Saturday, December 31, 2016

Building affordable housing: what can work in urban areas

OK, I've been away long, just busy and without much motivation lately, with this follow-up hanging on me, preventing me from starting other posts. Anyway, if anyone still follows my blog, here is this article I wanted to finish before 2016 is over... in the last article, I pointed out that the economics of building new housing is not really favorable to affordable housing. If you only had construction costs to contend with, it would be possible to have reasonably large affordable housing, however, in places where land prices tend to be high, these impose a premium that make it really hard to build affordable housing of decent size.

Furthermore, the post-WWII idea of affordable housing, the starter home for young families, is based on a "grow home" model, the idea that houses be sold with the basement or attic unfinished, with the expectation that, as the family grows, so will the house, thus keeping the initial price low but allowing for expansions further on. This worked well in greenfield developments, but the problem is that, once "grown" the house cannot be ungrown for the next generation. So a "grow home" is only affordable once, then it gets progressively more expensive.

All of that leads to a simple question: is it even possible to have affordable housing in an urban area, where land is expensive and greenfield development impossible?

So let's see a few approaches that can make economic sense...

Very high increase of density, without going too high

Replacing existing buildings by higher density ones is the traditional approach to increasing density. However, this isn't a silver bullet. There are two main pitfalls to that approach.

The first pitfall is if the density of the proposed building isn't high enough. That's because developers who want to redevelop a property have to compete to buy that property with people who would want to buy it to live in it.
Two major types of buyers competing for a given property
That means that the minimal price a developer could pay is the market value of the building, because if he's not willing to pay that much, a potential resident will just outbid him and buy the house to live in. And usually, if a developer wants to build something, it's because the area is reasonably desirable and the market value isn't too bad. That means that the land price the developer has to pay is AT LEAST equal to the market value of the current building, to which you have to add destruction costs (not that high usually) and the construction cost of the new building.

So let's say we take this house...

Which could be worth 270 000$ in market value.

And we want to replace it with this triplex:
So let's estimate things...

The current house's market value is 270 000$, let's say 30 000$ for removing the current house and 300 000$ for building the triplex. That's 600 000$, let's add another 10% for profit margin, that's about 660 000$, while the building has about 2 900 square feet of livable space. That's about 227$ per square foot, or 220 000$ per 969-sf 2-BR apartments.

That's not necessarily very affordable, especially when you consider that this replaces a 3-BR 1260-sf house with another 1260-sf of basement that sells for 270 000$.

Now, if you could replace the house with an 8-plex like this one...
Then we add the value of the house (270 000$) and the destruction cost (30 000$) and the construction cost (1 030 000$), that comes up to 1 330 000$, plus 10% of profits, that's about 1 500 000$. That comes out to about 150 $ per square foot, or a bit under 200 000$ for a 1200-sf, 3-BR apartment. That's more reasonable.

Of course, that's ignoring factors like parking minimums, regulatory costs of rezoning and the like.

The saving grace would be a house in disrepair, a "fixer-upper" which market value would be lower than similarly-sized houses. Thus allowing for a lesser increase of density to have affordable housing as the effective land price is significantly lower.

The other pitfall is when the actual building is high enough density that the only way to add density in any significant manner is to go for a mid-rise or high-rise (4+ stories) building, which is significantly more expensive to build as they require concrete, elevators, water sprinklers, and other elements. So if the cost of construction of the new building alone is 200$ per square foot, that imposes a high threshold for housing prices.

So another approach might be needed.


So building a new building in replacement of an existing building is not a great proposal economically speaking. But what if you used the second of the three Rs? Reuse. After all, if you have a perfectly suitable building, maybe you don't need to destroy it to increase density. How? Well, by subdividing the existing building. Simply rearrange the interior of a large home to create many units out of just one.

For example, here is an example of a house that could easily be subdivided into a duplex:
A raised ranch in Québec
This building has a door that opens on a small lobby which has 2 sets of stairs, one that goes up to the upper story, the other that does down to a half-buried floor.

If you put a wall in the middle of the lobby and add another door to the outside, you separate the floors, possibly creating two 2-BR or 3-BR units. So if you build a raised ranch with an unfinished basement, you can finish the basement, maybe for 70 000 or 80 000$, then convert the house into a duplex.

What if the house is too small to be subdivided into reasonably big units? Well, buildings can have additions built to them. You can build additions next to them, or even add another story in some cases. This seems to have been done relatively frequently in older American towns, like these examples in Lowell, Massachussetts:

As you can see, you have these long elongated buildings, often with dissimilar roofs and the front part of the building looks like a self-standing house, which is probably how they started. Additions aren't that cheap to add, as they are more complicated to build than a new building and so require more labor. I've seen estimates that they can cost 200$ per square foot to build in the case of an additional story. But the advantage of such additions is that the land cost is nil, unlike new constructions, since it's an addition, not a replacement.

Additions are also not a magic bullet, they are limited by zoning, like everything else, and the structure of the building must be able to accommodate the addition.

New building on the same lot

This is a reasonably frequent avenue used in many cities like Vancouver, taking the form of the laneway house for example. A laneway house is a possibility afforded by a lane behind rows of houses, as a second house can be built on the same lot. Like additions, the advantage of this approach is a nil land cost, since the owner can build it without replacing the existing building. In a way, it can be seen as a particular type of subdividing, where one subdivides the lot rather than the building.

Boarding houses

Boarding houses are a practice common in traditional American cities but that is now rarely seen. Basically, once upon a time, when an homeowner had empty bedrooms in their houses, for example, because their children were grown up and had moved out, they would advertise these bedrooms to rent. And thus, single people looking for affordable housing could rent out these bedrooms, often with meals being offered as part of the renting contract. That way, the occupancy of single-family houses could remain high, the "empty nesters" phenomenon didn't occur much and affordable housing options were available to singles, and maybe even some couples. This also solved an income issue for older people who were retired before the institution of government pensions... Hmm, interesting hypothesis: one of the reasons for the disappearance of boarding houses may be that government pensions reduced the need for older homeowners to find new income sources.

Famous boarding house occupants are Sherlock Holmes and John Watson, who rented out bedrooms in Ms Hudson's personal house.

What potential does renting out empty bedrooms have?

Well, going by the Canadian census, I estimate there are as many as 11 million unoccupied bedrooms in occupied dwellings in Canada. Enough to absorb a 30% increase in the population. Of course, most of these bedrooms may not be in desirable locations, but even in the city of Vancouver (NOT including the suburbs), there are probably around 82 000 unoccupied bedrooms in occupied dwellings, enough to increase the population by nearly 14%. In the city of Toronto, there are 340 000 unoccupied bedrooms, again, enough to absorb a 14% rise in population without building a single new home.

In older suburbs, the amount of unoccupied bedrooms is amazing. In my childhood's suburb of Boucherville, there are at least 47 420 bedrooms:

Meanwhile, there are 40 753 people living in the city, and 10 890 couples, married or common law. If we suppose couples sleep in the same bedroom, that means at most there are 29 863 occupied bedrooms, leaving 17 557 bedrooms unoccupied in occupied dwellings. That's enough to house a population increase of 43%, with a similar rise in density.


Filtering is a phenomenon where the construction industry in a metropolitan region is able to build sufficient amounts of housing to outpace the growth in population and households. Basically, supply grows faster than demand. In such a situation, even if newly built units are not affordable on their own, they may lead to a decline in market value of older units that are not as desirable, therefore their owners who are renting them out or trying to sell them have to lower their asking prices to find a buyer or a renter.

This is not that common in the major cities of Western countries, because though fertility rates have fallen, governments have compensated that fall with increases in mass immigration, and immigrants tend to gather in the major metropolitan areas. This factor adds up on the rural-to-urban migration and makes it often unlikely to see demand growing slower than supply, as the local industry fails to keep up with the population growth. The temporal reality of supply and demand and industry output must not be neglected.

However, it does happen in Japan, as these graphs I made from condo (mansshon in Japanese real estate parlance) prices per square foot in central areas of Sapporo and Tokyo reveal. Data from

When density in a city is high enough, filtering is the only approach that works, because subdividing would result in too small units for the needs of people and the current density makes it impossible to economically replace current buildings with walk-up low-rise buildings.

An alternative to the "grow home": the adaptable home

Finally, let's look at my own counter-proposal to the "grow home" model, that I've already derided. A reminder, the "grow home model" is the following:

First, a cheap home is built with an unfinished basement and/or attic, which is bought by a young couple.

As the family grows, the attic or basement is finished, or the house likewise improved, to accommodate the needs of the family at its peak (teenage or young adult children still in the home).
After the children leave, the parents still occupy the fully improved, now very valuable home.

When the parents have enough, they sell the house, at its current market value... likely twice what it was worth when it was built, because of all the improvement and additional living space.

So the house starts cheap, but doesn't remain cheap much longer... what to do?

Well, let me present the alternative, the "adaptable" home.

The two first look like duplexes, they have two doors side-by-side. The second looks like a triplex, with two doors on the first story, one going upstairs, the other opening on the 1st story unit. But here's the thing, the two side-by-side doors can in fact be easily converted into one single door, and the lobbies on the other side of them can be merged by removing the wall that separates them, thus integrating the stairway of the upper or lower unit into the other unit. This effectively merges the units together.

The building can thus be easily merged or subdivided as needed.

Raised ranches fit the bill, as does the triplex shown on this very page, since it has two adjacent doors that could be easily modified to merge both top units.

Imagine this but with one door instead of two on top of the stairs

So the life of these buildings can be the following.

First, they are built complete, all floors finished, and sold to a family that will occupy it. The family is small, so it first put the building into its duplex or triplex form: living in one unit and renting out the other(s).

As the family grows, it may lack space, in this case, the owner warns the renters to move out as they plan to merge two units to create one big 4- to 6-bedroom unit to accommodate their growing family.

As the children leave, the owners may decide the unit is now too big for them, they can then easily separate the units anew and rent out the now unoccupied unit.

This provides needed affordable housing to a city and a retirement supplementary income to the owner. Though the initial capital cost is higher, the revenue from renting out the other unit(s) more than makes up for the higher cost. In fact, if I compared a "grow home" scenario of a 200 000$ home which has its basement finished for 80 000$ 7 years afters its initial sale to an adaptable home that is built with a finished basement unit for 280 000$, renting it out for 7 years at a price of 600$ a month for a 2- or 3-bedroom basement unit, then converting into a single house, losing the revenue. The adaptable home is paid back 2 whole years before the grow home, thanks to the rent income.

In fact, an adaptable home can in fact start out as a grow home that ends up being subdivided later in the life of the family that occupies it.

The main issue here is purely regulatory. An adaptable home goes from single-family to multi-family and vice versa during its life, adapting to the needs of the owners. This would be illegal in most cities, and is just further proof of the idiocy of the current North American zoning practices.


  1. Glad to see you back, I still check in a couple times a week looking for new posts.

    Anyway, one thing I was thinking about is filtering, which you say isn't very common in major western cities. But with the exception of the sprawlburgs like Los Angeles or Atlanta, it seems like a lot of filtering happened in older established cities that were drained out after WWII. Metropolitan Detroit for example hasn't appreciably increased in population since 1970 while its footprint continues to expand. Metro Buffalo famously doubled its land area since 1950 without adding population. So it's not just that the construction industry in those cities outpaced demand, they kept going in markets with effectively no demand at all.

    Of course there's other economic, social, and racial factors at play here as well, not to mention that the construction industry itself plays such a big part in the economy it can become virtually self-sustaining as long as the ponzi-scheme nature of it doesn't collapse (the growth industry is growth itself). Still, even in healthier metros where the city itself lost population but the metro as a whole still grew in spite of that, the result is nevertheless that city properties filtered down, just usually due to external forces than inherent ones.

    1. There was demand, you're just looking at a slightly too large generalized picture where negative and positive changes in demand cancel each other out.

      In the 1900s we basically saw the invention of the cheap automobile, very cheap fuel (at least for the first part), and a huge investment in free-to-use motorways. At the sametime the US doubled down on oversized roads (in addition to freeways) throughout urban areas (newworldeconomics has a good post showing how the US had a "automobile inspired" urban design fetish prior to the automobile existing!).

      All this resulted in huge demand for larger, more dispersed housing in new suburbs, while the demand for accessible, convenient urban housing collapsed.

      Today as fuel prices, congestion, etc rise and urban areas reform we're seeing the opposite - increased demand for apartments and loss of demand in rural areas.

    2. Thank you for both your comments.

      Regarding filtering, I was referring to major US cities, not all US cities. I'm fully aware filtering is going on in a lot of second-tier or third-tier American cities. The construction industry wants to keep working, so it will build houses even when the population is stagnant, because the richer classes do prefer new houses than old houses. And unfortunately, it's often too expensive and complicated to build new houses to replace old ones because of zoning and planning policies. So much easier to just build on the periphery.

      Jed, I think it's important to point out that demand for housing in rural areas has been falling continually for more than a century. The car simply transformed some rural areas close to cities into suburban areas because of higher travel speed.

  2. good to see your back! more soon.

  3. Boarding Houses, Filtering and Adaptability three great ways to look at higher density. Regulations and the myth of an 'American Home' (my own castle) are holding back these kinds of communal living in North America. is building prefabricated homes that we would like to be able to adapt later, instead of remodeling a kitchen, we would like to remove and replace building modules so that construction only takes a week: remove and replace. It would also be a way to recycle the structure, and manage waste.

    Filtering is particularly interesting because it implies that owning one's own home is a money looser because the large financial companies will over-supply the market lowering the value of individual homeowners.

    Boarding Houses would be great in college towns where older people could supply housing for students rather than dormitories.

    Thanks again for you thoughts I hope to meet you some day.

  4. You don't see boarding houses these days because they're illegal

  5. So you made... the brownstone? The idea is that it can be subdivided or merged pretty easily is at least that old.

  6. Again, very tightly-knitted reflexion and very actual problematic.
    I follow this blog by email , so this new post got directly to my inbox - life flies by fast for everyone, and I hadn't realized it had been so long since the last post.