Forecasting Capital Formation in Residential Housing

Abstract

_HIs paper focuses attention upon one of the subdivisions of gross private domestic investment, the money value of new private residential construction. One of the salient characteristics of all the segments of gross private domestic investment is a tendency toward wide amplitude of fluctuation, particularly as compared with personal consumption expenditures. The construction of residential housing conforms closely to this pattern of fluctuation. As an industry it is characterized by extreme instability. It thus is a significant factor in the instability of gross private domestic investment and of the economy as a whole. Some analysts have found evidence of considerable periodicity in the data on housing construction. The record seems to indicate the existence of a short cycle, averaging about four years in length, and a major, or long, cycle of from fifteen to twenty years' duration. Between 1890 and 1950 either three or four long cycles can be discerned, depending on whether a relatively low peak in 1941 is interpreted as breaking the most recent cycle into two separate cycles. Definite peaks are found in 1909, 1926, 1941, and 1950. At this writing, 1950 is a peak, although the descent in the intervening years has not been steep. It would seem reasonable to expect, on the basis of factors specified later as being of importance in forecasting housing expenditures, that the 1950 level will not be surpassed in the near future. Although these cyclical swings and the current phase of the cycle cannot be ignored, it should be remembered that the data on past cycles show such wide dispersions around the averages that cyclical patterns cannot be followed blindly. The wide fluctuations in housing starts since 1920 are shown in Figure 1. The peaks of 1925, 1941, and 1950 are clearly evident. The number of new-housing starts increased during the early 1920's to a peak of 937,000 in 1925 and then fell off to only 93,000 in 1933. Expansion occurred in each subsequent year to 1941, after which construction was curtailed until 1945. Following the war, the industry increased its output rapidly, surpassed the 1925 level for the first time in 1949, and started a record number of 1,396,000 new units in 1950. The immediate problem of the forecaster is to evaluate the factors that have been responsible for the construction of more than 7,000,000 new nonfarm dwelling units since the end of 1945, to measure what strength remains in these forces, and thus to estimate the number of houses likely to be built in the coming months.