Vancouver Price Drop

Documenting Vancouver real estate price movements

Disproving the HPI?

I have been very skeptical of the HPI because these 0.1% drops just don’t seem to match the reality of the situation so I’ve been hoping to somehow get my hands on a large enough sales data set where I could prove that this number is massaged and manipulated as much as the other numbers that are thrown out by the industry each month.  The data set I was able to work with included most of the apartment sales in Vancouver West for the last 10 years although unfortunately not including 2012.

Crunching the data using the methodology they use in the calculation is a lot of work but the workflow looked something like this:

  1. Identify all the units that have sold multiple times in the data set
  2. Any units that had more than one sale in the same year had those 2 sales merged into a single sale with the average of the 2 prices.  I was only concerned with year to year HPI changes so this was a good enough estimation of the average selling price
  3. Starting back at the first year find any units that sold in both 2000 and 2001 and take the median price difference as the HPI increase for the year.  This value is then applied to every other listing that sold in 2000 but not in 2001 to set a new 2001 approximated value
  4. Repeat step 3 for each additional year finding units that sold in that year (2002 for example) and compare that to the HPI adjusted (or last year selling price if valid) then take this median value as the HPI and apply to all the listings that had a value at that point – etc, etc

The end result was that the numbers were very close – close enough for me to conclude that the feeling I had in my gut that this number is completely inaccurate is actually more conspiracy theory than fact.  It is really too bad that I didn’t have the same data for 2012 because that would be more telling but given that 2008 / 2009 seemed accurate I’d expect the same now.

However, there is still an issue with it.  How many people do you know who have lived in an apartment or home have spent money upgrading, renovating, rain screening, special assessmentizing etc etc during that time?  Just about everyone I would imagine.  A 20% gain in HPI for apartments over a period of a few years essentially means that the average apartment sold for 20% more but how much did the average apartment spend on updates, new appliances, repairs, special assessments etc?  In reality the apartment that is being bought 5 years later is generally better (through expenses the owner paid for) than the same one that sold 5 years earlier.  So this isn’t like looking at a stock price over time or an ounce of gold.  It seems like a small difference but the impact can be pretty large in making the HPI look bigger than it really is.

As a final comment, when going through the HPI values for Vancouver West Apartments I discovered something that might be shocking, at least it was to me.  The current HPI for Vancouver West Apartments is the same as it was at the 2008 peak in April…  That’s over 4.5 years with a return of 0%.  I’ll be doing some more HPI related analysis in the future so stay tuned.


25 responses to “Disproving the HPI?

  1. ernie December 22, 2012 at 5:20 pm

    Oh well, the realtypimps won’t have get their usual obscene amount of Xmas bonus this year, eh?

    Wishing you a joyous holiday season and a happy 2013 New Year, observer.
    Once again, thanks for sharing your knowledge on legit data that others contrive to confuse and manipulate the public for their personal gains.

  2. Alexcanuck December 22, 2012 at 9:15 pm

    Postulate: Properties that people are selling are more likely to have had improvements made than ones that are lived in with no plans to sell.
    Second postulate: Properties that do sell at any price point will be the best one for the price, keeping in mind that “best” refers to perception by emotionally involved people being influenced by a realtor whose income depends on them buying. This further excludes unimproved properties from your data set.
    Result on your data and conclusions: The majority of the improved properties show up in the data, the unimproved ones don’t. Especially since you limit your data set to multiple sale units, which are even more likely to be extensively improved. Cost of the improvements doesn’t appear anywhere in your data, although it should be added to the previous sale price. This skews the price higher, and is hard to quantify and correct for.
    This depends on the original premise being true. It certainly feels right, but that only makes it truthy, not true!

  3. Zerodown December 22, 2012 at 9:25 pm

    You may calibrate prices using landcor data which is presented by municipality and type. Yet another source, but one concludes that some muni condos are flat since 2007.

  4. Canadian Watchdog (@CanadaWatchdog) December 22, 2012 at 10:14 pm

    Ahh the HPI. When someone can show me any mortgage insurer or global institution (IMF, BIS, etc.) that uses CREA or Taranet’s HPI, I’ll start to believe its accuracy. It’s the wrong methodology to begin with because repeat-sales was originally developed to stitch and plug holes in historical data, not to measure 12 years of available data. Even Shiller and Case admitted at times how they don’t trust their own index.

    Sales and how much money buyers bring to the market is what makes value. Nothing else!

    • an observer December 22, 2012 at 10:48 pm

      At the end of the day each property has its own dynamics but as far as measuring the pricing trends of an area I think a repeat sales approach is probably the best approach we have despite its flaws – with condo’s it’s actually a very good approach, less so with SFH because of the additional factors that come into play. What was interesting to me was the ranges of values – even in late 2008 there were many apartments being sold that registered large HPI increases and in the biggest boom times earlier in the 2000’s there were apartments registering large HPI drops. Using the median however seems to do a good job with eliminating the noise.

      One concerning thing with the HPI methodology is they openly admit that they weight the sales pairs and adjust the data based on various factors. They don’t go into many specifics but imagine a case where they provide a 3x weighting to a +5% sales which then offsets 3 other -5% sales. I don’t like the fact that it is massaged without providing the raw data and reasoning for adjustments in a way that can be audited. This happened in the US and it was later found that the HPI-type numbers they were using were completely wrong during the decline because data was being “adjusted”

      What is pretty interesting is that when looking at some of the listings today that have been around for a while and then look at the the last sale date and price and apply the HPI methodology to get a “current value” and compare to the current asking price, there are HUGE HPI drops coming soon.

      • Canadian Watchdog (@CanadaWatchdog) December 23, 2012 at 12:47 am

        If the momentum is diverging from what’s actually happening in the market, then obviously it’s flawed. That’s why I don’t bother.

        My biggest issue with repeat indexes is that it doesn’t account for pre-market prices, where assignments are sold and flipped likes resales. I’m not sure how large Van’s pre-market is, but just to put it into prospective using Toronto’s figures: TREB’s November dollar volume sales was $2.8B, BILD/RealNet November new dollar volume sales (my estimate using RealNet’s price index) is $1.3B. That’s 47% of TREB’s volume, and this market is not included in CREA or Teranet’s HPI.

        This entire segment of the market is widely being missed and I compare it to some situations Shiller and Case encountered when exogenous activity (like natural disasters) affected their index.

        Our boards are inexperienced with providing proper statistics to buyers and should focus their resources on data such as price per/sq.ft., foreign investment, foreclosure data, etc., not tinkering around with some silly index to smooth out a line. This is absurd.

        We’re deprived of statistics here in Canada — this can make one who watches the market with their nose pressed against the glass believe the only statistics being provided should be trusted. I don’t buy it. Until I see our boards providing the aforementioned stats, I’ll ignore any HPI noise.

        I appreciate your dedicated work here as it offers more information then any HPI. Hopefully more analysts will realize their time is better spent collecting new data then waiting around for others. Then maybe we can all trade like hockey cards.

      • an observer December 23, 2012 at 12:13 pm

        A quote I like goes something like this: “Democracy is the worst form of government… except for all the others”. Right now the same can be said about the HPI in relation to other metrics.

        It isn’t close to perfect but in an environment where average selling price was the most widely quoted and used figure up until recently we are really dealing with a void of information and a dumbed down public and I’m not convinced that there is a magic stat that would be any better than the HPI (or at least an objectively calculated HPI managed by an impartial group) even if we had all the information.

        I don’t think $/sqft is much more useful than average sale price and foreclosure rates have more to do with external factors. Foreign investment – well good luck getting honest answers there when apparently 97% of buyers in Vancouver West are “local”. My own key metric of price drops is somewhat useless on its own but at least it is something that can be monitored effectively.

        Now, give us an environment where we have access to all of these things and all of a sudden a cleared picture can be painted but still far from perfect

      • Canadian Watchdog (@CanadaWatchdog) December 23, 2012 at 2:54 pm

        You have the best data and I if I had it, I would create an index to monitor price drops using an average SP/LP ratio (average sold price divided by the average original list price). The premium or discount between the two is a real value measurement of what buyers are willing to pay. TREB provides SP/LP ratio in their monthly report, but they use the list price from the last MLS#, not the original list. Surely you can see the problem if they used the original list price. No way in hell they’ll publish that.

        As for data disclosure, I think Canada lacks enough corporate interest willing to spend money to access information or litigate institutions for disclosure, but then again, prices have been going up so there was no need for detailed information, until now. This type of practice is everyday business in America.

  5. RFM December 23, 2012 at 10:35 am

    Observer, thanks for the significant work you undertake and report.

    Currently, I have real-estate investments in two markets, Vancouver and southern California. In California there are free and publicly-accessible data sources on every transaction, allowing you to track what is happening. Here, there is a complete lack of transparency. Data are not available to the public. ‘Statistics’ reported by the stake holders are not explained. The media are gullible and unquestioning, reporting the pablum being generated by the people who are the least objective. Further, the HPI and other indexes are ‘lagging’ indicators, not ‘leading’ indicators. They only show where we’ve been. So, I rely on self-collected anecdotal evidence, such as the below, which I posted on VCI a couple days ago.

    I monitor completed condo buildings to see how the final sales numbers worked out. In summer 2011 (yes, 2011) Rize Alliance Properties Ltd completed a 48-unit condo building at 2511 Quebec Street in the South Main/Mt. Pleasant neighborhood. Promotional materials at the time made it appear the development was sold out or close thereto. Throughout 2012, there were always 5-7 units listed for sale on the MLS by the developer, although there are none listed at this time. However, based on very reliable information, it appears the developer actually has 14 units unsold, all 1 bedrooms, from 639-723 square feet, priced at $357,900-$411,900 with strata fees from $262-$297. That’s 29% of the building unsold more than a year after completion. Small projects are having trouble finding buyers; there are dozens of these small developments currently under construction, all over the city, in all price ranges. It’s worse than it may appear and probably not getting better any time soon.

    One day soon, interest rates in Canada are going to rise. The result will not be pretty. I believe that all the worthwhile gains in lower mainland prices are history; accordingly, it makes sense to liquidate investments here as soon as feasible, take any losses and move to markets outside Canada where I see more positive prospects. The strong Canadian dollar makes this especially attractive right now.

    Carry on and remember, it’s just money.

    • Ford Prefect December 23, 2012 at 4:14 pm

      RFM what you say about unsold units appears to uncannily parallel a situation in Courtenay. A development called “Piercy Creek Estates” (actually it is a dry ditch about 8 months of the year) consistently shows 5 or 6 units for sale but driving by in the morning there are suspiciously few units showing any sign of life. Further at least one unit, which has never been lived in, is advertised for rent at about $14k per year, yet ask price for units is around $240k.

      As to the HPI indexing problems discussed here, it seems to me that there is basically no market at present in this area and no way to give an accurate idea of HPI. The number of units sold is very small and the asking prices are still very high and not declining. As a result sales are virtually nil. In these circumstances it seems to me almost a guessing game as to where the market will go next. Up seems out, but how much down? How do you estimate for example the impact of the enormous debt levels the public is currently carrying? Or, if you follow the work of Jeremy Grantham, the impact rapidly rising commodity prices will have on housing prices? Presumably commodities up will equal house prices down given that more income will have to go to living expenses and less to mortgage payments. But almost impossible to quantify.

  6. Real Estate Tsunami December 23, 2012 at 11:33 am

    I find it incredible that with all the technology available, no one is able to gain access into CREA’s database. I believe sales data is public information and should be made available to the public.
    Maybe somebody should take the RE cartels to court. This is ridiculous.

    • an observer December 23, 2012 at 12:00 pm

      Does anyone know how long Zillow-like information has been available in the US? Has it always been public or did something change during the housing meltdown that caused government to mandate that the information be public?

    • an observer December 23, 2012 at 12:17 pm

      Also remember that this isn’t a technological problem – this is about a monopoly holding on to this information and deciding who can access it and how. Their argument, I believe, is that it is their information and they can do what they want with it.

  7. Julia December 23, 2012 at 3:53 pm

    the idea of using inflated asking prices to measure price decline is a little misleading. As you have illustrated so many times, people have delusions of riches in their asking prices. What did properties SELL for compared to assessments, and how many weeks did it take to make it happen. In the Vancouver market I am still seeing product listed from the spring that appears to be rationally priced. That is more alarming than the 3-500,000 corrections in expectations.

    We have new assessments coming out. Year over year comparisons will be interesting on some of your featured properties. Those numbers will reflect recent sales.

  8. Canadian Watchdog (@CanadaWatchdog) December 23, 2012 at 5:47 pm

    the idea of using inflated asking prices to measure price decline is a little misleading.

    If you're looking at individual property prices, yes, it appears to be misleading, but here's the problem when measuring entire cities, districts or areas. Suppose in a rising market, 100-200 homes inflate their prices 15-20% over their original purchase price or assessment. If they've all inflated prices, then collectively, the difference from their purchase price or assessment is a new premium to buyers who have no other choice to buy in that area.

    Now in a declining market, let's suppose one seller drops the price and sells at original list. This is now a discount from the premium asking price, therefore it can and should be considered a price drop even if no record of sale or stats will show it. This is why I focus on what is between active listings (offer) and monthly sales (bid).

  9. Damien December 23, 2012 at 7:13 pm

    Hello Obeserver, I just want to wish you happy holidays and say thanks for the info in 2012 ( it sure changed the way I see a ” RE cartel ” in Vancouver and Canada in general ). You should create a “donate app” on the homepage – I’d be more than happy to donate some cash for X-mass ” nogg & rum”.

  10. Kelly December 23, 2012 at 8:59 pm

    Merry Christmas Observer and thanks again for all your hard work. 2013 should be a very interesting year!!!!!!!!!!!

  11. Village Whisperer December 24, 2012 at 2:58 am

    Merry Christmas

  12. an observer December 24, 2012 at 9:52 am

    Merry Christmas everybody!

  13. elme December 24, 2012 at 2:16 pm

    Not only the CREA, all governments in Canada are not as transparent as a democratic system. And access to information requests are always refused! No wonder people like this kind of blogs which give more insights than any of governmental/institutional sources!
    Keep on great work!

    • Real Estate Tsunami December 25, 2012 at 9:50 am

      Do realtors have to swear an oath as to not disclosing certain RE data, unless it is to a client?
      Somehow they are acting like a secret society. Someone should go in undercover and report.

  14. micker07 December 25, 2012 at 7:42 pm

    Get 2012. prices didn’t start dropping in 2011.

  15. Landbaron December 26, 2012 at 3:42 pm

    Belated Merry Christmas to all! Doh, I forgot to leave season’s greetings earlier on this site.

    Thanks for all you hard work observer. This blog is very much appreciated as not only a great tool to moniter RE trends in Vancouver, but alwasy good for a laugh as well.

    Keep up the good work!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: