The number of San Francisco homes that sold in the first half of this year totaled a little under 2,700. That’s down a little over 8 percent versus the roughly 2,900 sales in the first half of 2018 and the fewest first-half sales in San Francisco since 2011 when a little over 2,600 homes traded hands.
In fact, the sales volume in San Francisco is down around 13 percent since the current cycle peak in the first half of 2014, when roughly 3,100 homes traded hands with lower average inventory levels, and 10 percent below the transaction volume through the first half of 2008 when 3,000 homes were sold.
And the pace of sales is on the decline.
The MLS displays 2019 is actually up, 1/1 – 7/31, with 2925, over both 2016 (2784) and 2017 (2903).
And yet recorded, versus self-reported, sales are actually down (as reported above).
I question your data.
The underlying data is by way of Corelogic. And while you’re welcome to question our compilation, it’s correct.
The MLS is backed by Corelogic aggregated tax data. My hunch here is the discrepancy is largely down to tenancy in common, joint, deed recordings. But what I’m seeing for example, right now, is that between SFRs, condo + TICs, 2-4 units, vacant lots, and 5+ unit buildings July 2019 has so far 509 sales. July 2018 has 518. If I were to guess July 2019 will ultimately surpass July 2018 as late filings trickle in …
We do exclude wholesale multi-unit building and vacant lot sales. But if you’re working with recorded deed data, you should know that single-family home, condo and TIC sales were down in the aggregate through the first half of the year, as reported above (and will be down through the first seven months as well).
In addition, we separately track contract activity, which is how we knew that June sales would be down (not up), and that the pace of sales in the city had dropped.
Don’t presume to tell me what I should know, as you report flawed information. Thanks. Nobody said June wasn’t down year over year. But April and May were up.
As to the other point, skip 5+ and vacant lots. The story remains the same. And it is not the one you wish to narrate on here. Sorry.
Like it or not, our report is correct. And while April sales were up, as expected, and May was a push, a one month bump wasn’t nearly enough to overcome the year-over-year declines in January, February, March and June.
And yes, the current pace of sales is down.
Well I don’t like it or dislike that you’re reporting flawed data due to recorded deeds vis a vis TICs, seemingly. It merely stands to reason. Otherwise, your point is that June was everything?
And what of the 2019 versus 2015 volume superiority? Because you spoke of trends from 2015 above. Or did your data hole get that wrong too? Are group recorded TIC deeds not a thing, Socketsite? Because it sure looks like you’ve under-reported a market sector, from here.
We appreciate the concern, but there aren’t any holes in our data (which we’ve compiled on a monthly basis going back 15 years and is correct).
And we’re not sure why you think we’re focused on “June” when the first half of the year has six month, four of which recorded year-over-year volume declines (ranging from 8.5 to 21.1 percent), one of which was a push (up 0.5 percent) and only one of which was up 5.1 percent (which was back in April with the pace of sales having since dropped).
Edit, that was 2019 vs 2016 and 2017 volume superiority. Actually so far 2019 1/1/ – 7/31 has been 3183 and 2015 was 3324 for a 4.2% better 2015. But 2019 has been 4.9% better than 2016 and 1.1% better than 2017. Regardless, there are no year over year volume dips trending per the MLS. There are volume wise a better 2015 and a better 2018 (3352, 5% better).
I did an edit for that 2015 v 2019 typo but it disappeared into the ether. Meant to say 2016, 2017, as 2015 was higher volume …. disregard this one if that one shows up … site is acting weird
How do you explain the discrepancy, then?
Or are you now conceding that 2019 so far has more sales than the first seven months of both 2016 and 2017? Because that renders your lede inaccurate?
Quite simply, we can’t account for your calculations.
And no, the accuracy of our lede remains.
Indeed. They’re flat statistics for SFRS, 2-4 units, and condo/TIC/Lofts, 1/1 – 7/31. 2019: 3185, 2017 3146, 2016 3026. You’re simply wrong.
It’s Deja Vu all over again.
The same thing happened at the end of the last real estate cycle. CoreLogic noticed a large persistent discrepancy between their data and the NAR/MLS data: “There are several reasons for the divergence, including benchmarking drift, more sales going through MLS systems due to consolidation and a lower share of for sale by owners (FSBO) home sales. Net, NAR’s existing home sales data are overstated by about 15% to 20.”
There was some back and forth, but eventually NAR ended up downward revising five years of data: ““All the sales and inventory data that has been reported since January 2007 is being downwardly revised. Sales were weaker than people thought,” NAR spokesman Walter Malony told Reuters.”
In Chicago, which I followed at the time, the IAR/MLS data for both price and volume was significantly overstated: “The Illinois Association of Realtors said Monday that the median price it reported for home sales within the city of Chicago was inflated in May and mistakes in its reports may go back more than three years.”
What was actually a decrease in median price was erroneously reported as an increase: “The size of the Realtors’ errors is statistically significant, at least based on the May median price for condo sales wtihin the city. In its official report that has now been discredited, the trade group previously said that the median price of an existing condo sold in Chicago in May was $299,000, compared with $271,150 recorded in May 2010. In fact, the median price was $243,000, compared to a year-ago price of $265,000, according to data from Midwest Real Estate Data LLC, the multiple listing service for the Chicago area.”
If you look at how days on market and price/list are manipulated along with past “mistakes” which always are in a direction which makes the market look better than it is, the Realtor data is a joke. I would not spend too much time worrying about discrepancies.
Statistician: Note that NAR members earn more on deal flow than price. Their interests are be best served by communicating a rising market to buyers and a falling market to owners.
Yeah, I don’t know there’s a lot to unpack there really. First, we’re able in the SFARMLS to look at the sales history of any individual sale. And it’s individual people uploading each individual sale. It is not a gross meta-manipulation. Also, what of the historically CoreLogic covers 85 to 90 percent aspect? And then what of the “elevated” ? Elevated to where? To even? How was that done? Because again you can back out any given data point and examine it minutely.
Also, finally, what do you make of the TIC component I’m surmising is also a culprit? Because it’s a fact that sometimes let’s say a 4 unit building will trade. That will be 4 individual sales. But Corelogic will view that as one deed, and one transaction in that given month. There’s no way that this is not a factor albeit, OK, in the margins, in the 100s per six months or 200s per year. But again, I’m only arguing marginal differences in numeric values here with the editor anyway … All these years still remain close in count at the end of the day. There’s no falling off a cliff, anywhere, to be seen.
Other thing is, while I questioned the CoreLogic figures. I’m not talking about CoreLogic versus MLS year over year. I’m submitting data that’s MLS year over MLS year. Where’s the distortion there?
SocketSite — your stubborn, self-righteous attitude undermines the trust I might have had in your data and claims. I’m starting to doubt whether I can trust any of the stats you post. Why are you being like this?
We definitely don’t suffer fools gladly. And if that leads one to doubt whether they can trust any of the stats we publish, versus simply not liking what the stats suggest or who they contradict, that’s a burden we’ll have to bear. But you might want to judge our “data and claims” based on their actual accuracy (such as when we point out that inventory levels are staring to rise (and then they do) or that contract activity is actually down, not up, and then sales really drop).
And all snark aside, we place a premium on accuracy and will quickly admit and correct any mistakes that we’ve actually made (which simply isn’t the case here).
You use the MLS data for pending content but reject MLS data for sold content. You refer back to your own editorial opinions to validate subsequent editorial opinions. That people question these editorial choices remains only normal.
Yes, we do use adjusted MLS-based data in our pending sales model but rely on recorded data for actual sales figures, which is what we report (and are down). If it makes you feel better to characterize our data/analysis as “editorial opinions,” or ignore the actual links (such as from our “opinions” to the actual results, which happen to align), so be it. But now back to the actual trends and data at hand…
CoreLogic DOES use MLS data, but, if I recall correctly, they double check with the recorder’s office to see if a deed was recorded that indicates it was a true sale. So if a “sale” is entered into the MLS where all that happened was the placement of a property into a trust, CoreLogic strips it back out of the sales numbers. So CoreLogic “cleans” the MLS data to make it more accurate, while your random real estate agent has every incentive to put questionable sales into the MLS to goose their stats so they can announce an inflated number of home sales to prospective clients. Just watching you in this thread should make it clear to everyone how important it is to a real estate agent to show increasing stats, even when the market is declining. We’re not going to count on your own self policing of your own statistics, when there’s every motivation to lie.
I’m not sure what CoreLogic does with respect to TICs, but I’d believe that they want to be as accurate as possible. Even if they removed all TIC sales, which I seriously doubt, the TIC sales would have been removed from the earlier periods as well.
Tipster, your take has nothing to do with reality. Nobody is entering fake sales into the MLS to goose his or her stats. The board of realtors governs the MLS and you simply cannot enter false data into it. Doing so would lead to fines, suspensions, and ultimately banning. It’s too important of a tool for any given realtor to abuse. Seriously, you do not know what you’re talking about whatsoever. And you don’t even grasp the TIC aspect I mentioned seemingly.
I grasp it just fine. If someone sells a 4 unit building as a TIC, the deed shows only one sale. And because CoreLogic just fell off the Turnip Truck, they don’t understand that its 4 sales and they shouldn’t disregard the other 3. And if the TIC is a billion unit TIC, they’d miss 999,999,999 sales! But first of all, that assumes that they disregard the other 3 sales (unlikely – it’s just as likely that they would identify the unit as a TIC and just follow the MLS as being the best data available), AND that the number of disregarded sales was any different in the period being compared. If they disregarded 100 sales this month, it wouldn’t matter, as long as they disregarded 100 sales in the prior period. And there’s no indication that the disregarded sales are disregarded, no indication that they are different, and no indication that the differences would be meaningful. So I think that argument is too speculative.
Finally, regarding the MLS as being a bastion of accuracy, I’ve seen enough “for comp purposes only” notes to determine that a number of sales are recorded in the MLS that were not true sales. Furthermore, when every photo is distorted, days on market are reset to 0, pricing is changed right before a lower offer is accepted to keep up the stats that tell all buyers that they have to pay 102% of the asking price because “everyone else does”, I think you might consider the rest of us have little faith in that system’s accuracy. Obviously CoreLogic has made a business out of everyone else knowing it’s nearly complete theater.
So it’s settled. The market is tanking and every buyer should cut their offer price in half. The realtors will make less money, which is why they will argue vociferously that the verified data is a lie.
Holy cow. You do not understand “for comps purposes only” ? That maneuver completely belies the very point you think you are attempting to make on here. But yes, good job grasping the TIC aspect. I feel glad, for your parents I suppose, that you understood at least that. The Corelogic ability to dive in and determine various divergences though? Why does that not apply to the MLS in general if you wish to make that argument? No, you do not know what on earth you’re talking about. But you like being cynical on here for fun. The editor doesn’t suffer fools though. He says. Please. You get to be a foolish cynic on here if you’re bearish, clearly.
“I’ve seen enough “for comp purposes only” notes to determine that a number of sales are recorded in the MLS that were not true sales”
“Holy cow. You do not understand “for comps purposes only” ?”
But that proves that it is possible to enter data in the MLS not backed by true sales. What if someone just “forgets” to check the “for comps purposes only” box?
“The board of realtors governs the MLS and you simply cannot enter false data into it. Doing so would lead to fines, suspensions, and ultimately banning.”
You say there is self policing, but these women in Miami altered and manipulated MLS data for years with no repercussions. Crazy story, read the whole thing:
“As Tomlinson dug deeper, he discovered that the Jills—or someone in their office—had made 552 data manipulations on 51 of their listings to erase a cumulative 23,740 days that the properties had gone unsold. The move allowed the Jills to “keep control of their expired listings,” Tomlinson later wrote, and earn “an extra $3 million in commissions.””
“Most of all, it’s about what one industry insider calls “a self-serving, greedy cesspool of unethical behavior,” all set against some of the world’s most expensive residential real estate.”
““It’s an insult to the real-estate community that there have been no repercussions against the Jills for this highly publicized violation of our rules and ethics,” says one broker. “It proves that no one’s watching, and anything goes.””
“But that proves that it is possible to enter data in the MLS not backed by true sales. What if someone just “forgets” to check the “for comps purposes only” box?
You don’t understand this on multiple levels.
First, what would constitute a “true” sale, to you?
Then, merely check a box? No. That’s not how it functions. One needs to enter an address that has corresponding Corelogic backed tax data.
Then once the item is entered, the entry is subject to anonymous error violation reporting by the entirety of the subscription base.
I’ve generously given correct data here and I’ve had a series of mistaken, disingenuous really, challenges. The editor speaking of not suffering fools, while allowing the likes of the poster Tipster to spout his or her sheer nonsense. I’ve had enough for now. Enjoy your bad information.
That’s funny, because lower interest rates compared to a year ago have given buyers around 10% more in purchasing power in terms of monthly payment. The fact that the market is not up around 10% actually means it is down, in terms of what buyers are willing and able to pay each month. And what happened to the much touted IPO boom that was supposed to provide a huge boost to the local real estate market??
My friend who works at a massive tech firm says they are hiring all of the lesser positions in secondary cities like Phoenix, and many of those in higher positions have one foot out the Bay Area door. Meanwhile the NAR reports that foreign buyers of US real estate are down around 50% since 2017, although some readers of this site claim that they were never a major factor here.
Some will jump to blame the trade war, but perhaps they don’t realize that the biggest threat to the US economy is actually higher interest rates, as evidenced by the market melt down last October.
Ten percent more buying power, you say. As prices have only increased? How so?
Not up 10% equals down, you say.
What happened to the much touted IPO boom, you say. As if much more nuanced takes haven’t been described in this space ad nausea.
Your friend who works at a massive tech firm says Phoenix, you say. Yet SF unemployment rates are at record lows.
Foreign buyers you say. I’ll leave that one.
Then you conclude as if something negative has actually occurred, and call last October a “melt down” ?
I don’t think you’ve said one thing that’s valid there.
Well, have a nice weekend.
SF unemployment rates are at record lows, but that doesn’t mean as much as you think it does.
The majority of the newly-created jobs pushing down the unemployment rate are service jobs that don’t provide meaningful benefits, much less the amount of income necessary to purchase a home in San Francisco or indeed anywhere in the inner Bay Area. So indicators from the relatively small sector of the economy that is producing jobs that do pay enough to purchase a home are more valuable to our understanding than aggregate stats like the overall unemployment rate.
SF unemployment rate is also low by default due to the lack of housing and high price of whatever housing is available, such that if you can’t find employment or you get laid off, you quickly leave (perhaps to wherever you came from to “make your mark” here) since you can’t afford to stay here and there’s probably no point in putting up with high fees, dysfunctional government, poor infrastructure, etc. if there wasn’t a high paying job to make it worth it.
Interest rates were going up and the economy kept growing when Obama was president and pre-trade war Trump.
Of course it’s not only the trade war. It’s also Trump’s deficit spending. Increasing the deficit in a growing economy is a terrible idea.
The shift of mid-level tech jobs to “satellite” campuses is going to only continue. Boise and Colorado Springs are two locations HP is shifting workforce to from the BA. The top paid techies will remain in the SV but they are not the majority of tech workers. As good paying tech jobs leave the Bay Area the smaller top paid tech tier will not be enough to hold prices up.
As the tech economy grows more quickly than the Bay Area builds housing, it will continue to expand, although HP is hardly the best example. HP is a former great which is shriveling.
At least for the foreseeable future, though, the indisputable center is in the Bay Area. More of the hottest companies are here than anywhere else.
over priced houses are not selling that’s a surprise,i noticed we are going into a recession,look at empty storefronts for clues,also Chinese from china have stopped buying America,reports 80% have stopped buying,also rents are at peak and are dropping,so investors cant get money back in rents,expect hugh recession after 2020 election,as companies and home owners have huge debt,trade wars are not helping,
“over priced houses are not selling”. Overpriced is in the eye of the beholder. The last 6 months have seen more over $3M and over $4M sales than the last few years saw.
Inner parkside saw over $1200/ft several times in the last month. Then there is this house which got nearly $3000/ft.
Remember when the editor says “homes” he means houses, condos, lofts, and multi-units. “houses” means single family houses and they are not operating the same way. Lots of new condos are getting built being one big factor in the decoupling.
That’s true. 2M and up SFR sales 1/1 – 7/31: 2019 411, 2018 409, 2017 311 , 2016 265, 2015 299.
3M and up: 2019 174, 2018 150, 2017 133, 2016 114 2015 128.
By the way, July sales 2019 vs 2018 for SFRS, 2-4 units, condo / tics, vacant lots, 5 + buildings, and unlisted sales is up now YoY. Tanking that is not.
You’re welcome, every reader not named Tipster or Editor.
Ignoring the fact that you’re still quoting a query of the MLS, unless July sales were up 50 percent on a year-over-year basis (spoiler alert, they weren’t), recorded home sales in San Francisco are still down through the first seven months of the year and the current pace of sales has actually slowed.
July sales would also need to be up at least 38 percent on a year-over-year basis for sales through the first seven months of the year not to total the fewest since 2011.
Given that it took 9 days into August for the MLS sales to beat YOY, I’m assuming we’re talking a “beat” like by 1%, not 38% or 50%.
Better luck next bubble.
Not true, as 2019 is up over both ’16 and ’17 through the first 7 months of the year, as I’ve said like six times by now.
In which case, for like the sixth time, you’d be incorrect (at least based on recorded sales data, versus your queries of the MLS, as outlined above).
And to put a finer point on the actual trend, which didn’t just emerge, it might be worth taking a look at the dotted blue line right here, which is based on recorded sales data from CoreLogic/DataQuick.
Indeed. This website owes its very existence to the MLS.
Despite what one’s query of the MLS might have suggested, recorded home sales in San Francisco were, in fact, down in July. And yes, year-to-date sales remain at an 8-year low, which shouldn’t catch any, or at least most, readers by surprise.
Nope. They weren’t, they were up, and you use the MLS data for pending data but not for sales. And that is all I have to say on the subject, other than you’re likely missing TIC sales.
No, they were actually down. As expected.
Citing your own post for a closed feedback loop remains as nothing as last month. You’ve clearly got a problem with your data. Too bad you’re not interested in putting out a better product for your readers.
While it’s our post, it’s Corelogic’s data with which you seem to have a problem. Of course, the year-to-date roll-ups are ours, as are the pending sales stats, which is how we’ve known the direction in which sales have actually headed (as have our readers that have been following along).
I have no problem with Corelogic. The problem lies with your postulation of Corelogic data as definitive sales results, and the narratives that are so based. It’s all rather obvious if you dive into the details. But it doesn’t match the narrative you wish to yarn here. Oh well. Your Socketsite persona lacks any shred of magnanimity. What else is new.
Separately, where would you be without MLS content? Photos, data? nowhere.
It’s striking how overhyped “IPO effect” has been and how little impact it’s had on the market. The headline progression in the Chronicle is pretty funny/tragic:
March: Bay Area home prices edge up as IPO anticipation builds
April: Bay Area median home price drops for first time in 7 years
May: Bay Area home prices flatten as buyers are ‘getting their sanity back’
June: Bay Area home prices fall 1.7% in May, biggest year-on-year drop
July: Bay Area home sales fell sharply in June – prices mostly fell, too
August: Bay Area homes got slightly more affordable in July as prices and mortgage rates fall
What do people expect the IPO events to look like as reflected in the SF real estate market?
Since Q2 there have been 527 sales of SFRs and condo + TICs above 2M. They’ve averaged $1261 per square foot.
Over the same timeframe in 2018 there were 492 such sales, averaging $1239 per square foot
I don’t know, Uber’s IPO price was what, $45 a share? Now it’s trading at over 27% lower. Lyft priced its initial public offering at $72 a share, it’s now almost 32% lower. I keep trying to figure out what that means for insiders who borrowed against their shares for liquidity just prior to the IPO. Did they face margin calls? Or are they better off because they got the money “in hand” before the price plunged and essentially just created more leverage at the expense of the lender?
Does the drop so soon after the IPO event increase of decrease the marginal propensity to consume (or invest, if you think high-end real estate is an investment rather than a consumption good)?
Anyway, when the real estate agents started talking about how, due to the IPO effect, “there’s going to be ten thousand millionaires overnight,” several commenters on this site mentioned that the people who were in a position to benefit most from the newly-created wealth either already owned their homes or were going to buy homes outside the San Francisco real estate market.
And here we are: Garrett Camp, the co-founder of Uber with Travis Kalanick paid $72.5 million for a brand-new Beverly Hills estate. Travis Kalanick himself just bought a new New York City penthouse for $36.5 million.
Fine, but you didn’t answer my question. Uber is down, Lyft is down, but capital was raised. And other IPOs, acquisitions, mergers, etc fared better. The Uber guys bought homes elsewhere. (Do you know they did not also purchase locally? I don’t. But I bet they did.) Anyway, locally we’ve seen that the $2M and up market has been trading at higher volume and price than any other year. That sector is unarguably up over 2018. So again, on a granular level is that not what it would resemble? You see all sorts of headlines about the Bay Area is down, market cooling, etc etc. Yet simultaneously this has happened, and there is more to come such as Airbnb … well, we think anyway.