In my office we have attempted to tackle this problem by means of a different approach. Using data from various sources on production and prices of general merchandise durable goods and separately for nondurable goods-we have computed an index of the value of production of such goods. This index of production and the index of retail sales of the same class of goods are plotted together. A period during which production and retail sales were in balance-i. e. inventories were approximately equal at the beginning and end-was chosen as the base period. Accumulation of inventory is represented in the chart by a rise in the production line to levels higher than for sales. A reverse relationship indicates liquidation of inventory. Our experience with this use of sales and production data has been surprisingly favorable considering the perverse design of the statistical source material with which we have to contend.

Exhibit G is offered as an illustration.

One of the particularly useful series of reports providing specific information on the extent of the market for selected consumer goods in terms of physical units, as measured by manufacturer's inventories, is the Facts for Industry series of the Bureau of the Census. They have additional value because of their almost complete industry coverage and because of their frequency-monthly or quarterly. Only a few trade associations have been able to match the comprehensive coverage and accuracy of these reports. Consideration should be given to the supplementation of these series by reports on other consumer products of importance for which data are not now available.

Supplying the right merchandise at the right time is only part of the problem for production and marketing people. The other part of the problem is to determine the right place. Stores, warehouses and other marketing facilities must be planned from the standpoint of the retail sales potential of the local community. The retail market is not distributed homogeneously over the map nor is the distribution a static one. Many communities have grown rapidly in population and income in recent years. Other communities have experienced changes in one or the other, or both, that are relatively less favorable than the average. The Census of Business is still the only source for reliable retail sales data on a comprehensive basis for local areas. This means that the latest authoritative figures for even the largest metropolitan areas bear a 1948 tag. Exhibit H is a worksheet developed by my office on which is listed the names of the cities in which Sears stores are located. Using the best available estimates of retail sales in each of these cities in total and for general merchandise, Sears relative sales performance in each is determined. Analysis on the basis of such data reveal possible locations for expansion of facilities or clues to localities in which Sears is competitively weak. The instituting of a series of reports giving comprehensive retail sales on a yearly basis for all cities with populations of 100,000 or over would go far toward closing this gap.

SOURCE OF DATA: Computed by Dept. 733C from data issued by various Federal agencies
and private trade associations.

The index of retail sales of durable goods of the general merchandise type represents monthly sales, seasonally adjusted, as a percentage of the 1948-49 monthly average. The index of the value of production of the same kind of merchandise, is also in the form of a percentage of the 1948-49 monthly average. It is calculated from a number of statistical series on physical production and shipments of various kinds of consumer durable goods, and production manhours of labor in other consumer durable goods manufacturing industries, converted to value equivalents by use of the nearest appropriate price index series. The use of the 2-year period 1948-49 is due to the fact that average monthly production and average monthly sales for this period were in fairly close balance. Revision of this index of production in accordance with the new production index series of the Federal Reserve Board, is in process.

(A similar chart has been prepared showing production and retail sales of nondurable goods-general merchandise type.)

In the course of my remarks, I have had occasion to refer to statistics on manufacturing production and on wholesale trade where common spheres of interest with retail trade were involved. There is yet another statistical series of especial interest to the analyst of retail trade, namely the Personal Consumption Expenditure series. In the absence of data on retail sales by type of merchandise it is common practice to resort to the use of personal consumption expenditure estimates as the nearest substitute. Often the limitation and restrictions of the data are overlooked in the process. The lack of detailed definitions item by item for this series contributes to the misuse of the data. But the greatest source of misunderstanding, in my opinion, derives from the unusual concept of the term "expenditure." Exhibit I is a chart which portrays the changing relationship between retail sales and the rising costs of consumer services. Among the various consumer services, shelter is the most important in terms

The relation of consumer spending for services to spending for merchandise in retail stores, is the subject of this caart. Although the former has shown greater percentage increases since 1947 than has the latter, the prewar balance between the two has not yet been achieved, hence it is quite likely that service expenditures will continue to gain ground in the battle for the consumer's dollar. All rates are expressed on an annual basis using seasonally adjusted estimates of the Department of Commerce. The breakdown of retail sales into the merchandise classifications indicated has been estimated from Department of Commerce data on retail sales by type of store and on personal consumption expenditures.

of actual expenditures. We find, however, that it is not the actual cost of shelter which is being reported but a hypothetical figure which is the sum of rental costs for tenants and the theoretical space rental value of owner-occupied dwellings. More than half of the families are homeowners, a larger percentage than prewar. The ratio of this "imputed" rent for nonfarm homeowners to rentals paid by nonfarm tenants doubled between 1940 and 1952. The point I want to emphasize is that we do not really know what change there has been and is occurring in the actual costs of shelter to consumers. And if that is true of shelter, it is also true of service expenditures as a whole, of which total "imputed" rent for nonfarm homeowners constituted more than 20 percent in 1952.

In this material, I have tried to present the viewpoint of the user of retail market data. The scope of Sears retail operations, reaching as it does into most large population centers, as well as rural areas, and embracing a wide assortment of merchandise, has made us particularly aware of the need for detailed information about the consumer market. Some of the points I have raised involve fundamental revisions but their practicality and worth will, I believe, be apparent upon deliberation. The problems involved are certainly within the competence of the Government statistical experts who have demonstrated their capabilities in other fields.

Mr. Chairman, that is the statement I prepared prior to the receipt of the material from the Bureau of the Budget and Council of Economic Advisers. I have a brief supplementary statement which I believe can be put in the record. I don't think it needs to be read at this time.

Representative TALLE. Thank you, Mr. Rosenbaum. That will be included in the record.

(The supplementary data submitted by Mr. Rosenbaum is as follows:)

SUPPLEMENTARY DATA SUBMITTED BY ARTHUR ROSENBAUM

In reviewing the observations of the Bureau of the Budget and the Council of Economic Advisers, neither of which were available in sufficient time to be taken into consideration in connection with my statement, I was of course particularly interested in remarks pertinent to the areas of my principal interest. With respect to the need for data on sales and inventories by kind of commodity, the Bureau of the Budget indirectly acknowledges this need although with a different purpose in mind-improved estimates of consumer expenditures by commodity. The Council of Economic Advisers felt that it would be desirable to develop a series on "sales and inventory holdings of consumer durable goods at successive stages of the distributive process, to match the new index of production of consumer durable goods by the Federal Reserve Board." Both of these suggestions are steps in the right direction but are piecemeal approaches toward the real goal as I see it.

I also find support for my recommendations for the expansion of the monthly Facts for Industry series of the Bureau of the Census, which provides data on physical volume of production and inventories of individual commodities. The Council of Economic Advisers believes this would improve FRB production indexes, and the Bureau of the Budget, that it would also contribute toward the development of better information on productivity trends. As I have previously stated, such data delineating market potentials are per se valuable to business.

All users of economic statistics are naturally interested in improving the accuracy of the data, within reasonable limits. Any doubts held by statistical experts familiar with the details of collection and manipulation of the data, regarding the soundness of any particular series, should be resolved as an indispensable part of any program aiming for the betterment of economic data.