Blessed by the Potato

…or Real Estate: Changing Sales Mix and the Effect on Averages.

The average is a very important way for us to reduce the complexity of a large number of things down to one measure that we can then compare to other groups of things. It’s a simple grade-school concept, yet we must at times remember that there are different ways of creating an average, and they may give different results depending on how the raw data is distributed. There are issues of weighting data, and that we have to be aware that sampling subsets of a large population does happen, and that how data is sampled can affect the outcome.

For the real estate boards (e.g., TREB) they report the “average house price” which is a simple average of all the property sales in a given region within a given month. Now for an organization that is primarily concerned with how much commission-generating activity is taking place, a simple average is great (multiply the average by the number of sales and you get the total sales volume). However, it doesn’t truly answer the question many people have, which would be something like “how much more/less is my house worth?”

The sales mix — how many condos versus detached houses, which neighbourhoods have more activity — affects the average, especially when it changes. You see, the average price in a month or a year is not a census; not the average of all properties across the city, weighted by how many of each there are, but rather the average of those that sold. With a large enough city over relatively normal periods the distinction shouldn’t matter that much: the properties that sold should be a random and consistent sample across the city. Maybe condos turn over more often than houses, but as long as that’s a constant influence over time then you can still make year-to-year comparisons.

But Toronto has seen a massive real estate boom over the last decade that included an insane amount of condo construction: 50,000+ units built and another 60,000+ under construction*, in many cases removing single-family houses in the process. And the sales mix has changed even more, as condos become investment commodities to flip rather than places for long-term occupation.

The result? The increase in average price that so many find concerning may actually be understating the degree of price increases that have occurred (though price-to-rent analyses sidestep this). Moreover, as this effect unwinds it will mask price decreases. For an example of the change in weightings affecting the average, see the May condo numbers: 416 condos were up 1.2%, 905 condos up just 0.6%, yet the overall GTA average was up more than either at 1.6%. This seemingly illogical result came about because the sales volume dropped so much more in the 905 — these lower-priced units counted for less in 2013 than in 2012 so the overall average was up much more than any single component. With a large enough swing in sales (the weighting factors) you could see average prices rise even if each individual component had decreasing prices.

My thinking is that condos are the most speculative component in Toronto real estate, so as the market starts to correct the sales volume (and prices for that matter) on condos will be affected more. Indeed, sampling** a few months shows that this year is showing a decrease in the ratio of condo sales to detached sales from the previous few years (detailed data doesn’t go back far enough to show how the mix changed as a bubble inflated). So as we get through the Wile E. Coyote phase of the correction, the change in average prices may continue to look positive (or less negative) due to changes in the sales mix, even if the actual prices on individual properties are going down.

Now, it’s not a huge effect: a few percent one way or the other, and it doesn’t compound. But when you have a market on the edge, with large changes in sales mix happening, where a few percent one way or the other would have large psychological effects, then it’s an effect worth keeping in mind.

In a similar vein, there seemed to be a lot of media attention around the last two Urbanation rental reports, which described steep increases in Toronto rent costs. However, I seriously question the reliability of these reports, particularly when it comes to their sampling.

Ideally, when looking at any population we’d have the full set of data to analyze. But you can’t measure the length of every fish in the lake without draining the lake to catch them all, and for large populations even if you could measure everything you’d end up with a massive amount of data to crunch. It would be nice to have the rent rates of every rental agreement everywhere, or to run an appraisal on every property every year so we could analyze the data as we see fit. But that wouldn’t be worth the cost and effort of doing so, so realistically we have to sample: pick a few fish out of the lake, phone up a few potential voters, or get information on some rental agreements out of the city and find out what the average of those are, estimate how our sample differs from the overall population and go on with our lives. A good sample should be representative, random, and appropriately sized.

Urbannation doesn’t provide their raw data (indeed, they charge an arm and a leg to get at the report), but the information in the press tells us that with several thousand data points it’s likely large enough. You could probably get a good sample with just a few hundred representative rentals. However, it is not representative. Their report is based on the sample of all rental agreements that go through the MLS system. When you’re talking resale housing transactions MLS is a great sample — it captures almost the entirety of the market, and there isn’t necessarily an obvious bias to what’s missed. For rentals however, MLS is a small slice of the market and is definitely not representative. Many landlords advertise their rental using low-cost methods such as ViewIt, Craigslist, Kijiji, or the local hospital/university bulletin board. The standard charge for an agent to list a rental on MLS is one month’s rent. For a 1-year lease that might get renewed for 2-3 years that’s ~3-8% of the gross — quite the cut for advertising and an illegal “custom” lease on what is already a loss-making “investment”. So we can’t really expect any old random landlord is going to go and get an agent to list their rental on MLS with some equal probability. In my anecdotal experience, the MLS listings are largely from cases where the landlord already has a relationship with the realtor, either a recent sale (“now that I’ve sold you the place, how about I get a tenant for you?”), or a current listing that’s failing (“geez, 45 days and not a single bid, how about I help you rent it out and we can see if the market improves in the spring?”), or is completely out to lunch/out of town.

The Urbannation rental report might say less about the increase in rental demand than it does about the increased use of MLS for rentals. The headline could just as well have been “rental demand unchanged in Toronto; Sales collapse means more people with more expensive units resort to renting out via MLS.” We don’t know how the represented areas and units changed over time (perhaps a few years ago cheap CityPlace units were over-represented while it was newly flipped; now maybe more inherently expensive developments are dragging the average up).

Now I don’t personally track the downtown condo lease market so I can’t say what correlation there is between the Urbannation report/MLS data and what’s actually happening with like-to-like rents. I can say that it does not look like the outskirts of the 416 (North York in particular) are experiencing anywhere near the same level of rent inflation as they suggest. Indeed, anecdotally I’ve heard a few cases of landlords voluntarily freezing rent north of the 401, as even the 2.5% rent control increase can’t be justified by the market rents that could be sought if a tenant left (which is likely reflected in the 2014 cap of 0.8%).

* – the 2nd figure is across the GTA, so perhaps closer to 40,000 units under construction in Toronto proper.
** – I wasn’t able to find the data in a spreadsheet for ease of analysis, so I had to manually parse a few months and build my own spreadsheet. I wasn’t going to go through that for more than a few sample months.

I’ve mentioned speculative holding before as something that underlies a bubble, but haven’t really gone into any depth on the subject. Basically, it’s a speculative behaviour that isn’t as obviously speculative as buying something purely in the hopes of future appreciation: instead you hold something you may have bought for other reasons on that hope.

One of the more typical examples is to hold on to an old property to rent out after you buy a new place to live in. The purchase you make at the time may not be speculative, but often the decision to hold on to excess property is — if you weren’t counting on large future gains, you would have sold off the old place, rather than take the risk and hassle of becoming a landlord.

Less obvious is buying preconstruction while owning. Even if you plan to sell as soon as the new place is finished, you have double the real estate exposure for the duration of the construction, which could be a few years. When people are advised not to time the market, that means they should be selling a their old place as soon as they buy a new one — even if the new one isn’t built yet — in order to limit risk. Many may chafe at that advice, in which case they shouldn’t speculate in the preconstruction market unless they have the capacity to take on the risk, and instead shop around for homes that are already built.

One of the more subtle effects on supply is the decision of whether to buy first and then sell, or go the other way around. Deciding to buy first then sell has a small effect on supply and is like a minor version of buying preconstruction (for a time you are exposed to double the risk). It is a small effect, and the market adapts to whatever way is accepted as the norm (or even some mixture of methods). But when the shift happens from one scheme to another en masse, it can sway the supply. Take the case where everyone considers that the way to transact in real estate is to buy your new place first, then go out and list the old one. If then everyone changes their mind, perhaps deciding that the market is softening and the old way was too risky, and sells first, it could shift months of inventory over all at once: suddenly new houses are coming on the market while inventory sits. If “the” way to transact is different in a “buyers'” market than in a “sellers'” market, then once the shift is proclaimed, it could lead to a short-term swing in inventory and put pressure on prices.

I think one of the largest effects of speculative holding is the shift that occurs in the supply curve under the influence of rising prices, the holding on in the face of steady price increases. Imagine if someone comes out of the blue and offers you $1M for the house you paid $500k for just a few years before. Many of you would be all over that deal, telling your neighbours about it who would rush out to list their houses and take advantage of the opportunity. You’d think the buyer was nuts and that it was a one-time, not-to-be-missed opportunity. Hey, you could go move to the next town over (where houses were still $500k) and practically retire on that kind of money. But if you got to the $1M offer via a succession of offers: one at $550k that you turned down, then again a bit later the buyer comes back to you offering $600k, then again a few months later with an offer of $650k… by the time you got to $1M in a few years, you would have been expecting that price, and possibly even projecting out to the $1.2M offer you were sure was in the works for you. The price itself was still just as insane, just as much of a windfall, but because of the path to get there you’re less likely to actually put your house on the market and take it. You were inoculated against the crazy, so it seemed right.

And finally, the incarnation of speculative holding that made me think to write this post: taking the house off the market for “when it recovers in the spring”. Sales in Toronto and Vancouver dropped double-digit percentages over the past year, to the lowest levels since the financial crisis. Prices barely budged, but were down a bit in many sectors. Thus it’s quite common to hear on the subway, in the restaurants, the newspaper articles and the chat boards that famous idea of trying to relist later when the market looks better. That is perhaps the baldest speculative holding of all: the person wants to sell, but will hold on in the hopes of higher prices later.

I posted yesterday about some bullish (or not-bullish but not-bearish) arguments on housing and why I think they’re overlooking important factors. Michael James hosted an interesting discussion on his blog where he called me his favourite writer (well, not quite). Larry MacDonald left a comment there about an upcoming article he’s writing for the Globe. I hate to preemptively publish, and want to read it and give it a chance before formulating a response, but unfortunately tonight is the only time I’m going to have this week to write.

Larry’s comment was that “interest rates don’t go up in isolation, as he [referring to Ben Rabidoux, but could equally apply to me] appears to assume. Looking at business cycle dynamics over history, interest rates and household income tend to rise together.”

This is another case of something that is true but not helpful. Yes, the economy will likely be doing better when interest rates do finally go back up (next year, or next decade), and that likely will bring wage growth. But that doesn’t obviate the risk of buying an over-priced house now: the impact of rising rates and that of rising wages and employment are vastly different:

• Rates can rise very quickly, increasing payment obligations equally quickly, whereas even robust wage growth takes time to compound enough to influence affordability metrics.
• The impact of modestly higher rates on affordability/mortgage payments is in all likelihood going to be much greater than the impact of the associated wage growth.

Let’s work through a concrete example: say that you’re an approximately average Toronto couple. Together, you pull in $100k/year, and recently bought a house at $575k, taking on a $460k mortgage fixed for 5 years at 3%. Your monthly payment of $2180 is a touch over 26% of your gross pay: with heat and taxes it’s still (barely) below 32%, so this place is officially affordable!

About halfway through your mortgage term, this ridiculous not-quite-a-recession we’ve found ourselves mired in ends. Job growth picks up, and inflation rages at 10%/year. The Bank of Canada (and the bond market) is forced to respond to this double-digit inflation threat, but in this dream scenario mortgage rates merely go back up to 6%.

You don’t pay much attention because your mortgage isn’t up for renewal until 2018, and by that time surely wage growth will take away the sting. Well, following the first two and a half years of pay freezes at work, things indeed started looking up: with consecutive raises of 10% you’re now grossing $128k, and you’ve paid your mortgage down to a mere $393k.

Then you get the bad news: your monthly payment is now $2800, still representing a touch over 26% of your gross pay. If property taxes and heating costs have also increased 28% due to inflation, then your place is still borderline affordable at 32% of your income. But you were one of the lucky ones: what if your wage growth had merely paced interest rates? Using 2.5 years at 6% wage growth would mean your mortgage alone was 29% of your income — with taxes and heat your GDS would be over 34%. Your house would have to fall in value by 14% (in nominal terms) in order for a buyer in the new environment to buy it with the same affordability metrics as you enjoyed back in 2013.

What if you were one of the unfortunate few whose mortgage renewed the very year the economy picked up and rates increased? You’d have been afforded no time for the inflation you heard so much about to increase your wages… so your mortgage payment alone would top 33% of your pay, and the affordability pressure would attempt to push house prices 23% lower.

So what I’m saying is that the effect of small changes in interest rates on affordability is very likely to be much greater than the offsetting effect of wage growth. Interest rate increases raise the cost of buying a house immediately, but wage growth takes time — and we’re unlikely to see the BoC or the bond market allow wage inflation to rage for a few years before getting around to lifting rates off the zero bound.

Perspicacity is one of my favourite words. The dictionary definition mentions understanding and discernment, and I think of it as more specifically referring to the ability to discern what is and is not important from conflicting data. Part of what helped rocket it to near the top of my favourite word list is that it was the defining trait for NASA astronaut selection during the space race:

The quality most needed by a scientist serving as an astronaut might be summed up by the single word ‘perspicacity.’ The task requires an exceptionally astute and imaginative observer but also one whose observations are accurate and impartial. He must, from among the thousands of items he might observe, quickly pick out those that are significant, spot the anomalies and investigate them. He must discriminate fine detail and subtle differences in unfamiliar situations, synthesize observations to gain insight into a general pattern, and select and devise key observations to test working hypotheses.

There have been many articles on the state of the housing market — more every day — but for today I will pick on Larry MacDonald. In part because his dig (or his headline editor’s) at bears for making “unsubstantiated claims” was highly unfair: housing bears are some of the most data-driven people I know. (Though speaking of editors, he may have just drawn the short straw in taking sides for a manufactured debate). And in part because his articles (like many bullish ones) seem to lack perspicacity.

We had the one where he tried to set up an esoteric monetary policy criteria as being necessary for a housing correction, though left unsaid was the market’s vulnerability should such an inversion in the yield curve arise, or how changes to mortgage insurance might have the same effect. The affordability index is a perennial favourite, though it is highly interest-rate dependent. In short, lots of focus and analysis on the measures and factors that are — IMHO — not as fundamental.

The most egregious is also the most recent: “Is Canada talking itself into a housing crisis?” He tries to take a paper by Shiller — Professor Robert “Irrational Exuberance” Shiller! — to make the case for there not being a bubble in Canada, and that all the negative media stories may cause a downturn when the fundamentals are ushering in a soft landing.

In that piece, he picks bits out of the stories to come to strangely opposite conclusions. You can find the Shiller paper online, explaining that buyer expectations help set house prices — and that the media may have helped change those expectations as the market turned in the US around 2006. But that’s negative press precipitating a downturn in an over-valued market that’s primed for it, quite a different matter from Canada “talking itself into a housing crash.” Indeed, in that same article Larry cites a CBC interview with Shiller from September, summarizing Dr. Shiller’s points as “Canada should be spared.” Yet for many others, Dr. Shiller’s take-home message from that interview was “I worry that what is happening in Canada is kind of a slow-motion version of what happened in the U.S.”. He’s not at all saying that housing prices won’t correct or somehow be spared — the suggestion is that such a process won’t take the banks and the rest of the world economy down with it.

Dr. Shiller argued — in advance — that the fundamentals were out of line and that a correction was due in the US. He has not been as vociferous about Canada, but has several times said that Canada in general, and Vancouver in particular, are worrisome. Hell, even when trying to be bullish the bit about the RBC affordability index can’t support the insane singularity that is Vancouver. The paper Larry cites is about perception and media reports affecting the timing of the correction, not causing it. If anything, it’s just as much about how expectations helped fuel the bubble in the first place.

The real world is a messy place, and markets particularly so, with a great deal of data to parse, much of it conflicting. Hell, differing perspectives and valuation schemes are what make a market. So one must proceed with a degree of perspicacity: seeing what is significant, understanding what conflicting data imply, and acting with all due caution.

The Globe has an article out this weekend on Emili, CMHC’s automated housing appraisal system.

To break it down, when someone wants to take out a mortgage, they go to the bank and say something like “I’d like to borrow $500,000 for this house that I just purchased for $550,000, and I’ll pay the other $50,000 with my own money.” The bank then has to make sure that the $500k they lend will be paid back, by looking at the income and creditworthiness of the borrower, and also at the value of the house, so that if there is a default that the value will cover the mortgage. Even with 10% down, a loan for $500k is not very secure if the property is only worth $400k. Emili is an automated system to determine that house value (and, as I understand it, some of the other aspects of the loan), which makes the whole process a lot faster and more efficient than sending an appraiser to check out the house.

The problem pointed out by the Globe is that Emili is too generous. There are lots of reasons given in the article as to why, including this gem:

A CMHC spokeswoman said that staff are aware of “a handful of cases” in which Emili approved a mortgage for a non-existent house.

But let’s think about it logically. Emili could be perfect, always assigning the correct value to the house. Personally, I find that unlikely for many reasons mentioned in the article, such as that Emili can’t see inside the house to assess the state of repair or the level of renovations, and that known errors exist.

Emili could be good enough, but with a few mistakes made here and there. That’s a fairly likely scenario, after all many times those small matters of internal shape are not that important to the valuation, especially if there’s a decent margin of safety built in or if the land value is a significant component of the valuation. If Emili is mostly accurate with a few inescapable random errors, then we should see mistakes made in both directions. We should hear reports of buyers caught in the emotion of a bidding war, only to find their mortgage rejected because Emili won’t support the valuation, or of Emili erroneously denying a mortgage because it reports a vacant lot after a house was destroyed by fire (not having the record of the replacement), or coming in too low on valuation for some other reason. I’ve been keeping my eyes open for these sorts of anecdotes for years, and haven’t seen them.

That leads me to what I believe is the actual situation: Emili systematically over-values real estate.

Now, some over-valuation is to be expected. Lenders don’t want to turn away business by cutting the appraisal too fine, and the insurers will build this into their models. They may also assume that prices generally go up (at a modest rate), and since it generally takes time for a default to occur some upward bias can be tolerated. But when extreme dislocations occur — such as bidding wars leading to “winning” offers hundreds of thousands of dollars too high, or certain neighbourhoods seeing annual appreciation way above the norm (like 20%/year) — then the system should be flagging those as problems and denying the loans. That would act as a natural brake on a bubble.

Though it is important on a national, system-wide basis to avoid bubbles, nobody on an individual level has much of an incentive to implement brakes. The banks want to lend (especially if they have CMHC covering their butts), the buyers want to buy, the sellers want to sell, and the ancillary agents want transaction volume. CMHC is one of the few entities that could play the role of a disinterested, rational appraiser, yet they too have no political will or desire to stop housing momentum and break deals by being strict with qualifying criteria. Indeed, contrary to Robert McLister’s opinion that “CMHC knows the risk of it botching property valuations en masse. It has the public, press and regulators breathing down its neck around the clock,” I’d say that the public, press, banks and mortgage brokers are breathing down its neck to allow transactions to proceed. So instead, Emili accounts for many things including the “…housing market conditions in which the property is located…” which to me reads as “becomes loose and permissive in hot housing markets.”

Rob McLister of CMT counters:

“Inevitably, people will read the Globe’s story and think that CMHC is using some back-of-the-napkin formula to judge property risk. That’s so far from the truth. Emili is not some 100-line computer program written by a college intern. It is multi-million dollar mission critical technology benefiting from the best available data and over two decades of R&D.”

For what it’s worth, I don’t doubt that. But it’s not open source, and we don’t know the assumptions that went into making that expensive, sophisticated valuation engine. For example, does it implicitly assume that buyers are rational? A single over-heated bidding war might raise a flag, but would 3 or 10 in an area upgrade the valuations of everything, as it’s then a pattern? Though fraud becomes less likely, the loans really are no more better supported in the long run. Similarly, in assessing risk the core assumption seems to be that valuation changes affect severity, while unemployment affects default rate — the two factors combining to make up the losses that CMHC may face, and the two factors being completely separate and orthogonal. Yet in the aftermath of a bubble, valuation changes also affect default rate as speculators walk away (even though they may remain gainfully employed), and unemployment as well (as construction grinds to a halt). But if that wasn’t observed in the dataset used to build the models, then it may not be accounted for.

As a parallel, consider the subprime mess in the US. I’m sure the ratings agencies had expensive teams of people and fancy computer systems to come up with the “mission critical” ratings for CDOs, yet every AAA handed out was in error, due to some flawed underlying assumptions and a lack of checks. For instance, an underlying assumption of building many of the CDOs and securitized portfolios is that not all the crappy subprime debt goes bad at once, so you can have a AAA slice from something made up of junk. Michael Lewis also highlighted one of the other flawed assumptions: that “average credit rating” meant something, when in fact a pool of 100 mortgages to people with a credit score of 650 is rather different than 50 mortgages to people with a 700 and 50 to those with 600.

I think that based on first principles and the housing market insanity we’ve seen in the last few years, it’s clear that whatever is inside the black box that is Emili is biased to the upside in its valuation methodology. While that doesn’t cause a housing bubble, it allows it — a tragedy given that CMHC is the ultimate holder of risk and should have its systems tuned to be more conservative.