Rehabilitating Walmart

Walmart has been in the news more than usual of late, with the announcement that it was closing 269 of its stores as well as backing out of building two stores in lower-income areas of Washington, D.C. Add this to Walmart's stock price decline of nearly 30% during 2015, and the picture begins to look somewhat bleak.  Although Walmart's financial prospects for the future are an interesting topic, it's the chain's position within urban environments that has been relatively little explored.

To set the context, it's crucial to note that Walmart has become in essence a bloated grocery chain, with 56% of its sales coming from groceries.  The growth in the grocery sector has been achieved in spite of the fact that Walmart earns by far the lowest customer satisfaction rating for its supermarket offerings of any major chain.  Moreover, the grocery business is more localized then the variety store
business. According to one recent study on the market effect of Walmart on the supermarket business:

"We find that Wal-Mart’s impact is highly localized, affecting firms only within a tight, two-mile radius of its location. Within this radius, the bulk of the impact falls on declining firms and mostly on the intensive margin. Entry of new firms is essentially unaffected. Moreover, the stores most damaged by Wal-Mart’s entry are the outlets of larger chains. This suggests that Wal-Mart’s expansion into groceries is quite distinct from its earlier experience in the discount industry, where the primary casualties were small chains and sole proprietorships that were forced to exit the market." 

It seems that, for grocery purchases, most consumers are unwilling to undertake longer journeys than necessary even for somewhat lower prices.  Walmart's use of the "Supercenter" business model, however, means that it is impossible extend those two-mile radii over an entire metro area.  Moreover, the dependency on low land prices for these stores means that Walmart is often absent from entire central areas of cities.

In San Antonio, for example, there is not a single Walmart establishment in the central 50 square miles of the city.  By contrast, local chains like H-E-B and independent groceries have the flexibility to open stores within central areas.  The abandonment of Walmart's "Express" stores with the recent closings seems to have spelled an end for an attempt to compete with these smaller urban groceries.
Further, these supermarket chains have been edging in on Walmart's own territory, with chains like Kroger beginning to stock clothing and a wider range of household items in their most recently opened stores.  The competition from Amazon has been widely reported on as well.

------------------------------

From an urbanist perspective, it's easy to cheer all this news and condemn these stores as destroyers of local business and magnets for the car-dependent, but does it need to be this way?  I'm not so sure.  The person who can find no redeeming qualities in big-box stores ignores a central fact of the modern US retailing industry, and to his or her peril.  If the conclusion is that these stores are incompatible with urbanism, then a large portion of the retail business will be relegated to the car-dominated realm.

Could we instead imagine, rather than a mall anchored by a big box store, a community supported by one of these same stores?  There is no need to imagine, as such spaces already exist.  In a suburb of Madrid, for example, here is a Carrefour hypermarket (the local equivalent of a Walmart or Target, at center) surrounded by a residential development and with no surface parking:

Although there is some underground parking, there is no surplus, and hundreds or thousands of surrounding residential units are within walking distance of the market.  Rather than detracting from small business, the pull of the market draws traffic down arteries past other businesses, providing them with needed foot traffic.  It should be noted there is rapid transit in close proximity as well, just barely visible in the lower left hand corner, and there is also a network of segregated bicycle paths.

There are historic precedents for this sort of urbanism in the form of the souk or bazaar, itself a sort of mall at the center of a walkable city, as well as the ancient Roman forum, which was often little more than temple in the middle of a two-story arcaded and enclosed shopping mall.  Some structures, such as Trajan's market in Rome, were in fact enclosed malls of shops and offices with an appearance very similar to those of today.

Trajan's Market in Rome, one of the first "dead malls." Via Flickr.

 In a sense, the big box store offers unusual possibilities for walkability, and yet these have rarely if ever been taken advantage of within the United States.  Although there are some urban big box stores, the number of consciously planned "Walmart-oriented developments" appears to be close to zero (but wait -- see below).  Although there are many master-planned developments which incorporate retail, none are anywhere near as integrated as the above Spanish example.

A Target and supporting retail in Phoenix, AZ.

The typical Walmart or Target is designed around the assumption that near 100% of customers want to arrive and will arrive by driving, but this is also a self-limiting business model that virtually eliminates quick or impulse shopping trips, or trips by those without access to a car.  It is easy to envision how a combination of higher quality and more local food shopping options plus online shopping could threaten to weaken these chains' business model.

A greenfield Walmart or Target truly integrated with its surroundings and embracing the possibilities of pedestrian access -- even if it required these companies to dabble in residential real estate -- could have tremendous upside.  Using a five-minute walkshed, a sufficient density could be obtained using some of the residential forms visible in the Phoenix image above to provide the store with a reasonable portion of the customer base needed to economically sustain it.  Moreover, it would just be a more pleasant environment in which to live.

The challenge of incorporating parking, which would still be needed in lesser amount, was resolved in the Madrid example by a partially underground lot.  An alternative, rarely explored, is to place parking on the roof of a building.  The added construction expense might be compensated for by surrounding residential land use.

Rooftop parking at a Whole Foods in Washington, D.C.

This just in: just before I planned to publish this piece, Washington Business Journal published renderings of a newly-planned development in Washington DC that incorporates almost all the principles I discussed above:

Note the parking on the roof and the use of what would have been the surface parking area for shops and residential on narrow streets (!) closely integrated with and with easy walking access to the "big box" store at lower right in the rendering.  I have to think that this is the way forward for the big box store.  This design format makes allowance for car trips but embraces walkability as well.  It will be interesting to see how this project fares and if the concept catches on in other cities.

NIMBYism Under the Microscope

Chris Bradford has a post at his new blog, also picked up at Citylab, urging a better understanding of the so-called NIMBY phenomenon. Chris' thesis, as he sets it out, is that "NIMBYism is about monopolizing access to neighborhood amenities."  This is somewhat different from other perspectives which view NIMBYism (a term I use due to lack of substitutes) as an attempt to protect property values, to exclude undesirables or to guarantee privacy.  As Chris writes:

"There are plenty of neighborhoods ... in which the value of neighborhood amenities is capitalized into home prices. These are neighborhoods with valuable amenities that lack close substitutes and which are, for lack of a better word, “under-zoned.” Homeowners in these neighborhoods have strong incentives to obstruct increases in zoning entitlements and to agitate for down-zonings in order to protect the value of club membership. And they do."

The post is very timely for me, as the city I live in is currently considering a proposal to upzone a single-family residential area around a commuter rail station that is surrounded by neighborhood commercial uses.  I won't get into the specifics of the location or the local politics, as the story could be repeated a thousand times across American towns and cities, but suffice it to say that the comments from neighborhood residents on a petition opposing this proposal are enlightening and, I think, strongly supportive of Chris' thesis.  I've excerpted a few of them below:

"I've lived here for 31 years and you are and in some ways already have destroyed the essence of our quaint village of [Anytown]." 

"I'm signing because our lovely [Anytown] is being destroyed, one building at a time. Please stop this madness. The only people benefitting from all this construction are the builders and developers." 

"Stop the intrusion to our community of apartment buildings and additional retail businesses. Our neighborhood is over-saturated. Traffic is horrendous, parking is limited, schools are over-crowded." 

"We have everything within walking distance; movie theater, skating rink, pet store, library, ice cream shop, restaurants, a great ball field. This is why our residents have chosen to live here for that small community feel. We do not need anymore apartments."

"I grew up in [Anytown] and it is a charming village and should remain that way. The streets cannot handle the traffic that would be created from any further development."

"[E]nough is enough. Think about the tax paying homeowners and not the greedy developers."

There is the clear sense here of a valuable amenity, a "quaint," "lovely" village with "everything within walking distance" that has been subjected to competition from new residents.  Because I always prefer a helping of data with my anecdotes, I charted out all the "NIMBY" comments -- more than 80 of them in this case -- based on the concerns they raise in the chart below:

Increased car traffic is far and away the most mentioned concern, with overpopulation/overcrowding second, neighborhood character third, school overcrowding fourth and parking fifth.  Notably, there were no comments citing home values or privacy concerns.  One or two comments did refer obliquely to people deciding to sell in response to the arrival of apartments, but these were linked to changing neighborhood character (a less family-friendly environment) rather than falling neighborhood property values, and implied greater rather than lesser demand.

Again, I think this is strongly supportive of Chris' thesis.  The neighborhood in question is "under-zoned" relative to its capacity, particularly in light of transit access, and enjoys amenities that are underutilized.  The repeated references to traffic, strain on schools and overcrowding are simply different ways of stating opposition to more intensive use of neighborhood amenities by additional residents.  Issues such as increased noise, crime, anxiety over renter populations and similar concerns were cited much less often.

The tone of the comments also provides a window into NIMBY psychology.  Comments are rarely measured in their language and frequently employ hyperbole.  The majority are devoid of any optimism, and one gets the overriding sense that there can be no positive change, only a constant battle against further decline and decay.  This NIMBY mindset, pessimistic in the extreme, appears to be behind the despairing tone evident in much of the commentary.  It may be that these attitudes are most prevalent in those areas where upzoning would be most beneficial for precisely the reasons Chris mentions.

Overall, these sorts of findings suggest that addressing concerns over property values may have little impact in many cases of neighborhood opposition to densification.  Understanding what motivates this opposition will require looking closely and in good faith at the concerns that are raised.

Homeownership Postscript: An Even Grimmer Appraisal

In a recent post, I pointed out how, when adjusted for age and life expectancy, homeownership in the United States is lower than is has been for many decades.  In the comments, it was further pointed out that not only is homeownership itself lower, but that free and clear ownership has been on the decline as well.

There is some data available here, enough to put together a chart over time first using the non-adjusted homeownership rates (blue is homeownership, red is the percentage of all homes that are owned without a mortgage):

As can be seen, there has been a dramatic change over time, with over a majority of owned homes being owned free and clear in the 1920s and 1940s, and thereafter steadily declining to the present figure of just under 30 percent.  The increase in the 1940s may be related to the effects of wartime rent control and the shift in cash investing to real estate during those years.

Today, the majority of these non-mortgage encumbered homes appear to be owned by senior citizens who have paid off their 30-year mortgages.  In light of this, and the growth of the over-65 demographic since the 1940s, it should be somewhat surprising that this statistic has declined.

Here is the same analysis using my age-adjusted homeownership figures (1920 is omitted):

In addition to the above, of the majority of homes that are encumbered with mortgages of all types, the percent of equity in those homes has been steadily falling as well, as shown on the below chart:

The chart speaks for itself, with a decline in the 1950s and 1960s likely related to a growing use of the 30-year mortgage with its lower down payment requirements, a stabilization through the early 1980s, and a precipitous decline during the cresting and burst of the housing bubble in the late 2000s.

The implied conclusion here, that a dramatic expansion of debt has been necessary just to maintain the illusion of a stable homeownership rate (setting aside the explosion of debt in the 2000s necessary to support an increase in homeownership), puts an even more negative spin on the figures from the preceding post.  In short, a decline in homeownership has until the past few years been masked by shifting demographics and an increase in household debt.

One last point here is that although homeownership fell back to its earlier baseline (in non-adjusted figures) following the real estate bubble, equity has not risen back to to the prior 60-70% range, even in spite of the many institutional cash buyers on the market.  The debt legacy of the bubble appears as though it will be around for many years to come.

Related posts:

• "Affordable Housing," part II

Are Millennial Families Really Seeking a Car-Based Suburban Lifestyle?

A recent article by Lyman Stone makes the argument that the return to cities observed during the late 2000s, rather than being primarily a reflection of increasing preferences for urban living, was a temporary phenomenon caused by a bubble in suburban real estate which for a brief time made city renting significantly less expensive than suburban buying.  Under this theory, there was no great change of preferences among the so-called millenial generation or others, but only a temporary price inversion caused by fleeting and unsustainable cost factors.

The general idea that land rents are higher toward city cores, and lessen in a concentric pattern outwards from the center, is not new.  As the "bid rent theory," it was developed by William Alonso in the 1960s, and has applicability to cities from the distant past all the way down to the present.  Modifying but not necessarily contradicting this theory is the concept of the "favored quarter," in which bid rents are determined by cardinal direction from the core rather than by distance alone.  In contemporary cities, both the bid rent and favored quarter can be easily found and mapped.

Income distribution map of Dallas/Fort Worth by Bill Rankin.

In general terms, the bid rent theory holds that commercial uses will compete more intensively for space in central areas, resulting in higher density (and higher cost) housing forms as residential uses are forced to bid against non-residential uses for scarce land.  By contrast, the favored quarter may represent a wealthy neighborhood using the zoning power or self-rule to insulate itself from non-residential or high-density residential competition, thereby securing what is in effect a subsidy for a valued central location.  This sort of abuse of the zoning power, at city-wide scale, has been the subject of a tremendous debate in recent years.

None of this is new or particularly controversial.  The bid rent theory does not presume anything about residential preferences, so far as I am aware, though we might imagine that the same proximity and centrality that is attractive to commercial uses is also appealing to residents who could enjoy that same immediate proximity and centrality as a major amenity.  At the same time, the hustle and bustle of commercial uses are repellent to those who, for their residential spaces, crave some degree of quiet enjoyment.  Central areas are also likely to be disfavored by those, such as young families, who value large living spaces and high quality public schools above even immediate conveniences where both preferences cannot be met simultaneously.

In a post a while back, I also doubted as to whether these underlying preferences had changed, and that inflated costs in suburban areas and/or foreclosures, had driven higher demand to rent in urban areas.  This is not really contradicted by the National Community Preference survey, which continues to show that people highly value the single-family home and immediate, walkable convenience:

The most undersupplied areas, relative to demand, are the "suburban neighborhood with a mix of houses, shops and businesses," the "small town," and the "rural area."  These patterns seem to be amplified for the millenial generation, with a particular emphasis on walkable neighborhoods.  Oversupplied, relative to preference, are the mostly residential suburb and city (although I suspect many of the "city" residential areas are largely "suburban" in character).  I do not know exactly how "small town" differs in form from the mixed suburban area, but one imagines the category to be inspired by the fictional New England village of Stars Hollow from the Gilmore Girls show, with its vaguely New Urbanist mix of houses and shops with quirky and eccentric independent proprietors immune from the long arm of Walmart:

Stars Hollow set, and also, I believe, for Hill Valley from Back to the Future.

The allegedly unrealistic image of the town in the show has been critiqued here, but I still think the popularity of the series has something to tell us about the environments and lifestyle people idealize, even if the particular example in the show may not be economically plausible.

Lest we imagine that these choices are in fact economically or spatially incompatible, or that what is being sought is the unobtainable single-family house in Central Park, this is the essence of Japanese market urbanism: an extremely compact assemblage of small single-family homes (and some apartments) that is pedestrian and bike friendly.  This must be the case, since lower densities will result, for the majority, in the perception of a "residential-only" neighborhood.  Naturally, Japanese-style development (as described by Nathan Lewis here) is one thing which the American cities have almost entirely failed to provide, although a few neighborhoods here and there, generally developed before 1930, provide a reasonable facsimile.

Tokyo neighborhood.

Quite a bit has also been written as to whether the millennial generation will, in time, leave urban areas, as though there was some question as to whether this particular age bracket would buck the trend of all groups before it.  As Joel Kotkin wrote two years ago:

"The millennial “flight” from suburbia has not only been vastly overexaggerated, it fails to deal with what may best be seen as differences in preferences correlated with life stages. We can tell this because we can follow the first group of millennials who are now entering their 30s, and it turns out that they are beginning, like preceding generations, to move to the suburbs.
....
These trends can be seen on a nationwide basis. Among the cohort of children under 10 in 2007, the number who lived in core cities as of 2012, when they were 5 to 14 years of age, was down by 550,000. Families are the group most likely to move either to the suburbs or smaller towns. This movement, plus the high degree of childlessness in large urban cores, suggests that many of those who are leaving the core cities in their early 30s are parents with young children."

Now, for families with young children approaching kindergarten age who lack the resources for expensive private tuition, school quality emerges quickly as an important preference, subordinating almost all other concerns.  But this does not mean that these families do not desire an urban lifestyle, or, by their housing choices, are rejecting such a lifestyle.  The survey data seems to broadly refute that idea.  Rather, the cruel spatial economics of exclusion favor low-density, restrictively-zoned places for "good schools," and American cities offer few other intermediate options.  Abandoning an urban life, with its high costs, is a sacrifice for one's children rather than, necessarily, a pursuit of an ideal.  The choice is reinforced by the panoply of incentives the US tax code offers to those who would buy rather than rent.  Many other families with financially limited choice, shut out of suburban options by restrictive zoning and other exclusionary policies, must remain in urban areas regardless of their preferences.

Seen in this context, I do think the New Urbanism has tapped in to something important in the American psyche.  Only, as Nathan Lewis has written about, it has generally (but certainly not always!) done so too literally, using lackluster American examples as inspiration rather than successful ones from abroad.  I do not think the American imagination is so literal, though.  Stars Hollow passed as a New England town even though it bears no resemblance to the typical Connecticut small town with its large central green and sprawling layout.  What was important was not the specific form, but rather the walking lifestyle, the spontaneous interaction and the community as a whole.  The set simply provided the urban form necessary to sustain the belief that this lifestyle was possible for the characters.

If a real-world development does not offer a sufficient density, or sufficient flexibility in terms of mixing of uses, these things will not occur, and you will have little more than a film stage set.  Imitate a Japanese neighborhood, on the other hand, and you may have more success.  Perhaps clad the buildings in Georgian and Colonial facades for the tastes of American buyers, but leave the form alone.  Do not obsess over mixing of uses or "apartments over the shop" -- these things take time and happen gradually, not all at once and from the beginning.  A suburb built to this form, odd as it may seem, will meet the stated preferences of American buyers, families included.  The demand is certainly there.

The New Urbanism, whatever its failings, has at least recognized the situation, changed the conversation and opened new possibilities not only in terms of building, but regulatory reform and making possible traditional forms of urbanism under contemporary city codes.  This blog is intended as a sort of continuation of this new conversation using a slightly different vocabulary.   

As to what such a new neighborhood might look like, Nathan Lewis has already written at length, but in another post, I'll reiterate some of his findings along with an example that could be done today.

Is the Decline in the Homeownership Rate Even Greater Than Thought?

One of the most heavily reported stories of the post-housing bubble economy has been the decline in the American homeownership rate, which has supposedly fallen to levels not seen since the days of the Lyndon Johnson administration.  Although ownership data is often broken down every which way, including regionally, across ethnic and income groups, over time, by immigration status and by age bracket, one analytical quirk I've noticed is that the non-age comparisons are rarely themselves controlled for by age.

This is highly significant, since homeownership is, obviously, heavily correlated with stage in life.  While only around 33% of 29-year-olds own homes, by age 39, this number leaps to around 49% (according to age bracket data from the 1990s to present).  This fact has major consequences for comparisons among nearly every demographic group, but even more so for comparisons over time.  How so?   Consider that a major age-related change since the 1940 Census has been an increase in the median age of around 10 years due to a decline in the birth rate and a substantial increase in life expectancy.

What is the significance of these changes for the homeownership rate?  The chart I prepared below uses averages of ownership by age bracket for 1994 and 2014 to estimate interval changes and make a complete graph showing that, in general, the rate increases until around age 66 (one would think related to post-retirement home sales) and then gradually tapers off thereafter, while still remaining well above the overall average.

As a result, the older a population grows, the higher its homeownership level should be, all else being equal.  Intuitively, it should be shocking that today's homeownership rate is comparable to that at the height of the baby boom era, when the national median age was only 28.  The Census Bureau has apparently made some adjustments to more recent ownership data to reflect changes in age structure, but I was interested in doing the same for the decennial data going back many decades.

The data has been adjusted to account for 1) changes in the median age, normalized to 2014, to reflect that an older population is centered around a much higher ownership level, and 2) changes in the life expectancy independent of median age, to reflect that the older people live, the higher the homeownership level is likely to be.  First is a plot of the non-adjusted decennial Census data (plus 2014), showing the familiar pattern of a large jump in the 1940s due in part to wartime rent controls, a slowdown in the 1960s, and a general stagnation since that time.

Once we account for the demographic changes, however, it becomes obvious how the aging of the population has masked a major decline since 1970.  The chart below shows the data adjusted for change in median ages only.  Note also how the slowdown in the 1960s has disappeared and been replaced by a substantial increase, as the population was growing younger during that time.  This is more consistent, I think, with the conventional wisdom that the 1960s were a boom time for new buyers as much as the 1950s were.

Finally, if we account for life expectancy as well, the trends are further amplified, showing that, in age-adjusted terms, today's homeownership rate is lower than that of 1940, which was itself a low point only a little above where it had been in the depths of the Depression.  A peak is reached around 1970, after which there was and continues to be a decline (the upward blip in the mid-2000s is not represented here).

My attempt at analysis here is definitely not the last word in examining this data, and no doubt there are flaws, but it at least leaves open the possibility that today's homeownership rate, in the context of age adjustments, is lower than it has been for a very long time, at least as long as 80 years and perhaps as far back as 120 years or more.  With more data on earlier Censuses and a more rigorous approach more refined and complete results could be calculated, but I would doubt whether the overall picture would be greatly different.

The question as to why homeownership peaked around 1970, in these charts, and declined thereafter, is one which has been discussed in many other and related contexts.  Nathan Lewis has written extensively about the abandonment of the gold standard.  Ben Ross has discussed the changes in land use policy in the 1970s, including the rise of environmentalism, which threw up new barriers to development and reduced the supply of new housing coming to market.  Seeking a full explanation would be far beyond the scope of this post, but I think it is noteworthy that the results are broadly consistent with other economic markers, including inflation-adjusted income.

Finally, it it may be interesting to compare these trends with those of Europe, as shown on this graph which I grabbed from Twitter (via Antony Slumbers):

Although the US continues to have a lower homeownership rate than the Eurozone, and far lower than many individual countries, I suspect this difference may disappear or even reverse once median ages are taken into account (the median age of the Eurozone being over 41).

Spreadsheet with data and charts here.

Related posts:

• Was the Rise of Car Ownership Responsible for the Midcentury Homeownership Boom in the US?
• Homeownership, Unemployment and Economic Growth: Looking at the Trends

That 70s Urbanism

1970s suburb, Birmingham, AL.

The 70s, as much as any decade from the 1920s to the 1990s, has gotten a bad rap in contemporary urban planning circles.  It was the golden age of the suburb and the nadir for many urban areas around the United States, not to mention a boom time for highway construction and the development of very low density suburbs on non-gridded street patterns.  Surely there is nothing we can glean from this time period that has applicability to the urbanism of today?

Well, this is perhaps not entirely true.  The 1970s, as well as the decade before it, arguably represented the first time in the urban history of America that planners self-consciously pursued entirely pedestrianized environments, although the inspiration for these areas went back to flights of imagination from the 1920s if not earlier, and the 1950s saw the genesis of several of these ideas, at least in their embryonic forms.    I refer not only to pedestrianized main streets, but to the five great pedestrian creations of the age: the enclosed shopping mall, the international airport, the vacation resort, the convention center, and the network of tunnels and/or skyways.  Each of these represented a vision, if incomplete, for pedestrian-only circulation and commerce on a grand scale, often on the footprint of a small town.  

Do I jest?  No.  The future, if the human imagination is a key to the future, was pointing toward pedestrianism in the 1960s and especially 1970s.  This was, after all, the age of Paulo Soleri's implicitly pedestrian-centric arcology.  Gas prices were soaring by the early 70s.  The 1939 World's Fair vision of high-speed highways was out, and a more refined tecnho-utopianism involving futuristic megastructures was in.  The great majority of the buying public rejected this vision for their private lives, but as the success of enclosed malls shows, they gladly embraced it in many other contexts -- in commerce, for employment, in travel and for recreation.  

The arcology: no highways here, or at least, not the central element! Source.

The tunnel system of Houston and the skyway system of Minneapolis, both begun in earnest in the 1960s, were built out through the 1970s.  Enclosed mall construction, which began in 1956, was in such a boom that four enclosed malls opened from 1969 to 1980 in the city of Toledo, Ohio alone.  Air travel multiplied through the 1960s and airports were expanded on an unprecedented scale.  The very idea of the convention center, although it had been around for decades, attained gargantuan scale by the latter half of the 20th century.  

The hostility toward these type of environments in recent years, has, it seems to me, been based on two primary grounds: 1) the environments, being private and privately controlled, are corporate and sterile; and 2) although the environments themselves may be pedestrian, they are dependent for their business on car-driving customers.  Of these two critiques, however, the second is not inherent to the form of the structures themselves.  Rather, a pedestrian environment such as a tunnel system or an enclosed mall is, as the cities of Japan or Korea bear witness, a better partner to transit and walking trips than it is to car transport.  That the United States largely squandered its opportunities to integrate these structures and systems into its transit networks is an indictment of city planning, but not necessarily of these forms themselves.  

The Japanese, arguably the best city-builders in the world, have not missed this point, and typically have built shopping malls either adjacent to or, in the case below, on top of railway stations:

Osaka's Keihan Mall, over the Kyobashi railway station. Bing Maps.

In the case of American cities, even where the option is obvious and available, no such plans are made.  In Stamford, Connecticut, where nearly the entire downtown was condemned and bulldozed in a vast urban renewal scheme, the new enclosed mall built by the city's hand-picked developer was inexplicably not located next to the city's heavily-used Metro North railway stop:

The mall, at top right, and station, at bottom left.  Bing Maps.

What were they thinking, one could ask.  Note also the very low value placement of the interstate highway just north of the railway line, although this area probably had higher property values than the south side of the tracks at the time of the highway's construction.  In truth, the Stamford planners had given up entirely on any idea of pedestrian circulation outside the new megablocks.  The curated spaces within the new mall were designed to be accessed by car alone, and the structure to this day presents a fortress-like appearance to the outside.  The decision cannot even be explained by a desire to exclude train riders, who then and now tend to be well-heeled commuters to New York and are in general better off than most.

Had the planners of the 1970s left the city alone, and simply redeveloped the train station as a mall, who knows what might have happened by today?  Yet these questions need not be hypotheticals.  Cities have the power to make these changes today.  Stamford cannot move its mall, but it can consider a new infill station closer to it.  It can consider how to improve pedestrian mobility.  It can revamp its bus network.  It can do many things to address past mistakes, but it must be able to understand its mistakes, and even more importantly understand how these creations of the 1960s and 1970s can be a force for good, not merely magnets for a car-driving public.

There are a few easy steps that can be taken at first.  If Stamford wants to fix its mall, tear down one of the car ramp entrances, already duplicated by two others, and install a prominent pedestrian entrance, like so:

Easy?  In the grand scheme of things, yes.  Expensive?  Not terribly.  A bolder step would be an entirely new infill station, one closer to the mall, perhaps at one of the locations highlighted with a green arrow:

Plan a new station.  New Haven already did it, twice, and Bridgeport is doing it, so this need not be some impossible fantasy.  The overpasses need to be replaced anyways, so merge the projects.  Use the gains in real estate value to improve pedestrian connections.  Narrow the streets.  Add bike lanes.  It need not be that difficult!

If hostility toward the idea of enclosed malls, or skyway systems, or convention centers is retained, however, then cities may or may not succeed at revitalization, but they will squander again an opportunity to use these forces for good rather than ill.

---------------

Other reading:

See Mallville by Andrew Price for a related perspective on the traditional city aspects of the American mall.

Lot Sizes: Regional Trends and Causes

In a prior post, I examined minimum lot sizes as a general concept with reference to a few examples.  Another way of approaching the subject is to examine regional and national patterns and trends on lot sizes.  Although the subject doesn't get a great deal of attention, and Census information is not as detailed as one might like, it is possible to cobble together some statistics.

Based on data from the MLS as compiled by Realtor.com, here is a map showing the median lot size of properties offered for sale in each state (for 2012).  As we'd might guess, the patterns reflect both complex historical influences as well as current degree of urbanization:

More relevant to current debates over urban expansion and development are the median lot sizes of new homes, as compiled by the Census.  Unfortunately, the data is not broken out by state, but the regional trends are nonetheless interesting and helpful:

Source

As far as I can tell, rather than primarily being the product of planning and development policies, these differences are substantially driven by agricultural land values as well as groundwater availability.  Federal land ownership may also play a role in some western states.  A map of well-water supply is below, showing a fairly good correlation with lot size:

Source

The outlier value for New England in the median lot size map seems to fit the well-known historical account of agriculture in that region, where after the arrival of railroads carrying midwestern grain in the later 19th century, farms were rendered unprofitable and were abandoned en masse, leaving only stone walls -- 250,000 miles worth -- through new-growth forests as a sign of their former presence.  During the height of this abandonment in the 1890s, some New England counties, like Tolland County in Connecticut or Rutland County in Vermont, actually declined in population.   Post-1950 tract homes in New England, built in forested, rocky terrain, prefer to sit Walden-like in their own little forest clearings, as though each homeowner were an 18th century pioneer carving out a homestead in the wilderness.  Much of New England, as William Fischel also notes in Zoning Rules!, with a wet and cool climate, is heavily well-watered, with the result that homes need not be clustered closely together.

Outer suburbs of Hartford, CT.

Where agriculture yields the highest-value crops on a per-acre basis, such as in the fruit groves and orchards of Florida or California, the smallest lot sizes are found.  A list of the densest urban areas in the United States is surprisingly dominated by small California cities, such as Davis, Woodland and Delano, most of which sit within the fertile farmland of the Central Valley.  In Florida, meanwhile, the rapid development of The Villages retirement community has produced some of the densest single-family subdivisions in the entire country:

The Villages: homes built for former residents of Hartford, CT.

This ingenious map shows the distribution of cropland and pasture by type throughout the US, highlighting the concentration of agriculture in certain areas and its near-absence from wide swaths of the country, including the majority of New England, much of the southeast apart from coastal lowlands and the mountain and desert west:

Map by Bill Rankin.

In most of Europe and Asia, by contrast, pre-industrial towns and cities of necessity grew in immediate proximity to valuable agricultural land:

Farm towns in Heibei province, east of Tangshan, China.

A few exceptionally large cities of the ancient world imported grain by ship, but for the most part cities were dependent on their agricultural hinterlands.  Until the early industrial era, the high value of this land for crops probably did not impose significant constraints on city expansion (as opposed to the limits of sanitation, transportation, the need for a defensible perimeter, etc), but in an age of rapid population growth and urbanization, the cost of land acquisition gained greater significance.

Where land has little value for agriculture, we should not be surprised to see greenfield development taking lower-density forms.

Urban Governance: Merger and Fragmentation

Let's consider two hypothetical cities. For convenience, I'll call them "Hartford" and "Nashville."  Both are state capitals.  Both are favorably located on bluffs overlooking large rivers and are surrounded by abundant buildable land.  As of 1960, both cities proper had comparable populations, with Hartford at 162,000 and Nashville at 170,000.  Hartford was in 1960 by far the wealthier of the two, being located in one of the richest states of the union and hosting the headquarters several of the nation's largest insurance companies, yet housing costs were reasonable in relation to Boston and New York.

From the standpoint of a 1960 observer, Hartford appeared to have the brighter prospects for population growth.  Hartford's metropolitan area, represented by the county of which it was the seat, was in fact growing at a more rapid clip than Nashville's Davidson County, with a population increase of 28% during the 1950s compared to 24% for Nashville, even though Tennessee at the time had a higher birth rate than Connecticut and Davidson was absorbing heavy in-migration from much of the central Tennessee region.

Around 1960, however, two developments reshaped the governance of both cities.  In that year, Connecticut's abolition of county government went into effect, leaving only the state government and 166 town governments, one of which was the rump city of Hartford, a jurisdiction of 17 square miles.  Only three years later, Nashville moved in the opposite direction, merging itself with the 504-square mile Davidson County and forming a consolidated city-county government.

Since 1960, Hartford County has grown 23%, while Davidson County has grown 40%.  By 2010, Nashville's urbanized area population exceeded Hartford's.  As of 2014, Davidson County with 668,000 inhabitants held fully 62% of the greater Nashville urbanized area population.  By contrast, the city of Hartford had lost population since 1960, and with only 125,000 residents held only 13% of the greater Hartford urbanized population.  Nashville's downtown area is today booming with apartment construction, while Hartford has seen little multifamily development since the 1960s.

The factors in the success of Nashville relative to Hartford obviously are more complex than the arrangement of city government, and involve developments as varied as the economic rise of health care and higher education and the general increase in prosperity in the American south relative to the nation during the mid and late 20th century.  Nonetheless, the relative fragmentation or centralization-by-annexation/merger of city governments is of major importance in how cities are run, regulated and taxed.  For instance, consider that a home in Hartford today bears a tax burden more than three times that of a home of identical value in Davidson County.

For the 2010 Census, the list below shows the top and bottom 15 rankings for the population of cities proper expressed as a percentage of total urbanized area population (out of the top 50 largest urbanized areas):

The list of metros with the highest percentages of their urban population contained within their cities includes several of the fastest-growing and most economically resilient cities in the United States: of Brookings' list of the top 10 performing cities over the past several years, five are on this list.  Only one metro on Brookings' list, Salt Lake City, appears on the bottom 15.

Simply because a metro appears on the low end of the rankings, however, does not necessarily mean that it is highly fragmented overall.  Washington D.C., for instance, is surrounded by large and relatively powerful county governments, including Fairfax (1.1 million, 407 sq. mi.), Montgomery (1.0 million, 507 sq. mi.), Prince George's (890,000, 598 sq. mi.), as well as the smaller but populous Arlington County and city of Alexandria.  Others, such as Bridgeport, CT, Providence, RI and St. Louis, MO do have a highly fragmented governing structure throughout the metro area, with a large number of small towns each guarding their own local taxing and land use prerogatives.  County governments fractured into townships tend to replicate, in approximate fashion, the metro areas of non-county states such as Connecticut and Rhode Island.  The proportion of people living within the central jurisdiction is suggestive of degree of fragmentation, however.

I can hear the reader say right now: but correlation isn't causation!  While that is certainly true, and while I also bear in mind Jane Jacobs' comment that a region "is an area safely larger than the last one to whose problems we found no solution," those who have studied this issue have found definite advantages to a more centralized form:

• Zoning.  Although one might imagine that a multiplicity of jurisdictions might compete with each other for residents, thereby raising the net quality of life, what is observed instead is that jurisdictions attempt to compete for wealthy residents by imposing restrictive zoning regulations.  This results in a zoning race to the bottom, as towns enact ever-stricter regulations to keep prices high.  The result is high citywide housing costs relative to income.  More centralized cities tend to have more permissive zoning.
• Property Taxes.  A highly fragmented metro tends to have high property taxes as well, due to lack of economies of scale.  Central cities often suffer the most, as they have the largest share of public or non-profit land, and must impose nearly ruinous taxes on the remaining devalued private land to cover basic municipal services.  The taxes further drive down property values, resulting in a situation where (as with Hartford, see below), the situation nearly becomes unsustainable.
• Education.  As I've written about before, overall educational outcomes in county-wide educational systems funded out of general revenues can be as good or better than even the best and most lavishly-funded local school systems.  
• Transportation.  In general, regional transportation planning would be expected to have a smoother political course in a centralized city.

There are disadvantages too.  Very local governments, for better or worse, can be (or at least are perceived to be) more accountable to the individual citizen.  These governments may have long histories and impart to an urban area much of its character.  Some of the advantages of the centralized form can be overcome through adopting elements of a centralized form, such as pooling of certain services, without relinquishing all local authority.  

Still, where all else fails, consideration of more drastic measures may eventually become necessary.  In its election just a month ago, Hartford elected a new mayor, Luke Bronin, who defeated the incumbent in an earlier primary.  Although the local paper has wished him well, it closed a post-election op-ed with words that may soon need to be uttered more forcefully:

"The steep challenges facing Hartford and Bridgeport [which also elected a new mayor] raise the question of whether these cities, which have heavy costs and little property to tax, can continue to be viable. The canary in the coal mine on that may be New London, where Democrat Michael Passero, a city council leader and former firefighter, was elected to replace Mayor Daryl J. Finizio.

At seven square miles, New London is the second smallest geographical municipality in the state. Like Hartford and Bridgeport, it bears a disproportionate amount of its region's social costs. 

If the new mayors cannot make these cities work, we shall have to rethink them."

Whether Connecticut has the capacity to "rethink" the organization of its centuries-old towns is a reasonable question, as is the question of whether any such reorganization would work to solve central city ills.  That the question is presenting itself at all, however, is indicative of the scale of the problems at hand.

Demise of the Duplex

The New York YIMBY website has complied Census building permit data to reveal how construction of single-family and small multifamily dwellings in New York City's five boroughs has plummeted since reaching a peak in 2004.  Of the potential explanations advanced for this collapse, contextual downzoning appears to the most likely to me, as the decline began four years before the peak permit year of 2008.  In general, however, small multifamily buildings have widely fallen out of favor not only in New York but across the country over the past thirty years and more.

Using the same Census data, this time calculating for all 50 states and the District of Columbia, it can be seen that small multifamily dwellings (those with between two and four units) fell from providing around ten percent of all new residential units in the early 1980s to a low of just under three percent in 2013.  Even as multifamily construction has rebounded since 2009, increasing its share of all units from 21 to 34 percent from 2009 to 2013, these smaller multifamily units have actually continued to decline as a proportion of the total (the chart shows the number of units, not the number of structures):

Triple deckers in Bridgeport, CT.

The permit data only begin to capture what is, I suspect, a much longer-term decline in this housing typology. Anyone familiar even in passing with the larger cities of the Northeast will immediately recognize the heavy predominance of the type in their older neighborhoods, as represented by the wood-framed triple-decker, or three-decker, house.  Whether built up to a flat roof, as in the typical image of the Boston triple-decker, or with the third story sheltered under a pitched roof, these are large and bulky structures that generally provide three spacious units with windows on all sides.  Despite the popular narrative of urban-dwellers fleeing cramped apartments for more spacious suburban homes, these units rivaled or exceeded in size the modest Cape-style single-family homes built in the 1940s and 1950s, were conveniently set on a single floor like the ranch houses that later became popular, and offered gracious architectural features such as bay windows and front porches that were often lacking in the new single-family homes. It would not be until the 1960s that the average new single-family home would significantly exceed the size of the ordinary triple-decker apartment.

The vertical axis shows the number of  two  and three-family
structures sold over the past three years within the metro area.

Census data provides housing statistics both by year built and by housing type, but unfortunately does not combine these statistics, making it impossible to determine what proportion of the small multifamily stock was built at what time. A workaround can be had, however, by using year-built information from housing listed on the MLS, which should provide a neutral sampling of the overall housing stock. Performing this exercise for the Bridgeport-Stamford-Norwalk area, an area with housing from all eras of American history and continuing high demand, reveals an exponential increase in the type from the 1860s to the 1900s, then a slower increase to an all-time peak in the 1920s.  Small multifamily construction collapsed during the Great Depression, along with most other housing construction, but unlike single-family construction did not rebound, even though the Bridgeport area experienced a manufacturing boom in the 1940s and early 1950s that attracted many factory workers.  Instead, it continued a gradual descent into irrelevance by the 1980s and 1990s.

The appearance of these structures therefore coincided both with the industrialization of American cities and a wave of immigration from rural America and from foreign countries, with the particular architectural style of multifamily housing in New England perhaps influenced or inspired by the multiplexes common in French-Canadian towns and cities (a building type which Urban Kchoze's Simon Vallee has recently explored).  These buildings had particular appeal to new immigrants, who could with sufficient savings purchase such a building and rent the upper two floors out to other immigrant families (often, members of their own extended family) to defray the cost of housing or perhaps even earn some additional income.

Why these dwellings ceased being built after the 1920s, never to return in any great numbers, is a difficult question, but fortunately it is not one that I need to guess at.  MIT graduate student Jacob Wegmann authored a 115-page thesis entitled What Happened to the Three Decker, supervised by none other than Sam Bass Warner, which explores that very question and offers a series of possible explanations which are equally applicable to other forms of small multifamily structures:

• Above all, exclusionary zoning, particularly in the 1970s and later, that restricted small multifamily housing from being built in those places where it would have been most desirable.
• Concentration of the real estate industry in the mid-20th century, resulting in the production of large-scale tract subdivisions that were able to exclude multifamily housing altogether.
• Federal involvement in mortgage finance starting in the 1930s and an emphasis on homeownership paid for by way of extended mortgage terms rather than over a shorter period with the aid of rental payments from tenants.
• Other regulatory barriers, including parking requirements, disability mandates from the ADA and state laws and enhanced fire safety requirements that have increased the construction cost for small, non single-family structures.
• A negative image of small multifamily dwellings that had always simmered among the native middle-class and which intensified during the 1920s and later, and which contributed to attempts to exclude these dwellings from newer areas of cities.

The last point, although it may seem less important, does indicate to me a genuine underlying problem with the three decker or stacked duplex form.  As an article on Worcester's three-deckers puts it, "to look at a three-decker means ... appreciating the attempts of three-deckers to echo freestanding single-family dwellings even in the midst of an undeniably urban setting and the effort to create an illusion of space for residents."  

Bridgeport, CT rent map showing lowest rents in duplex/
triple decker belt between downtown and outlying
single-family neighborhoods.  From Trulia.

That is, the small multifamily house was apologetically urban, and offered as its apology an attempt to mimic the outward form of the cottage or farmhouse style of housing prevalent in New England prior to the 1870s.  This reticence to adopt an unambiguously urban form left such structures appearing to be second-best, a characteristic that was not shared by rowhouse neighborhoods or those composed of larger apartment buildings.  The attempt to leave small gaps between buildings, rather than using shared walls, only served to emphasize the scarcity of space and lack of light and air as compared to larger lot single-family homes.  The virtues of having natural light on all four sides of a dwelling, however, were very much real and have been appreciated for decades by the residents of these apartments.

The limited appeal of these structures has had the upside of keeping them as relatively affordable housing options down the present day even in otherwise expensive cities.  Neighborhoods composed of them are not immune to gentrification, but their form and physical location -- typically in the inner-ring areas once served by streetcar lines -- combines to keep their rents among the lowest in the metro area.  

Although the building type has not come back into vogue, the notion of using a second residential unit on one's own property to help with mortgage payments has, through the New Urbanist revival of the so-called "granny flat" (a concept which goes by countless names, but never "duplex").  The strategy of appealing to homeowners' financial interests rather than to the need for more low-cost rental apartments is politically astute and has probably helped the idea gain traction.  In the meantime, in areas where demand is high, many single-family houses continue to undergo illegal conversion to multifamily use, indicating how the America of 2014, by banning the construction of small multifamily buildings and the division of homes into multiple units, once a commonplace process, is in some areas and in some respects doing an inferior job of housing the poor and recent immigrants than was the country of a century earlier. 

NYC Suburban Demographics: Choice or Fate?

The New York Times recently profiled a study from Community Housing Innovations which noted a demographic collapse among the young adult population in the wealthy suburbs of Long Island and Westchester County since the 2000 Census. The decline is attributed largely to high housing costs rather than an immutable preference for urban surroundings:

"The villages and towns suffering the largest losses of their young workforce have historically restricted multifamily development in favor of single-family homes. The practice, known as exclusionary or snob zoning, makes living in these communities unaffordable for much of the Long Island and Westchester County workforce."

The study paints an interesting picture of the unintended consequences of restrictive zoning, which in this case may be hindering adults at a prime age for starting a family (25-34) from living in what, in many cases, are probably the very same towns they grew up in. Although the Times quotes a Westchester County source disputing the figures and noting that the "county’s own enrollment data shows that more children are attending its schools, a telling sign of young families," a glance at the Census figures reveals that the under-5 population in many of these towns is plummeting as well.

For instance, while Westchester County saw a decline of 11% in this group from 2000 to 2010 despite experiencing overall population growth, Scarsdale's under-5 population fell by 28%. Jericho, New York, mentioned in the article, also saw a decline of 27%. For further comparison, here are the figures for part of nearby Fairfield County mapped over median housing prices:

These can be compared to the same figures for the 25-34 population:

Although the numbers don't match up precisely, the correlation is strong. There is an overall decline of around 10% in both population groups, compared to an increase nationally of 3% for both, but there has been a rapid assortative process as young families have apparently been priced out of many locations and restricted to the few remaining jurisdictions, which include the urban markets of Stamford and Bridgeport, without stratospheric housing costs.

Whether this is also due to a preference for urban living conditions is unclear, since the urban markets also happen to be the low cost markets, and are the only places in Fairfield County permitting the construction of significant quantities of new housing. The fact that population loss of these demographics was somewhat less in relatively more affordable suburban towns (such as Fairfield or Trumbull) suggests that cost is at least as important a factor, a pattern which is also noted on Long Island in the New York Times article.

The overall losses among the 25-34 age group indicate an accelerated outmigration in addition to the internal assortative process. The study cited by the Times suggests, in the vein of the Great Inversion, that this migration is a product of relocation to New York City proper, and to particular areas within the city, although many must be leaving the region entirely. In this broader context, though, I'm not satisfied with a sample period of only ten years and certain areas, and so I compiled median age data for the greater New York area from 1970-2010, dividing it into three regions*:

Although there is certainly a great inversion here, it appears to be a multi-decade process, rather than an sudden shift which occurred over the last ten or fifteen years. These trends can also be expressed as a function of the US median age in each Census period so as to show which regions were growing "younger" or "older" relative to the population as a whole:

Rather than the city abruptly changing course in recent years, a long term downward age trend actually decelerated after 2000. Trends have not always been consistent, with the inner suburbs actually growing relatively younger from 1970 to 2000. The continued aging of the outer suburbs is also notable. Several counties in this region now exceed a median age of 40, with individual towns highly coveted for their excellent public schools exceeding a median age of 45. These median age figures in some cases approach or exceed those of Florida counties known as retirement havens (cf. Westport, CT, at 44.6, with Palm Beach County at 43.5). With little new construction and limited housing turnover, combined with greatly elevated prices, these towns await a generational shift.

Just how big will that shift be? The Westport town newspaper, at it happens, suggests dramatic changes after 2015. According to one quoted source, "there's going to be major turnover in property. People who currently own homes and have for a long time, are going to decide to downsize. When turnover does occur, you'll get a flood of students entering elementary school." 

The problem for towns such as Westport, however, is that due to their resolute opposition to any new construction – even including senior housing, as described in detail in Snob Zones – there are not many options to downsize to. It seems more likely to me that the inhabitants of these homes will, by and large, remain in them to the end, as they will generally be unwilling to abandon their towns and long-established connections simply to find smaller accommodations elsewhere. The turnover point is therefore more likely to be in the 2030s than by 2020, but this will obviously come too late for the Boomer generation's own children seeking to start families. It's also not clear, given the widespread declines in the under-5 demographic, where a "flood" of new students could come from several years down the road.

An apparent denialism about the effect of high prices on young families seems to be rampant among many of the wealthy towns profiled in the CHI study and in Fairfield County. Wishful thinking about floods of new students that are just around the corner, or analysis of falling school enrollments that fail to diagnose high housing costs as a contributing factor, rules out consideration of the supply-based solutions that have apparently worked to some degree to retain young people in certain Long Island suburban towns discussed in the CHI study. These towns need not be victims of fate, but unless changes are made, their futures are already inscribed on their current demographics.

-----------------------------

Related links: Remember this New York Times article from 2013 covering young families moving out of Brooklyn to suburban towns in Westchester County? As it turns out, Brooklyn experienced a decline in its under-5 population from 2000 to 2010, but only by 3%. Hastings-on-Hudson, by contrast, saw that group decline by 19%.

*By county, "Urban" = Manhattan (New York), Bronx, Brooklyn (Kings), Queens, Hudson; "Inner Suburban" = Westchester, Nassau, Bergen, Union, Passaic, Staten Island (Richmond); "Outer Suburban" = Suffolk, Fairfield, Rockland, Morris, Somerset, Middlesex, Monmouth. Exurban counties are omitted.