unweighted regional indexes which were then converted to a base of 1935-39-100. Finally, the national average was obtained by weighting the regional indexes according to the proportion of dwelling units built in each region during the period 1935-40, as shown by the 1940 Census of Housing. Since 1940, the basis of weighting has been the number of nonfarm dwelling units started annually in each geographic region.

This index applies to all residential construction ranging from single-family units to six-family apartments. The indexes for larger apartments were not included in the average since units in multifamily structures normally represent a relatively small percentage of total nonfarm new construction.

An index derived by combining data for only a few localities has limited value in terms of representing typical cost movements throughout the whole country. It is felt, nevertheless, that this index is probably a more reasonable indicator of current residential construction costs than the various published indexes which measure national changes in industrial, commercial, or general construction costs.

The other series in this table, which presents the cost trend of a dwelling unit was obtained by applying the derived index for each year to a house that would have cost an average of $5,000 in the period 1935-39. All land costs are excluded.

National Housing Agency Index

As indicated above, the Federal Home Loan Bank Administration inaugurated an index of residential material-labor costs in 1935. Since the middle of 1946, this index has been prepared by the National Housing Agency. The index figures for 44 cities are presented in table 34.

In constructing this index, a hypothetical standard frame dwelling was developed which would have sold in 1935-36 for approximately $5,500, excluding land. The method of compiling the index was revised in 1942; the original and revised methods are described below.

Originally the index for this standard house was based on bills of specification listing 88 material items, the quantity and quality required of each, and the number of work hours required for the various crafts to construct this six-room frame dwelling. Every third month material price quotations were secured from local building ma

For floor plan and sketch of dwelling, see Federal Home Loan Bank Review, July 1938, p. 355.

terial dealers by field employees of the Home Owners' Loan Corporation. Reports also were obtained on prevailing hours and wage rates for the principal construction trades. After 1942 the HOLC field staff was unable to handle the data collection work; since then the index has been compiled on the basis of data on retail prices of 44 building materials collected by the Bureau of Labor Statistics. These items represent about 75 percent of the total materials bill and the trend in prices for them appears to be almost identical with the trend for the 88 HOLC items.

In securing its price quotations, BLS queries a group of distributors on a quarterly basis. One representative dealer is designated for each item in each community and his price is considered as typical for that time. The material prices used are not mathematically computed averages, but are representative of actual prices.

The new source of price quotations does change the concept of the cost index somewhat. Instead of reflecting the price changes of a precisely specified bill of materials required to construct a hypothetical standard house, the index now shows price changes in a limited number of important construction items.

1936 37

LABOR COSTS

39

40

41

42

160

150

140

130

120

MATERIAL COSTS

110

100

90

45 1946

43

38 44 SOURCES NATIONAL HOUSING AGENCY, FEDERAL HOME LOAN BANK ADMINISTRATION AND OFFICE OF THE ADMINISTRATOR

Quotation on gross pay roll and number of work hours for each craft are secured through a canvass of builders engaged in privately financed residential construction. Average hourly earnings for each craft are computed for each builder and a representative over-all rate chosen for each trade. These representative rates are not "average," but the actual earnings of workers employed by a specific builder.

The wage-rate data relate to journeyman wage rates only. Residential builders' disbursements to journeymen, however, represent approximately 90 percent of the total labor bill.

Table 35 presents the NHA indexes of building costs for the standard house by census geographic divisions by quarters, for the period 1936-46. These indexes are derived from individual National Housing Agency indexes for 44 cities for which material and labor costs are collected. The indexes of all reporting cities in each geographic division are averaged with equal weights assigned to each division. These averages for each geographic division are not clear-cut quarterly indexes since the 44 cities are divided into three groups, basic data being collected for just one group each month. However, during every quarter ("cycle"), each of the three groups-comprising the 44 cities is surveyed.

From the foregoing discussion it can be seen that no completely satisfactory index of construction costs is available. The Boeckh indexes serve most usefully for the cities they cover but provide no composite picture for the country as a whole. The NHA index which is derived from the Boeckh

city data attempts to fill this gap, but is necessarily based only on the 20 Boeckh cities for which data are shown in tables 29-32. A much larger number of cities would be necessary for the construction of a sound index. The FHLBA-NHA index covers only the more important materials and labor items. In common with the Boeckh index it does not include builders' overhead and profit. It is therefore not truly a construction cost index.

Certain other difficulties must be recognized in connection with the attempt to measure the changing cost of the typical house, or mix of houses.

1. Many changes in the nature of the house as a commodity have occurred during the past two or three decades, and, in fact, such changes are occurring constantly as the homebuilding industry responds to buyers' style preferences and to improvements in plumbing, heating and electrical equipment, and in structural components such as insulation. These changes have come progres

sively and in many cases, unobtrusively but they have effected such marked change in the commodity as to suggest a high degree of caution in the use of comparisons between the present and the past.

There is thus the danger that the "typical" house, and its related bill of materials, may become obsolete and no longer resemble any of the dwellings actually being built. Or, even if the character of the house itself remains unchanged, the bill of materials may be altered, as was commonly the case during the period of war and postwar shortages. An index which could in no way be tested against actual current building experience would have little to recommend it.

2. Certain difficulties are inherent in the collection of data on prices, wages, overhead and profit and relating them to the building process. Although these data may be solicited on a confidential basis, there is a natural reluctance on the part of many people in the industry to disclose such information. For example, hourly wage rates actually paid may differ from quoted union rates and care must be taken in adjusting for these differences. Bonuses and special overtime payments undoubtedly inflated construction costs during the war, but such payments are not likely to have been reflected uniformly, if at all, in the reported wage rates. Similarly, black market prices are not likely to have been reflected in costs quoted during the period of price control.