# Comparing Homeless Counts, BC Edition

We most commonly hear about homelessness as a big city housing issue. But are big cities where people are most at risk of becoming homeless? Comparing homeless count data enables us to start answering this and related questions.

Homeless counts draw upon volunteers and service-providers to provide point-in-time (one night) estimates of people without regular access to long-term housing. People are typically defined to be experiencing homelessness, as in the BC Homeless Count from 2018, “if they do not have a place of their own where they pay rent and can expect to stay for at least 30 days” (p. 11). People counted as homeless include both those staying in shelters and transition houses (counted by service providers) as well as those sleeping in “…alleys, doorways, parkades, parks, and vehicles or people who were staying temporarily at someone else’s place (couch surfing)….” (p. 11-12). Suffice it to say, this is not an easy population to find or track on any given night, and people are often also asked about where they spent the prior night during visits to service providers the next day. As a result, the “hidden” homeless population is always going to be larger than the number of people counted through homeless counts, meaning counts are always underestimates.

Homeless counts are also a lot of work, and even with the generosity of volunteers, they require significant funding and coordination to carry out in a defensible manner. We tend to know a lot more about homelessness in big cities in part because they’ve got more resources to direct toward tracking the issue. So it’s great news that BC Housing has been working with partners to provide counts for smaller communities. The BC Homeless Report, delivered in December of 2018, summarized much of what’s been learned so far.

The report is worth a read, and the count data, all by itself, is useful in assessing where urgent need for more supportive housing can be found. Here’s a lovely summary map of the data for BC, bringing together new counts funded for smaller communities with the most recent (at the time) data from other counts, funded by the Federal Government or independently (often from larger communities like Metro Vancouver and the Fraser Valley).

We can clearly see from the numbers that more people are homeless in Metro Vancouver than anywhere else in BC. But Metro Vancouver is larger than anywhere else, so this doesn’t seem too surprising. What’s more, adding all of the communities covered, more people appear to be homeless outside of Metro Vancouver (3,904) than within (3,605), despite the fact that Metro Vancouver contains over half of the total provincial population within its boundaries. And we haven’t even got homeless counts here for several large communities in BC (e.g. Squamish, Whistler, Powell River, Trail). So already we know homelessness seems bigger outside of Metro Vancouver than within. Maybe not just a big city problem after all!

But we can try and do better than that. Let’s try and create a rough baseline risk of experiencing homelessness at a given point in time for each community covered by a homeless count. We can do this by dividing the number of people counted as homeless by the total number of residents in each community. This seems pretty simple, but there are actually a number of considerations that go into creating this baseline risk (which is perhaps why the report itself does not attempt it). First, are most people experiencing homelessness coming from the community where they’re being counted? In fact, we know that they are. Check table 3.9 (p. 37). In no community studied in the report do the majority of people counted as homeless report living there for less than a year. In most communities studied, the majority counted as homeless have lived in the area for five years or longer. Homelessness is mostly local. So we’re on sound footing assessing the risk of homelessness as local.

But what do we mean by local? Local can easily cross municipal boundaries to include broader catchment areas (e.g. metropolitan areas). And there may be clustering of homelessness within broader catchment areas, following services and shelters. We know, for instance, that while the City of Vancouver contains around a quarter of the Metro Area’s population, it includes well over half of the region’s homeless counted, (table 34, p. 39). So fitting local base population to local homeless count isn’t entirely straightforward. Still, outside of metro areas these problems are diminished.

In the chart below, I draw upon homeless count data while making a best guess as to what constitutes a local population to set a baseline risk of experiencing homelessness in each community where counts took place. I mostly use municipalities here, but switch to metro area or regional district where suggested by the Homeless Count Report. For BC communities, I order by population size. But I also include, for comparison purposes, baseline risks calculated from homeless counts for a few other big cities (Calgary, New York City) as well as King County (Seattle) and LA County (Los Angeles). Data for US cities come from a big report to Congress also released in 2018.

Pulling all the data together, it appears that homelessness is definitely not just a big city problem. Tiny little Merritt, BC, appears to have the same baseline risk of homelessness (1.4 in 1,000) as Metro Vancouver. More strikingly, the little communities of Nelson and Salt Spring Island seem to have nearly 8x the risk for people experiencing homelessness as Metro Vancouver. These estimates reveal greater prevalence of homelessness for these places than we get from population-adjusted counts in New York City, Los Angeles, or Seattle.

Let’s put some bands around these estimates and put them back on the map, where the southern half of BC (rightfully) takes its place as centre of the world. In comparative perspective, while Metro Vancouver contains a LOT of people experiencing homelessness, the overall risk of experiencing homelessness at any point in time seems strikingly low, putting the area on par with other communities like Cranbrook, Merritt, and Comox Valley. The risks of homelessness seem higher in other large communities, including Nanaimo, Greater Victoria, and Kelowna. But it’s the high risks in small communities; Smithers, Terrace, Prince Rupert, Port Alberni, Nelson, and Salt Spring Island, that really stand out. Each of these little communities looks like the big cities to our South in terms of the base risk of experiencing homelessness.

For comparison’s sake, let’s see what happens if we use the (more generous) base populations of Metro Areas (CMAs and CAs) from BC Stats in dividing count data to assess risks of homelessness. Does much change?

Not really. A few communities (e.g. Nanaimo, Williams Lake, Vernon) move down a category, but we don’t see major shifts, which is encouraging. Still lots of caveats remain with respect to the data: is the quality the same across communities? How do count methods differ? Check the reports for these and other details, and by all means have a look at the reports and play with the data yourself! I’ll park my little excel datasheet here in case anyone wants to check my work or use it.

(And yes, yes, in case you’re wondering I’m still hoping to transition to a nice transparent R system with GitHub support later this year, but I’m… slow… and sometimes excel with hand-entered data – it’s artisanal! – works ok too).

# A visit to Union St.

The resumption of Spring in Vancouver found me biking down to Union St. to grab some delicious Portuguese sweet bread from Union Market (near Hawks Ave). Highly recommended! Union Market is one of these old store fronts that popped up along an otherwise residential street prior to the advent of zoning. It was grandparented into the neighbourhood’s current RT-3 zoning through allowing:

“Dwelling Units, up to a maximum of two, in conjunction with a neighbourhood grocery store existing as of July 29, 1980, subject to the provisions of section 11.16 of this By-law.”

For anyone in love with these little corner grocery stores (and there are a lot of us), it’s striking and bizarre that we don’t allow them in residential zones anymore. But Union Market isn’t actually a corner store. Why? Because this beautiful old townhouse complex right next to Union Market occupies the corner.

To those of us who love townhouses, it’s striking that this form of housing is also forbidden from the majority of residential zones in Vancouver. Here it’s grandparented in with RT-3 zoning to preserve the pre-1920s cityscape of Strathcona. It’s even got heritage designation. It’s clearly a valued streetscape. So why isn’t this form of housing allowed on other residential streets?

Just down the street I passed another old store front. This one is also heritage (if you squint, you can see the marker near the door). But it’s no longer being used as a store. It’s been turned over to residential use. The whole lot was redesigned to support three different residential units about twenty years ago (1999). One in the back (along the laneway), and two up front. (UPDATE: as noted by an observant twitter user, the storefront portion seems to be AirBnB‘d, so it retains an ironic (?) commercial use…)

The history of this lot is fascinating, as revealed in a staff recommendation to City Council from 1999 supporting variance from existing RT-3 zoning to enable its renovation:

Heritage Value: The site at 658 Union Street is listed in the “B” category on the Vancouver Heritage Register and is noted as being an “unique example of [an] early multiple structure”. Three distinct structures were built on the site between 1893 and 1913:

· middle dwelling: the earliest structure was a one-and-a-half storey dwelling built in 1893 and situated in the middle of the lot; only the foundations and wall fragments remain;
· rear dwelling: the next was a one-and-a-half storey end-gable frontier-style structure erected sometime between 1901 – 1912, along the rear property line; and
· grocery store: the final addition was made in 1913, after street levelling activities in Strathcona were complete; it is a two-storey clapboard-sided grocery store with tin cornice, which abuts the Union Street edge of the property.

The latter two extant buildings are representative of architectural and historical themes unique to Strathcona including: Strathcona’s working class heritage; urban change/street levelling activities and community response; and vitality of neighbourhood through grocers and other home-based business.

Of note: the valuable “working class heritage” of Strathcona is precisely what was zoned out of single-use residential neighbourhoods of all sorts (RS, RT, and RM) in subsequent years. And supporting the renovation of the old store front required all sorts of variances from the RT-3 zoning currently in place. The administrative staff helpfully catalogued all the variations proposed:

That’s a lot of variance! It also demonstrates nicely just how many restrictions around development are currently in place in zoning by-laws. And this, of course, is simply to restore the “working class heritage” of the lot. Were there objections?

Oh yes:

As part of the Development Application review process, a sign was placed on the site and 47 surrounding neighbouring property owners were notified. Eight neighbours responded. Four support the project in its entirely, including the immediate neighbours to the west. The four others support the conservation component of the proposal, but have the following principal concerns:

· the proposed site coverage leaves little useable outdoor space at grade;
· the building length next to the east property line is excessive;
· entries, decks and coach house configuration would significantly detract from the privacy of the property to the east;
· the extent of proposed changes to the existing rear structure is excessive;
· the configuration of the front unit lends itself to conversion to an illegal secondary suite;
· the current difficulty of finding parking on the street will be exacerbated; and

· the proposed development is too dense relative to the single family dwellings typical of this block.

Strikingly, the neighbourhood association supported the retention (with some caveats) and the half of respondents supported the project in its entirety, with the remainder supporting parts. The most pertinent objections came from the property to the East of the lot. A variety of alterations were made accordingly. But the basics of the renovation remained, and ultimately the lot provided room for three units, subsequently stratified, and now assessed as worth $832k,$868k, and $1,100k. That’s a lot for twenty year old dwellings, and probably well beyond “working class” territory. But most of the value, as always in Vancouver, is in the land. What do we get in allowing three households to split the cost of the land rather than one? Well, the single structure beat-up house next door (to the West, partially obscured by the tree above) is assessed at$1,568k. Even taking into account a bit of land lift, each of the three twenty year old units created remains far cheaper than the nearly hundred-and-twenty year old unit (likely in need of some repairs?) next door.

What about that persnickety neighbour to the East? Well… about that… some dozen years after the old store front building was re-done, the lot to the East was entirely re-developed through a lot assembly with the adjoining house. Now the redevelopment of the two lots support and serve as their own heritage infill case study.

It’s pretty fancy! What’s striking is that the two lots together now support SEVEN different dwelling units, centred around an interior courtyard.

And how much are these new (2013) units? They’re assessed from $523k all the way up to$1,259k (I’m assuming for the big laneway house at the back). In other words, none of these practically brand new units reach the price of the run-down old house on the lot two down. Why? Because they’re sharing land costs. Here’s what the four lots in question look like from the back, via Google Maps satellite view:

Even though there’s been uplift in the land value with the permission of extra density on the two redeveloped lots, the uplift still doesn’t come anywhere near cancelling out the benefits of sharing. From left to right for the lots centred above, beginning with the partially shaded lot containing the old car covered in vegetation, here are the assessed land values for the lots:

• one lot with one dwelling*: $1.523 million =$1,523k/unit
• one lot with three dwellings: $2.291 million =$764k/unit
• two lots with seven dwellings: $3.878 million =$554k/unit

Despite the benefits of sharing land, none of the ultimate unit prices (ranging from $523k to$1,259k) seem likely to provide stable and affordable housing for the working class households of today. For that, we’ll need more purpose-built rental and social housing. But both of these things become more viable when land costs can be shared across units. Maybe the best way to insure that the working class heritage of Vancouver continues on into the future is to enable and support purpose-built rental and social housing everywhere – especially in the places this kind of housing has historically been excluded. Vancouver’s re-legalization of duplexes on RS zones and moves forward on Making Room are probably good steps along the way.

*- It’s actually unclear how many dwellings are contained in the dwelling with the car in the backyard because we don’t know whether it’s been subdivided to contain one (or more) suites. Legally there is only one dwelling available to be owned.

# How are condos used?

Comparing How Condos are Used Across Canada

Co-authored by Jens von Bergmann; Nathanael Lauster; Douglas Harris (Cross-posted at mountainmath.ca)

Condominium apartments are fascinating! At their heart lies a relatively recent legal innovation enabling individual ownership of units in multi-unit developments. Since their arrival, condominium apartments have become places to build homes, sources of rental income, sites of speculative real estate investment, and experiments in private democratic government. They’re also in the middle of many on-going debates about housing and the future of cities in Canada and around the world. In 2018, we formed a team to study condominium apartments and how they were being used in order to better inform public and academic debates. Team members include data analyst and mathematician Jens von Bergmann, sociologist Nathanael Lauster, and law professor Douglas Harris. We recently presented some preliminary findings at the National Housing Conference in Ottawa and we’re looking forward to continued research collaboration.

Here we make public some basic information about the development and use of condominium apartments across different metropolitan areas in Canada.

The first thing to note is that the legal architecture of condominium is deployed across a broad range of structure types. In addition to apartments, developers commonly use the condominium form to subdivide row houses, and occasionally single-detached houses (as in some gated communities). Nevertheless, condominium is used most commonly to subdivide ownership in low-rise and high-rise apartment buildings, and that’s what we focus on here.

The next thing worth noticing is that condominium is much more common in some metro areas than others. Vancouver jumps out for the proportion of its apartments – and housing stock overall – owned within condominium. Calgary and Edmonton also rely heavily on condominium to subdivide apartment buildings, although these sprawling metro areas are dominated by single-detached houses, much more so than Vancouver, reducing the overall prevalence of condominium.

We know that condominium apartments are exceptionally flexible forms of housing, but how are they being used across different metro areas? What proportions are owner-occupied? Rented? Occupied temporarily? Unoccupied?

We couldn’t extract data to answer the last two questions from the census because condominium status is recorded by respondents. However, using a variety of datasets, we figured out a transparent and replicable (if somewhat complicated) method for estimating temporarily occupied and unoccupied condominium units.

The answers to these questions about how condominium apartments are used speak to important elements in popular discourse and public debate. Since provincial governments introduced a statutory form of condominium in the late 1960s, developers have built condominium buildings rather than purpose-built rental apartments across much of Canada. Does this also mean that the proportion of owner-occupiers increases while that of renters decreases in cities where condominium developments proliferate? Or do owner-investors rent out their condominium units, augmenting the existing rental stock?

Our findings on how condominium apartments are used are really interesting! In all the metro areas we analyzed, the modal use of condominium apartments is owner-occupation. As a result, it appears that condominium apartments are enabling more homeowners to live in increasingly dense cities.

However, condominium apartments also make up a substantial proportion of the rental stock in many metro areas. While many condominium apartments are rented, relatively few show up as vacant (i.e. empty but listed as “for rent”) at any given point in time. Here we distinguish these rare vacancies, which are good for renters, from unoccupied condominiums. In tight markets such as Vancouver and Toronto we see effectively non-existent condominium apartment vacancy rates, comparable to purpose-built rental vacancy rates.

The least common use of condominium apartments is as a temporary residence (where owners declare their principal residence as somewhere else in the census, but occupy the unit occasionally).

Finally we get to the “empty condos,” or those that show up as unoccupied in the census. Overall, we estimate that between 10% to 23% of condominium apartments were unoccupied in 2016, depending upon the metropolitan area. We don’t know why so many condominium apartments appear to be unoccupied, but it likely relates to their newness and to their inherent flexibility as property. Flexibility can show up in the census as “unoccupied” directly, as when owners use condominiums as second homes, and indirectly, as when condominium apartments are left empty in order to facilitate transactions between uses. We suspect that condominium apartments may cycle more frequently than other forms of property between different uses and occupants, thus creating transition periods without occupants and inflating the proportion of unoccupied units. For instance, condominium apartments can more plausibly be re-claimed for owner’s use than purpose-built rental apartments, cycling in an out of rental supply and potentially creating less stable rental housing.

Strikingly, Vancouver and Toronto stand out as having the lowest proportion of unoccupied condominium apartments, a finding that may be somewhat counter-intuitive given the public attention that vacant units have received, rightly or wrongly, in both cities. When metropolitan areas rely upon condominium apartments as a key form of new housing supply, they should take the flexibility of the form into account. However, it appears that the proportion of unoccupied units in the housing stock will rise as the proportion of condominium apartments in the housing stock increases because condominium apartments are more likely to be unoccupied than purpose-built rentals, a pattern also noted with respect to other flexible housing forms, such as secondary suites (especially basement suites, which show up as units in a “duplex” in the census). This means that even though a smaller proportion of condominium apartments are unoccupied in Vancouver than elsewhere in Canada, a larger proportion of Vancouver’s housing stock shows up in the census as unoccupied.

In Canada’s three largest metropolitan areas, a pretty simple rubric applies: for every ten condominium apartments built, six are owner-occupied, three are occupied by renters, and one is unoccupied. In Calgary and Edmonton, add a renter and take away an owner-occupier. The data for the other cities we surveyed is available in the graphic above. As a bonus, we also provide a comparison with estimations from 2011 data to show changes over time in the graphic below.

In Vancouver, where condominium apartments have been an established part of the housing market for longer than in the rest of the country, there is very little change in the occupancy pattern between 2011 and 2016. In other big metropolitan areas, it appears that condominium apartments are increasingly used as rental stock. In most cases, the proportion of empty condominium apartments appears to be decreasing, something that may reflect the lingering effects of the 2008-09 property market crash. However, this is all very preliminary. But we’ll keep looking at the details as we proceed!

## Methods

We mixed two data sources to arrive at these estimates–the Census and the CMHC Rental Market Survey–and that made coming up with the estimates a little more complicated. There are several assumptions that go into the estimates, and there are several issues with mixing the data that we set out below.

### Overview

We cut the condominium stock into five different categories. The numbers of units occupied by owners and renters are straight-up census estimates from 98-400-X2016219 and 99-014-X2011026. To estimate the unoccupied units and the units occupied by temporary residents we used a custom tabulation of Structural type by Document type. We received this cross tabulation from Urban Futures, which one of use has worked with before on secondary suites. Both of those variables–the categorization of the dwelling type as well as the decision to label a unit without a census response as empty or occupied by someone who did not respond–is made by the enumerator. This allows us to ascertain the structural type of unoccupied units, and we can also get that information for units that are temporarily occupied.

So, we know how many apartment units were classified as unoccupied or temporarily occupied. To estimate how many condominium units fall into that category we need to make some assumptions. First, we assume that the apartment stock consists of three distinct type of units: condominium units, purpose-built rental units and non-market housing units. That’s not quite accurate. For example a single-family home with two secondary suites will be classified as an Apartment, fewer than five storeys if the census found the suites. These do exist in Vancouver, and elsewhere, but their numbers are small.

Given those three types of apartment units, we need to understand how many of the unoccupied and temporarily occupied units fall into each category. The CMCH Rental Market Survey has annual estimates of vacancy rates and universe size for the purpose-built rental stock. We take those estimates, only counting apartment units, to attribute unoccupied units to the purpose-built rental stock. In Vancouver, with its extremely low vacancy rates, this is a fairly small number. In Halifax, that number is comparatively larger. Further, we assume that the non-market units have a vacancy rate of zero, so that there are no empty non-market units. What’s left over we assign as empty condominium apartments.

Finally, we use the estimate of vacant condominium apartments and those on the rental market from the CMHC Secondary Market Rental Survey, using their estimates of the condominium vacancy rate and the condominium rental universe. The vacancy rate is not available for all years and all CMAs. We have marked the CMAs with an asterisk in case the data was not available and back-filled it with our estimate of the condo rental universe and the Rms vacancy rate. We have seen previously that the Rms vacancy rate tracks the secondary market vacancy rate reasonably well.

Attributing the temporarily occupied units gets even harder, but the numbers are smaller so getting things a little wrong has less impact. Here we again assume that no temporary residents live in non-market housing, and we assume they are equally likely to live in a condominium apartment (as owner or renter) or rent in purpose-built. That is a bit of a judgement call, but the details of these assumptions don’t make much of a difference to the numbers, and we invite people to grab the code if they would like to adjust the assumptions.

There are several issues when mixing CMHC Rms data with census data. For one, both are point-in-time estimates for slightly different times. The census is pegged in early May, the Rms for October. There may be fluctuations in temporary and unoccupied units, in particular in areas dominated by universities such as Waterloo, with the census being outside of the regular semester and the CMHC survey within.

Next comes the geographic problem, with CMHC switching to new census geographies at the end of the year, so the rental universe still reflects the previous census geography. Montreal is one such example where the CMA changed 2011 to 2016 as we have explained before. That leads to problems when estimating the rental universe, but the effect is moderated when focusing on the empty units.

Another issue is that the definition of apartment that CMHC uses differs slightly from the census.

Finally, for estimating the vacant condominium apartments that were on the rental market we used the CMHC rental condo universe estimate and not the one we derived from the census. There appear to be some differences in how CMHC and the census estimate rented condo units, with CMHC relying on surveys of property managers. In BC that likely involves tallying up units for which Form K was filed, likely leading to CMHC under-estimating strata rentals.

It is instructional to compare the two different estimates.

With the exception of Hamilton, the census condominium rental estimates are higher, in some cases substantially so. To shed more light on this we also compared the estimates of overall condominium apartments.

We looked at two separate census estimates: the occupied (by permanent residents) units that come straight from the census by filtering occupied units for apartments that are stratified, and the overall condo estimate that we derived by adding in vacant and temporary units. With the exception of Montréal the census estimate of occupied units only comes quite close to the CMHC condominium universe estimate. The differences are worth looking into in more detail at some point.

### Waffle graphs

To communicate the makeup of condominium apartments we settled on a custom version of a waffle graph. Displaying proportions on a square grid makes it easier to read them compared to pie charts or tree graphs. The 10×10 layout rounds numbers to percentage points, which is the appropriate level of accuracy given the uncertainty in the data and is intuitive to understand. When rounding to the nearest percentage, the numbers don’t always add up to 100. So we don’t do traditional rounding but round with the constraint that the total adds up to 100 while minimizing the $$l_\infty$$ error.

This does introduce potential problems when comparing across time or across geographies, where theoretically we could see an increase in the number of squares in one category although the actual estimated share dropped. This will only happen under very specific circumstances, and we checked that this did not occur in our graphs.

## Reproducibility

The code underlying this post is available on GitHub, as are the parts of the custom tabulation for 2016 and 2011 used in this post. Part of the Statistics Canada data we used requires conversion from XML into more manageable data format which, for performance reasons, requires python to be installed next to R that runs the rest of the code.

# Urbanism Axis & IMBY Allies

Last post I talked about how Vancouver’s election provided possible insights into the relationship between Urbanism (or IMBY-ism) as a political axis running perpendicular to more traditional Left-Right axes. Now we have results!

But first: a clarification. Initially I drew from the Cambie Report‘s clever crowd-sourcing of placements for political parties and prominent independents in Vancouver along both a municipal urbanism axis and social and economic left-right axes (ultimately combined), as follows.

Positions were allocated (and defined) by the wisdom of crowds. Not surprisingly, I received some pushback for accepting this wisdom – not everyone agrees with crowds! And that’s fair. Many policy positions and histories, especially within my area of housing, were actually more nuanced. Here I’ll provide a – still very rough – breakdown of how I see the axes providing important information about different positions and histories, which may be of use both for interpreting Vancouver and thinking through IMBY coalitions more broadly.

Starting with the Urbanist Right in the upper-right quadrant, we have a relatively familiar market urbanism: anti-zoning and libertarian inspired. Pro-housing everywhere “the market” wants it. Market urbanists tend to extol the virtues of density and disruption. YIMBY everywhere.

Moving clockwise, below we have the Preservationist Right. To add a bit of nuance, this is a position that I’ve argued actually much better characterizes the North American tradition: Rigid zoning for exclusive single-family neighbourhoods and more flexible market allocation of housing within a constrained urban core. Right leaning municipal coalitions offer a grand bargain between middle-class detached homeowners’ relatively conservative desires to be left alone and developers’ interests in making money downtown. The mantra goes something like: “Strong protections for me and the market for thee.” Or NIMBY in the Great House Reserve, YIMBY in the Urban Core.

Continuing around the clock to the Preservationist Left, we move toward the left-leaning reaction against the North American tradition. Anti-poverty alliances frequently identify developers as villains. This makes sense insofar as many alliances have borne repeated witness to the displacement that can result from unleashing market development upon the marginalized neighbourhoods of the urban core. Anti-developer politics can seem like a progressive end in their own right and can sometimes also win over middle-class voters (think “All neighbourhoods matter”). For a farther left subset, the socialization of housing seems the best bet for protecting those marginalized by the market. Lots more social housing is in order – but often concentrated in and meant to preserve neighbourhoods viewed as under development pressure. The orientation runs from NIMBY everywhere to PHIMBY (Public Housing in My Back Yard).

Moving up to the Urbanist Left, we find alliances that often view urban growth as good, both in terms of promoting diversity and in terms of reducing environmental impacts. Many accept that disruption is part of living in a city. But it shouldn’t be imposed unequally and policies should work to avoid displacement. Those I’ve also termed Inclusive Urbanists set their sights on returning exclusionary neighbourhoods to the urban fabric by reforming single-family zoning. They look to introduce social housing and diverse rental options to every neighbourhood. The tendency is YIMBY, but reform-oriented, with an egalitarian emphasis directed at diversifying single-family exclusionary hoods and large helpings of PHIMBY.

Ok, now let’s get back to what happened in Vancouver, where we had parties occupying each of these quadrants. Who won? And what does it tell us about IMBY-coalitions?

The mayoral race ended up a showdown between two strong Urbanist Left candidates (Kennedy Stewart and Shauna Sylvester) and the strongest of the Preservationist Right (Ken Sim from the NPA – historically the epitome of the grand bargain party: NIMBY for detached home-owners, YIMBY in the urban core). Stewart beat Sim by less than a thousand votes (half a percentage point) in a real squeaker of a race.

Broadly speaking, both the Urbanist Left and the Preservationist Right fielded strong candidates. But what happened to the Urbanist Right and the Preservationist Left? The Urbanist Right candidate was clearly Hector Bremner, and at 5.7% of the vote, his Yes Vancouver party failed to attract much support beyond its passionate young base of market YIMBYs. The Preservationist Left initially had a party-supported candidate in COPE’s Patrick Condon (who early on made clear he wouldn’t run if the Green Party’s Adriane Carr had decided to try for the mayorship). But Condon dropped out after experiencing a stroke. Late in the race he endorsed independent candidate Sean Cassidy, who failed to attract much other support. Even adding in support for the fringe candidacy of IDEA Vancouver’s Connie Fogal (widow of progressive hero Harry Rankin), the Preservationist Left failed to crack 2% in the mayoral race.

So can we collapse the field, calling YIMBYism a left-wing phenomenon and NIMBYism mostly driven by more conservative impulses? Not quite so fast… let’s turn to Council! Here I note the average % of total council votes accorded to candidates in each party, highlighting the parties that actually won council seats (no independents won seats, despite overall respectable showings). I also provide the range of averages for major parties in each quadrant.

While the mayor’s race seemed to reduce relatively neatly to a singularly important Urbanist Left – Preservationist Right axis, the council race sees a real and strong split between the Preservationist Left (COPE and the Greens) and the Urbanist Left (OneCity and Vision), with the Preservationist Centre-Left ultimately receiving the most support. Indeed, the new council will be made up of five members from the Preservationist Right (NPA), four members from the Preservationist Centre-Left (three Greens and one COPE), and one member from the Urbanist Left (OneCity).

So what does this all this tell us about IMBY coalitions?

First: it’s important to distinguish those BYs: the backyards of Single-Family House neighbourhoods are treated differently from the backyards of the Urban Core.

Second: Right-leaning coalitions tend to do well in cities only when they leave the back yards of Single-Family House neighbourhoods alone. So far there’s little evidence that a right-leaning YIMBY coalition can win, though this could change in the future, as single-family neighbourhoods continue to lose population.

Third: Most YIMBYs lean left by quite a large margin, and left-leaning YIMBY coalitions can win. A Lefty Urbanist won the mayoral race in 2018, despite the competition from an exceptionally strong contender in the same quadrant. Moreover, Vision Vancouver’s coalition held power for the last ten years until their organizational implosion in 2018, and still placed well ahead of most other parties even if they won no seats.

Fourth: There’s a real and consequential split between Lefty Urbanists and Lefty Preservationists. I think this is often about perspective. From the point of view of anti-poverty activists working in the urban core, developers almost always look like villains (non-profit developers MAY be exceptions). From the point of view of people feeling excluded from cities’ vast tracts of single-family neighbourhoods, developers look like potential allies. On the flip side, the path to political success often runs through middle-class homeowners, and it’s easier to get them on your side by promising it won’t inconvenience them much than by suggesting they might need to sacrifice some parking or sunlight on their gardens. Vancouver’s Green Party, in particular, has walked this line to great success.

Theoretically, this election should put to rest the notion that all or even most YIMBYs are mostly market-oriented. After all, if they were they would’ve come out in mass for Yes Vancouver. Instead YIMBYs seemed to support Urbanist Left candidates in numbers easily surpassing support for the Urbanist Right. But to be fair, this was also a really messy election, witnessing the organizational implosion of the reigning party (Vision) and a confusing profusion of new parties. This likely benefitted those older organizations that managed to avoid imploding (Greens, COPE, and above all the NPA, who recovered strong from a shaky start), above and beyond informed platform comparisons.

Final Question: Can those Urbanist Leftys who made it into office this year work with their Preservationist Left or Preservationist Right colleagues? I’m guessing efforts to Make Room in single-family neighbourhoods are going to slow down again after Vision’s successful last-minute drive to introduce duplexes (with suites!), opening up all nearly all lots to four potential dwelling units (2x owned, 2x rented) across the City. Given housing plans put forward by the NPA and Greens, maybe we’ll eventually get matching legalization of an additional main unit rental suite (1x owned, 2x rented, 1x laneway rented), legalizing what’s already happening on the ground in many places. It’s less clear what will happen in the urban core, where alliances may shift project by project (remember, NPA councillors look pretty market YIMBY outside of single-family zones).

Let’s animate that GIF:

Addendum: (preliminary) vote tallies from the City of Vancouver obtained here! Looking forward to the voting location breakdown we got from the 2014 election.

# How to Become a Resort City

I think many people are concerned that Vancouver is turning into a “resort city” – a playground for the rich – rather than a diverse and thriving city for all. But is this really happening? Yes. And I’m concerned too!

Certainly we see a lot of luxury cars and retail outlets, but we don’t really track wealth very well in Canada. Nevertheless owner-occupiers report their home values in the census and that’s where most wealth ultimately lies for Canadians. We can ask a simple question: where do the millionaires live? By millionaires I’m referring to anyone who reports owning and occupying a home worth a million dollars or more – not a perfect proxy, but not bad. The census uses self-reporting to get at this, and Statcan Table 98-400-X2016232 – in conjunction with total household numbers from Census metro area profiles – enables me to generate the following figure.

Vancouver contains almost as many millionaires as Toronto, despite being less than half the size. Together the two metropolitan areas account for less than a fifth of Canada’s total number of households but over three-quarters of all owned dwellings worth on million dollars or more. To put matters differently, nearly 25% of Metro Vancouver’s households own a dwelling worth a million dollars or more, compared to just over 1% of the Rest of Canada (outside Toronto). We are literally concentrating the One Percent in terms of Canada’s wealthiest households.

Maybe Vancouver’s not quite a resort city yet, but it’s definitely in the neighbourhood!

So how did we get to this point?

In some ways Vancouver was predisposed to growth because it’s got lots of things people want: ocean, mountains, one of the mildest climates in Canada, a thriving port, a railroad line, lots of jobs, diverse ex-patriot communities, parks, strawberry-picking in the Agricultural Land Reserve, etc. This could explain growth overall. But everybody wants these things, not just rich people.

If it’s not just amenities that attract specifically rich people to Vancouver, then how are we becoming a resort city? Some people blame the fact that Vancouver’s amenities have been heavily marketed to rich people around the world in recent years. They have a point: this has certainly happened. But lots of other places try to market themselves to the rich as well. What makes Vancouver special?

We might not be attracting the wealthy so much as we’re systematically excluding everyone else who wants to live here. It’s about changing the composition of in-movers, so that wealthier and wealthier people tend to come (incidentally, this fits with research on neighbourhood change suggesting this is mostly how gentrification occurs).

How are we systematically excluding everyone who’s not rich? Easy! Under conditions of growth, all we have to do is preserve lots of urban land for millionaires and largely prevent anyone else from competing with them. Effectively this is what residential single-family (RS) zoning accomplishes in places like Vancouver, which we can see by comparing what proportion of detached and duplex housing is evaluated at over a million dollars.

Across Metro Vancouver, nearly 60% of detached or duplex housing is evaluated at over one million dollars, yet the vast majority of urban residential land is zoned to support only these forms of housing. This is how you get a resort city, fit only for millionaires (with a little bit of room for their servant-tenants living in basement suites below – which is what duplexes mostly consist of in Canada). This compares to around of quarter of detached houses worth over a million in Toronto, and just over 2% in the Rest of Canada.

We can flip the question around to ask what percent of million-dollar dwellings are single-family detached or duplex dwellings. Strikingly, the answer in Metro Vancouver, Metro Toronto, and the Rest of Canada is pretty much the same: almost 90%.

Luxury zoning is almost entirely detached zoning. Across Metro Vancouver, the converse is also increasingly the case: detached zoning is becoming luxury zoning, affordable only to millionaires.

In places with lots of amenities – including jobs! – where people really want to move, growth is mostly limited by housing. If we only make housing for millionaires, we’ll increasingly have a city of millionaires. If we want to keep Metro Vancouver from becoming a Resort City, we’re going to have to tackle the zoning issue.

Thanks to Frances Bula, Jens von Bergmann, and Chad Skelton for inspiring today’s post and/or snatching away my afternoon! And if you’re interested in what you can do about reforming zoning, look into the platforms of municipal parties, like OneCity, that advocate for inclusivity across our urban landscapes.

# A Rough Working Guide to Housing Crises and Policy Levers

The housing market: it’s all about supply and demand, right?

Not quite. States and markets – especially markets for housing – grew up together. Housing is heavily regulated in ways that generate, constrain, and ultimately channel supply and demand, creating not one, but many sub-markets and also non-markets. The policies governing housing have been layered one atop the other through history.

Where do these policies come from? They often respond to perceived crises. For instance, Vancouver enacted its first Fire By-law on July 19, 1886, just five weeks after the newly incorporated city burned to the ground. As a result, the policies set in place to deal with various crises are frequently reactive in nature. They also tend to be crafted in the image and interests of those most powerfully situated to govern.

Since the late 19th Century, policies across North America (and elsewhere) also tend to respond to an ideological background of “market fundamentalism,” or the idea that the market governs best (sometimes referred to as “neoliberalism”). As Karl Polanyi diagnosed early in the 20th Century, markets are terrible at governing some things (people, nature, productive capacity). As I argue in my book, urban land and the housing built atop it is one of these things. This sets up an interesting dynamic whereby ideological attempts to govern by a singular market create all manner of housing problems, which in turn generate reactive policy responses and explain why housing is so heavily regulated. The most frequent policy response, I think, has been to set up protective partitions within markets. This produces sub-markets and non-markets and helps explain the distinct nature of our various housing crises.

In what follows, I’m going to attempt to provide some insights into Vancouver’s current housing crises in a way that gets at the history of this partitioning of markets. The past barriers we’ve created to generate, constrain, and channel the market forces of supply and demand continue to shape policy levers available to us today.

Here’s a working visual guide (link to larger version). Sorry: it’s still pretty rough and necessarily messy, so it might not “work” for everyone!

Ok, so what’s going on here? Just to get it out of the way: “SFD” = Single Family Dwelling and “PBR” = Purpose-Built Rental. Combined together with “Condo” and “Non-market” housing, we’ve got our (heavily simplified) major basic forms in which housing is provided.

We can link these basic housing forms into how they relate to our perceptions of four distinct housing crises. I’ve distinguished these housing crises in terms of need, primary group of interest, and how Vancouver is doing addressing these crises in comparative perspective (more here).

Crisis #1: Homelessness involves the greatest need for housing, mostly affects the poor, and in comparative perspective, Vancouver actually has a not-terrible track record addressing this crisis. Though it’s gone up in recent years, the prevalence of homelessness remains relatively low. But the homeless are those most likely to be at risk of dying due to their housing situation. Everyone else’s crises pale by comparison.

Crisis #2: Rent Access involves high and immediate need for housing, mostly affects the working class and/or young, and Vancouver has a mediocre track record addressing this crisis. Rents remain relatively middle-of-the-pack for North American cities, but this is largely due to rent control and vacancy rates are very low, making it difficult for new entrants into the rental market. Failure to provide rentals can lead to real hardship and (ultimately) homelessness.

Crisis #3: Housing Price to Income reflects relatively low need, mostly affects the aspirational middle class, and Vancouver looks awful in comparative perspective (at least within North America). Since it affects the middle class & Vancouver does pretty terrible compared to elsewhere, this crisis sucks up most of the attention in Vancouver’s debates, despite reflecting relatively low need compared to the rental and homelessness crises.

Crisis #4: Return to Investment is unlike the other housing crises insofar as it reflects a more general crisis involving finance. Correspondingly, I place it as the lowest of needs, and it tends to affect an investor class already most protected by their assets. Recently, at least, Vancouver’s housing market has provided a very good return on investment. No real crisis here (despite recent attempts to set in motion a property tax revolt).

I’ve tried to cram as much of the history of reactive housing policy as I could into describing the barriers channeling supply and demand, with particular reference to Vancouver. Each letter describes a partition in markets and also acts as a policy lever. The letters identifying each lever / barrier are ordered in rough historical fashion. When was each barrier or lever put in place? Meanwhile, the arrows follow the production of housing from its monetary backing (capital – in green) through its development (in blue) to one of the four major forms of housing described above, and finally to its end use (consumption) as non-market, primary rental, secondary rental, resident owner-occupation, and empty or other.

It’s complicated! And I’m still leaving a lot out (e.g. non-profit organizations) and simplifying in places (e.g. treating resident-owners as strongly distinct from investors).

Nevertheless, a few points are worth making:

1) there is no such thing as a singular “housing market.” Instead there are many little sub-markets and non-markets produced by the layering of policies.

2) sub-markets are still connected to one another, but the connections are shaped by policies and determine how processes like “filtering” are variously enabled to work.

3) this reflects real concerns that there can be a “right” and “wrong” kind of supply. The overall history of housing policy insures that there really are different kinds of supplies that end up addressing different kinds of crisis. At the same time, supplies and crises do relate to one another. The trick is that the relationships aren’t at all straightforward, and some kinds of supply (e.g. condos) have been made more flexible than others.

4) earlier and wider ranging policies (e.g. A & B, covering tax, finance and land use) can have really wide-ranging effects on how housing works. For instance, shifting from extensive to intensive land use policies potentially unlocks a wide range of new housing stock that could help address all kinds of housing crises. Later policies tend to have more limited effects, carving off new little sub-markets (e.g. Accessory Dwelling Units) and specifying their links to other little sub-markets. But these links can be important!

5) sometimes the effects of policies can be complementary or conflict, depending upon the crisis at hand. For instance, Empty Home Taxes (policy lever F) diminish the range & value of benefits for investment (crisis #4, lowest need), but encourage the channeling of existing supply toward resident-owners (crisis #3, low need) and rentals (crisis #2, high need), ideally benefiting both the middle and working class. By contrast, a broader tax on use of housing for investment (policy lever A) without any exemption for rental tenancies would channel even more existing supply toward resident-owners (crisis #3, low need), but possibly exacerbate the options available for renters (crisis #2, high need). Rental restrictions within strata associations do the same thing. One concern with enabling the stratification of ADUs (policy levers D & G) – currently limited to rentals – is that it may have a similar effect, pitting the working class and middle class against one another.

6) am I already at six?!? I have lots more to say about this working model, which I began sketching months ago just to begin thinking through some of housing policy change we’ve seen and continue to see in Vancouver. With luck, I’ll work toward turning it into a paper! But as it is, this blog post is already too long. What’s your working model look like?

# Visualizing Succession

After my talk at the Vancouver Historical Society, I’ve been playing around with some of the old fire insurance maps for Vancouver. As it turns out, fire insurance maps are super-awesome resources for historical research. Here I want to explore ways of using them to visualize succession as an urban process.

Succession, meaning the gradual outward expansion of an urban core, was thought by many observers of the early 20th Century to be the key process driving neighbourhood change. For sociologists like Ernest Burgess and Roderick McKenzie, succession was a process of “invasion” from the centre outward through a series of concentrically organized ecological habitats. The central business district (Habitat I) was always attempting to move into the surrounding industrial and commercial areas (Habitat II), which in turn were always invading the areas of dense working class housing beyond (Habitat III), which were always trying to muscle in on the middle class and wealthy residential suburbs on the outskirts of the city (Habitat IV).*

“Succession” was a science-y and neutral sounding concept (having been drawn directly from plant ecology), while “invasion” was quite threatening, especially to the middle class and wealthy suburban neighbourhoods hoping to hold the urban core at bay (very much including its working class & minorities). With the invention, legalization and rapid spread of single-family detached zoning, the middle class managed to mostly surround and contain the urban core, as I describe in my recent talk and book. As a direct implication of halting the process of urban succession, constrained urban cores mostly grow up instead of out, and now we mostly talk about gentrification instead of succession as a key process affecting our cities.

But let’s go back to succession. Can we see it in action? Let’s start with an early residential development consisting of two blocks (subdivided by alleyways into six) on the outskirts of a rapidly growing city in the late 19th century.

The base layer here is an old fire insurance map from the Vancouver archives, though I’ve faded it out a bit to draw over top of it. Let’s highlight all of the houses.

We can see this is pretty much just a residential neighbourhood at this point, with the only non-residential use a church (lower-left). Most of the other buildings are sheds. But wait a minute, are these all single-family detached houses?

Trick question! Those weren’t invented yet! So nobody was recording whether or not single families lived in these houses, nor how they might have been subdivided. That said, it’s clear that four houses in the upper-left block were all connected to each other, making them recognizable as rowhouses. Meanwhile, there were at least two houses connected side-by-side on other lots, making them semi-detached houses (by modern Canadian census categories). Also notable: at least four lots had both houses fronting onto major streets and houses fronting onto alleyways – with the latter now identified as “laneway houses” in Vancouver. Lots of “missing middle” housing!

Importantly, these residential blocks were located on the outskirts of a growing urban core in 1889. What will become of them? Let’s jump ahead 25 years and see…

I’ve highlighted in red all of the buildings that disappear over the course of 25 years. Oh no! There goes our church! Over half of the houses are also gone. Devastation in Habitat IV! But at least the townhouses remain. Let’s fill in the neighbourhood…

Some of those old houses remain, and new ones have been added too, meaning that we don’t see a complete replacement of seemingly middle-class residential uses (Habitat IV). That said, for the most part torn-down houses have been replaced by much larger buildings that take up their lots and add abundant industrial, commercial, and office uses to the neighbourhood. At least three hotels (Alcazar, Cadillac, and Canada) join numerous rooming houses, with the latter being an especially important source of housing for the working class and poor (Habitat III). Industrial uses (machine shop, workshop, numerous printers, warehouse) also crowd the block, along with commercial uses and services (second-hand shop, steam laundry, tailor, restaurant). There’s even an undertaker with lodging above! Lots of what Burgess termed a zone in transition (Habitat II). And offices… so many offices. We can make a strong claim for central business district uses (Habitat I). In effect, the entire urban core has moved into a former middle-class residential district over the course of 25 years, give or take, without completely destroying it. On a first pass, pre-zoning urban succession looks more like urban mixing than invasion and replacement. But maybe we just didn’t give it enough time.

What does this marvelous neighbourhood look like now, some 100+ years later?**

It looks like the two blocks bounded by Pender, Hamilton, Dunsmuir & Richards, right smack in downtown Vancouver. The 19-story BC Hydro Centre Tower dominates these blocks. But remarkably, at least nine of the old buildings that replaced the original houses in the neighbourhood during the 1906-1914 era remain intact! Alas, the townhouses in the upper-left have been replaced by a parking lot. Overall, the neighbourhood still has a mix of low-income housing, commercial and service space, and office space, though probably not a lot in the way of middle class housing (have to seek out some condos nearby for that).

I grabbed the dates on all the buildings I could find from VanMap‘s Assessment data. Old Fire Insurance Maps can be found in the Vancouver Archives.

Here’s the Fire Insurance Map from 1889, with zoomed in panel below:

Here’s the Goad’s Atlas Map from 1910-1920, with zoomed in panel below:

This is a first-pass, of sorts, at drawing upon Vancouver’s old fire insurance maps to get at patterns of urban succession in the days before zoning. I’d welcome suggestions, collaborations, and better visualizations! (This was pretty much all done with powerpoint on the fly). In the meantime, here’s my cheap-o little animated gif of the slides above. Enjoy!

*- The Chicago School of Sociology modeled this process on Chicago itself – though McKenzie, it should be noted, also detoured to Seattle in describing “The Ecological Approach” to the study of The City (1925), in a volume for which Park & Burgess are usually given first billing.

**- The maps date from 1889 and somewhere between 1910 and 1920, which is why I say “give or take” in assessing the length of time that’s passed! Though the Goad’s Atlas dates itself as 1910, it was reproduced (with consolidated updates) in 1920, and includes buildings up to at least 1914, which is why I can narrow it down to around 25 years after 1889.