32 Operating Costs in Public Housing ear relationship of cost per unit to input prices is appropriate. Strong substitution possibilities might make a non-linear relationship-for example, a logarithmic relationship-more appropriate. The cost of providing a unit of housing service may also depend on characteristics of the tenants who are receiving the services. If some groups of tenants are harder on the stock of housing than others, causing larger maintenance outlays, then an increase in the proportion of tenants in these groups ought to raise costs. An additional reason for including tenant characteristics in the analysis is that available operating cost statistics do not measure costs per unit of housing service, but rather costs per dwelling unit. The two measures differ whenever the flow of housing services is not proportional to the number of dwelling units in the stock. It seems likely that a higher number of persons per dwelling unit increases the demand for services per unit and thereby raises costs per dwelling unit. Other tenant characteristics could also affect the amount of housing services provided per dwelling unit. The relation of costs to structural characteristics of the units also requires some elaboration. Age can be expected to increase costs because many of the subsystems of which a housing unit is made up-plumbing, wiring, flooring, and so forth-take more repair or require replacement as they get older, if the unit is to be kept in stable physical condition. If each subsystem has a predictable "cycle of deterioration and replacement," costs ought to (a) rise with age until age equals the largest replacement cycle of any subsystem, and then (b) flatten out as age enters the range in which each subsystem repeats over and over again the cycle of deterioration and replacement. On the other hand, if the unit as a whole functions less and less reliably even with regular replacement of subsystems, costs might rise with age continuously. Both possibilities are kept in mind in the statistical testing that follows. The effect of number of dwelling units on costs per unit (either per dwelling unit or per unit of housing service) is a question of economies of scale. If there are economies of scale, then Costs per unit ought to decline with number of units under man The Cost of Public Housing Operation agement. If there are diseconomies of large scale, costs per unit ought to rise with number of units. There seems to be no reason to expect one or another tendency to predominate; in fact, it is possible that economies of scale predominate in some size ranges and diseconomies in others. To sum up: viewing the provision of housing services as a production process, costs per unit of housing service can be expected to depend on input costs, age of units, number of units, and certain tenant characteristics. Costs per dwelling unit (rather than per unit of housing service) ought to depend on these same variables and on any tenant characteristics which might cause variations in housing services per unit. 33 There are, of course, many other cost influences not covered in the list above. They will show up as errors or residuals in the statistical work below. Causes of these errors include variations in the degree of efficiency with which different housing authorities operate. They also include variations in the amount of housing services per unit which housing authorities provide. An authority may have operating costs per unit which seem exceptionally high in view of its local input costs, age, or number of minors per unit. This could be either because it operates inefficiently or because it provides extra services for the tenants. THE VARIABLES AND SPECIFICATION OF EQUATIONS Variables. The focus here is on the total operating costs per housing unit in the 23 housing authorities for the four years 1965 through 1968. Besides total costs per unit, four important subgroups of costs have been analyzed, so that there are five cost variables in all: Ctot: Cadm: Cutil: Total operating costs per housing unit for each Administrative costs per housing unit for each Utility costs per housing unit for each fiscal year. Crmain: Routine maintenance costs per housing unit for each fiscal year. 32 Operating Costs in Public Housing ear relationship of cost per unit to input prices is appropriate. Strong substitution possibilities might make a non-linear relationship-for example, a logarithmic relationship-more appropriate. The cost of providing a unit of housing service may also depend on characteristics of the tenants who are receiving the services. If some groups of tenants are harder on the stock of housing than others, causing larger maintenance outlays, then an increase in the proportion of tenants in these groups ought to raise costs. An additional reason for including tenant characteristics in the analysis is that available operating cost statistics do not measure costs per unit of housing service, but rather costs per dwelling unit. The two measures differ whenever the flow of housing services is not proportional to the number of dwelling units in the stock. It seems likely that a higher number of persons per dwelling unit increases the demand for services per unit and thereby raises costs per dwelling unit. Other tenant characteristics could also affect the amount of housing services provided per dwelling unit. The relation of costs to structural characteristics of the units also requires some elaboration. Age can be expected to increase costs because many of the subsystems of which a housing unit is made up-plumbing, wiring, flooring, and so forth-take more repair or require replacement as they get older, if the unit is to be kept in stable physical condition. If each subsystem has a predictable "cycle of deterioration and replacement," costs ought to (a) rise with age until age equals the largest replacement cycle of any subsystem, and then (b) flatten out as age enters the range in which each subsystem repeats over and over again the cycle of deterioration and replacement. On the other hand, if the unit as a whole functions less and less reliably even with regular replacement of subsystems, costs might rise with age continuously. Both possibilities are kept in mind in the statistical testing that follows. The effect of number of dwelling units on costs per unit (either per dwelling unit or per unit of housing service) is a question of economies of scale. If there are economies of scale, then costs per unit ought to decline with number of units under man The Cost of Public Housing Operation agement. If there are diseconomies of large scale, costs per unit ought to rise with number of units. There seems to be no reason to expect one or another tendency to predominate; in fact, it is possible that economies of scale predominate in some size ranges and diseconomies in others. To sum up: viewing the provision of housing services as a production process, costs per unit of housing service can be expected to depend on input costs, age of units, number of units, and certain tenant characteristics. Costs per dwelling unit (rather than per unit of housing service) ought to depend on these same variables and on any tenant characteristics which might cause variations in housing services per unit. 33 There are, of course, many other cost influences not covered in the list above. They will show up as errors or residuals in the statistical work below. Causes of these errors include variations in the degree of efficiency with which different housing authorities operate. They also include variations in the amount of housing services per unit which housing authorities provide. An authority may have operating costs per unit which seem exceptionally high in view of its local input costs, age, or number of minors per unit. This could be either because it operates inefficiently or because it provides extra services for the tenants. THE VARIABLES AND SPECIFICATION OF EQUATIONS Variables. The focus here is on the total operating costs per housing unit in the 23 housing authorities for the four years 1965 through 1968. Besides total costs per unit, four important subgroups of costs have been analyzed, so that there are five cost variables in all: Ctot: Cadm: Cutil: Total operating costs per housing unit for each Administrative costs per housing unit for each Utility costs per housing unit for each fiscal year. Crmain: Routine maintenance costs per housing unit for each fiscal year. 48-279 O 70 pt. 2 11 34 Operating Costs in Public Housing Cxmain: Extraordinary maintenance costs per housing unit for each fiscal year. The unit of measurement for all of these variables is dollars per unit per month. One group of variables to which costs are related are local prices and wages. Four local price variables have been tested and two local wage variables, but the results presented below involve only two of the price variables and one wage variable. These are: Ptot: Putil: Wgov: The cost of living, in dollars, for a "low-budget" The cost of utilities, in dollars, for a "low-budget" The average earnings, in dollars, of a full-time The first two variables were calculated by multiplying the BLS city A second group of variables to which costs are related are characteristics of the stock managed by each local housing authority. Of three variables measured, the following were used in the analysis: Units: Age: Number of units under management. To calculate the second variable, expected to have a positive rela |