Eleven months ago Paragon Real Estate attempted to define the decline from peak for 12 segments in San Francisco.

Based on a comparison of average dollars per square foot for sales at “peak” versus an average of sales from 10/15/08 to 1/30/09, Paragon concluded that declines to date in February had ranged from 6 percent for single-family homes in the Richmond to 25 percent for single-family homes in Bayview/Excelsior.  In-between were single-family homes in Noe/Eureka Valley (down 10 percent at the time) and condos in Hayes Valley/Alamo Square/NOPA (down 11 percent at the time).

In November, Paragon repeated their analysis but changed to 18 segments.

As a plugged-in reader noted last month, Paragon’s declines from peak to November (actually averages for sales from May to October) ranged from 9 percent for Pacific Heights/Marina condos to 45 percent for single-family homes in Bayview (no longer combined with Excelsior).

The decline from peak for single-family homes in Noe/Eureka Valley fell to 21 percent while the decline for condos in Hayes Valley/Alamo Square/NOPA fell to 18 percent.

And while the average price per square foot decline from peak to November for single-family homes in the much maligned District 10 averaged 35 percent, the average decline from peak for single-family homes elsewhere in San Francisco averaged 17 percent. The average decline for condos? 17 percent.

According to Paragon’s analysis, peaks for each area ranged from the first half of 2006 (bottom end of the market) to the first half of 2008 (top end of the market), with the majority in 2007 or before. As we wrote in 2006, “Get ready for what we’re going to call a real estate ‘flight to quality’.” And nobody that’s plugged-in should have been caught by surprise.

Keep in mind a comparison of average selling prices per square foot is far from perfect especially when painting with such broad brush strokes. We can’t vouch for Paragon’s methodology or results. And as Paragon correctly noted in February (but not in November), price per square foot comparisons in a down market tend to understate actual declines in no small part due to changes in mix (might “beauty pageant effect” sound familiar?).

All in all it’s just another metric to consider, some extra context for our apples (think drops from pre-2006 values), and food for thought and reflection at the end of the year.

Paragon’s November numbers versus peak (and “April”) for all eighteen areas:

Decline from an estimated peak in the first half of 2006:

  • Bayview SFR -45% (down 5% from “April” 2009)
  • Excelsior/Portola SFR -25% (-1.5%)
  • Ingleside/Heights/Oceanview SFR -23% (-1%)

Decline from an estimated peak in the first half of 2007:

  • Mission (Inner) Condo -20%
  • Sunset (Central/Outer) SFR -20% (-6%)
  • SOMA Condo -18% (+2%)
  • Miraloma/Sunnyside SFR -19% (-8%)
  • Saint Francis Wood/West Portal/Forest Hill SFR -15%
  • Richmond (Central/Outer) SFR -14%
  • Potrero Hill SFR -14%

Decline from an estimated peak in the first half of 2008:

  • Noe & Eureka Valley SFR -21% (-6%)
  • Noe & Eureka Valley Condo -18% (-9%)
  • Hayes Valley/Alamo/NOPA Condo -18% (-7%)
  • South Beach Condos -18% (-6%)
  • Bernal Heights SFR -13% (+2%)
  • Russian/Nob/Telegraph Hill Condo -13%
  • Pacific Heights/Marina Condo -9% (-4%)
  • Most Expensive ($1.5-4.0M) North Houses -18%
43 thoughts on “Another Market Metric and Food for Thought at the End of the Year”
  1. As far as I can discern, Paragon has calculated these declines based on places that actually sold. That is a valid and useful metric. It would, however, miss all the places that were withdrawn or which simply have failed to sell.
    The second table here gives a nice picture of the decline in $/sf in listing prices for all units for sale (not just those that actually sold) citywide for SFRs and condos:
    http://www.altosresearch.com/paragon/latest/paragon_market_update_zip_based_cmid_55_zipd_none.html

  2. In some respects 2009 has been has been a fantastic year. Thank you, SocketSite, and all frequent posters. In a year in which it was not uncommon for families (especially those tied up in real estate) to lose half of their net worth, my wife and I more than doubled ours. I’m young, so our regular income contributed some, but I attribute a very healthy percentage to investment decisions heavily influenced by the regulars on this site (we miss you, LMRiM). Happy New Year to all, and I hope to be able to contribute to the discussion myself in 2010.

  3. disagree that 2009 was a year in which it was not uncommon to lose half of net worth.
    stock markets are significantly up on the year.
    i think housing may be up slightly as well, using Case-Shiller as a guide, I would expect that to show prices up for calendar year 09 the way things are heading. certainly significantly up for the past 6 months.

  4. Valentino wrote:
    > In some respects 2009 has been has been a
    > fantastic year.
    It sure was for those who put it all in the stock market back in March. I was just listening to the radio after the market close and thinking that and I never would have guessed the S&P 500 would be up 65% and the Nasdaq would be up 79% from their March lows. In the apartment rental business 2009 was not a great year with overall flat to down rents and a lot more turnover than average due to job losses. The one bright spot in the family real estate business was that I was able to double the monthly rent on my Grandparent’s former house in Balboa Terrace after the long term tenants moved out and I convinced my parents to spend about $50K to remodel the kitchen and bathrooms (that have not any real work since my Grandfather bought the place for ~$5,000 back in 1932).
    > My wife and I more than doubled (our new worth)
    How about a little info on how you did this (investing in the stock market at the low this year? Flipping condos South of Market? Shorting the Dubai real estate market?
    > We miss you, LMRiM
    I was just talking to a friend about blogs this past weekend and let him know how much I miss LMRiM/Satchel, but I also said that I would miss anon/fluj if he were to move back to Ohio and stopped blogging (and giving me a hard time on a regular basis)…
    P.S. Happy NY to all and remember a cab ride is a lot cheaper than a DUI…

  5. but I also said that I would miss anon/fluj if he were to move back to Ohio and stopped blogging
    People born in California can be so provincial/funny the way they view people born elsewhere. I have lived here in SF longer than I’ve ever lived anywhere. Most likely longer than you ever did as well, Mr. Peninsula. LOL. Happy NYE to all haters and lovers everywhere.

  6. btw that Altos link by anon above does show how misleading the concept of ‘peak’ can be – the peaks for median $ per sq ft around 7/1/06 and 1/1/07 especially .
    anyone who bought just a few weeks after these dates would already be showing over 10% down from “peak!”
    Just shows clearly how the changing mix factor will always distort these peaks upwards, and averages are even more susceptible to these things than medians. Bring in smaller data sets like districts/sub districts and its pretty out of hand!!
    anyway, happy New Year!

  7. FAB,
    Well I can assure you that anonn is not going to move back to Ohio. He is here to fight the good fight on SF blogs as long as it takes.
    Here’s to Summer ’10 seeing the 1000/ft. comeback we are all hoping for.

  8. It definitely wouldn’t be as much fun without anonn to keep us on our toes.
    Here’s to a return to sanity in the financial markets in 2010.

  9. disagree that 2009 was a year in which it was not uncommon to lose half of net worth.
    Maybe not as common as I made it sound, but going from -10% to -21% on your single-family Noe Valley home could be a very substantial chunk of net worth.
    How about a little info on how you did this (investing in the stock market at the low this year? Flipping condos South of Market? Shorting the Dubai real estate market?
    We were basically 100% in cash through most of 2008. At the beginning this was mostly dumb luck as we were considering buying a home and sold off all stocks to do so. As I gained a modicum of savvy it became apparent that my timing couldn’t have been much better, and buying a home was not the right thing to do. I did a whole lot of day trading during the craziest times of September to December… buying and selling every 10 minutes kind of stuff… double long and short ETFs. That sure was a blast. I made quite a bit of money off of DXO while that was still around. I shorted REITs when a few posters on this site were doing the same. I took LMRiM’s advice and bought into many overseas market ETFs such as EWY, EWS, EWH (should have bought EWZ, but did not). Lately (the past 3-4 months) I have been trading U.S. company options and have been mildly successful, but it’s peanuts compared to the previous 9 months. Probably because I don’t have LMRiM around to give me any more new ideas 🙁 Also, the whole of 2009 has been very good as far as my personal business goes, and I’ve made more money “working” than ever before.
    I doubt I’ll be able to repeat the performance but I’m happy to be where I am at the moment. We’ve got enough to buy a cabin in the mountains and that’ll be one life goal checked off the list.

  10. An American home no longer affords the owner a call option on a future of robust wage growth–something which has been roughly true since the end of WW2. While some homes, and some towns, and some neighborhoods in America will offer a way to retain stored capital built up over the past 50 years–even these will not see rising values in the next decade.
    The game is over: replacing real GDP and real energy inputs with credit no longer works. Real estate is a dead asset class, and will be for a long time to come. At least until a broader swath of the nation converges around the 250K level for a house. It’s not coming back. Ever. The prices of the last decade are generational highs.

  11. Here’s to Summer ’10 seeing the 1000/ft. comeback we are all hoping for.
    Happy New Year from (as LMRiM affectionately called it) Shytzville. Sparky-b, if your bottom call is correct, I’ll bring you back some of the finest the region has to offer. Looking next door I see: dual incomes, one “over-improved” home and lot, one job loss, and now a for sale sign. Same story, different state…
    We entered 2009 with the Marina still prime but the tide going out; will be interesting to see who’s swimming naked in 2010.

  12. 2009 was a great year for us here in the netherlands of the San Joaquin Valley. We paid off our house and socked away cash. Now we’ve signed a lease for a very tiny SOMA studio as a second home until later when we hope to move full time to SF. You all are delightful to read and learn from. Cheers for 2010.

  13. One place where many (of the northern neighborhood and Marin-residing) bears on here were wrong (at least, so far): Bernal houses did better than single family houses anywhere else, on the above list. Several folks here predicted Bernal prices would follow the Bayview and collapse. In fact, prices for Bernal single family homes fell less than in Noe or in the most expensive northern neighborhoods.

  14. I remember being impressed by this data when Paragon came out with it previously. I only have two quibbles — (isn’t that what’s so great about statistics? You can always quibble!) First, it would be useful to know on what basis they decided what was the “normal” price range for each segment. Clearly, the decision about how, and how much you decide to cut off at the top and bottom ends will have a significant impact on the results you end up with. Secondly, I don’t like the fact that they use the list price as the sale price on properties where the sale price is confidential. This can be really misleading, for example, in the SOMA condo market, where lots of the developers are not publishing their selling prices and everyone knows that they’re not close to list. I’ve taken the opposite approach when I run my numbers and simply exclude properties that don’t have a published sales price.
    Predictions for 2010?
    Happy New Year everyone!

  15. The biggest problem with these numbers is the fact that everything sold at the peak, even properties in terrible shape. So he median at the peak represented properties in not the best shape.
    When the downturn hit, only the best properties sold and much of the rest was withdrawn. So the median now represents a much higher quality property. People don’t buy fixers as readily on the way down because by the time you fix them up, the price can fall further and it’s initially tough to get the seller to lower the price to compensate. So properties selling while things are headed down are of higher quality.
    So they are comparing a lower quality product (the median at the peak) with a higher quality product (the median now), which will mask the declines and make them seem smaller.
    So the stats are basically worthless. They could have blindly (i.e. without knowing if it was a peak sale or after peak sale) graded the condition of the homes sold from the listing photos and then done a better job on the median ppsft by eliminating outliers and trying to match up the same quality. Instead, they let the inherent market bias goose the stats on the more recent sales.
    It may be that Bernal is so undesirable, almost no one will touch a home there unless it is in perfect shape. Or it may be that Bernal was considered so desirable at the peak, that lots of homes in bad shape got sold. Maybe none of these things are true and the comparison of the before and after is fine, but if so, that has occurred by coincidence.

  16. “Predictions for 2010?”
    Real SF saw price declines of roughly 10% in 2008 and another 10% in 2009. Non-Real SF saw much bigger declines in 2008 and a bit smaller declines in 2009. (Obviously declines are not uniform in a non-transparent and non-fungible market like this). The difference in 2008 is explained by the different loan products and buyer profiles used during the bubble, with non-Real SF buyers quickly defaulting as their loans fast became unaffordable and they had no savings to draw from, and Real SF bubble buyers using loans with longer recast periods — which continued to be available after subprime loans disappeared — and generally some savings to draw from to pay the “owner’s premium” for a while.
    The trillion dollar bailout in 2009 largely, but not entirely, stemmed the declines in Non-Real SF as these efforts were capped at conforming and FHA limits. These did not help Real SF much, if at all, other than to lower interest rates.
    For 2010, further declines in Real SF are baked in with continuing economic weakness, growing recasts, underwater owners more willing to walk away, and foreclosures greatly accelerating at these higher price levels from 2009. My prediction is another 10-20% down in 2010. Conforming loan limits will provide somewhat of a floor for primo properties.
    Non-Real SF is harder to predict as it entirely depends on government action. My best guess is that mortgage rates will rise somewhat and foreclosures will continue to accelerate as they did in 2009 despite the outlandish sums being dumped into the system. Non-Real SF has already fallen much further and is closer to the bottom, but my best guess is another 5-10% down in 2010.
    Many on this site warned would-be buyers in 2007, 2008, and 2009 not to jump in or they would face big losses. They were right (OK, we don’t yet know about 2009 buyers). 2010 will be less expensive, but my advice is that those buying in 2010 are still going to see further declines in 2011 and 2012, and you better factor in that risk when calculating whether you can afford the still-high premium to buy vs. rent.
    Happy 2010 to all those happy renters who avoided a devastating financial loss! And to the realtors out there, the best development would be if prices fall much faster than I am predicting, so you can once again start moving homes as buyers are not going to return in large numbers until they are confident they will not lose big money on their purchase.

  17. Predictions for 2010: prices flat and trending upward, roughly tracking inflation. CSI for top tier SF at 169 in Dec.2010.

  18. I think lumping Sunnyside and Miraloma together is as erroneeous as lumping Miraloma and Glen Park together would be.
    Two completely distinct neighborhoods. Miraloma has a much better housing stock with many large homes, views and a fair number of detached residences.
    Sunyside is in the valley below and flanked on one side by City College with attendant parking problems and a lot of “through” traffic. The housing stock is not so great and the grid street pattern leaves a sterile feel to the area.
    Just a guess but I suspect Miraloma did not do as bad as 19 (8) and that Sunnyside did worse than that.
    It’s an averaged figure across two distinct neighborhoods and an especially valid measure because of that.

  19. otheranon: nice analysis, and I largely agree, except that I’d put the downside risk for real sf at 10% — barring a major global surprise (significant unrest in China, Iran getting bombed by Israel etc.).
    I know that many people seem to think that realtors can be nothing but self-serving, but I’ve been predicting a double-dip market since the September 09, despite the sterling run-up in prices over the summer: http://www.pegasusventures.net/wordpressblog/2009/09/16/alphabet-soup-what-shape-will-the-recovery-take/

  20. @BCCB: Given that the ‘districts’ they seem to create are geographically continuous, I imagine the western addition/fillmore would be part of the Hayes Valley/Alamo Square/NOPA tract they list.
    I’m not sure those areas represent a homogenous real estate market, as I don’t think people looking to live in a certain type of SF neighborhood would be looking within that entire area. Put another way, how much overlap is there between buyers/sellers in NOPA and hayes valley?
    NOPA, as defined, is a bit strange actually, as people east of Lyon probably go to Divisadero more frequently for neighborhood businesses and people west head to Masonic or Stanyan. The major streets should be used as neighborhood centers, not dividers, in many cases. Alamo Square and Eastern NOPA are more like one continuous neighborhood than NOPA is (in my opinion)
    @anon: regarding the Mission, they have a category called Mission (Inner), which is what I believe you are looking for.

  21. Toured 5 houses on the Peninsula today — tons of open house traffic, weird mix of estate sales (heavy traffic and selling fast with multiple bids — selling in a matter of a few days) and short sales / underwater sellers languishing on the market for months and months at high prices. And nothing in-between. No “regular” un-forced sales anywhere in my price range (3BR/2BA sub-$1M).
    If it weren’t for the fact that the probate and estate sales are the only properties selling (and therefore “available” inventory is very low), I would say the market is really hot, but there is clearly a ton of supply lingering out there in the form of short sales and other overpriced distress situations.
    Maybe its just the time of year …

  22. Jimmy,
    When I do a redfin search for San Carlos for 3/2 under a million, I get 53 sales in one zip code (94070) in the last three months. Prices look to be down at least 15% in the last year and are somewhere around their 2004 levels. Looks like there has been plenty of inventory that sells in that price range
    Looking for houses between Christmas and the Super bowl is largely futile. Not impossible, though.

  23. House-hunting is a little sobering at times; one of the homes (for sale due to a divorce) had an “in-law” unit underneath where the tiny shower was hidden behind a rack of clothes. The shower door was about 18″ wide at most… pure horrorshow (that plus the cement floors down there reminded me of ‘Saw’). The tenant living down there had a book on his desk titled ‘Personal Bankruptcy.’
    The owner of that property paid $785k in 2000 and was getting offers in the low 8’s now. Ouch.

  24. The paragon analysis reflects what I have been saying since 9/08. Namely, except for d10 most other props are down 10-20% from peaks.
    I slightly disagree with their -20% drop for the mish. I think SFR’s fell ~ 10% and condos 10-15%. A) inventory was never high in the mish B) gentrification continues at a strong pace here C) most props are sub $1mil, which have taken a lesser hit (than say Noe).
    Also disagree with tippy about mix. Less than stellar homes still sold post boom. There were also plenty of flips during the boom period (usually renovated). The biggest change in mix is in the > $1.5 mil catagory. we simply had less higher end sales post boom, mostly effecting Noe and northern part of town. This could have some psf effects but not enough to make the analysis untenable IMO.

  25. How is it that they determined “peak” again? Remember the numbers at the top are never the large sampleset some of you numbers, numbers and more numbers folks want. The determination looks a whole like “pick a few months where seven or eight excellent properties happened to sell, and call it peak.” Over a four year period that grouping could also be called “coincidence” instead of peak. Also, if you look at top Eureka Valley versus top Noe SFRs, you’ll definitely see Castro has slid more. Especially if you concede that Liberty Heights is by and large its own thing, and often more thought of as Noe Valley than Eureka Valley as far as most buyers are concerned.

  26. “How is it that they determined “peak” again?”
    You’re preaching to the choir boss, but that doesn’t tell us what we should be using for an algorithm to determine where the peak was.

  27. ^ I think what anonn was trying to get at is that “peak” is a fuzzy thing (pun intended) and thus open to interpretation/deliberation/misrepresentation/etc.

  28. I think everyone gets that peak is somewhat open to interpretation/deliberation, partly because the quality of the data is so terrible. Honestly, I don’t really understand why the amount of debate over the exact timing of peak is as high as it is here. Generally, I think people do the best they can to determine when the peak was, and the statistics aren’t very good on these things, so it’s even harder for that reason. After all, if we had decent statistics, there wouldn’t even be much of a debate on this.
    Paragon’s stats seem better than most, and I don’t at all buy this baseless claim from anonn that it happened to coincide with the sale of “seven or eight excellent properties.” That’s just another way of saying, “it’s all micro, bro” and trying to cast doubt on actual data without saying anything of substance.

  29. No. You can doubt the validity of “it’s all micro bro” when it comes to small samplesets if you like. But you need to explain why. From where I’m standing, it is you who are inserting disingenuous language. It’s a fact that the datasets are small. It’s a fact that numerous high ticket sales in a particular quarter, over a 4 1/2 year period, can be down to coincidence. (It’s also a fact that the very period called “peak” in the Paragon study is a time period in which the Socketsite bears were saying Katy bar the door, SF is about to get killed. But I digress.)
    No. What you and others want to do is take the Paragon data, and some of its dubious neighborhood linking too, and seize upon it. “Better than most,” you say. Fine. But why is that? How are small datasets not easily skewed? How is it that that skewed result can be applied unilaterally to a proper dataset?
    You can argue against it if you like. But if six of the very best houses in Noe were to sell right now, first quarter 2010 would look pretty close to peak. And we all know that isn’t true. The volume is never what you want it to be.
    Some of these same voices were the very voices who cry out against small datasets as exemplar in an upward market. Funny how that works.

  30. Apparently you wouldn’t know that 1Q2010 isn’t peak, anonn, since you said in mid-2008:
    “It’s over. Sorry. Scare tactics are dead. San Francisco never really took a price hit and it won’t, either.”
    Sounds like we shouldn’t take you seriously when it comes to interpreting data…

  31. No one disagrees that it’s a relatively small dataset for certain neighborhoods — $1M+ homes are a smaller market than in, say, Contra Costa. But how small a dataset is it? Can you tell us, or would you just like to assert that it is?
    The peaks estimated by Paragon were determined based on a 6 month period, so they are certainly based on more than one quarter. Using half a year as peak seems more than fair, in any case, when comparing sales between years — it’s not like they picked a single month. Furthermore, these are per-square-foot numbers, which are less susceptible to so-called coincidences within these neighborhoods than, say, medians.
    Again, if you have better data, please share, whether from DataQuick or somewhere else. We all know you don’t like numbers because they might actually reveal something, but the alternative is taking you at face value without proof, so I’ll take the numbers.

  32. Hey, annon, feel free to give us your own analysis of peak prices and where they are today in terms of that peak. You won’t do so, of course, even though you have full MLS database access, because you know what the results would show. Heck, you haven’t even pointed to anything concrete you think is wrong with Paragon’s methodology. You’re just noting an obvious point that nobody disputes, that hypothetically it might result in skewed numbers, which might be skewed in either direction I might add. This is just another example of your constant refrain that “You’re wrong, just because I say so.”

  33. Oh, you mean something different than the 5 to 10 percent off peak for SFRs any Socketsite reader would consider purchasing? The opinion I’ve voiced a thousand times on here? OK. I now think it’s more like 10, by and large.
    And “hypotheticaly” might result in skewed numbers?
    Read Paragon’s own disclaimers at the bottom. Are they saying anything different than, “It’s all micro, bro” ?
    Lovely pretzel logic, fellas.

  34. As I expected, anonn, you refuse to provide any specifics that can be be reviewed. “I’m right and you’re wrong because I say so.” All hat and no cattle.

  35. No. You’re calling something imprecise, precise. It acknowledges its own imprecision in the footnotes. Meanwhile, you’re asking me for precision. You’re stating that my take on what I experience every single day is somehow less than a heavily qualified report from obviously skewed numbers. What, it’s better because they used powerpoint? My predictions on this website have seldom been as much as 10 percent off. You think the Paragon disclaimers might account for at least 10 percent slides? Most people would.

Leave a Reply

Your email address will not be published. Required fields are marked *