages and reviewing problems inherent in the shrinking size of the Federal Housing Administration data base of house prices used in the CPI.

These technical issues, however, must not be confused with the goals of indexation. If the goal is to keep real income constant, one should not be surprised when the income share of the indexed group rises relative to a group whose income is not indexed. The statistical measure used as the mechanism should not be blamed for an indexation policy that produces unexpected results.

The CPI is one of the government's most important measures, affecting the incomes of millions of people who receive government payments or work under union contracts. It must have both continuity and credibility.

Before it is changed, there should be full advance notice and opportunity for public discussion. It has been customary to revise the index every 10 years or so. The last major revision was in 1978. The next is now in the early planning stage.