ling" had two aspects. First, the number of married couples and parent-child families without their own households declined by 1.2 million. Second, households were formed by young persons and households were retained by elderly persons who before the war might have gone to live with others. TABLE 1.-Change in number of households, by type, and new units built, 1946–50 The 5.6 million increase in households was some 2.1 million more than the number of new permanent nonfarm dwelling units built, as estimated by the techniques employed until recently. Some of the difference is accounted for by differences in the populations being measured; much more, however, reflects some reduction in vacancies (including use of seasonal quarters the year round), occupancy of temporary quarters made available by public agencies, and the use of trailers. The difference also reflects the informal and frequently temporary conversion, as well as permanent, structural conversion of nonresidential to residential use or of residential structures to house additional households. Some residential units, on the other hand, were demolished or converted to nonresidential uses. There are no satisfactory data representing the whole period for any of these additions to and subtractions from the housing supply. It seems clear, however, that the total number of additions from such sources during the 4-year period must have been substantially larger than the 2.1 million difference between the 5.6 million increase in households and the 3.5 million new permanent dwelling units built. In contrast with the immediate postwar period, between the spring of 1950 and the spring of 1955 the number of new permanent dwelling units built exceeded the number of households formed. The number of households rose 4.3 million while the number of new permanent dwelling units built was 5.9 million, as may be seen in the table. The composition of the increase in households also differed from the first period. The annual rate of increase in married couple households was less than half of that from 1946 to 1950, and more than a third of the total increase represented households headed by women. TABLE 2.-Change in number of households, by type, and new units built, 1950–55 The final period (1955-60) was like the second, as may be seen from the table. The increase in the number of households was slightly larger than in the previous period, reflecting slightly larger increases in married couple and "other male" households, offset in part by a slightly smaller increase in "other female" households. TABLE 3.-Change in number of households, by type and new units built, 1955–60 [In millions] A survey conducted by the Bureau of the Census late in 1956the national housing inventory-provides fuller information on the changes that took place in the housing supply between April 1950 and December 1956. According to the survey, new construction added 10.9 million units rather than the 8.2 million suggested by the Bureau of Labor Statistics (BLS) series then used as a measure of starts. Some of the difference is accounted for by differences in concept, coverage, timing, and similar factors. The major portion of the difference, however, appears to be attributable to a higher level of accuracy in the national housing inventory (NHI). The results of this survey have therefore been used as a benchmark for the new series on starts first published by the Bureau of the Census early in 1960. Both the total number of households in the United States at the end of 1956 and the increase between 1950 and 1956, estimated from the NHI, were larger than estimated by the Bureau of the Census from the current population survey (CPS). The NHI estimate of change in households in this period is above the estimate of change implied by the CPS data by about the same proportion as the NHI estimate of units built is above the BLS estimate roughly 25 percent. The BLS and CPS estimates have been used in the earlier tables in this section. The table that follows presents the data of the NHI. The general picture that emerges for the period after 1950 is much the same as that shown by the earlier figures-namely that additions to the housing supply were larger than the increase in households. In detail, the picture is different. Although the average annual increase in households is about the same from the two sets of data, the NHI shows formation of married couple households to have been more important as a source of the increase than indicated by the earlier data. TABLE 4.-Change in number of households, by type, and additions to the housing supply, 1950-56 1 Households headed by married couple, no nonrelative in household. * Includes some households headed by married couple with nonrelative in household. 10.9 -1.6 1.6 -.2 Furthermore, the change in the housing supply is seen from the NHI to have been substantially different from that suggested by the BLS data. The number of units removed from the supply (counting units lost by conversion and merger as removed) was two-fifths as large as the number of units added by new construction, and the number added from sources other than new construction was one-fourth as large. On balance, losses exceeded additions apart from new construction, so that the net increase in the size of the housing supply was only about 85 percent as large as the number of new dwelling units built. TABLE 5.—Changes in the housing supply, 1950–56 These gross changes are reflected in the composition of the housing supply at the end of 1956. Whereas the 1950 housing supply was 83 percent as large as the 1956, only 75 percent of the 1956 supply represented units that existed in 1950. Units built between 1950 and 1956 accounted for 20 percent of the supply, and miscellaneous sources such as conversion, merger, change of use from transient to permanent accommodations, and use of trailers as housing accommodations, accounted for the remaining 5 percent. TABLE 6.-Relation between housing supply of 1950 and 1956 Until the results of the 1960 census are available, we shall not know whether changes in the housing supply since 1956, apart from new construction and demolition, have been as rapid as in the period 1950-56. In any case, the excess of new units built over the increase in the number of households was about the same between 1955 and 1960 as between 1950 and 1955, and the increase in vacancies in the two periods was also about the same. The market effect of a rental vacancy rate of 71⁄2 percent (in 1960), however, is likely to be different from the effect of a rate of 4% percent (in 1955) or 21⁄2 percent (in 1950). FACTORS CONTRIBUTING TO INSTABILITY Change is typical of economic activity in general as well as of residential construction in particular. It occurs as growth or trend over long periods of years. It is manifested in the broad movements of employment, distribution, and production over business cycles. It shows up even more strongly in monthly variations which reflect seasonal influences. Normal seasonal fluctuations in many lines of activity tend to be larger than either year-to-year trend or cyclical changes, and the year-to-year trend changes are ordinarily smaller than the cyclical. Such a comparison indicates the importance of distinguishing various meanings we may attach to the idea of instability. The statistical magnitude of cyclical changes shown by monthly series will depend, sometimes very largely, on whether or not the series used is adjusted for seasonal variation. If it is not adjusted, the cyclical low may occur at the seasonal low and the cyclical high at the seasonal high, thus in effect leading us to count the seasonal movement as part of the cyclical. If the series is adjusted, the range from cyclical low to cyclical high for monthly data may be only moderately greater than the range from the cyclically low year to the cyclically high year. Similarly, the amount of aggregation reflected in the series is likely to affect the amplitude of movement. For example, the seasonal range of industrial production in total is small compared with the seasonal range of many of the components of industrial production. Since the timing of seasonal highs and lows in the component series tends to be distributed (though not evenly) through the year, the average seasonal movement of the total is smaller than the movement of many of the parts. Degree of aggregation influences the range of year-to-year and of cyclical fluctuations in much the same way. The magnitude of fluctuation does not necessarily provide a significant measure of a problem that must or can be rectified. Much will depend on the technology of the particular industry and on the market in which the activity takes place. In construction, for example, technological changes in the past generation have probably helped reduce the range of seasonal movement; the relative increase in markets for building in the South and West, however, has operated in the same direction. COMPARISON OF POSTWAR CYCLES Market factors in the postwar period have influenced fluctuations in residential construction. Each of the three most recent booms in housing starts appears to have grown out of economic developments in the period immediately before the boom. Each of these periods has been different in important respects, but the general similarities are striking. Similarities The fact has often been noted that each upturn in starts began after a decline in general economic activity had started, but before a general recovery was underway. This has sometimes been accounted for by the idea that residential construction is a countercyclical activity. Examination of the data already presented suggests that the countercyclical influence of residential construction has come as much from the market readjustments that took place in recession as from the continued strength of housing demand. Before or along with each downturn in general economic activity there was a slackening in housing demand that showed up in the market for existing houses or for new houses, or for both. This slackening which is not clearly or uniquely associated with changed credit terms-together with reduced economic activity generally, operated to produce adjustments in several ways: old houses became harder to sell, and prices softened; potential buyers, with more choice of both old and new houses, found new house prices lower or subject. to bargaining; builders, finding houses harder to sell, curtailed starts, bargained harder with suppliers, and made greater efforts to improve efficiency of operation. The resulting reduction in demand for construction materials tended to come at about the same time that demands from other lines declined; thus there was accumulation of inventories and reduction in prices, or a halt to price increases. Thus, at the time when credit demands generally slackened and financing was more readily available, builders were able to offer houses that were less expensive or more attractive in other respects. Such price 66225-61- -3 |