Among other things, insurance companies now use aerial photography, combined with AI, to better assess property-level risk. Here's an excerpt from Bloomberg Green:

“Weather and catastrophe losses are running ahead of the ability to manage them, and many insurers are having trouble sustaining their business because they’re not getting the right rates,” said Jay Guin, chief research officer of the extreme event solutions team at Verisk, a catastrophe modeling firm. “AI changes the equation.”

Zurich Insurance Group AG, one of the largest insurers in Europe, uses AI powered risk-modeling software to assess catastrophe risk and often tweaks it for its own purpose.

“If there’s fire hazard like vegetation, overhang or debris in your backyard that shouldn’t be there, we can tell you to lower the risk otherwise we may not be able to underwrite you,” said Ericson Chan, chief information and digital officer of the Swiss company.

What AI allows is a level of granularity that just wasn't possible when humans were the ones who had to do it. Insurers now talk about "continuous remote risk monitoring," meaning they can use AI-powered aerial imagery to constantly check on that risky debris in your backyard.

This feels like quite an improvement for the insurance industry. But when you more accurately price risk, I would imagine that it will lead to more insurers deciding to stay clear of certain risks and certain properties, as has already been the case in places like California.

Cover photo by Pim de Boer on Unsplash

If you recall, the largest buyer of single-family houses in the US last year was Opendoor. This is according to SFR Analytics. Opendoor is a so-called iBuyer, meaning they provide cash offers to sellers, close quickly, and then turnaround and sell each home for — hopefully — a profit. They also collect commissions, and make money in other ways (such as through home loans).

This process seemed to be working reasonably well up until 2022, but then the market turned. They then quickly discovered that they couldn't sell their homes for a profit and so they ramped down acquisitions. Here's a chart from another post by SFR Analytics showing the fall off in purchase volume in the second half of 2022:

The worst performing market at this time was Phoenix, which apparently accounted for around 50% of the company's losses. In some cases, their gross margins were -$60,000 per property. But once they recalibrated their models (I'm just assuming this is what they did), they were able to regain positive unit economics. Here's Phoenix starting from their January 2023 cohort:

Opendoor has not done well as a public company. But it is the biggest buyer of single-family houses and it seems to be back to positive gross margins. Maybe that's something.

Charts from SFR Analytics; cover photo by Chris Tingom on Unsplash

This past Sunday, Paris voted in favor of greening and pedestrianizing an additional 500 streets in the capital (5-8 per neighborhood). This will add to the 300 or so streets that have already received this treatment since Mayor Hidalgo started her second term in 2020. And as a result of this expansion, it is estimated that about 10,000 on-street parking spaces will be removed, which represents about 10% of the city's total inventory.

Exciting. But who voted for this? Of the Parisians who voted, 66% voted in favor of the initiative. And it carried in 14 of 17 arrondissements (with the 1st, 2nd, 3rd, and 4th counted as one). But similar to prior referendums, voter turnout was extremely low: only 4.06% of eligible voters showed up (approximately 56,500 people). And this is after the voting age was lowered to 16 years old for the first time.

For context, when Paris voted on whether electric scooters should be banned, 7.46% of voters showed up. So while low, this situation is not entirely unique. Though it does, once again, raise the question of whether the outcome of this referendum truly reflects public opinion. My outsider view is that it probably does. Because I take the apathy to mean some level of support, or at the very least, an absence of strong aversion.

Think, for example, about who shows up at community meetings for new development projects. The vast majority of people in attendance have concerns they would like to air. It's very rare for someone to show up and say, "I didn't have much going on tonight so I decided to come by and see everyone. I have no real concerns. Project looks cool. Carry on as you were."

If you agree with this logic, well then it suggests that many/most Parisians do generally support more pedestrianized streets, even if it means the removal of parking. That's an accomplishment in my books.

Cover photo by Maximilian Bungart on Unsplash

This is an important chart taken from this recent article by Steve Lafleur talking about the need for Canada to "bulk up." What it obviously shows is that housing completions and population growth have generally been diverging in Canada since the 1970s.

Back then, we were building about 200,000 homes a year and, today, we're building slightly under that. Of course, our population has also grown dramatically during this time period, as has the number of people who move to Canada each year. The result is that the Canada Mortgage Housing Corporation estimates that we'll have a housing shortage of approximately 3.5 million homes by 2030.

But we already knew this. Big numbers are often thrown around in studies. I think the more important question is: How do we reconcile this massive shortage with the fact that, in cities like Toronto, we have lots of zoned land ready for the construction of new housing (but that isn't financially feasible) and lots of unsold homes that aren't selling right now?

Do we really have a shortage?

Well, Toronto is just one specific market, and I can't speak to all the dynamics playing out across the country and the world. But it strikes me that what's missing from the above chart, and this discussion in general, are considerations around (1) housing type and (2) affordability. And by type, I'm largely thinking about size, as it's closely linked to affordability.

If what we're building is too expensive for most people and unsuitable for their household needs, then yes, I guess that would mean we have a shortage of housing.

Cover photo by Matt Wang on Unsplash

My recent post titled "Canada must become a global superpower" has quickly become one of my most-read posts in the almost 12 years that I have been writing this daily blog. Within a few days, it quickly got to 11x the number of daily views that I typically get.

One of the points that I made was about Canada's population, and specifically the target set by the Century Initiative of 100 million Canadians by 2100. Today I'd like to expand on this point, because I'm seeing more people talk about it on the socials.

At the time of writing this post, Canada's official population clock from Statistics Canada was sitting at 41,591,151 people. So to reach 100 million in the next 75 years, it would mean we would need to grow our population by 58,408,861 people. At first glance, this seems like a big number. And to some, it has proven to be an unsettling proposition. But 75 years is a long time for compounding to work its magic.

For us to reach 100 million Canadians by 2100 it would mean that we would need to grow our population by a compounded annual growth rate of just 1.18% per year. On our current population base, that would mean about 490,000 new people next year. To put this into perspective, since Confederation in 1867, Canada's population growth rate has averaged around 1.2% per year.

So by arguing that we want to reach 100 million Canadians by 2100, we are, in a way, just saying "we should continue what we've been doing since 1867 and not change a whole lot." The status quo should inevitably lead us to 100 million people during this time period.

Of course, history isn't exactly the same. Canada's fertility rate was much higher in previous years. At the beginning of the 20th century it was nearly five children per woman. And in 1960, it was 3.81 births per woman, which placed us ahead of the US.

Today, we are 1.26 births per woman (2023), compared to 1.66 in the US (2022). We are now among the countries classified as having "lowest-low fertility." Meaning, we're sub 1.3. What this means is that we are now more dependent on immigration to maintain the same growth rate as before.

At the same time, it's not like we're unaccustomed to high immigration. Between 1901 and 1921, Canada's population increased by almost 3% a year on average. This was in large part because of immigrants from Europe, specifically the British Isles. And between 1901 and 1911, alone, Canada welcomed 1.2 million people. This is at a time when we had just over 5 million people in the entire country.

So in the end, 100 million Canadians by 2100 is probably not all that ambitious. A compound annual growth rate of 1.5% would, for example, have us grow to over 127 million people. That would be more of a stretch. There's also the important question of how quickly are we growing relative to other countries.

Whatever the exact target, I stand by what I said before. We should be aiming to lower the cost of living for Canadians, and in particular housing costs. We should make it easier for families to have more babies, should they choose to. And we should continue to attract the smartest and most ambitious people from around the world.

Cover photo by Kevin LEE on Unsplash

I love Toronto. It's my favorite city. And every so often, there will be a moment that explicitly reminds me why I love it so much. That happened yesterday when I walked down a laneway off College Street to find a somewhat hidden sushi spot called Oroshi Fish Co. Right away I was delighted by the combination of the small street and the mix of uses housed on it. But then, I walked inside to find two guys manhandling the carcass of an enormous bluefin tuna, and that's when I really said to myself, "man, Toronto is awesome." The sushi is some of the best you'll find in the city.

Here are a few photos.

Solar energy's share of total US electricity generation was only about 3.9% as of 2023. So it's not powering all that much today. However, the cost of PV modules continue to come down and installed capacity is growing very quickly. Here's an excerpt from a recent post by Brian Potter about solar energy:

By some metrics, solar PV has been deployed faster than any other energy source in history, going from 100 terawatt-hours of generation to 1,000 terawatt-hours in just 8 years, compared to 12 years for wind and nuclear, 28 for natural gas, and 32 for coal. In the US, solar PV projects are by far the largest share of planned new electrical generation capacity.

And here's a chart:

It's also interesting to look at which US states have the highest "capacity factors." The average for the entire US is 23%, which means that, on average, solar panels produce 23% of what they would if the sun were shining 24 hours a day. You might also think that the "sunshine state" would be one of the highest. But in fact, the top states are Utah and Arizona:

I'm highly interested in solar and we want to deploy it as much as we can on our projects going forward. If you're also interested in solar and want to learn more, Brian's post is an excellent place to start.

Images: Construction Physics

Last week, we spoke about affordable housing in Paris. Today, let's talk about tourist rentals in the city. The city of Paris and Greater Paris (i.e. la Ville de Paris and la Métropole du Grand Paris) recently commissioned Apur (which is a non-profit that I regularly follow) to do two studies on this topic. The first was for Paris proper and the second was for Greater Paris. What they found is super interesting:

• In August 2024, Greater Paris had 149,936 tourist rentals, of which 124,988 were available for immediate booking. This represents an 84% increase compared to August 2023, which is a massive number, but maybe not entirely surprising given that Paris hosted the Olympics last summer.

• Paris proper had 97,975 listings in August 2024 and 90,299 in December 2024. Overall, the city sees fairly muted seasonality. It's also worth noting that 31% of these listings belong to hosts that own multiple properties (that is, at least two).

But let's put these figures into context. Here's a map showing the density of Airbnb listings:

Here's a map showing the number of Airbnb listings compared to the number of principal residences:

And here's a map showing the percentage of unoccupied homes in the city, which totalled 268,500 as of 2021:

The report defines an "unoccupied home" to be any home that is not used as a household's primary residence. So in addition to flat out empty homes, it includes homes that are used sporadically throughout the year for pleasure and/or for work. And as you can see, there are large sections of the center of the city where "unoccupied" and second homes make up over 28% of the total housing stock.

These areas also closely mirror the areas where tourist rentals are most popular, and where Airbnb listings make up over 20% of the housing stock. (See the second chart above.) And as far as I can tell, these are mutually exclusive classifications, meaning there are sections of the city where a large percentage of the housing stock (perhaps up to half?) is either a short-term rental or a second home.

This tells you a lot about the housing market in Paris, especially when you compare it to other global cities:

NYC, for example, is shown here as having 8.8 million people, compared to 7.1 million people in Greater Paris. And yet Greater Paris has about 4x the total number of short-term rental listings. The number of available listings (where the property was available for at least one day of the year) also increased by 84% from August 2023 to August 2024 in Greater Paris; whereas it dropped by 16% in NYC, likely because the city basically banned short-term rentals.

The two reports can be found here and here (note they're in French). And they're rich in data if you'd like to learn more about some of the dynamics impacting Paris' housing market.

Cover photo by Kris Atomic on Unsplash

Over the last few years, there's been growing concern around institutional buyers (namely "Wall Street") buying up too many single-family houses and then renting them out.

But as we spoke about last year, the number of homes owned in this way is actually quite small. The vast majority of homes are owner occupied. And the second largest share of owners is what you might call "small landlords." That is, people who own somewhere between 1-9 homes.

So if the specific concern is that people are out there buying houses and then renting them out, the more fruitful target would be these small landlords. But nobody seems too fussed by them, which leads me to believe that this is an instance of symbolic politics theory. In other words, it's the association with the big bad Wall Street that people don't like.

Whatever the reason, here's the data on the largest single-family house buyers in the US last year (2024) via SFR Analytics:

Here are the metro areas where they transacted:

And here's this same data in heat map form:

The largest buyer was Opendoor, which is a so-called iBuyer. We've spoken about this company a lot on this blog. They don't actually want to hold any of the homes they buy. Instead, they buy, renovate, and then resell as quickly as possible.

The second largest was New Western. They are a wholesaler or "double-close buyer." These buyers want to own for an even shorter period of time and sometimes never actually own the home; instead they just assign their contract. What they're trying to do is buy at a discount and then immediately turn around and sell for a profit.

Note: SFR Analytics believes that New Western's count might be meaningfully understated in the above data. The company uses lots of different LLCs and acquisition strategies and so it's hard to aggregate the data. Assigning a contract also doesn't show up in any county records, so it's kind of impossible to track these. It's just like assigning a pre-construction condominium agreement.

Even still, what this data suggests is that single-family rental funds aren't as dominant as some might think. The overall counts for all of the largest buyers also remain relatively small. Last year, over 4 million existing homes (including condominiums and co-ops) were bought and sold in the US. And this was a nearly 30-year low.

Cover photo by Michael Tuszynski on Unsplash

Back when Elon Musk was running Tesla, he was known for saying that LiDAR technology (basically laser beams that measure distances) was not needed to create full self-driving cars. And that's why their cars instead use a bunch of cameras to monitor the outside world.

Now, I'm not an engineer, but this never made much sense to me. Cameras can only see so far and they certainly can't see at night. So wouldn't laser sensing technology that can see 250-500 meters out — including at night — be greatly preferable when it comes to human safety, even if it costs more?

I'm reminded of what I said to my eye doctor before getting laser eye surgery many years ago: "This is not a transaction where I'm looking to be price sensitive. Get me the absolute best." And that's exactly how I feel when it comes to self-driving cars. I don't care if cameras are pretty good most of the time; I would prefer to have the best.

So which is the best? Damned if I know, but here's an interesting and also hilarious video by YouTuber and engineer Mark Rober where he compares the two technologies: cameras (i.e. Tesla) vs. LiDAR. I won't spoil it for all of you, but his last test is the "Wile E. Coyote test" and it's awesome.

At the time of writing this post, the video already has more than 11 million views and it seems to have been incredibly helpful to Luminar's stock price:

But now the internet is filled with speculation that he deliberately used the video to mislead people regarding Tesla's Full Self Driving capabilities and maybe even to pump's Luminar stock. (Full disclosure: I own a few shares, but this post is in no way any sort of investment advice.) I don't know if this is true or not. But I do think that the cars of the future will all come equipped with LiDAR.

Cover photo by Vlad Tchompalov on Unsplash