A method for evaluating publicly subsidized housing is presented and applied to New York City's Mitchell-Lama middle-income housing program. Households occupying the 57,000 subsidized dwelling units received net benefits of 25.6 million, or 450 per household, in 1968. The public subsidy required to provide these benefits was 46.9 million, or 824 per unit. Hence, rather substantial nontenant benefits of 21.3 million, or 374 per unit, would have had to be generated by the program for it to be considered an efficient use of resources. Net benefits to participants were found to rise with age of household head and family size, to be lower for female-headed households, and to show little relation to income level. However, net benefits varied considerably for otherwise similar households. Finally, program participants were found to differ little on average from those occupying private, unsubsidized rental housing in New York City in 1968.

One externality which has received little attention from economists is that connected with congestion (a recent treatment is Strotz, 1965). An increase in the utilization of a facility can result in longer waiting time or in a less appealing service. For example, an increased number of tows on a water- way can give rise to greater delays at locks; a larger crowd at a beach can lead to a lower satisfaction for each "customer." The former case, which involves production, is particularly complex. Here, delay is caused either by the random arrival of customers or by a random service (production) rate. Serving tows at a lock is analogous to serving shoppers at a super- market checkout stand. Similar problems occur with respect to allocating docking facilities in a port, runways and terminals in an airport, land to streets in an industrial park, and machines in a job shop