It seems strange that there are short price surges. Are they reflective of market conditions or is there some sort of index measuring error/bias? I don't recall there being movements like that in any of the NAR price graphs.
Hmmm - not a bad analysis, however I note that the current downturn is very very orderly, while the 90's downturn in LA was very haphazard. Thus, I dont think that one helps us much.
I do note however that it looks like an element of seasonality can be seen in the current CS graph. It seems strange that same house sales could have an element of seasonality to them, but could that be what we are seeing?
House sales and prices have seasonality. Sales and prices go up in the spring and down in the fall. I think it's mostly attributed to the school year, since people don't want to move in the middle of a school year. Weather plays a part too but even in Socal you have seasonality. The sales changes are large. Price changes are less, but still a couple of percent.
This just reflects an improvement for the spring selling season. We'll go back to the steep plummet in the fall.
We're batting inflation by not only fighting high prices, not only by murdering high prices, but by blowing the living shit out of high prices. Yessir, you heard me right.
F-E-, I disagree; when I ran the sales and sales price data for San Diego through MiniTab, I clearly saw seasonality for sales but no seasonality whatsoever for sales prices.
Also, using the 'eyeball' meter, I saw no seasonality in sales prices, just sales volume.
You can talk about cliff diving all you want, but in my limited experience serious or longer-term price declines in assets don't work that way.
There's a momentum of these things as they try to move toward economic equilibrium. They can be slowed part-way through by seasonal necessity-based buying and by bargain hunters who gamble that the bottom has already arrived.
But the momentum won't dissipate until the market has rolled downhill to equilibrium and has exhausted its forward momentum. The real bargain-hunting comes when momentum carries the market temporarily beyond equilibrium in the other direction.
exactly -- this is obvious, is it not? or am i missing something?
the relevant comparison to detect acceleration/deceleration in house price changes is to compare the trailing-3-month annualized rate of change to the same period from last year.
this may not be as responsive as people want, but it eliminates seasonality and gives a much more honest picture. and on this basis, house price declines are clearly still accelerating.
I would not read anything into the first graph. Month-to-month is a first derivative, so the slope of this graph is the second derivative, or acceleration (which is pushing the bounds of meaningful, but is still recognized as meaningful). The slope for Feb-July 08 is in the same direction as Jan-June 07. In other words, this looks like (unaccounted for) seasonal change to me.
not even that -- a predictable pattern of rising demand in the february-june period that puts upward pressure on prices pretty much every year.
i think a lot of people, including obviously CR and tanta, could have told you last year that the rate of change would improve from january to june.
the problem is that the rate is still intensifying downward from last year at this time -- implying that price declines in the coming autumn and winter months may be even steeper than they were for the same months in 2007.
IF, however, we end up this december seeing slower declines than we saw in december 2007 -- THAT would be news.
I noticed the peak month-to-month price declines occur pretty consistently around December-January, and that the best support for prices is in the May-July summer period.
I'm not much of a chart reader/analyst but it seems to be a pretty consistent pattern.
As any statistician knows, the noise in any series increases as the sample size decreases. Since we are now looking at much lower sales numbers than during the peak it is not surprising that these numbers are becoming noisier.
I have friends who are buying houses right now and I am really kind of afraid for them. Few people outside the Street and the blogosphere seem to know how bad this is.
Birds nest in the spring?
Not to say humans are just dumb animals, but this pattern is genetic.
Does Australia/New Zealand have a Case-Shiller equivalent? I'd be willing to bet big on a similar Sept/Oct/Nov. peak.
Check out the ratio of homes selling for more than asking, it has the same seasonal pattern; and the number of homes sold likewise oscillates.
Once the spring frenzy of mating rushes passes on, the price of nesting materials declines again.
The seasonality aspect still seems strange to me. Median prices do change with the seasons, but I always thought it was due to the "mix" of houses put in the market during the summer versus winter season.
I assumed Case Shiller would take care of that because we are dealing with same house sales. However, if there is true seasonality to same house sales, a that very same bungalow that would sell for 240K in december can sell for 260K in May. Again, that just doesnt seem right.
Seems that the smallest declines correspond to traditional peak season for sales?
Psy 101: perhaps sellers "bet" to get a better price then, and only compromise after they realize that if the house did not sell in peak season they have no choice but drop the price,
wondering what the money flow variation, total number of dollar sales (instead of just price) looks like, ie if there is "no spring season" this year this reduces the significance of the reduction in price decline this year as compared to last year.
Red writes: Birds nest in the spring?
Not to say humans are just dumb animals, but this pattern is genetic.
Genetic? WTF?
Seasonality in the residential real estate market is not really some biological mystery. In the areas that experience this thing called 'winter' and 'blizzards' and icy steps and whatnot, moving is a much more pleasant undertaking during the warmer months. Also, people prefer to move when the kids are out of school for the summer. There's nothing 'genetic' about it.
I plotted historical M-M change for a couple of cities and the two Composites, binned by month, and found a significant seasonality, peaking in June in all cases.
Over C-S published history for LA, Portland and the two Composites, the peak M-M change has been 12-14% (annual basis) in June and minimum typically -1% to +3% in mid winter.
Typically M-M change from April-May has been about 5% higner (more precisely, 500 bps higher) (annualized) than the full-year change.
I meant to convey that over C-S history the annualized M-M change over all months was typically ~6%, peaking @ ~12% in mid-summer and dropping near zero in mid-winter.
So "5% higner" than average is really double the average. That's why I tried to use bps to convey the difference.
Hi, Shnaps.
As a biologist, I don't think of instinctive behavior in response to seasons as "mysterious".
But if homes are cheaper in December, wouldn't rational humans in mild seasonal areas (like those in LA, San Diego, San Jose, etc) buy then? How about people with no kids?
Shouldn't buying correspond very tightly to finance conditions? Why would anyone buy if prices are going down; why not wait until they are at least stable?
No, I contend that this seasonality is driven by fundamental biological urges; it is stronger than the rational mind and overrides nearly all other possible inputs into the system.
I assumed Case Shiller would take care of that because we are dealing with same house sales. However, if there is true seasonality to same house sales, a that very same bungalow that would sell for 240K in december can sell for 260K in May. Again, that just doesnt seem right.
Not exactly. It's like saying a bungalow in would be worth 4.45% less in May rather than worth 9.9% less in May than December.
And, I still agree, there is seasonality that CR is ignoring. It shows up in the data 2002 to 2005 to the upside pretty clearly (look at the bumps in the chart), so it's logical that you'll see it on the downside. Nothing moves in a straight line.
I can't speak about other markets, But my anaylsis of the Denver market data (10 years worth) indicates that roughly 75% of the annual price increase occurs in the second quarter.
I don't think there would be nearly as much volitility from a 20 city index than there would be from just the LA index.
A find it odd that there is such a long lag time in the data (unless I am reading the chart wrong). The most recent data point represents mid May, and we are now at the end of July. The lag time makes the data a lot less useful.
If you knew nothing else about the national and LA price charts other than that the first is from a much larger and more diverse sample, you'd expect the first to be smoother. The volatility in the LA chart may serve as an example for the national experience, but it may not. Even if it does, we'd expect less volatility and more persistence in national price swings.
Red writes:
"No, I contend that this seasonality is driven by fundamental biological urges; it is stronger than the rational mind and overrides nearly all other possible inputs into the system."
how about : schools closing, vacation time easier to "schedule" at work in summer , moving cross country on icy roads with truck and car in tow (have done that, would avoid a repeat), carrying furniture under the rain or snow...
Prices "presumably" adjust to demand... or the perception of demand, may be the people who are selling in winter are those who HAVE to sell (or they would have chosen summer...)
leftback writes:
As any statistician knows, the noise in any series increases as the sample size decreases. Since we are now looking at much lower sales numbers than during the peak it is not surprising that these numbers are becoming noisier.
Actually, the numbers in CR's first chart (national month-to-month) don't look all that noisy. There's a pretty clear sine-wave-ish pattern going on, with peaks in the late spring/early summer and troughs in the late fall/early winter. In fact, the only data point which really bucks the trend is March 2006.
So, I'd say it's pretty clear that same home sale prices are seasonal, period. At least on a national level.
I do notice that the Los Angeles chart from the early 90's is considerably noisier, which makes sense, being a smaller sample size.
Fair enough, Red, but as a Finance and Economics d00d, I must warn you that I think of most biological stuff as mysterious.
Maybe you should run your theory by Steve Levitt, it might make his next book, but it's got volumes of competition coming from the more conventional wisdom that I alluded to: the seasonality is attributed to the rational, convenience-based factors that increase demand at rather predicable times of the year.
Granted, the mild climates may have lesser degrees of seasonality, and not everyone has kids to worry about, but still -- there are plenty of other cultural components we should consider before turning to photoperiodism and vitamin D levels to explain it all. For instance, first-time homebuyers are mostly young couples, right? When do most weddings take place? Google that one, would ya? When do most apartment leases expire? They coincide with the academic year, right? I'm thinkin' September. And don't forget abut the 'holiday season', nobody wants to move then - they're already too chock full of travel plans and so forth. Hence, more downward pressure on demand in Nov and Dec.
When you aggregate the effect of all these non-biological/'rational' factors, they collectively are the reason that homes are slightly cheaper in December and January as opposed to May and June.
Lastly, don't think that the mild-climate markets function in their own vacuum. Let's suppose we have a retired, empty-nester couple from Wauwatosa, WI who'd like to move out to Pasadena, CA to live out their golden years. In all likelihood, they'd plan to buy in the Springtime (not just beacuse Wisconsin garage sales are so lightly attended in January). Hence, demand in Pasadena feels some of that seasonality indirectly even though it's 80 degrees and an otherwise fine day for a flowery parade on January 1st.
In general prices are still too way too high, and will continue to fall for some time.
I agree. Historically, home price appreciation has been a 1-2 percentage points above inflation. If we go back to 2000 (as the Case Schiller 20 City index does) and increase home prices by 4% over a 10-year period, we would come out with a 48% rise (index at 148, with 2000 = 100). Currently, the 20 city index is at 168. For it to come back to normal, that would mean an 11.9% decrease over the next two years. Of course, that's all cities. Cities like Charlotte and Dallas will probably be flat to slight increases, but Los Angeles and Miami will probably have doubled that decrease. Decreases of that magnitude would also bring home prices more in line with the traditional relationships to rent and incomes. Look for decreases to continue to lessen, finally flattening out in a year to 18 months.
Chuck Ponzi - my point is this, assume case shiller showed there was no movement up or down in a market - prices were static. In that case if Case Shiller had an element of seasonality in it, it is almost as if the bungalow would sell for 240K Jan 2008, 260K May 2008, 240K Jan 2009, 260K May 2009, etc. etc. Again, that just doesnt seem right.
Median prices in places like boston, NY, DC, etc move higher in the spring, and lower in the fall/winter. I always assumed this price differential had more to do with the houses being offered at that time (i.e. different houses available in the spring, different buyers are out looking, etc.
If thats true, then I would assume case shiller would not show any seasonailty because it compared the same house over time.
Ah, well, we're off of the "spring selling season" so homes will vanish from the market and return the Shadow Inventory and prices will stagnate and slowly plod downward since fewer sales = less price discovery. And that is a good thing if you're a crook or a doomed "homemoaner."
All of this could be fixed if those derned Bloggers would get out there and buy an overpriced house that nobody can afford instead of spreading the truth over the intertubes.
...
It seems strange that there are short price surges. Are they reflective of market conditions or is there some sort of index measuring error/bias? I don't recall there being movements like that in any of the NAR price graphs.
Hmmm - not a bad analysis, however I note that the current downturn is very very orderly, while the 90's downturn in LA was very haphazard. Thus, I dont think that one helps us much.
I do note however that it looks like an element of seasonality can be seen in the current CS graph. It seems strange that same house sales could have an element of seasonality to them, but could that be what we are seeing?
But last year the price decline slowed during the same time period, through May. Then they accelerated until Feb.
Looks like too identical a pattern to believe it's a true equilibration.
Look at the earlier graph... peak selling season sees any gains that occurred.
"peak selling season sees any gains that occurred."
exactly, and this coming off season will have worse mortgage availability and rates.
House sales and prices have seasonality. Sales and prices go up in the spring and down in the fall. I think it's mostly attributed to the school year, since people don't want to move in the middle of a school year. Weather plays a part too but even in Socal you have seasonality. The sales changes are large. Price changes are less, but still a couple of percent.
This just reflects an improvement for the spring selling season. We'll go back to the steep plummet in the fall.
We're batting inflation by not only fighting high prices, not only by murdering high prices, but by blowing the living shit out of high prices. Yessir, you heard me right.
Very nice work, C-R-.
F-E-, I disagree; when I ran the sales and sales price data for San Diego through MiniTab, I clearly saw seasonality for sales but no seasonality whatsoever for sales prices.
Also, using the 'eyeball' meter, I saw no seasonality in sales prices, just sales volume.
You can talk about cliff diving all you want, but in my limited experience serious or longer-term price declines in assets don't work that way.
There's a momentum of these things as they try to move toward economic equilibrium. They can be slowed part-way through by seasonal necessity-based buying and by bargain hunters who gamble that the bottom has already arrived.
But the momentum won't dissipate until the market has rolled downhill to equilibrium and has exhausted its forward momentum. The real bargain-hunting comes when momentum carries the market temporarily beyond equilibrium in the other direction.
House sales and prices have seasonality.
exactly -- this is obvious, is it not? or am i missing something?
the relevant comparison to detect acceleration/deceleration in house price changes is to compare the trailing-3-month annualized rate of change to the same period from last year.
this may not be as responsive as people want, but it eliminates seasonality and gives a much more honest picture. and on this basis, house price declines are clearly still accelerating.
Bob Dobbs is dead on. No market goes straight down, either. There are always up legs between the drops.
so, the consensus is--white noise?
Wow that LA 90's chart is scary. I just threw up a little in my mouth.
I would not read anything into the first graph. Month-to-month is a first derivative, so the slope of this graph is the second derivative, or acceleration (which is pushing the bounds of meaningful, but is still recognized as meaningful). The slope for Feb-July 08 is in the same direction as Jan-June 07. In other words, this looks like (unaccounted for) seasonal change to me.
Get back to me in December.
so, the consensus is--white noise?
not even that -- a predictable pattern of rising demand in the february-june period that puts upward pressure on prices pretty much every year.
i think a lot of people, including obviously CR and tanta, could have told you last year that the rate of change would improve from january to june.
the problem is that the rate is still intensifying downward from last year at this time -- implying that price declines in the coming autumn and winter months may be even steeper than they were for the same months in 2007.
IF, however, we end up this december seeing slower declines than we saw in december 2007 -- THAT would be news.
I noticed the peak month-to-month price declines occur pretty consistently around December-January, and that the best support for prices is in the May-July summer period.
I'm not much of a chart reader/analyst but it seems to be a pretty consistent pattern.
The high season is now over. The rest of the year is going to be ugly.
Blogger: Blog not found
As any statistician knows, the noise in any series increases as the sample size decreases. Since we are now looking at much lower sales numbers than during the peak it is not surprising that these numbers are becoming noisier.
I have friends who are buying houses right now and I am really kind of afraid for them. Few people outside the Street and the blogosphere seem to know how bad this is.
Birds nest in the spring?
Not to say humans are just dumb animals, but this pattern is genetic.
Does Australia/New Zealand have a Case-Shiller equivalent? I'd be willing to bet big on a similar Sept/Oct/Nov. peak.
Check out the ratio of homes selling for more than asking, it has the same seasonal pattern; and the number of homes sold likewise oscillates.
Once the spring frenzy of mating rushes passes on, the price of nesting materials declines again.
The seasonality aspect still seems strange to me. Median prices do change with the seasons, but I always thought it was due to the "mix" of houses put in the market during the summer versus winter season.
I assumed Case Shiller would take care of that because we are dealing with same house sales. However, if there is true seasonality to same house sales, a that very same bungalow that would sell for 240K in december can sell for 260K in May. Again, that just doesnt seem right.
Thanks Shnaps. Marshall Lucky was a fantastic break.
Seems that the smallest declines correspond to traditional peak season for sales?
Psy 101: perhaps sellers "bet" to get a better price then, and only compromise after they realize that if the house did not sell in peak season they have no choice but drop the price,
wondering what the money flow variation, total number of dollar sales (instead of just price) looks like, ie if there is "no spring season" this year this reduces the significance of the reduction in price decline this year as compared to last year.
Red writes: Birds nest in the spring?
Not to say humans are just dumb animals, but this pattern is genetic.
Genetic? WTF?
Seasonality in the residential real estate market is not really some biological mystery. In the areas that experience this thing called 'winter' and 'blizzards' and icy steps and whatnot, moving is a much more pleasant undertaking during the warmer months. Also, people prefer to move when the kids are out of school for the summer. There's nothing 'genetic' about it.
Hi CR --
Is there a Case Schiller national index that excludes California and Florida?
It would be interesting to see what the rest of the country looks like without the high-flyer hard-crash states.
where everyone go? only 25 comments so far?
I plotted historical M-M change for a couple of cities and the two Composites, binned by month, and found a significant seasonality, peaking in June in all cases.
Over C-S published history for LA, Portland and the two Composites, the peak M-M change has been 12-14% (annual basis) in June and minimum typically -1% to +3% in mid winter.
Typically M-M change from April-May has been about 5% higner (more precisely, 500 bps higher) (annualized) than the full-year change.
picosec - are you plotting prices or volume?
picosec - bps == basis points?
In the context of housing prices, what is a "basis point"?
Price using C-S.
Could send Excel sheet but don't know how to post it.
Yes bps = basis points.
I meant to convey that over C-S history the annualized M-M change over all months was typically ~6%, peaking @ ~12% in mid-summer and dropping near zero in mid-winter.
So "5% higner" than average is really double the average. That's why I tried to use bps to convey the difference.
Hi, Shnaps.
As a biologist, I don't think of instinctive behavior in response to seasons as "mysterious".
But if homes are cheaper in December, wouldn't rational humans in mild seasonal areas (like those in LA, San Diego, San Jose, etc) buy then? How about people with no kids?
Shouldn't buying correspond very tightly to finance conditions? Why would anyone buy if prices are going down; why not wait until they are at least stable?
No, I contend that this seasonality is driven by fundamental biological urges; it is stronger than the rational mind and overrides nearly all other possible inputs into the system.
Burbed wrote:
I assumed Case Shiller would take care of that because we are dealing with same house sales. However, if there is true seasonality to same house sales, a that very same bungalow that would sell for 240K in december can sell for 260K in May. Again, that just doesnt seem right.
Not exactly. It's like saying a bungalow in would be worth 4.45% less in May rather than worth 9.9% less in May than December.
And, I still agree, there is seasonality that CR is ignoring. It shows up in the data 2002 to 2005 to the upside pretty clearly (look at the bumps in the chart), so it's logical that you'll see it on the downside. Nothing moves in a straight line.
Chuck Ponzi
I can't speak about other markets, But my anaylsis of the Denver market data (10 years worth) indicates that roughly 75% of the annual price increase occurs in the second quarter.
I don't think there would be nearly as much volitility from a 20 city index than there would be from just the LA index.
A find it odd that there is such a long lag time in the data (unless I am reading the chart wrong). The most recent data point represents mid May, and we are now at the end of July. The lag time makes the data a lot less useful.
CR,
If you knew nothing else about the national and LA price charts other than that the first is from a much larger and more diverse sample, you'd expect the first to be smoother. The volatility in the LA chart may serve as an example for the national experience, but it may not. Even if it does, we'd expect less volatility and more persistence in national price swings.
Red writes:
"No, I contend that this seasonality is driven by fundamental biological urges; it is stronger than the rational mind and overrides nearly all other possible inputs into the system."
how about : schools closing, vacation time easier to "schedule" at work in summer , moving cross country on icy roads with truck and car in tow (have done that, would avoid a repeat), carrying furniture under the rain or snow...
Prices "presumably" adjust to demand... or the perception of demand, may be the people who are selling in winter are those who HAVE to sell (or they would have chosen summer...)
leftback writes:
As any statistician knows, the noise in any series increases as the sample size decreases. Since we are now looking at much lower sales numbers than during the peak it is not surprising that these numbers are becoming noisier.
Actually, the numbers in CR's first chart (national month-to-month) don't look all that noisy. There's a pretty clear sine-wave-ish pattern going on, with peaks in the late spring/early summer and troughs in the late fall/early winter. In fact, the only data point which really bucks the trend is March 2006.
So, I'd say it's pretty clear that same home sale prices are seasonal, period. At least on a national level.
I do notice that the Los Angeles chart from the early 90's is considerably noisier, which makes sense, being a smaller sample size.
Fair enough, Red, but as a Finance and Economics d00d, I must warn you that I think of most biological stuff as mysterious.
Maybe you should run your theory by Steve Levitt, it might make his next book, but it's got volumes of competition coming from the more conventional wisdom that I alluded to: the seasonality is attributed to the rational, convenience-based factors that increase demand at rather predicable times of the year.
Granted, the mild climates may have lesser degrees of seasonality, and not everyone has kids to worry about, but still -- there are plenty of other cultural components we should consider before turning to photoperiodism and vitamin D levels to explain it all. For instance, first-time homebuyers are mostly young couples, right? When do most weddings take place? Google that one, would ya? When do most apartment leases expire? They coincide with the academic year, right? I'm thinkin' September. And don't forget abut the 'holiday season', nobody wants to move then - they're already too chock full of travel plans and so forth. Hence, more downward pressure on demand in Nov and Dec.
When you aggregate the effect of all these non-biological/'rational' factors, they collectively are the reason that homes are slightly cheaper in December and January as opposed to May and June.
Lastly, don't think that the mild-climate markets function in their own vacuum. Let's suppose we have a retired, empty-nester couple from Wauwatosa, WI who'd like to move out to Pasadena, CA to live out their golden years. In all likelihood, they'd plan to buy in the Springtime (not just beacuse Wisconsin garage sales are so lightly attended in January). Hence, demand in Pasadena feels some of that seasonality indirectly even though it's 80 degrees and an otherwise fine day for a flowery parade on January 1st.
In general prices are still too way too high, and will continue to fall for some time.
I agree. Historically, home price appreciation has been a 1-2 percentage points above inflation. If we go back to 2000 (as the Case Schiller 20 City index does) and increase home prices by 4% over a 10-year period, we would come out with a 48% rise (index at 148, with 2000 = 100). Currently, the 20 city index is at 168. For it to come back to normal, that would mean an 11.9% decrease over the next two years. Of course, that's all cities. Cities like Charlotte and Dallas will probably be flat to slight increases, but Los Angeles and Miami will probably have doubled that decrease. Decreases of that magnitude would also bring home prices more in line with the traditional relationships to rent and incomes. Look for decreases to continue to lessen, finally flattening out in a year to 18 months.
Chuck Ponzi - my point is this, assume case shiller showed there was no movement up or down in a market - prices were static. In that case if Case Shiller had an element of seasonality in it, it is almost as if the bungalow would sell for 240K Jan 2008, 260K May 2008, 240K Jan 2009, 260K May 2009, etc. etc. Again, that just doesnt seem right.
Median prices in places like boston, NY, DC, etc move higher in the spring, and lower in the fall/winter. I always assumed this price differential had more to do with the houses being offered at that time (i.e. different houses available in the spring, different buyers are out looking, etc.
If thats true, then I would assume case shiller would not show any seasonailty because it compared the same house over time.
"homes are cheaper in December"
no mystery here, people are distracted with christmas shopping.
Ah, well, we're off of the "spring selling season" so homes will vanish from the market and return the Shadow Inventory and prices will stagnate and slowly plod downward since fewer sales = less price discovery. And that is a good thing if you're a crook or a doomed "homemoaner."
All of this could be fixed if those derned Bloggers would get out there and buy an overpriced house that nobody can afford instead of spreading the truth over the intertubes.