The Supply of Rental Housing: Reply
Abstract
Ronald Grieson comments on two aspects of our paper: the plausibility of its results, and the interpretation of its empirical findings. On empirical findings, we have no strong preference for our interpretation over Grieson's. He shows, in essence, that if the general price level enters the supply function as well as the demand function, then our regression coefficients may imply elasticities of supply somewhat higher than the ones we presented. The elasticities we presented ranged from .3 to .7; the ones he presents range from .4 to 2.2. Both Grieson's interpretation and ours lead to the same principal conclusion; namely, that . . . the long-run supply of housing services is less than perfectly elastic (see de Leeuw and Ekanem, p. 812). Both sets of elasticity estimates suffer from the problems of measurement error which our article discussed. An advantage of Grieson's interpretation is that it implies higher reduced-form coefficients for the price level than for income per household, which fits the data better than our specification. One disadvantage of his interpretation is that there is no clear theoretical basis for including the general price level in the supply function; if the supply function is the result of profit maximization subject to a production function for converting capital and operating inputs into housing services, then the only relevant prices would seem to be those of housing services, capital inputs, and operating inputs (all of which were already included in our specification). Possibly the general price level serves as a proxy for prices of inputs not covered by our measures. But surely the general price level does not belong in the supply function simply because it appears in the demand function, as Grieson seems to argue. On the plausibility of results, we have reservations about Grieson's position. He finds a long-run rising supply price for housing services plausible because of the inelastic supply of land. Both our elasticity estimates and his, however, refer to the elasticity of supply holding constant the price of land as well as prices of other inputs. (See de Leeuw and Ekanem, pp. 808, 810.) Hence, the explanation of our findings cannot be the tendency for the price of land to rise as housing demand shifts upward. We suspect that the explanation of our findings lies in diseconomies of scale in the maintenance, improvement, and conversion of existing housing capital. To understand the connection between these activities and our study it is useful to distinguish between two meanings of in the housing market. The first is a time period long enough for the total quantity of housing capital in a housing market to respond fully to changes in underlying conditions-a period whose length has been estimated to be of the order of magnitude of six years. (See Richard Muth, p. 76.) The second is a time period long enough for not just the quantity, but theform of existing housing capital number of units per building, architectural style, location pattern within a housing market area, and so forth-to respond fully to changes in underlying conditions. The second long run is surely a great deal longer than six years. We believe that the results of our study apply to the first rather than the second of these long runs. In the second long run, at least under competitive conditions, prices per unit of housing service presumably approach new construction prices (plus local land rents) for all types of housing. Since the construction of new housing is a replicable process, we might expect (input prices aside) constant returns to scale to characterize, at least approximately, this long run. In the first long run, however, there could be important quasi rents-positive or nega* The Urban Institute.
- Sources
- openalex