and 1930. If, however, farm and nonfarm areas are analyzed separately the divergent trends in the years preceding World War II become apparent. In nonfarm areas owner-occupied units became relatively more important, increasing from 37 percent in 1890 to 46 percent in 1930. Mainly as a result of depression foreclosures the proportion of owner-occupied units slipped to 41 percent by 1940. On the other hand, in farm areas there was a steady decline in the percentage of owneroccupied units from 66 percent in 1890 to 53 percent in 1940. The impact of World War II forced a sharp increase in owner-occupancy in both nonfarm and farm areas. In nonfarm areas owner

occupancy rose from 41 to 51 percent of all occupied units between 1940 and 1945, and in farm areas from 53 to 65 percent during the same period. Whether these gains will be held is yet to be seen, since many home purchasers bought only out of the necessity of finding living quarters.

The 1,743,000 nonwhite tenant occupants comprised 76 percent of all nonwhite nonfarm families in 1940 as compared with 57 percent for white families. These proportions varied widely be

tween urban and rural nonfarm areas and from one geographic division to another. In urban areas 20 percent of the dwelling units of nonwhites were owned by their occupants as compared with 36 percent in rural nonfarm areas.

Data Available on Rent

Information concerning the rent or rental value distribution of the total supply of dwelling units, including owner-occupied and vacant units, is available only for 1940. Tables 50 and 51 reflect conditions in that year. More current data from the November 1945 survey showing only the tenant-occupied segment of the supply indicate that substantial declines have occurred in the number and proportion of units renting for under $25 per month. A comparison of the rent distributions of tenant-occupied units in 1940 and 1945 is presented in detail in chapter IV of this Part Two entitled "Characteristics of Occupied Dwelling Units."

The rent data shown herein reflect the contract rent reported for tenant-occupied dwelling units and the estimated rent for all other units. Even

Not reporting year built...

Percent not reporting

Reporting year built.

1859 or earlier.

8.0

364, 763 7.8

191, 542 7.2

89, 183 6.3

1860-79.

1, 106, 472

57, 028 57,575 25, 368

15, 803

9,415 6,946

10,899

17,952

110,340

104, 472

39, 675

20, 872

12,367

7,997

61, 126

191, 530

281, 448

275, 304

225, 313

1890-99.

10,996

2,803, 530

22, 387

113, 307

1900-09

4,796, 184

265, 068

554, 301

660, 282

687,295

665, 121

601, 640

699, 213

288, 473

132, 824

77, 892

46, 393

47, 584

70, 098

The gross rent is computed by adding utility and fuel charges to the estimated shelter rent after subtracting certain miscellaneous charges for items such as furniture that may have been included in the contract rental.

Of the 29.1 million nonfarm dwelling units for which rent was reported in 1940, almost 15.1 million, or over half had contract or estimated monthly rents of under $25 per month. The age of dwelling units has an appreciable effect on the rent distribution, as is evidenced by table 50. With the exception of the under $5 rental class, the highest median ages of dwelling units are found in the rental value classes of under $30 per month, ranging from 27 to 33 years. Among units with a rental value of $30 per month or more, the median age ranged from 18 to 22 years. Units with a rental value of under $5 have an extremely low median age. This is accounted for to some extent by the presence of make-shift structures which do not have a long life expectancy, as well as a relatively large number of units built between 1935 and 1940 which did not have standard facilities. A special analysis of 11 areas made by the Census Bureau indicates the characteristics of units built between 1935 and 1940.1

Almost

Differences in tenure and rent are also associated with differences in the number of rooms. 40 percent of the tenant-occupied units reported three rooms or less while only 10 percent of the owner-occupied units contained so, few rooms. Rent and Tenure by Number of Rooms

In the cross-tabulation of rent and tenure by number of rooms, about 80 percent of the owner

! Bureau of the Census, Characteristics of Homes Built in 1935 to 1940, for Selected Areas, 1940, H-45, No. 3, August 1945.

occupied one-room units had a rental value of less than $10 per month. Of the tenant-occupied oneroom units somewhat less than 43 percent had a monthly rent of under $10. At the other end of the scale in the $75 and over class, there were found 44 percent of the owner-occupied dwelling units of 11 rooms or more. Tenant-occupied units with the same number of rooms in that class, however, constituted only 30 percent of all units of 11 rooms or more. If a medium-sized group is analyzed it appears that the disparities between the owner and tenant distributions narrow down. Thus, 10 percent of the five-room owner-occupied units had rental values of under $10 and 8 percent of the tenant-occupied units were of similar size and rent class. In both owner and tenant groups five-room units in the $75 and over rental value class represented about 1 percent of all units in that size class. The median rental value class was $20-$29 for both owner- and tenant-occupied dwelling units of five rooms. Therefore it appears

that in moderate-sized units owner- and tenantoccupied units tend to resemble each other in their rental value distribution while the smallest owner-occupied units show a heavier concentration in the lowest rental value class than do tenantoccupied units. Conversely, the largest owneroccupied units show a heavier concentration in the highest rental value class than do tenant-occupied units.

These data have implications important for builders and investors since they indicate what the effect of market forces has been in determining relationship between rental and sales units and the size unit associated most frequently with given rental value and tenure characteristics.

748221-48-6

63

The level of vacancies is one of the most significant factors for measuring the balance or lack of balance between the supply and demand for housing. Yet, until the census of 1940, no comprehensive picture of the vacancy situation by localities was available. A few localities it is true, had conducted vacancy surveys regularly as far back as the middle 1920's. In others, sporadic studies had been made using the facilities of the post offices, chambers of commerce, and local realestate boards.1

While these surveys had provided a great mass of valuable data, the sponsors of such projects were often handicapped by a lack of funds for undertaking complete and accurate enumerations. The surveys also lacked the benefit of any central clearing house for statistical techniques, methods and definitions. As a result it is often difficult to make intercity comparisons and in many cases even to establish trends over periods of time within a given city. Real property inventory surveys which were undertaken during the 1930's yielded good vacancy data. Intercity comparisons were difficult to make, however, because of differences in the timing of the surveys. From the Census of Housing in 1940, it was possible for the first time to get for the same date comparable information on the vacancy rate not only for the country as a whole but for regions, States, counties, cities, and in some instances, even for specific blocks and tracts within cities.

Sample Surveys Collect Vacancy Data

When the defense program began to bring about a heavy influx of workers and their families into centers of population, the need arose for current figures on available vacancies as a means of gauging the magnitude of the housing shortage in population-swollen areas. At the request of the Defense Housing Coordinator, therefore, a program of sample surveys in specific problem areas was begun

1 For information on local vacancy surveys conducted during the 1930's see the article in the Survey of Current Business, Urban Residential Vacancies 1930-38, August 1938 by Stuart B. Barber. Also see the report Residential Vacancy Surveys 1928-39, Department of Commerce, May 1940, by Stuart B. Barber.

• Rounded figures have not been forced to add to totals.

1 Because of changes in enumeration practice, between April 1940 and November 1945, the components of the habitable vacancies for the two periods are not strictly comparable. Thus, the treatment of units rented or sold but not yet occupied and units held off the market for various reasons was not identical in 1940 and 1945; some units which would have been classified as for rent or sale in 1940 would not be so classified in 1945.

Data for November 1945 are based on the sample survey conducted in November 1945 in connection with the Monthly Report on the Labor Force. Source: Bureau of the Census.

picture of the number of vacancies and of the intensity of utilization of occupied units. It was also decided that data on additional characteristics of vacant units would be highly useful in determining the effective supply of vacancies. Therefore, in recent years, in addition to the regular vacancy information, data have been secured on type of structure, age of structure, conversions, and duration of vacancy for vacant dwelling units. All told, more than 1,200 such sample surveys have been made since the summer of 1940 first by the Work Projects Administration and subsequently by the Bureau of the Census and the Bureau of Labor Statistics. In addition to these local surveys, three national vacancy surveys have been conducted since 1940 on a sample basis in conjunction with the Monthly Report on the Labor Force to provide benchmarks and frames of reference to which individual area figures could be related. The first of these was made in October 1944. Subsequent studies were made in November 1945 and April 1947.

Procedure Used in Making Surveys

Sample surveys were undertaken instead of total enumerations in order to secure the necessary information as rapidly as possible at a minimum of cost. The information was collected by means of a canvass of a carefully selected sample of blocks or segments in the survey area. The information was obtained by trained enumerators, directly from the occupants themselves for the occupied units and from the owner or responsible neighbors for the vacant units.

Both the Bureau of the Census and the Bureau of Labor Statistics used about the same survey procedures in providing housing information for the National Housing Agency. Since the vast majority of the vacancy and occupancy surveys have been made by the Census Bureau, however,

only that agency's procedure will be described in detail.

The first step was to determine the sample design and the size of the sample to produce the desired degree of accuracy with a minimum of cost. Only those sample designs were considered which made possible the measurement of the amount of sampling variability. In the case of vacancy surveys, the sample was designed to provide gross vacancy figures which, in two cases out of three, would differ from those in a complete census by less than 10 percent of the vacancy rate or 0.2 percent, whichever was the greater.

In designing the sample, several factors peculiar to the particular locality were taken into account. Most important of these were the available maps, the possible concentrations of vacancies in a few blocks and, if requested, the requirements to provide separate data by race. Sanborn maps 2 were used whenever they were available, for all surveys conducted in June 1946 and after. Prior to June 1946 the most common procedure used was as follows: The local field supervisor obtained a large scale up-to-date block map of the area to be surveyed. He stratified the survey area on the map into subareas as homogeneous as possible using wards or election districts as a first approximation, and then subdivided these into smaller areas in which the residents had similar racial, social, or economic characteristics. All of the blocks within the survey area were then numbered consecutively, the blocks being numbered completely within one subarea before proceeding to the next subarea. A random selection of the numbered blocks was made for the sample. To take care of the possible concentration of vacant units in blocks having large apartment houses, in blocks containing new con

1 Sanborn maps are large-scale maps showing considerable detail, including the location of each structure. They are issued and kept up-to-date by the Sanborn Map Co., for most cities over 25,000 population, and for some smaller

ones.