Saturday, March 21, 2015

Single Family Zoning: It's All About the Lot Sizes

Contemporary critiques of zoning take several forms.  First, and probably most common, is the critique of strict use segregation in the Euclidean manner, with commercial and residential areas segregated to greater or lesser degrees.  An increasing focus has lately been placed on the relative preponderance of single-family detached residential zoning within urbanized areas.  A third line of critique, which has received somewhat less attention although it has been the subject of numerous academic studies over the years, looks at minimum lot size requirements within both single-family and multifamily zones.

Although it may seem like a minor subject next to the first two critiques mentioned, the second critique is incomplete without examining permitted lot sizes.  For instance, even though a city may have substantial areas set aside for multifamily housing, if the minimum lot size per apartment unit (or floor area ratio equivalent) is approximately equivalent to single-family zones, the density and/or affordability difference may be minimal.  Alternatively, the single-family detached ("SFD") zoning designation, by itself, tells us little about the density of the area.  Depending on the lot area required and right-of-way widths, SFD densities for a given household size can range from 300/square mile to as much as 25,000/square mile or more, all without the need for any party walls.  

A few concrete examples can help illustrate the point.  In the American context, despite the presumed consumer preference for SFD homes, neighborhoods of homes on very small lots and using narrow rights-of-way, such as are abundant in many other countries, are quite rare.  In the entire New York metropolitan area, for instance, there are only a handful of such neighborhoods, mostly those intended as beachside retreats along the southern shore of Brooklyn and on the Rockaway Peninsula.  One such neighborhood is shown here:

This is Gerritsen Beach, in Brooklyn, showing parcels of 1,800 square feet on a right-of-way (property line to property line) of 28'.  The 2010 Census gives a population density of over 17,000/square mile, about half of the average for Brooklyn, although this may be affected by the presence of second homes.  As Nathan Lewis has shown, similarly scaled neighborhoods in Tokyo attain densities exceeding 30,000/square mile.  Even so, it is about 80% denser than the densest SFD suburbs built in Nassau County during the 1940s-1970s.  It is more than twice as dense as Levittown itself, and denser even than the three-decker neighborhoods of cities like Worcester, MA or New Haven, CT.  The New York metro area has a conspicuous absence of neighborhoods of this density and type, with Census tracts swiftly falling off from around 25,000/square mile to 10,000/square mile.

Beach neighborhoods seem to be popular settings for this type of design, as a similar (but slightly lower) density is present in Long Beach, California, shown below.  Lots are around 2,400 square feet with a 40' right of way and alleys of 13' (comparable density appears to be achieved through presence of small apartment buildings):  

And here is Levittown itself, with lot sizes of around 6,000 square feet and right-of-way (between outer sidewalk edges) of 50'.  Population density is approximately 7,500/sq. mi.:  

The Levittown homes in their initial form, intended for the large families of the Baby Boom era, were actually somewhat smaller (800 sq. ft.) than the 1920s vacation cottages of Gerritsen Beach (1,000 sq. ft.).  The contrast of very small houses with lots the same size or larger larger than those typical of the pre-1940 period seems to have been characteristic of the time period, perhaps related to the increased cost of labor following the Great Depression combined with stable land values.  The more spacious lots may have offered some compensation for the modest interiors.

At the extreme is the ultra-low density of interior but non-rural New England, showing 2- and 3-acre lot zoning (here, in Easton, Connecticut), accompanied by wetlands regulations, with population density of around 260/square mile:

Now, if a rapidly urbanizing country wished to offer the possibility of ownership of a SFD home to the greatest possible portion of its population, the obvious policy goal would be to allow construction of a SFD home on any size lot desired, and on streets as narrow as possible so as to minimize economic waste.  The result is very much what we see in most of the wards of Tokyo as well as in Gerritsen Beach: very small lots but with practical (square-shaped) dimensions along narrow streets and no alleys.  Where multifamily buildings are not prohibited, these will be interspersed here and there, sometimes occupying two or three lots.

In practice, however, American zoning does not take this approach, with the sole exception of mobile home parks (which I've discussed here).  Not only are lot sizes strictly regulated for SFD zones, generally to standards far in excess of what is needed for a comfortably-sized home, other policy measures are in place that make small lots difficult to build as a matter of economics, such as:
  • Apparent aesthetic preference for large, widely space multifamily over densely packed single-family in multifamily zones.  Although small multifamily buildings have fallen out of zoning fashion, if they were ever popular, those zones that survive have some curious features.  For instance, one Connecticut town's code, typical of the type, effectively grants density bonuses for building duplexes and triplexes over building two or three standalone houses on the same lot.  In the densest zone it is possible to build a single-family home on 5,000 sq. ft. -- the smallest SFD lot permitted in any zone -- but a duplex requires only 7,500 sq. ft. and a triplex only 9,000.  The policy intent here is not obvious, but implies aesthetic favoritism for retaining large dwellings with generous spacing even at the expense of the single-family ideal.  
  • Minimum street widths.  Although these do not affect the lot size directly, mandated wide streets make small lots for SFD a less economical proposition.  For instance, were the Gerritsen Beach lots placed along the Levittown right-of-way, each 1800 sq. ft. lot would look out onto 3,000 sq. ft. of sidewalk and roadway!  The neighborhood, overall, would have a ratio of 1.2:1 of private land to right-of-way, or, in other words, only 55% of the land would be in private lots for sale.  Levittown's ratio is a far better 3:1.  With a 28' right-of-way, Gerritsen Beach's streets are already about as wide as economically possible in light of high land values, and streets of Japanese dimensions would yield a far better ratio.
The design differences produce some noticeable differences even at the very large scale.  The prefecture of Tokyo, with population 13.2 million, has a homeownership rate of 45%.  The city of New York, with population 8.4 million, has homeownership of 32%.  Although the prefecture's density is lower than New York's (16,000 vs. 28,000/sq. mi.), Tokyo's outer suburban areas are so much denser than New York's that its metropolitan area density is higher.  For the entire metro area, Tokyo maintains a 56% homeownership rate with 47% share of detached houses, as compared to New York with 52% homeownership and 36% SFD share.  Los Angeles, to pick another example, has 49% homeownership with 50% SFD stock.

A metro area of more comparable size to New York's, such as Kyoto-Osaka-Kobe, has a homeownership rate of 58%, higher than the entire state of New York, and not much lower than far smaller American cities commonly associated with low-density SFD housing such as the Houston MSA (60%), Dallas (61%), Atlanta (63%) and Phoenix (63%).

Interestingly, the issue of minimum lot sizes appears to be one on which there is general agreement between the Smart Growth faction and the defenders of suburbia, as Wendell Cox wrote some years ago in response to critique from the Brookings Institute:
"I was even more surprised at the claim that I defend 'anti-density zoning and other forms of large lot protectionism.' Not so. 
Indeed, I agree with [Jonathan] Rothwell on the problems with large lot zoning. However, it is a stretch to suggest, as he does, that the prevalence of detached housing results from large lot zoning. This is particularly true in places like Southern California where lots have historically been small and whose overall density is far higher than that of greater New York, Boston, Seattle and double that of the planning mecca of Portland."
I think that Cox is correct here, but not even as correct as he could be.  Large-lot zoning not only does not cause a prevalence of SFD housing, it limits it, as shown in the New York metro area.  Los Angeles' metropolitan density is substantially higher than New York's (a point often raised to incredulous reaction), but its share of SFD stock is much higher.  To point out in response to an allegation that a city is heavily zoned for SFD that a majority of its housing units are in multifamily structures (in the case of Seattle, at the link) only underscores the point.

Although most studies have found that minimum lot sizes do affect overall development density (that is, developers appear to build at or near the minimum lot size allowed), one study of Maryland suburbs of Washington, D.C. found that, except in the case of areas with very large minimum lots sizes, developers were subdividing into lots larger than the mandated minimum, including in cases where planned unit development options left them with a free hand to build denser than the zoning code superficially allowed:
"We then examine the extent to which lot size is being constrained by regulation by comparing actual subdivision density to the allowable density under zoning rules. This analysis is done for three counties with different degrees of suburbanization. We find that only in the areas with the very large lot zoning does zoning seem to be constraining actual lot[] size. There is a good deal of excess capacity in the density that could be built, especially in the more densely zoned areas." Lot Size, Zoning, and Household Preferences:Impediments to Smart Growth?
There might be no concern with low-density development at the fringes of metro areas except that such areas, once built up, are politically almost impervious to change.  Some cities have implemented maximum lot sizes, but these are generally very generous and only apply in certain areas.  The misunderstood achievement of the New Urbanism, and the Maryland suburb of Kentlands in particular -- whatever its faults in design detail -- was that small lot single-family housing developments not only need not be qualitatively inferior to the 6,000+ sq. ft. tracts that characterized the post-1940 suburban era, and that they could even offer amenities that made them superior residential environments.  Where regulatory mandates fail, leading by example can succeed.

Kentlands, with SFD lots from 5,600 to 2,600 sq. ft, and with ROWs ranging
from 45 ft. to 12 ft. on alleys.
Related posts: None of mine, but Nathan has written several great pieces on the topic which you can find at his website.

Sunday, March 8, 2015

Auto Costs and Housing Costs, or, One Reason the Suburbs are So Appealing

Simon Vallee has a post from some time back about filtering vs. gentrification in which he analogizes the process of gentrification, in North America, to the car market in Cuba, noting that restrictions on supply will tend to boost prices and limit availability of a desired good.  Although the comparison is intended to be illustrative, I think it also highlights a substantive difference which, in effect, subsidizes automobiles at the expense of housing.  First, though, some background.

Going back for a moment to the subject matter of a Nathan Lewis post, we can note that, land costs aside, the sticker price of manufactured housing as compared to a new vehicle is not as different as one might think.  A two-bedroom manufactured home, for instance, of about the size of the average new home of the 1950s, costs only around $41,500 as compared to the price of a popular new sedan (I chose the Altima, one of the best-selling cars in the United States) at around $27,000:

However, when car costs are compared to overall home values, including site-built as well as manufactured homes, a different story emerges.  In 1940, the median home was valued at only 2.3 times the retail price of the average new car.  By 2010, in spite of the crash in home prices, this ratio had risen to 6.4.  Car operating costs have also generally fallen as fuel efficiency and vehicle reliability have improved.  Median rents, not shown here, have grown at an even faster rate than home values.

In short, over the last seven or so decades, car ownership (or leasing) has become dramatically less expensive relative to home ownership or tenancy.  Partly this must be due to labor-saving technologies that have affected car production more than homebuilding: even manufactured homes still require extensive human labor, which has become much costlier (though more productive) since 1940, whereas the formerly labor-intensive car assembly process has been heavily automated and accelerated.  The process of robotically assembling houses, or even apartment buildings, remains in its infancy.

Are long-term, over-inflation increases in home values also linked to increases in land values caused by general urban population increase and restrictive zoning?  It goes without saying that rural land values are lower than urban land values, and the Census homeownership figures show that housing values are lower, and homeownership higher, in more rural states in spite of lower incomes.  Relatively poor and rural West Virginia has the nation's highest homeownership rate, while 100% urbanized Washington D.C. has had its lowest in every Census since 1930.  As Luis Bettencourt writes:
"There are several important consequences for general land use in cities. First, the price of land rises faster with population size than average incomes. This is the result of per capita increases in both density and economic productivity, so that money spent per unit area and unit time, i.e. land rents, increases on average by 50% with every doubling of city population size! It is this rise in the price of land that mediates, indirectly, many of the spontaneous solutions that reduce per capita energy use and Carbon emissions in larger cities. Cars become expensive to park, and taller buildings become necessary to keep the price of floor space in pace with incomes, thus leading to smaller surface area to volume."   The Kind of Problem a City Is.
Urbanization in the era of the automobile in turn causes frictions which lead to pressure for zoning.  American municipal zoning, in its initial formulation and as is still practiced today, is fundamentally a device to politically manage these frictions by restricting the intensity of residential land use.  Though not its stated purpose, it has the effect of increasing land scarcity that is already inherent in the urbanization process, and thereby provides a positive feedback mechanism that puts additional pressure on housing values.

What does this all have to do with cars?  As noted above, the cost of a manufactured home, in isolation, is only slightly more than that of a typical sedan.  As urbanization increases, however, the increasing value of land makes cars, which do not have their land storage cost bundled into the sticker price (unlike Japan does, effectively), seem like a relative bargain.  Some time ago, Cap'n Transit wrote a fascinating series on how New York came to tolerate and eventually permit free overnight on-street parking in the late 1940s and early 1950s.  We would find it ludicrous if someone were to purchase a manufactured home and to drive it into Manhattan on a flatbed expecting the city to provide free land on which to site it, but that was how the story went with cars:
"This [middle-class] conception of the benefits of car ownership has always had a huge bait-and-switch component to it. In New York City in the 1940s it was no exception. When people looked at the price of a car, they didn't figure in $20-35 per month in garage rental. When they got their cars, many couldn't afford to pay and took their chances on the street. Garage owners now had to compete with free street parking and lowered their rates accordingly, which meant that they didn't have enough income to expand their facilities, and resorted to bribing the police. 
"These social-climbing drivers felt cheated, but they didn't take their anger out on the car dealers. No, they felt that the city owed them the free parking necessary to make their cars as affordable as they thought." The right to free parking in 1940s New York
There are therefore two clashing trends: as cities grow in size, the cost of a buying a car declines relative to increases in income and housing cost, yet the actual cost of storing a vehicle is, or should be, increasing rapidly, since cars, like houses, occupy a significant amount of valuable space.  Rather than taking the common-sense Japanese approach of the shako shomeisho (proof of parking), however, American states and cities have engaged in onerous mandatory inclusionary zoning for cars (parking minimums), zoning exemptions (e.g. not counting garages toward FAR limits and allowing parking, but not housing, in mandated setbacks), tax exemptions (only 16 states maintain a personal property tax that covers automobiles) and fringe benefits (the commuter parking benefit), in addition to rent-free public housing for cars (overnight on-street parking).  Whereas in 1940, buying and operating a car to escape urban housing costs simply shifted the balance of expenses, with a car costing almost half as much as the median home, in 2010 the prospect of doing so was much more economically feasible.  No doubt many of those New Yorkers of the 1940s and 1950s eventually drove those cars out of their subsidized parking spaces and off to the far reaches of Nassau, Bergen and Westchester Counties, and who could blame them?

Perhaps the biggest effect of all though, going back to the beginning of the post, relates to the obvious but important point that while housing production, and particularly in-city housing production, is subject to political constraints, car production is not (well, mostly not).  There even seems to be a difference in Americans' moral characterization of those who build homes and cars for profit: while a search for the phrase "greedy developers" returns over 60,000 hits, "greedy automakers" returns only 1,000.  From that perspective, the so-called "drive 'til you qualify" phenomenon, much questioned and criticized, is an entirely reasonable response to this economic reality, particularly given widespread lack of highway tolls. 

Making a full accounting of the political choices that have been made with regard to both housing and transportation is a daunting task, but it does help illuminate the residential patterns we see without the need to resort to moral judgments about those choices.

Related posts: Was the Rise of Car Ownership Responsible for the Midcentury Homeownership Boom in the US?