Price changes from some past reference date, in percentage terms, are averaged for the various goods and services. The resulting index number is the measure of price change from that past period (expressed as 100) to any later date. Through December 1952, the index was calculated using the average of the 5 years 1935-39 as a base. It was then changed to the base of 1947-49 = 100 to correspond to other indexes published by Government agencies, as recommended by the Office of Statistical Standards, Bureau of the Budget. Index numbers as published from month to month compare prices at each date with the average level of prices in these 3 years.
Prices are obtained in 46 cities so selected that their populations are representative of the entire population of the 3,000 cities in the United States. Prices in all 46 cities are then combined into the National Index.
Separate indexes are calculated for the 20 largest of the 46 cities— monthly for the 5 largest, and quarterly for the 15 others.
The index measures the effect of price changes on the cost of the goods and services in the family "market basket." The contents of the "market basket"—that is, the quantities and qualities of goods and services that represent what families bought in 1951-52—is assumed to remain the same, so that the change in cost from month to month is the result of changes in prices. The index does not purport to measure the changes in spending of families that result from changes in their standards of living.
It measures only the change in spending caused by changes in prices.
In each city the Bureau selected a representative sample of families from the entire population, including all family types and income classes. Interviewers visited and interviewed each family and obtained a complete record of the kinds, qualities, and amounts of foods, clothing, furniture, and all other goods and services the family bought in 1950, together with the amount spent for each item. These records for all wage-earner and
clerical-worker families of two or more persons were averaged together for each city, to form the basis for index weight determination.
The development of index weights from these survey results involved two major steps: (1) the averaging of variations in spending patterns reported by individual families, correcting the data for sampling and reporting errors and adjusting for unusually high purchases of automobiles, TV sets, and other consumer durable goods in 1950, and (2) the adjusting of survey data for price and income changes that had occurred after the survey year 1950.
The first step required the determination of stable relationships between average family income and average family expenditures on major categories of consumer goods and services, and also the development of relationships between expenditures on major commodity groups and expenditures for subgroups and items within these major groups. These relationships were discovered through a detailed analysis of the 1950 Consumer Expenditure Survey results which compared average expenditure patterns among the 91 cities included in the survey. This analysis defined the relationship between family average income and average expenditures for three major categories—food and alcoholic beverages, housing, and all other goods and services combined. Next, distributions of average expenditures on these major categories to successive groups and subgroups of items were determined; for example, the proportion of expenditures on all goods and services other than food and housing going to clothing, the proportion of clothing expenditures going to men's clothing, the proportion of men's clothing expenditures going to outerwear, and so on.
Finally, the distributions of expenditures to individual items included in the smallest subgroups of goods and services were based on average ratios of item to subgroup expenditures within groups of cities that showed approximately the same relationships. Adjustments for unusually high 1950 spending for automobiles and some other items were based on time series studies which estimated normal rates of growth in family expenditures on these items.