Prices of the 300 items are collected at regular intervals—some monthly, others less frequently—and the successive prices are compared to calculatc price changes. It is important to be certain, lum ever, that the calculati(m shows only price change and not changes which arc due to other factors, such as quantity or quality differences. For example, the price of pork. chops can vary considerably, depending upon how near the center of the loin they are cut. 11 center-cut pork chops are priced month and end-cut the next, the price of pork chops might seem to have gone down. But in fact this would reflect a difference in quality, not change in price. To prevent this, insofar as possible, the Bureau has prepared detailed specifications to describe the items that are priced for the index. These specifications arc carefully written with the advice of manufacturers and retailers of the items. For example, the following is a specification for one of the menss shirts priced: Sty lc Business, fused or similarly constructed collar, attached. barrel cuffs.
Fabric cotton broadcloth, wihte.
Yarn combed
Thread count 136*60 or 128*68
Finish Residual shrinkage 1 percent or less.
Construction Full cut, clean workmanship; 31 to 32 yards per dozen based on 36-inch fabric.
Size range 14 to 17 inch neckband.
Brand manufacturer's nationally advertised.
When the Bureau's agents price these shirts, they examine the merchandise in the stores to make sure that the prices they record meet this specification.
Limitations (I) Limitations of Measurement. The Consumer Price Index is not an exact measurement. It is subject to the many kinds of limitations that are always present in statistical calculations. The Bureau makes every effort, within the resources allotted to it, to keep the total effect of statistical errors to a minimum, and continually studies the nature of the errors, looking for ways to improve the index.
One kind of limitation arises from sampling procedures. Expenditures for items in the "market basket" are based on interviews with about 8,000 wage-earner and clerical-worker families; price changes are based on prices of about 300 items, collected in about 2,000 food stores and 4,000 other retail stores and establishments; rents are obtained from about 30,000 tenants. Prices are collected in 46 cities, some every month, some every 3 months, and some every 4 months. Thus, the index is based on samples of families, items, stores, and cities that are only a fraction of the total. There is even a "sample" of time, since information is collected only at certain periods.
The degree of error introduced into the index through sampling depends primarily on the amount of variation in price change that exists within groups of items and between stores and cities. To gain about the
same degree of accuracy throughout the index, therefore, the number of price observations obtained for any item is conditioned by its price variability and its importance in the total index. For example, prices of fresh vegetables, which are important in the family food budget, change frequently and have different seasonal patterns in different cities; they differ considerably from item to item and from store to store. To measure the average change in prices of all vegetables satisfactorily, a large number of them must be priced in a fairly large number of stores each month in every city. On the other hand, prices of men's nationally advertised brand shirts do not change often, and the same kind of shirt sells for the same price in almost all stores. These can be priced less frequently and in fewer stores, and still measure the price change satisfactorily.
Another kind of error may occur in the index because people who give information cannot always report exactly. In making surveys of consumer expenditures to determine the basic "market basket," the Bureau has found that families can estimate very well what they spend in a year for rent, or electricity, or automobile, but not all can remember all their expenses for men's shirts, or women's hose or other things that are bought frequently. Some report less than they actually bought, some more, so that these errors of recall tend to cancel out. The Bureau uses the most modern survey methods and highly trained interviewers to keep these reporting errors in the "market basket" to a minimum, and the survey results are compared with information obtained by other organizations whenever possible so that the figures can be adjusted if necessary.
The Bureau takes precautions to guard against errors in pricing which might affect the index most seriously. Price agents are well trained to recognize qualities of goods on the store shelves, and they take the selling prices as marked or as reported by the merchant. Sales prices and discounts are recorded, as well as premium prices and special charges. However, sales prices are included in the index calculation only if the goods offered on sale are in good condition, meet the Bureau's specifications, and the sale is of sufficient duration to allow most consumers to take advantage of the price reductions. Discounted prices are used if they apply generally to all customers; they are not used if offered to special types of buyers only, or if affected by the bargaining skills of the buyer.