Ever since the 1920's, and especially since 1933 when housing became a matter of governmental concern, there has been a growing interest in statistics measuring the level, trend, and economic significance of residential construction activity. Both industry and Government have sought to find, in quantitative data, some guidance in the development of public and business policies which might produce a steady flow of new housing adjusted to the market in both volume and cost.

This growth of interest has led various agencies, both public and private, to devote increasing attention to the technical problems of current reporting, and continuing efforts have been made to widen the coverage of reporting sources, to strengthen techniques where estimating was necessary, and to provide additional internal classifications as new questions arose for which the answers could not be found in existing data. No less important than the improvement of current reporting have been the efforts of research groups such as the Twentieth Century Fund and the National Bureau of Economic Research to provide historical data with which current trends might be compared. These organizations have prepared estimates of new nonfarm family dwelling units started by years from 1900 to 1929. These estimates were necessarily based on incomplete data compiled without the benefit of later improvements in current reporting systems. They are subject to some error in the distribution as between years but the decade totals are believed to be fairly reliable.'

Using techniques similar to those employed by the National Bureau of Economic Research, the Bureau of Labor Statistics, Department of Labor, has prepared estimates for the years 1930-39. The Bureau, which is the official Government source of statistics on residential construction,

1 See appendix A, Bibliography for Part One.

now keeps these data current through a system of estimates which are prepared for each month and then recast at the end of each year into a final annual estimate.

With certain omissions which are discussed later, it is thus possible to put together from the estimates of these three agencies, a picture of residential construction since 1900. This has been done in the insert table. (Public housing, shown separately in the table, is reported to Bureau of Labor Statistics by the Federal Public Housing Authority.)

A brief look at the history of residential construction since 1900 as revealed by these figures will serve as a background for further discussion of statistical developments.

The wide fluctuations which characterize home building activity are apparent. Approximately 4,000,000 dwelling units were added to the Nation's housing supply during each of the first two decades; the second decade undoubtedly would have shown an increase over the first had it not been for the relatively low activity during the World War I years.

The third and fourth decades are in marked contrast with each other as well as with the two earlier periods. More than 7,000,000 dwelling units were added in the years 1920-29, less than 3,000,000 during the 1930's. World War II with its severe shortages of labor and materials, interfered seriously with housing production during the first half of the 1940 decade, the total new supply for the first 7 years being less than 3,500,000 units including war housing.

Beginning with World War I the flow of housing into our supply has come not as a steady stream, but as a series of sharply contrasting peaks and valleys. Low levels of activity characterize both World War periods; sharp rises following the end of each war testify to housing shortages and

consequent high demand. The peace years between the wars show first an abrupt break resulting from high prices, then a rise to a high production level during the 1920's with each of the 7 years from 1922 to 1928 showing additions well above 700,000 units. Production tapered off from the peak of 937,000 units in 1925, and 8 years later, in 1933, construction amounted to only 93,000 units, or less than 10 percent of the peak. The climb from this low point was long and slow, and it was during this period that public housing first assumed some importance as a contributor to the Nation's housing supply. During the war, public, financing of housing production assumed substantial proportions; privately financed housing has nevertheless continued to constitute the mainstay of our housing supply. Total new nonfarm family dwelling units started: By type of financing, by year, 1900 through 1946

Excludes 5,998 publicly financed dwelling units completed in 1918 and 1919 by the U. S. Housing Corporation. Data as to when these units were started are not available.

A large proportion of this housing is comprised of temporary wartime units. Data for 1945 and 1946 adjusted for lapsed building permits and for lag between issuance of permit and actual start of construction.

Excludes veterans' housing units provided by conversion, stop-gap family accommodations, and veterans' housing developed with local funds which is not otherwise part of the Federal Public Housing Authority Title V program.

Sources: Twentieth Century Fund-1900-1919; National Bureau of Economic Research-1920-29; Bureau of Labor Statistics, U. S. Department of Labor for privately financed, Federal Public Housing Authority tor publicly financed-1930-46.

Current Residential Construction Data

The Bureau of Labor Statistics has been developing its present program of residential construction statistics since the late 1930's. It now prepares a group of interrelated monthly estimates covering

the number of new nonfarm family dwelling units started, their estimated costs, and the on-site employment and value put-in-place during the course of construction. The methods used in preparing these estimates are described in chapters I and II below. It should be noted here that they are estimates and that there are no current series which represent an actual count of dwelling units constructed or which provide data on their costs taken from actual accounting records. The Bureau has attempted to approach this ideal by enlisting the cooperation of municipal authorities in reporting data on permits issued for dwelling units, and an encouraging proportion of permits issued by urban places is now so reported. However, a large amount of construction takes place in outlying areas where permits are not required, and continuing and relatively costly field checks have had to be employed to account for this factor. While continuing to emphasize the improvement of its reporting sources, the Bureau believes that most users of residential construction data do not have the facilities for compiling their own estimates from the raw data reported and therefore has designed its program so as to produce, from the combination of reported information and field surveys, a set of "ready-made" estimates consistent with the broad estimates for earlier years referred to above.

While for this reason the BLS data receive primary attention in this book, it should be noted that data on residential construction volume also are compiled by the F. W. Dodge Corp. for the 37 States east of the Rocky Mountains. The data represent a compilation of the construction news reports of its field agents. In contrast to the BLS, the Dodge corporation presents only the unadjusted data and does not estimate for any undercoverage in reporting States or for construction in the 11 Western States where its field agents do not operate.

It also should be noted that current data on residential construction activity fall short of present requirements in certain other respects. Current reports, for example, provide very little data on farm dwelling construction. Even in the rural nonfarm classification, which they do attempt to cover, no satisfactory data are regularly collected on repair and maintenance construction despite its obvious importance, nor is there any permanent system for measuring the number of existing structures which are converted to provide additional dwelling units or are converted from

residential use to nonresidential purposes. Still another gap in housing statistics is caused by the fact that no systematic data are available to show the loss of residential structures through demolition, fire, floods and other disasters.

During World War II and later during the period of the Veterans' Emergency Housing Program, when critical needs developed for adequate bases upon which to make programming estimates of materials consumption, the data on residential construction activity were put to a severe test, and had to be pressed into service in connection with many problems which do not exist in normal times. Certain of the gaps referred to assumed major significance. For example, the lack of adequate information on repair and maintenance activity, and on conversions, left wide areas of doubt in material requirements estimates. The probability of an unusually high rate of lapse in the use of local building permits after issuance, or of local differences in those rates, raised many questions about the reliability of permit reports as indicators of actual starts and therefore of material needs. In some instances temporary reporting systems based on Federal priority operations, made it possible to obtain measures of some items, such as conversions, which are useful even though they cover limited periods. Perhaps more important from the statistical viewpoint, these temporary systems provided experience with the technical problems and cost that must be expected in any attempt to make basic improvements in the data collection program in this field. The emergency reporting systems were aided by priority systems which simplified the selection of samples and the control of coverage. No such convenient foundation for a reporting system exists in peacetime. Given the highly decentralized nature of the residential construction industry and its large and constantly changing membership, only a portion of which reports building plans to local building departments, the systematic collection of data on a permanent basis is a far from simple task. It seems clear, therefore, that supplementary field studies such as the BLS has conducted in the past will continue to be needed for some time to come.

? Between January 1940 and August 1942, for example, complete information on the volume of privately financed residential construction was gathered in 148 of the principal centers of war activity. The data were obtained through the Defense Housing Survey, a field study conducted by the Bureau of Labor Statistics in cooperation with the National Housing Agency and the Work Projects Administration. For a statement on the purpose and scope of this survey, see monthly Labor Review, December 1942 and September 1943.

BLS is also making sample surveys to develop data on the intended occupancy of privately financed dwelling units. Information on these surveys may be obtained from BLS.

Arrangement and Scope of Data

Not all of the available data on housing production could be included in this volume, but an attempt has been made to include all those which appear to be of major importance on a national basis. In some cases data also are provided by census region and for selected localities.

Chapter I is a discussion of privately financed housing production, in terms of numbers of dwelling units. Chapter II treats residential construction in terms of dollar volume. Chapter III is devoted to public housing, and covers both numbers of units and expenditures for them. Chapter IV is a discussion of residential construction cost indexes; although they are not measures of housing production their relation to construction activity indicated their inclusion in this part of this volume.

Several points should be noted concerning the data:

(1) The series on housing production represents only new nonfarm family dwelling units. The terms "nonfarm" and "family dwelling unit" are defined in chapter I under the discussion of the estimating procedure employed by the BLS, and these terms must be understood in connection with the data on physical volume as well as dollar volume of construction. This publication provides no data on the volume of farm construction, very little of which is available. With respect to the types of accommodations such as trailers, dormitories, etc., which are excluded from the basic series, however, several sets of data on such war and emergency housing are presented in the appendix. It also is possible to obtain, from the F. W. Dodge Corp., current data on construction contracts awarded in the 37 Eastern States for hotels, dormitories, etc.

(2) No break-down of units by sale or rent is included, since no such break-down is made in any of the regular series. BLS does provide a break-down by type of structure and rough minimum guesses as to the number of units placed on the rental market may be made by assuming that all multifamily units are for rent and that one unit in all two-family structures will be rented. There are, of course, a substantial

number of single family units which are placed on the rental market, but no sound basis exists for estimating their number.

(3) Until 1946, no series has been available on the total number of nonfarm dwelling units completed. In that year, however, the Veterans' Emergency Housing Program stimulated an extremely sharp rise in residential starts. Since this rise occurred before adequate supplies of

For the period July 1941-September 1945 monthly data were compiled on the number of dwelling units completed with priority assistance. During 1942 priority completions probably accounted for about 85 percent of the nonfarm total. Thereafter the proportion fell off as it became progressively easier to build without a priority.

building materials were generally available, an urgent need developed for data on completion progress. Under the sponsorship of the National Housing Agency and Office of the Housing Expediter, the BLS conducted a series of special field studies to develop factors by means of which estimates of completions could be made. The cost of those field studies and the lack of widespread demand for such data in normal times make it unlikely that such estimates will continue as a permanent part of the body of residential construction data after the end of the emergency period.

The introductory discussion above has served to disclose the fact that available data do not adequately cover certain aspects of residential construction, such as farm construction, conversions, and repair and maintenance activity. A distinction also was made there between the statistical program covering normal types of construction and the emergency reporting systems covering such nonconventional types of shelter as trailers and reuse dormitories.

While these are important items, the construction of new permanent housing is by far the most important component in the provision of shelter. Again, as the introductory table shows, new Table 1.-New privately financed nonfarm dwelling units started, United States totals, by urban and rural nonfarm area, by year, 1920-46

housing financed by private investment is by far the most important element in permanenttype housing.

It is the purpose of this chapter to describe in more detail the methods developed by the Bureau of Labor Statistics to prepare its estimates on the volume of such privately financed nonfarm new housing in terms of family dwelling units. Volume in terms of cost is discussed in chapter II.

Most of the estimates presented are for the country as a whole, but certain of the data have been provided for the nine major census geographic divisions. Data also are presented for the 59 selected industrial areas and urban counties for which BLS conducted its regular area housing surveys. The national estimates are of course most reliable, the difficulties of precise estimation increasing as the area covered decreases. It should be noted that the 59 industrial areas and urban counties were selected by BLS as sample areas from which data essential to its national estimates could be obtained. They do not necessarily represent local housing markets.

The family dwelling unit, as used in these statistical series, is defined as a living accommodation intended for occupancy by one household. Occupancy may be "year around" or it may be seasonal; the unit, however, must not be located on