In a move which shouldn’t catch any plugged-in readers by surprise, the number of homes actively listed for sale in the San Francisco (765) jumped 39 percent over the past two weeks and is now running 9 percent higher versus the same time last year (700) and 25 percent above its mark at the same time in 2015 (610), the year which remains an inflection point for the current cycle.

At a more granular level, the number of single-family homes currently listed for sale in the city (305) is now running 33 percent higher versus the same time last year while the number of listed condominiums (460) is currently running 3 percent lower, a total which doesn’t include the vast majority of new construction condos for sale across the city, the inventory of which has been hovering around 500.

In terms of pricing and expectations, 14 percent of the active listings in San Francisco have undergone at least one price reduction, which is even with the same time last year, while 35 percent of the homes on the market are currently listed for under a million dollars, up from 34 percent at the same time in 2017.

Expect inventory levels to continue to climb through the end of September or mid-October. And keep in mind that recorded home sales in San Francisco dropped to a seven-year seasonal low in July.

18 thoughts on “Inventory of Homes for Sale in San Francisco Jumps”

Single family home inventory YoY number seems significant.. see how the market reacts..

Prices have gone up for so long, I’d be surprised if we didn’t see a flattening or decline. Of course, those proclaiming that the peak has passed have been wrong for about three years running, so taking the “over” on my bet may not be a bad idea. One (of many) examples is 45 Hancock #2, just sold for $1,800,000, $1500/sf and $200,000 more than 3 years ago. And MLS single family home prices just hit a new high in August, and places continue to sell extremely quickly. But even with a booming local economy, there is a limit.

And with many more examples of recent apples-to-apples sales which were much lower, if not negative in terms of returns, where are the dozens, if not hundreds, of 30-percent-plus gains since the fourth quarter of 2015 which would be necessary for the average gain to simply match the median price gain in terms of actual values?

Right, as you point out, this place on Hancock underperformed the market by a little bit. Median price is up. Case shiller is up. $/sf is up. Some places will underperform, and some will outperform. A big mistake, and very misleading, to just focus on those at the edges.

I’d be surprised if those buying today see the decent gains we’ve seen the last three years.

A much bigger mistake is to continue to conflate changes in the median price with changes in values, changes in a Bay Area index (Case Shiller) with changes in San Francisco proper when the area’s segments aren’t moving in lockstep, and to ignore a growing body of apples-to-apples results.

But once again, with many more examples of recent apples-to-apples sales which were much lower, if not negative in terms of returns, where are the dozens, if not hundreds, of 30-percent-plus gains since the fourth quarter of 2015 which would be necessary for the average gain to simply match the median price gain in terms of actual values? And yes, we’re giving you free rein to cherry pick your points.

I think you need to do some basic research on statistics. Try articulating why it would require pointing to dozens, if not hundreds, of 30-percent-plus gains since the fourth quarter of 2015 to be necessary for the average gain to simply match the median price gain in terms of actual values. Go ahead, and I think that in the process perhaps you’ll see your error in statistics and logic.

Our apologies, we assumed you understood how averages worked. If there’s a “below average” point in a set, there needs to be an “above average” point as well. And with a growing set of (well) “below average” actual, apples-to-apples, results, there would need to be dozens (if not hundreds) of (smaller) “above average” results to yield an average change in actual value to simply match the 20 percent change in median sale price since the fourth quarter of 2015.

Got it – the huge flaw in your logic is that the average and median price data are not ascertained by reference only to “apples to apples” results. Indeed, a very low percentage of sales in any period are recent resales. So you don’t need to show “dozens or hundreds” of over-performing “apples to apples” to demonstrate anything. Indeed, you’re mixing up apples and oranges, in a sense. You could have 1000 sales in Year A with a certain median and average, and 1000 sales of completely different homes in Year B, with a certain median and average. And the results are reliable. “Apples to apples” resales have little to nothing to do with broader measures of medians and averages. Monthly data comparisons are not great because the sample sizes aren’t bug enough. But annual data or moving averages are reliable and show significantly rising prices since 2015 by any measure.

It’s true that “apples to apples” sales are illuminating, but not if you only mention a small, non-random and skewed sample of those. There are far more “apples to apples” showing gains than losses. You’re also blowing hot and cold as you argue case shiller data are all wrong because the low end has been outperforming the high end concentrated in SF. But then you argue SF medians are all wrong because the high end has been outperforming and changing “the mix.” You’re getting crossed up by your own leaps of logic (and with no evidence at all). It’s perfectly fine to have an agenda, but don’t pretend that you don’t have one with faulty logic and transparent sophistry based on a misunderstanding of simple statistics.

Interestingly enough, the change in Median Sale Price we’re quoting is a quarter-to-quarter average from 2015 to 2018, not simply month-to-month. Unlike some (ahem), we pick our potential “apples-to-apples” sale pairs prior to knowing their outcome. And unfortunately, simply typing “there are far more “apples to apples” showing gains than losses,” over and over and over again doesn’t make it so.

But you are correct, “apples to apples” resales in San Francisco “have little to nothing to do with broader measures of medians and averages” as our growing aggregate of apples-to-apples transactions tracks the actual change in San Francisco home values versus simply measuring what’s selling (median price) or the Bay Area as a whole (Case Shiller).

So forget the hundreds, or even dozens, of apples-to-apples sale pairs you’d need to average out the growing body of “below average” outcomes in our cart, how about a few “above average” examples to simply average out the “below average” point you picked to match the median price gain? As there are “far more examples,” it should be a piece of cake!

You’re demanding dozens or hundreds of “apples to apples” resales since Q4 2015? Go read my previous post, then maybe spend some time with Khan Academy on statistics. But here you go. 506 Cole Street. Sold for $1,375,000 in May of 2016. Just sold again for $1,750,000. Up 27.3 % since Q2 2016. Usually you delete posts of resales with nice gains, but maybe you’ll let this one remain.

It is really all about the 2018 buyers paying less. The fact that a property is an Apple just makes the value change obvious. With non-Apples, you can factor out the value of renovations and price changes from years back and get an estimate of recent price movements. But this is not an exact science and the element of subjectivity lets people bend these estimates up or down. With an Apple it is completely clear. The exact same good now selling for less than it did before.

doc java, “2018 buyers paying less” than whom? I agree “apples to apples” are useful. The big error is when someone tries to take a sample size of 1 (or 10 or 20) and claim that reflects “the market.” That is invalid, whether it is pointing to one example of a 27.3% gain in two years or a 10% loss in two years. That is why the whole world uses broader measures, like means and medians, or case shilller (or the S&P 500 in the stock realm). None is perfect, but cherry-picking is always useless.

We certainly agree that “cherry-picking is always useless,” which is precisely why we don’t. But a funny thing happens when the collection of so-called “exceptions” continues to grow, and grow, and grow – it represents the market overall while catching the leading indicators and segmentation along the way.

And yes, broader measures are great! Except when one confuses what they’re actually measuring and the underlying trends at hand.

““2018 buyers paying less” than whom?” Less than the previous buyers.

People take population samples all the time in statistics. Why do you think Apples are not a representative sample? Nothing stops you from looking at non-Apples as well. But you need to have some model to account for renovations, legal changes, past appreciation and the like. IMHO you are going to have an order of magnitude more error trying to estimate the 2017-2018 value change from a property with a 10 year hold and multiple renovations versus a 2017 to 2018 Apple.

The modeling error on the 10 year hold will overshadow any sampling error for the Apple. And the editor has a good point about the mathematics of averages. If you claim that averages prices are up 20% from 2015 and the editor has a large list of Apples with prices at or below 0% from 2015, mathematically there would have to be an equal weight of resales far above 20%.

West Portal has seen huge gains in average sale price in the last 3 years as buyers priced out of Noe and Cole Valley markets buy there instead. There have been a couple $3M+ sales recently and more close to that level.

Sincere response: San Francisco real estate is like the weather. You must account for microclimates. There are very, very few true apples-apples in any of our micro real estate markets.

So transaction count is down and inventory is up, suggesting people who don’t have to sell are choosing to hold on while people who have to sell are not getting the prices they want from the market or buyers are just being more picky. On average people sell every 7 years, so the vast majority of owners are still well in the money from their purchase price.

Of course owners are trying to unload SFH…. Prop 10 is on the ballot! Owners are not stupid. Being a rent control landlord is for suckers. Right now, the pre-1979 rental unit market is broken. When prop 10 passes, the market for the entire city will be broke.

Single family home inventory YoY number seems significant.. see how the market reacts..

Prices have gone up for so long, I’d be surprised if we didn’t see a flattening or decline. Of course, those proclaiming that the peak has passed have been wrong for about three years running, so taking the “over” on my bet may not be a bad idea. One (of many) examples is 45 Hancock #2, just sold for $1,800,000, $1500/sf and $200,000 more than 3 years ago. And MLS single family home prices just hit a new high in August, and places continue to sell extremely quickly. But even with a booming local economy, there is a limit.

Up 12.5 percent since the fourth quarter of 2015? But the median sale price in San Francisco is up nearly 20 percent since then!

And with many more examples of recent apples-to-apples sales which were much lower, if not negative in terms of returns, where are the dozens, if not hundreds, of 30-percent-plus gains since the fourth quarter of 2015 which would be necessary for the average gain to simply match the median price gain in terms of actual values?

Right, as you point out, this place on Hancock underperformed the market by a little bit. Median price is up. Case shiller is up. $/sf is up. Some places will underperform, and some will outperform. A big mistake, and very misleading, to just focus on those at the edges.

I’d be surprised if those buying today see the decent gains we’ve seen the last three years.

A much bigger mistake is to continue to conflate changes in the median price with changes in values, changes in a Bay Area index (Case Shiller) with changes in San Francisco proper when the area’s segments aren’t moving in lockstep, and to ignore a growing body of apples-to-apples results.

But once again, with many more examples of recent apples-to-apples sales which were much lower, if not negative in terms of returns, where are the dozens, if not hundreds, of 30-percent-plus gains since the fourth quarter of 2015 which would be necessary for the average gain to simply match the median price gain in terms of actual values? And yes, we’re giving you free rein to cherry pick your points.

I think you need to do some basic research on statistics. Try articulating why it would require pointing to dozens, if not hundreds, of 30-percent-plus gains since the fourth quarter of 2015 to be necessary for the average gain to simply match the median price gain in terms of actual values. Go ahead, and I think that in the process perhaps you’ll see your error in statistics and logic.

Our apologies, we assumed you understood how averages worked. If there’s a “below average” point in a set, there needs to be an “above average” point as well. And with a growing set of (well) “below average” actual, apples-to-apples, results, there would need to be dozens (if not hundreds) of (smaller) “above average” results to yield an average change in actual value to simply match the 20 percent change in median sale price since the fourth quarter of 2015.

Got it – the huge flaw in your logic is that the average and median price data are not ascertained by reference only to “apples to apples” results. Indeed, a very low percentage of sales in any period are recent resales. So you don’t need to show “dozens or hundreds” of over-performing “apples to apples” to demonstrate anything. Indeed, you’re mixing up apples and oranges, in a sense. You could have 1000 sales in Year A with a certain median and average, and 1000 sales of completely different homes in Year B, with a certain median and average. And the results are reliable. “Apples to apples” resales have little to nothing to do with broader measures of medians and averages. Monthly data comparisons are not great because the sample sizes aren’t bug enough. But annual data or moving averages are reliable and show significantly rising prices since 2015 by any measure.

It’s true that “apples to apples” sales are illuminating, but not if you only mention a small, non-random and skewed sample of those. There are far more “apples to apples” showing gains than losses. You’re also blowing hot and cold as you argue case shiller data are all wrong because the low end has been outperforming the high end concentrated in SF. But then you argue SF medians are all wrong because the high end has been outperforming and changing “the mix.” You’re getting crossed up by your own leaps of logic (and with no evidence at all). It’s perfectly fine to have an agenda, but don’t pretend that you don’t have one with faulty logic and transparent sophistry based on a misunderstanding of simple statistics.

Interestingly enough, the change in Median Sale Price we’re quoting is a quarter-to-quarter average from 2015 to 2018, not simply month-to-month. Unlike some (ahem), we pick our potential “apples-to-apples” sale pairs prior to knowing their outcome. And unfortunately, simply typing “there are far more “apples to apples” showing gains than losses,” over and over and over again doesn’t make it so.

But you are correct, “apples to apples” resales in San Francisco “have little to nothing to do with broader measures of medians and averages” as our growing aggregate of apples-to-apples transactions tracks the actual change in San Francisco home values versus simply measuring what’s selling (median price) or the Bay Area as a whole (Case Shiller).

So forget the hundreds, or even dozens, of apples-to-apples sale pairs you’d need to average out the growing body of “below average” outcomes in our cart, how about a few “above average” examples to simply average out the “below average” point you picked to match the median price gain? As there are “far more examples,” it should be a piece of cake!

You’re demanding dozens or hundreds of “apples to apples” resales since Q4 2015? Go read my previous post, then maybe spend some time with Khan Academy on statistics. But here you go. 506 Cole Street. Sold for $1,375,000 in May of 2016. Just sold again for $1,750,000. Up 27.3 % since Q2 2016. Usually you delete posts of resales with nice gains, but maybe you’ll let this one remain.

It is really all about the 2018 buyers paying less. The fact that a property is an Apple just makes the value change obvious. With non-Apples, you can factor out the value of renovations and price changes from years back and get an estimate of recent price movements. But this is not an exact science and the element of subjectivity lets people bend these estimates up or down. With an Apple it is completely clear. The exact same good now selling for less than it did before.

doc java, “2018 buyers paying less” than whom? I agree “apples to apples” are useful. The big error is when someone tries to take a sample size of 1 (or 10 or 20) and claim that reflects “the market.” That is invalid, whether it is pointing to one example of a 27.3% gain in two years or a 10% loss in two years. That is why the whole world uses broader measures, like means and medians, or case shilller (or the S&P 500 in the stock realm). None is perfect, but cherry-picking is always useless.

We certainly agree that “cherry-picking is always useless,” which is precisely why we don’t. But a funny thing happens when the collection of so-called “exceptions” continues to grow, and grow, and grow – it represents the market overall while catching the leading indicators and segmentation along the way.

And yes, broader measures are great! Except when one confuses what they’re actually measuring and the underlying trends at hand.

““2018 buyers paying less” than whom?” Less than the previous buyers.

People take population samples all the time in statistics. Why do you think Apples are not a representative sample? Nothing stops you from looking at non-Apples as well. But you need to have some model to account for renovations, legal changes, past appreciation and the like. IMHO you are going to have an order of magnitude more error trying to estimate the 2017-2018 value change from a property with a 10 year hold and multiple renovations versus a 2017 to 2018 Apple.

The modeling error on the 10 year hold will overshadow any sampling error for the Apple. And the editor has a good point about the mathematics of averages. If you claim that averages prices are up 20% from 2015 and the editor has a large list of Apples with prices at or below 0% from 2015, mathematically there would have to be an equal weight of resales far above 20%.

West Portal has seen huge gains in average sale price in the last 3 years as buyers priced out of Noe and Cole Valley markets buy there instead. There have been a couple $3M+ sales recently and more close to that level.

Snarky response: read this.

Sincere response: San Francisco real estate is like the weather. You must account for microclimates. There are very, very few true apples-apples in any of our micro real estate markets.

So transaction count is down and inventory is up, suggesting people who don’t have to sell are choosing to hold on while people who have to sell are not getting the prices they want from the market or buyers are just being more picky. On average people sell every 7 years, so the vast majority of owners are still well in the money from their purchase price.

Of course owners are trying to unload SFH…. Prop 10 is on the ballot! Owners are not stupid. Being a rent control landlord is for suckers. Right now, the pre-1979 rental unit market is broken. When prop 10 passes, the market for the entire city will be broke.