Why dwelling units per acre is not a good way to limit apartment buildings

I realize that zoning laws traditionally limit density based on the count of dwelling units per acre.  But treating all dwelling units alike strikes me as a wrong-headed way to go about this, for the simple reason that smaller dwelling units will typically have fewer people living there.

Counting a four-floor 3200-square foot townhouse the same as a 700-square-foot apartment is not really sensible.  Most likely, you’ll get a family in one, and a single individual in the other.  Traffic impacts, sewer and water needs, and the like probably scale better with the number of persons, not the number of units.

Worse, limits based on dwelling units can create perverse incentives for the builders.  They can actually lead to MORE persons per acre, not fewer.  Let me illustrate how.

Suppose for a moment that every 800-square-foot one-bedroom apartment has one occupant, and every 1200-square-foot two-bedroom apartment has two occupants.  So, please note that the two-bedroom has more people per square foot.  In terms of population density, it is the more dense dwelling.  So:

Suppose a proposed apartment building exceeds the density limit (defined by dwelling units) with a mix of one-bedroom and two-bedroom units.  The builder can get that building under the limit merely by shifting to two-bedroom units.  The size of the building is the same, the dwelling units per acre falls — but the population density (persons/acre) actually increases.

In other words, if the builder is building a mix of units, and larger units have higher population density, then forcing the builder to shift toward larger units actually brings in more people than would have been the case otherwise.

The obvious way to fix this is to weight the units by likely population density, or similar.  In the example above, if the law were stated in terms of expected residents per acre, and each unit was counted based on the expected number of residents, the builder would be indifferent between the mix of one- and two-bedroom units.

Once you take this into account, you can see that the proposed Tequila Grande/444 Maple West development has 2.6 times as many dwelling units per acre as the Marco Polo/Vienna Market development.  But it will have only about 1.5 times as may persons per acre.  That’s because the dwelling units at Marco Polo are much larger, and so are more likely to have more occupants per unit.  (For this projection, I used the Town of Vienna average occupants per dwelling for Marco Polo, and the builder’s estimated occupants per dwelling for Tequila Grande).