widow, widower, or divorced person with children will usually give up separate housing accommodations only under strong economic compulsion. An increasing number of aged persons appear to prefer to maintain separate accommodations after the death of the spouse, and this is particularly true where the accommodation is suited to the need of aged persons. Old-age and survivors insurance, private pension plans, aid for dependent children, and veterans benefits for widows and orphans, are steadily raising the ability of these classes of families to maintain separate households.

Current estimates of future family formation range from as low as 275,000 per year for the remaining years of this decade to the 2 million per year estimate of the National Association of Homes Builders for the later years of the next decade.' The marriage rate has fluctuated from under 1 million to over 2.2 million in the last 20 years. The range of net new social family formation shown is from 6 million to 10 million per decade. This does not include single-person families which are discussed below in connection with undoubling. If these were to be included in the estimate, the high estimate should be increased. Migration

The migration of people from one area to another may result in increased housing requirements where in-migration is not balanced by out-migration. Those moving into the country from other countries, and those moving from areas of stable or declining population to areas of rapidly growing population create such needs. From 25 to 30 million persons move annually, and about a third of these, 8 to 10 millions, move across county lines. There is a steady movement of persons into the country, and of the rural or farm population to urban areas. Of the 10 million persons who moved across county lines in 1950, 9 million were members of families who presumably created demands for about 3 million dwellings. Since a considerable part of this movement is to urban communities in the South and West where it results in net in-migration, it must be presumed that at least between 10 and 25 percent of this movement creates new housing requirements.

TABLE 1.-Factors affecting nonfarm housing requirements per decade1

1 See text for basis. Low may be low estimate or early in the 1955-65 period. High may be high estimate or late in 1955-60 period.

United States Ce sus Series P-20, No. 42, 1952, high. The census low of 2.75 could arise only under conditions of disastrous economic depression.

Cf. U. S. Census Series P-20, No. 42, 1952; NAHB Correlator, February 1954, pp. 4, 5. The National Association of Real Estate Boards estimates these at 700,000 per year currently Cf. address of Charles B. Shattuck, president, 46th annual convention, November 10, 1953.

Cf. E. E. Ashley, Mobility and Migration as Factors in Housing Demand, Housing Research, October 1953.

U. S. Census Series P-20, No. 39, 1952.

Replacement of current losses

The National Housing Agency and the Housing and Home Finance Agency estimate that 40,000 homes are lost each year as a result of fire, windstorm and other types of demolition. To these annual losses there is usually added 300,000 temporary war and veterans units required to be removed during the current decade if only by their physical deterioration. Thus it is customary to estimate 70,000 units per year as current losses requiring replacement.

These estimates require radical revision if only in the light of current urban highway construction. A modern urban superhighway built through developed areas will require clearance of from 250 to as high as 1,000 homes per lineal mile. If each metropolitan area built only 1 mile of such highway each year, and if the demolition requirements in the 5 largest cities were at the upper limit and all others at the lower limit suggested above, the total would be in excess of 40,000 homes per year. In some of our largest cities, demolitions resulting from highway construction have greatly exceeded demolitions from all other causes throughout the postwar period. With urgent highway needs measured in tens of thousands of miles, and current Federal aid programs including approximately 30,000 miles, it is clear that losses from this source alone will be exceedingly high for the next two decades. For current purposes it is assumed that losses from all causes will be 1,100,000 per decade in the 1950's and 1,600,000 during the 1960's equivalent to previous estimates of loss plus 1 mile per metropolitan area in the former period and 2 miles per area in the latter.

Other losses not accounted for here include abandonments and conversions to nonresidential use. One expert estimates that the latter alone exceed all extra units gained by conversion. One group of Government officials concerned with housing statistics has reached the tentative conclusion that 200,000 units are lost each year through demolition, abandonment, and conversion, and that 100,000 are gained each year through conversion, for a net loss of 100,000 units. This estimate accords with the estimate above arrived at on other grounds.

Obsolescence and deterioration

More than 6 million of our present housing units were built before the beginning of this century, and will be 60 or more years old by the end of this decade. Many of these are now dilapidated and should be replaced immediately, but many of them are merely old, obsolete, and now deteriorating and at a rate reflecting their age. Some 1,400,000 of these units, now standard but becoming 75 years of age or older by 1960, should be replaced during the next decade.

With our present housing stock of over 50 million homes, it will be necessary to replace 500,000 units per year in order to replace homes at 100 years of age. Many hundreds of thousands of fine old homes will doubtless continue to be well maintained, and will retain historical, architectural and other qualities worth preserving. But the speculative homes of the Gothic period, and millions of drab shacks built since 1900 lack these fine qualities. Their useful life as structures, and the useful life of the neighborhoods they comprise, will be long past at 60 years. Applying this standard would necessitate the replacement of 2.8 million homes during the 1960's and 4.8 millions in the succeeding decade." Substandard housing in 1950

Housing standards are not fixed and invariable. A wide range of judgment is involved in determinations of substandardness. What is standard for a primitive economy (mud huts) will be substandard for a more advanced economy. Standards are therefore in part determined by resources. In a society of abundant resources and high output, standards should rise steadily. There is no apparent reason why all American families should not have good homes within the next generation. This was particularly apparent during the 1930's when idle labor and unused materials led to the adoption of Federal aids to housing. The Congress has subsequently adopted a national goal of "a decent home in a suitable living environment for every American family."

Table 2 reveals the most serious inadequacies of our housing supply in 1950. For many years official agencies have used the standards of structural soundness and lack of plumbing facilities as measures of substandardness. Opinion has varied as to whether farm or country homes which lacked running water

10 The NAREB estimate of demolition requirements is 300,000 to 400,000 per year. Cf. Shattuck, op. cit.

should be considered substandard. Under the assumptions of this study, of a rising standard of housing, it is assumed that all families, rural and urban, should have interior plumbing facilities.

According to the most recent census of housing, 10 million nonfarm dwelling units were dilapidated or lacked running water. This number includes farm residences in standard metropolitan areas. In addition to these dilapidated or substandard structures, 1.9 million other units were located in blocks containing more than 50 percent substandard units. These units contain structural and plumbing deficiencies not sufficiently serious to be recorded by the census, but their environmental substandardness is clear. They would almost certainly be demolished in any program of slum clearance."

11

Only 23 percent of our farm housing units meet these urban standards. Thus by census criteria and urban standards, 4.9 million farm homes are substandard. The Department of Agriculture surveys of farm housing suggest, however, that by farm standards, only 3.4 million farm homes are substandard, 2 million of which contain serious deficiencies requiring replacement, the remaining 1.4 million being remediable.12

Not all substandard dwellings need be demolished and replaced. A basically sound structure, lacking running water or a toilet, may be brought up to standard by relatively minor repairs and the installation of plumbing facilities if the structure is located in a sound neighborhood. On the other hand, where the structure is located in a slum, the installation of plumbing or heating facilities may be economically unsound. The rents required to finance the improvements may exceed the levels which renters are prepared to pay in slum areas. While precise data is lacking on the location and character of all housing units which are substandard because of plumbing deficiencies, data are available on 2.7 million urban units which are located in substandard areas. It is reasonable to assume that these must be replaced. The remaining 1.9 million are in relatively scattered locations and may be brought up to standard. It is further assumed that two-thirds of the rural nonfarm units which are substandard because of plumbing deficiencies may be rehabilitated. One-third would then require replacement.

13

Of the 3.4 million substandard farm units, 2 million with serious deficiencies are beyond repair but 0.5 million of these may be abandoned, leaving a 1.5 million replacement goal." It is assumed that all of the remaining substandard units can be rehabilitated. Thus, of the 15.3 million substandard units in 1950, a total of 10.2 million must be replaced, 4.6 million must be rehabilitated and 0.5 million abandoned.

Undoubling

In recent years the number of doubled-up families has been reduced steadily. Nevertheless, over 1.7 million social families were still without separate housing accommodations in 1950. An unknown proportion of the doubled families and single persons prefer to share housing accommodations with others for health or other reasons. For many, however, doubling-up continues because of economic necessity or housing shortage. Under conditions of sustained prosperity and more adequate social security and old age allowances, the number of doubled families should be reduced steadily. It may be assumed that under these conditions, upward of two-thirds of such families might prefer separate accommodations.

In addition to these families, there were in 1950 some 10 million adult single persons not in families of whom about one-third occupied separate households. This number may be expected to increase sharply with sustained prosperity. An important future influence upon single-person families is the growing number of aged persons able to maintain separate accommodations. If a number equivalent to one-tenth of these persons, those with incomes of over $3,500 per year, were to establish separate households-a million additional housing units would be required. Thus the combined undoubling of families and single persons during the next decade might vary from 1.2 million (two-thirds of married doubled, no single) to 2.7 million (all married doubled and one-tenth of single).

Slum-clearance projects to date involve clearance of 20 percent of standard units and 80 percent substandard units. The proportion in the blocks mentioned above is 28 to 72 percent.

14 This report draws upon tabulations of census and Agriculture Department data prepared for a new edition of America's Needs and Resources, by J. Frederick Dewhirst and associates, Twentieth Century Fund, to be published in fall, 1954.

13 Census figures suggest 3.2 substandard farm units of which 2.4 are dilapidated.

Overcrowding

14

In 1950 over 6.6 million families (census households) were living in dwelling units which provided more than 1 person per room.' Over 2 million families were seriously overcrowded with more than 1.5 persons per room.

TABLE 2.-Substandard housing units in the United States requiring replacement or rehabilitation by condition and location, 1950

Farm units outside of SMU's. Deficiencies based on U. S. Department of Agriculture data. $500,000 units abandoned and not replaced.

During the postwar years, several million young couples purchased small twobedroom houses with floor areas far below those considered acceptable in prewar years. The continued high rate of births and the steady rise in the number of second, third, and fourth children born suggests that a large proportion of these owners of too-small houses already are seriously overcrowded as measured by number of persons per room, and that this number will increase.

Because of the overlap between families overcrowded and families living in substandard homes, only a quarter of the number of overcrowded homes are shown with other measures of need in the summary table. The range of estimates is from one-fourth of units containing more than 1.5 persons per room to one-fifth of units containing over 1 person per room. Under the assumptions of this study, a rising standard of housing should not require five-person families to live in two-bedroom homes. The estimates therefore leave three-fourths of the undoubling problem for future decades.

Vacancies

The number of vacancies required to permit freedom of choice in the market, and to allow for mobility has been variously estimated at from 3 to 6 percent of supply. The number of available vacancies in recent years has been only about one-fourth of the nominally vacant units, a large proportion of which are dilapidated, seasonal, or not on the market. To achieve an available vacancy rate of 3 percent it might be necessary to have at least a total of 6 percent vacancies. A further consideration is raised by our annual volume of family movements, involving nearly 28 million persons including about 3 million families in 1950-51.5

14 U. S. Census. Housing Census, vol. I.

15 A 5-percent rate is considered desirable by the realtors; cf. Shattuck, op. cit.

PROGRAMING HOUSING NEEDS

The total housing requirements shown in Table 1 suggest that we must build from 1.3 to 2.4 million units per year to meet our growing housing needs and must in addition replace accumulated deficiencies of the past of from 9 to 16 million units. Clearly all of these needs cannot be met at once, nor would it be economically wise to do so. Some of these needs tend to overlap, i. e., new family formation and undoubling, replacement of substandard houses and provision of homes for overcrowded families. Some orderly basis is required for estimating the changing volume of current housing requirements and for scheduling the replacement of existing substandard units. The first of these requirements will be served by the measure of household formation.

Household formation

Census projections of household formation involve circular reasoning to some extent since household formation is in some degree dependent upon the volume of residential construction and remodeling. Nevertheless, census definitions of households have been sufficiently loose to reflect the improvised housing conditions which families desiring separate accommodations have adopted in the past. Thus, the room or rooms with an electric hotplate and a shared bath and occupied by a man and wife are a household to the census taker. These conditions reflect an active desire for separate dwelling accommodations whether or not that goal has been reached in some more refined sense. The census household thus reflects in part the housing shortage.

Household size has declined steadily for the last generation. The decline was approximately the same in periods of prosperity and depression as is shown in table 3. This decline in average household size reflects both the smaller size of families containing two or more persons, and the larger number of single person families who desire and can afford to maintain separate housing accommodations. It is a measure also of rising standards of housing space and privacy. Continued declines in household size would reflect continued increases in housing standards in this sense.

The total future population of the United States is also shown in table 3. This is the most recent high estimate of the United States Bureau of the Census, and is thus consistent with assumptions of sustained prosperity. It should be noted that census estimates for the last 20 years have generally underestimated future population growth. Absolute and specific birth rates and marriage rates have been higher, and death rates have been lower than those used in even the so-called high projections. Under conditions of sustained economic growth, these estimates may well prove below actual growth.

Total population and average household size have been used to project housing requirements by 5 year periods from 1955 to 1970 in table 4. Using this method, it is first necessary to reduce the total population by the number of persons not in households. An arbitrary percentage, approximately that of 1950, has been used as a constant. A 20 percent increase in this constant would reduce annual construction requirements by 4 percent in 1960. The resulting population in households is divided by a straight line projection of the 1930-50 trend in number of persons per household. The result is increased by 4 percent to allow for necessary vacancies. A note to the table gives alternative estimates based upon a higher average family size and upon the assumption that family size approaches some constant in 1970.

The method suggests the number of housing units which will be required at different future dates and the annual volumes of construction required during successive 5-year periods to achieve this stock. The population used is total population, farm and nonfarm. The housing units shown as required would be necessary to accommodate future population growth, future reductions in family size, increases in the number of single persons desiring separate accommodations, and reductions in the number of nonfamily households. The measure accounts for the undoubling of families now doubled. Finally, as used here, the measure assumes the same rate of progress in the relation between population and housing that obtained during the last 20 years.

The table indicates that we will require approximately 1.43 million new units per year from 1955 to 1960, 1.65 million new units per year from 1960 to 1965 and 1.74 million new units per year from 1965 to 1970 to house our growing population adequately. It should be noted that this measure produces results which are below the medium estimates of need in table 1. The method is used, nevertheless, because it suggests time periods within which houses are needed.