Single-Family Home Inventory Is Building UpDecember 12, 2022
While the number of homes listed for sale in San Francisco ticked down 7 percent in the absolute over the past week with typical seasonality in play, there are still 40 percent more homes on the market than there were at the same time last year, with 30 percent more condos and now nearly 80 percent more single-family homes. In fact, the number of single-family homes on the market in San Francisco is at an 11-year seasonal high, as is the number of reduced single-family listings.
At the same time, pending home sales in San Francisco are down over 40 percent on a year-over-year basis and the average list price per square foot of the homes which are in contract, which is a leading indicator, has dropped under $870 per square foot to its lowest level in five years.
Expect a rash of unsold properties to be withdrawn from the MLS over the next couple of weeks before returning to the market as “new” in the spring, particularly single-family homes, the market for which will be tested in the first quarter of the year.
Comments from Plugged-In Readers
In January of 2019, the median PPSF for all attached/detached homes in San Francisco County was $979.
In April of 2022 the median PPSF for all attached/detached homes in SF County reached $1,152.
For November 2022 the most recent data available (This is all from the SF MLS) shows transacted prices of $1,003 PPSF. So, (including all condos and SFR) we are still above pre-pandemic pricing, but falling fast.
People still reaching for prime pandemic $1200ppsf type values may be disappointed, but houses are still selling at or near pre-pandemic levels despite the increased mortgage rates.
Closed sales are a lagging indicator. And as November sales are primarily based on contracts that were written in September, your most recent data and insight is a quarter behind the actual market and trend(s).
You’ve regularly pointed out that listing prices are not reflective of value when houses sell for “over asking” because those list prices can be manipulated by sellers and real estate agents. Now I’m learning your perspective is that the same parameter is a useful leading indicator of the market.
I do not pay taxes on, finance, or enjoy capital gains or losses from a list price as opposed to a transacted price.
Buyers writing offers in September already had elevated mortgage rates to deal with, so if your contention that September is when most November contracts were written is true, those transactions would be comparable to today’s market from a financing perspective. I don’t think it takes 60 days to close most home sales in San Francisco.
I don’t dispute that the market is trending lower.
Again, the lagging data you’re quoting is a quarter behind. And while the market had already turned in September, as we noted at the time, versus three months later, the decline has since accelerated, and rather dramatically, none of which should catch any plugged-in readers, other than the most obstinate, by surprise.
“You’ve regularly pointed out that listing prices are not reflective of value when houses sell for “over asking” because those list prices can be manipulated by sellers….”
You’re making this seem like hypocrisy, but isn’t this really just a classic trade off of data accuracy vs data timeliness? Of course transacted price is more accurate data, but you only get it after transactions have closed.
Using repeat sales, like Case Shiller, is also obviously the best way to look at market appreciation/declines. But CS is very lagged and the released data covers a very wide areas (much of the SF bay area, not just the city of SF). So you also here have a trade of between data accuracy, timeliness and granularity.
Seems like its a question of using the best available tool for the job. If you are looking at or making an argument about historic wide area trends, then things like list price and median price are probably the wrong tools. If you are trying to make an immanent decision for a specific segment of the market, then waiting months for delayed CS data that covers the entire region will do you no good.
“a classic trade off of data accuracy vs data timeliness” Yes – this is my point. The inaccuracy of the “over-asking” prices are both mocked and relied upon.
Why is it only now that SuperSport is using the “future-predicting” data?
Is it because these values have been going up for years and only now serve the pessimist bias?
Present the data with historic context, tell us what the premium/discount to later-reported typical transacted prices.
That there has been no adjustment for inflation/real pricing in all this “serious” work on real estate means that everyone is understating how much prices have come down.
I can only speak from my personal experience that well-presented, well-prepared houses still sell easily in San Francisco.
The pricing data is accurate and telling when property tracked, it’s its misapplication that we mock (along with the whiners and whingers).
Perhaps then you should properly track the data.
Terrific suggestion and fortunately we do, which brings us back to the trends at hand, none of which should catch any plugged-in readers by surprise…
We ARE talking about the trends at hand, and to the best of my knowledge you have never in the decade I have been reading this site used the concept of “average list price per square foot of the homes which are in contract” so if you are tracking that data properly, good on you. A chart with the premium/discount to transacted price would be illuminating.
soccrrmom – It is possible to get reliable trend information from an inaccurate data source if the inaccuracy is consistent.
For example if a data source consistently underprices something by 5%, you can still compute an accurate month to month trend because the underpricing applies to both ends of the sample interval and therefore factors out of the calculation.
Algebraically this is:
price_change = (price(November * 95%) – (price(October) * 95%)) / (price(November * 95%)
= ((price(November) – price(October)) * 95%) / (price(November * 95%)
= (price(November) – price(October)) / price(November)
In other words the consistent 5% underpricing factors out of the computation.
That’s a bingo!
No, the editor is being conservative. Your point is that because the editor usually argues that list price is irrelevant when prices are going up, because homes are generally underpriced in a rising market, it is a hypocrisy to use those same list prices when the market is going down. And the editor didn’t note the current list price to sale price ratio, which would have given a more complete picture. But Redfin has that and the sold price is currently running about 2.5% over list.
So the editor’s $870psft list price of homes in contract (he said under, but lets assume it’s at $870 on the dot), should translate to about an $892 psft sale price, down hard from $979 psft that you mentioned was the median in Jan of 2019.
The November 2022 last reported PPSF for all houses condos attached detached for November 2022 is $1,003.
You contend that value is likely to be reported for December 2022 at $892, implying an 11% drop in a single month.
We’ll see what happens.
Don’t fall into the trap of crossing data sets or (mis)comparing averages and medians.
Once again, the average list price per square foot of the homes which are in contract, which is a leading indicator, has dropped to its lowest level in five years, having dropped 15 percent since May.
A leading indicator of what?
A leading indicator of closed sales, which in turn (hopefully) become comps for setting prices for homes coming on market.
“A leading indicator of closed sales.” Can you unpack that? Do you mean the avg price/sf of closed sales? Okay, but that wouldn’t be very informative. Or do you (and the editor) mean there is some other metric (Case Shiller, median prices, etc.) of which avg list price/sf has a predictive value? Or something else? Something can’t just be a “leading indicator” in a vacuum. We just need more specifics so we can test the predictive value. This may indeed be a great leading indicator of something – but what is that something?
The editor doesn’t have a problem speaking for themselves.
Yes, what I mean is just what you summarized, that the average list price per ft.² of the homes which are in contract is a leading indicator of the average price per ft.² of closed sales. You might think that’s not very informative, but aren’t closed sales the most meaningful driver of future comps? (I am not a real estate agent).
In August of this year, 15.2% of home purchase agreements fell through because buyers backed out of purchase contracts, similar to 15.5% in July, according to a report from Redfin.
“Think mix.” I agree that pending avg $/sf should predict fairly well closing avg $/sf a short period later. But avg $/sf is not very often cited to measure the housing market, particularly at a point in time. Does this also predict the measures that ARE typically cited, like Case Shiller? Maybe. Or maybe not. That’s the question I posed – a leading indicator of what? If it is just a leading indicator of something that is not relevant, then it has little value. But maybe it does more?
@L&E – Ideally you’d have a repeat sale metric like Case-Shiller with a micro neighborhood granularity, timely updates and for non apple to apple re-sales a accurate model to compensate for changes to the property between re-sales. But this platonic standard doesn’t exist, so you have to deal with whatever imperfect metrics you can get that fit your needs for timeliness and granularity.
“Think mix” is right and it’s pointing you to one way to deal with imperfect data. Identify potential data issues and try and determine if they could be tainting your imperfect data. If ave $/sf is trending down it could be because the market mix is changing or it could be because prices are trending down in a repeat sale sense. It seems wrong to flat out dismiss changes in $/sf, but it seems very right to see if something like change in mix (or as soccermom points out inflation) is distorting that metric.
Back to the original issue, a drop in list $/sf for homes in contract could be an indication of a coming drop in closed $/sf or it could be an indication of a change in pricing strategy. If you are trying to make a timely decision you just have to weigh the probabilities of both possibilities.
I see you still have several stages of denial to go.
If people who read this are too stupid too ignore any RE analysis that relies on “median”, I guess that’s their business.
Nice SFH house (849 26th Ave) just sold at 10% below pre-pandemic price. Does that count as “near to pre-pandemic pricing”? Nearly 300K loss over 3 years still sounds like ouch.
That’s kind of a tough one, you know. I bet they were surprised it got nearly 500K over the list price at the time. I’m sure they were thinking overbids were a possibility of course. Yeah it’s a comp, etc etc. Anyone who responds with a rote take along those lines next I already know what you’re going to say, so don’t bother. Ahem. What would be good to know is what the other bids were like from that 2029 2.295M list price.
“I bet they were surprised it got nearly 500K over the list price at the time.”
This is a classic misuse of list price. As I mentioned above using list price in a historic context, in this case 3 years post sale, when you have the closed price available is pretty ridiculous.
But when all you have is list price data, before closed price data is available, then changes in list price can provide you with a leading indicator (providing that you keep in mind that you could also be seeing a change in pricing strategy vs a change in the market)
My active and engaged knowledge of this market dwarfs yours. Step off. Forever.
But merry Christmas, nonetheless.
2019, that was ^. Anyway, folks can say what they want about overbids and the market. But a massive overbid and a few years hold with no improvements will frequently result in a monetary loss. Good market, bad market, what have you.
UDPATE: The average price per square foot of the homes which are in contract in San Francisco has just dropped under $860 per square foot, which is 15 percent lower than in May.
Comments are closed.