But you tell me
Over and over and over again, my friend
Ah, you don’t believe
We’re on the eve
of destruction.
Don’t you understand what I’m tryin’ to say
Can’t you feel the fears I’m feelin’ today?
If the button is pushed, there’s no runnin’ away
There’ll be no one to save, with the world in a grave
[Take a look around ya boy, it’s bound to scare ya boy]
Eve of Destruction — Barry McGuire
The following test came to me from a reader named Zileas. It is great stuff, and it contains some sobering conclusions for those hoping to sell in Ladera Ranch…
.
{adsense}
.
What I did was try to build a predictive model using the same statistical
techniques used by economists and scientists to glean insights from data.
I did this because most of the time you just see “median price” or “low tier
median price” or whatever, and this tells you very little, and with such a
shallow market (low # of sales), these medians are all over the place… so
I wanted to get the best estimate I could, so I used the tools I know how to
use — statistical regression. The quick summary is that price may be
correcting a lot faster than the medians are letting on.
I took the last 6 months of condo sales in Ladera Ranch off of the MLS
database. I only used 1, 2 or 3 bedrooms condos to try to allow some
breadth of data, but to mostly be comparing apples to apples.
Anyway, before I get into the nitty-gritty of the model, here are my
important findings, the ones that I’m very confident about:
– 1/2/3 bedrooms in Ladera Ranch, on average, are losing $334 in value PER
DAY (over the last 6 months, and there is weak evidence this is
accelerating, and no evidence it is decelerating). This represents a 2.6%
value decrease per month on a 400k home!!! This is a lot higher than a lot
of other estimates, but ties in to all the talk about the low-tier market
falling faster.
– Each additional square foot you add to a property in this band adds $134
to the price, holding all else equal.
– Bedrooms and Bathrooms each add about 25k in value to a house, all other
factors held constant.
– Overpricing your house causes you to lose $0.23 on the final sale price
for every dollar you list it over its eventual sale price (if it sells at
all). Note that this is an anomaly — in good markets, over-pricing often
causes you to get more. Note 2 (sellers): price to sell!
Details:
My approach was to start with what I thought the major factors would be:
– Square Feet
– # of Bedrooms
– # of Bathrooms
– Garage Capacity
– Time at which it was sold (just raw market trend, not “seasonality” —
since I was only looking at 6 months of data its hard to do stuff like say
that Christmas is slow and prove it)
– Square of time at which it was sold (there is some rejiggering in here to
make this work, but this is to crudely capture some
accelerating/decelerating market trends)
– Fixed effects for development (this is an adjustment for whatever tract
the home was in)
– (there were other factors I wanted too look at but which were not in the
data set — HOA dues, taxes, quality of property, the development it is in,
if it was a REO, etc, but I had to work with what I had)
I was somewhat hampered by the lack of data points — I only had 61 to work
with, and more would allow me to make a much better model.
After messing around for a while, I realized I could only get meaningful
results using the following variables, given my shortage of data(if only I
had a full dump of the past year of MLS sales in OC…):
Square Feet
# of Baths
# of Beds
# of days ago the sale closed (I did this analysis on Nov 2)
Results (for Ladera Ranch):
1) As every day goes by, the average price of a house drops by $334. No
joke. On a $400k house that represents price going down 2.6% per month.
This means if you were offering on a house there, and your comparables were
telling you “$400k right now”, you would at the very minimum add 6 weeks of
depreciation in the offer to calculate when the house would actually close.
2) Garage capacity does not seem to predict housing price. This is probably
due to my limited data and because “garages” on MLS are not very descriptive
— you can’t tell if its tandem, or wide, or whatever. It’s also because
garages strongly predict baths and vice versa (bizarre!)
3) For every bed you add to a house, the value goes up about $25,000… But
this value tweaks around depending on how you do the model — it’s not super
accurate, but adding beds, holding all else equal, almost always increases
values.
4) For every bath you add to a house, value goes up about $25,000. This has
a lot of the same issues as the bed count, and has some other issues because
beds and baths “predict” each other.
5) For every additional square foot you add to a property, given that you
have a set # of beds, baths, etc, the value of the property rises by $143.
6) Yes, houses have some intrinsic value in this model at very small sqft
and 1 bed/1 bath, but the lower you go, the less accurate the model becomes.
7) This model still has a fair amount of swing for properties, as you might
expect… obviously, I’m not capturing a lot of other relevant factors and
my model only accounts for about 3/4 of the pricing factors.
Interesting other random results
– For every $1 you list your property over market price, the actual amount
you get for it goes down by about $0.23 in this market. This is
fascinating because other studies I’ve read have shown the opposite, that
every $1 you list over market price, you get $0.50 more. Note that I was
able to prove this result with high confidence, but it wasn’t in my model
because it had a minor interaction with the beds/baths variables and due to
my lack of data, I couldn’t separate it out enough.
– I can definitely show that different complexes sell for more or less, but
didn’t have enough data to make it work out with enough confidence for me to
include it in the model.
Technical notes for mathematicians:
This is an OLS regression. I whitewashed the results, though prior to that
heteroskedasticity wasn’t proven, with my Breusch-Pagan / Cook-Weisberg test
p value at 27%. Boxcox showed a theta of 1.8 reinforcing my choice of a
linear-linear model, which didn’t surprise me much since sqft to price
should be a somewhat linear relationship and sqft was the largest predictor.
Multicolinearity was weak – my VIFs were not exceeding 2, and I don’t see
any reason why my chosen variables should have high multi-colinearity. I
did have to filter out some weak predictors because I Felt they were likely
to have too much multi-colinearity.
Linear regression Number of obs = 61
F( 4, 56) = 79.56
Prob > F =0.0000
R-squared =0.7402
Root MSE =23976
——————————
———————————————-
—
| Robust
————– soldprice | Coef. Std. Err. t P>|t| [95% Conf. Interval]
————-+————————————————————-
beds | 25448.56 6264.813 4.06 0.000 12898.63 7998.49
baths | 26872.12 13132.92 2.05 0.045 563.7275 53180.52
daysago | 334.447 73.30785 4.56 0.000 187.5937 481.3003
sqft | 143.1072 23.48726 6.09 0.000 96.05653 190.1578
_cons | 105957 22440.9 4.72 0.000 61002.43 150911.5
—————————————————————————
.
{adsense}
.
One thing that jumped out at me was this statement:
Each additional square foot you add to a property in this band adds $134
to the price, holding all else equal.
To me this strongly implies that Ladera Ranch will bottom near $134/SF. If the incremental value is $134/SF, can the fundamental value be far behind?
What do you think?
“heteroskedasticity” and “Multicolinearity”…..whaaaaa ? I bet I’m not the only one this fine Sunday morning with my head spinning after reading this post. I’m sure our math braniacs will be all over it though. Dare I deem it Tanta-worthy ?
So if houses are losing $334.00 a day, that means by waiting a year I’ll be saving $121,190. I LIKE that data ! Thanks for your efforts.
—–
I skipped all the mathematcial disclosure. It might be interesting for someone to question the statiscal probability of your calculations, but it won’t change what is & what is yet to come.
The $134/ sq foot is very interesting. (of course, I am in Nashville & have no clue where in CA this place is, but I’m pretty sure it’s still in the “OC” so the price point seems remarkable considering.)
When the rollbacks you review are @ $300+, the $134 seems like you still have a lot more rollbacks to come. To get it near this point, where do you see this going? Would you guess @ pre-2000 levels? (even in Nashville, we’re rolling back to 2005 levels, without the 20%+ run-up.)
I have a completely different dataset, and used some rather different methods. However, some of my results are similar.
1. The 2.6% decline per month is reasonable in view of the data I have as well, and the raw rate of dropoff is increasing. However, there is a little bit of seasonality historically, and the rate is slightly less if seasonally adjusted.
2. I can give some informed speculation on why overpriced homes are fetching less than homes/condos which were priced at market when initially listed. In the academic research (perhaps the analyst has a copy of a paper on Boston condos in the 1990s), the price decline was slower. With slower declines people who listed above market often did get more. Those properties may have had some real difference not easily seen in the statistical analysis, like the best view, or exceptionally nice interior decor.
However, about half of the markets in the US have prices falling at the fastest rates since Case-Shiller started collecting data in 1987. In that case, selling earlier becomes more important. If the analyst tries a different approach, I think the components will become clear. Actual sales price/original list price is now mostly a function of number of days on market until sale. Overpricing gives you more days on market. Translation for anyone with a house/condo listed now: you will get a higher price if you sell 3% below market now than if you manage to sell at 3% above whatever the market price is in 6 months.
Don’t forget that if that house is sitting vacant, the negative cashflow of ownership costs will be another 3-4% of the price of the house.
3. The incremental value of $134/sf doesn’t seem that surprising to me. That is actually quite close to the cost of construction for that additional square foot for condos. Well, at least it was a few months ago, before there was so much spare construction capacity.
A different $/sf number would turn up for single family homes, probably much larger. The problem with detached single familiy homes is that the square footage of the structure correlates with the lot size. The price runup in the bubble raised land prices. The cost of constructing on that land didn’t go up nearly as fast, and is now falling for woodframe construction.
UHh-h-h-h, Huh?
There is a small condo listed in the newspaper for a cheap, but not slummy part of the space coast for $69,000. I don’t think it’s a type. The picture shows something inexpensive but not falling down. You
can’t get any lower than that. I don’t think so, anyway.
Great Analysis. Tons of 2005 and 2004 rollbacks in Ladera. Even more interesting is the number of new homes bought by realtors in 2004,05 and 06, all using negam option loans with 10% or less down. The vacancy rates are disturbing. Have the homes in the Covenant Hills gated section were flips with the negam financing.
Real shame, because its a really well designed community with great people. – also, a better value per sq foot than Irvine if you like SouthOC.
Meanwhile, I sold a year ago and renting a mcmansion at dirt cheap prices (relative). My landlord has bought four large homes in Ladera, all with Option Arm Financing in 2005 and 2006. (average sales price $1.2m to $1.4m), 5 or 10% down, then refied the 2nd mortgage to get the 5% ro 10% on the next home. They blanantly lied on the builder and mortage application (not that they were questionned), and then rent the homes at a price equal to their minimum option arm payment, plus taxes and mello roos. So, a new $1.2-$1.3m 4000 sq home (plus landscapping etc . . ), rents around $5,000, but their real carrying cost (with accumulating unpaid interest) is like $9,000 to $10,000. Even worse for them, they can’t refi or sell them without a shortsale.
These speculators drove the prices up, and actually wound up hurting transplants who moved into Ladera, and didn’t know better when they also bougth these homes with real 30 year or P&I mortgages.
There are so many cases like this, I don’t know where to start.
I am a real estate attorney and have followed these transactions, and short of a 1.0% feds fund rate again, I think we will see a 2002 rollback price by next year.
Local realtors are ‘talking up’ the spring 2008, after the superbowl line, just as they did last year for 07. Summer of 08 will be really ugly.
My only question is, will banks enforce the resets on the Option-Arms when the principal balance on these loans exceeds 110-115% (causing full P&I payments due)?
These Option-Arms are the subprime 2/28s of the middle-upper class
In my experience with construction prior to Y2K, the cost of construction for a SFH is around $200 per sq foot with standard materials,
If you spec upgrades: 30 AMP Romex, more electrical circuits, multiple HVAC zone, upgraded plumbing, etc…. the price quickly moves up.
Oddly, the cost of the stuff that people think as “upgrades”, base molding, crown molding, wood floors are not so expensive.
Of course, kitchens and bathrooms are a different story. A nice bathroom with travertine and good quality (not anything from Home Depot) will easily run $25K and up. And kitchens…. yes… kitchens…..
OTOH, a well design kitchen with quality stuff will help sell a house.
The way I see it, the bottom for my area is around 300/350 including the land. I believe at one point it had peaked over 500. Currently one of my neighbors has listed her rebuilt 55b/3.5ba home for just over 400. There a 4b/3ba dump down the street that also wants over 400 even though it badly needs a new roof.
The dump was bought two years ago. My neighbor who built the house has owned it since 98 and has tons of equity. She’s in no hurry anyhow.
That should be 5b/3.5ba.
A 55b/3.5ba would be a hotel with a horrible bathroom problem or a college dorm with large shared bathrooms.
I wonder what the cost per square foot is for a dorm?
“My only question is, will banks enforce the resets on the Option-Arms when the principal balance on these loans exceeds 110-115% (causing full P&I payments due)?”
The answer is YES, YES, YES.
Yes. The rampant speculation on new homes with cheap money was one of the prime reasons why prices skyrocketed.
Another was that the sellers did not check whether the buyer intended to live there or just rent it out. The contracts I saw stipulated that the home would be owned by at least one year.
I guess the builders didn’t really want to care because they made out like bandits as the price skyrocketed. Their costs went up somewhat as availability of construction trades tightened due to increased construction but their cost of land did not.
The interesting thing is that much of the damage was done by a collusion of real estate “professionals”, banks, lenders and builders. A number of home owners were caught in the middle.
And now we’re supposed to help the lenders and speculators? WTF?
We have Big problems with the r.e. market in Ladera Ranch. Get access to the MLS and look, its a mess.
“And now we’re supposed to help the lenders and speculators? WTF?”
I’m with you Tonye…if they didn’t do their Due Diligence, screw ’em.
“and short of a 1.0% feds fund rate again, I think we will see a 2002 rollback price by next year.”
Just a guess of course, but I tend to think that a Fed funds rate of 1% will drive mortgage and other interest rates higher as the folks who own our debt demand a larger return on a devalued currency. Actually, they are already dong that, aren’t they. It seems LIBOR has decoupled from the ten year and the Chinese are divesting out of USD.
It’s easy to say this, and I’m with you on the sentiment. But your financial health doesn’t opererate in isolation. Just as speculators (and primary residence purchasers) affected you by irrationally inflating prices, they are affecting you as prices deflate.
Speculators are going to take their financial lumps through serious credit hits and monetary losses. While this occurs, I’m more interested in an orderly decline than I am in speculators suffering pain. I suffered as they drove up prices and I don’t want to continue to suffer through a serious recession.
The FHA needs to finalize their reforms and Fannie/Freddie need to have their conforming loan limits increased in “high cost” states from $417k. I think it’s in everyone’s best interest that “knife catchers” are encouraged into the market as it returns to affordability.
If you want to be truly terrified look at Eff’s recent post under Headlines.
I would agree with Mark, except I think it’s too big to be fixed. I can live with saving some deadbeats, if I’m saved too. Along with other good guys.
The Fed could start a hyperinflation, I suppose.
We could default on the Chinese.
Both of those things may happen anyway, not on purpose.
Once way or another I think we will all be living much more cheaply than we previously did.
No, you’re not the only one. Even with the addition of coffee, I still didn’t get the details. Like any good exec, I just skipped down to the conclusion. 😉
Nice analysis. South OC is certainly a mess right now, Ladera in particular (Lake Forest is also exceptionally bad). A common argument is that since there wasn’t as much building in OC in the past few years compared to other areas, we won’t see the magnitude of problems as, say, the IE. Goes to show you what can happen in a newer community like Ladera that mostly went up during the boom.
Try the RSMeans Quick Cost Estimator. Put in a typical condo complex of about 70 units of 1400 sq ft each in an Irvine zip code. $105-132 per square foot of construction cost. Doesn’t include land. http://www.rsmeans.com/calculator/index.asp
RSMeans QuickCost Estimator
Project Title: condos
Model: Apartment, 1-3 Story
Construction: Face Brick with Concrete Block Back-up / Wood Joists
Location: SANTA ANA, CA
Stories: 3
Costs are derived from a building model with basic components. Scope differences and market conditions can cause costs to vary significantly.
Story Height (l.f.): 10
Floor Area (s.f.): 100,000
Data Release: 2007
Wage Rate: Union
Basement: Not included
Cost Ranges Low Med High
Total: $7,232,850 $8,036,500 $10,045,625
Contractor’s Overhead & Profit: $1,808,213 $2,009,125 $2,511,406
Architectural Fees: $488,729 $543,033 $678,791
Total Building Cost: $9,529,792 $10,588,658 $13,235,822
Ass you should know, price has nothing to do with cost. Even $134 is high, based on national averages. But I think it is a good call.
Sorry for the typo. AS you might guess, I was not addressing a pack animal.
I am sorry, but this model is retarded.
1. You sample size is tiny
With RE you will have lots of outliers so your results will be driven by them
2. You have heterogeneity bias in your regressors
And I don’t know which way it would go, so there is no way to tell if your estimates are upper or lower bounds.
Great Analysis! I am curious if you did the same analysis with SFRs instead of condos if the numbers would look comparable. My initial thought would be that 1 bedroom condos would be falling at the fastest rate and that condos in general are falling faster than SFRs, but I’m wondering if the data would support that.
I would expect multicollinearity. Housing is bed/baths/space dependent.
Just look at the comments on the forums. Too few baths -bad. Too many beds in too small of a space – bad. As a society as we age, we’re seeing a progression towards one bath per bedroom. Shared baths for children’s rooms are still acceptable.
Hence, 3/2 is good, 3/3 makes a great rental in addition to good home and a 3/1 is a problem. Likewise, a well laid out 2/2 in 900-1000 sf is good. A 2/2 is 2000 sf adds some value but overall, appeals to a smaller market segment hence, the 2/2/1000 shows up in multi-living situations and the 2/2/2000 shows up in the high-rise/luxury segment where ‘lifestyle’, ‘view’ or other amenity is the real driver.
Given Ladera’s family orientation and community model, I would still expect the driver’s to be bed/baths/space.
To think IHB started off profiling underwater specuflippers, and today IR takes IHB to another level with his econometric models. And, here I thought I was the ubernerdy data junkie. But…
What!? No charts? You have to have charts.
I think it would be hard for them not to enforce the resets since they count the accrued interest as income on their books.
Agreed. I think a plot of the data points and the regression line would be an appropriate visual.
Agreed on the charts.
Was just over at Calculated Risk, and I think with all due respect to Tanta and CR, this blog is better in its blogginess.
The comments are just so much better, and better organized too in the way they are set up in this blog.
Also, Tanta and CR are just so far above us, and their commentors, that in comparison the comments are just mush. Who is going to, who possibly could successfully challenge Tanta? Not me, I’d be too scared to try.
Altho Calculated Risk is essential, the overall learning experience is much better in this blog.
I doubt if CR or Tanta would let someone else start a thread and just let it go. Tanta would go off an a rant about something said that was so beautiful in its irateness that nobody else would dare to say something that was strong. Not to say I don’t adore her rants, I just wouldn’t want to be the subject of one.
My hub summarized it as a self-correcting research journal written and corrected in real time. He found that impressive.
The personalities of the people contributing stand out. This blog is evolving because so many people are pushing it in various directions.
None of us are as smart as Tanta (I guess), but Calc Risk is really the product of only 2 personalities, and the people whose articles are chosen, after the fact, to grace the sidebars. No commentator would ever reach the sidebars on CR.
The blogmeisters here are smart enough to be leaders, but also to let the thing grow in directions that they didn’t have the faintest idea it would go when the blog was started.
I think what I mean is that collectively, we are smarter than Tanta and CR, no matter how un-libertarian that concept is.
I likey.
Question: shouldn’t any house have intrinsic value (ie, land)? In other words, the slope of the line, $135/sqft, should intersect at the land value. Given the volatility of land value, this would be a terrible method to try to determine it–you’d need a cross-sectional regression at one point in time–but my point is that I wouldn’t count on $135/sqft to be the floor value of a property. The floor value, as you’ve correctly pointed out in previous analysis posts, should be the rental yield.
$135/sqft is the market value of construction, home only (no land). So, at any point in time, if you know (or have expectations for) land value, you can add $135/sqft, add/deduct bedroom and bathroom factors, and you’re good to go. Thanks for the effort.
1. The data does not represent a sampling, it is based on all of the values of those condos within a certain time period. There shouldn’t be the problems with outliers as with traditional real estate data sets because there isn’t that upper skew (I’m guessing there probably wasn’t a 12 million dollar condo in Ladera, a planned community, right?)
2. This is a pretty homogeneous sample. Condos, Ladera, all built pretty recently. Attached housing clumps a lot more with regards to value characteristics than detached. Think about the difference between 3000sqft on a 6000sqft lot vs the same house on an acre. When you’re attached, the property profile is less diverse, especially in one planned community! All of these are low- and mid-rise too. There’s more Homo for you.
But thank you anyway for your critique of this “retarded” model. If you can enlighten us with non-retarded version, I’d love to see it.
I enjoyed this analysis a alot, IR.
IR, thank you for the hard work you put into this thought provoking model.
I would like to address some of the concerns raised by readers.
Xy31 commented that, “I am sorry, but this model is retarded… You sample size is tiny…With RE you will have lots of outliers so your results will be driven by them.”
First, I don’t think that retarded is a technical or polite term. We might ask instead, whether the model is BLUE (Best Linear Unbiased Estimate).
Let us deal with the small n objection. If the number of cases is small, when compared with the number of independent variables, we might have cause for concern. However, I don’t see any evidence that the n is causing problems in the equation. Further, the dataset includes more than 5 cases per independent variable, generally regarded as the hurdle one needs to cross. Therefore, I do not consider the tests of statistical significance to be suspect.
I don’t see anything to suggest the presence of outliers. (Outliers are data points which lie outside the general linear pattern of which the midline is the regression line). For example, do any of the houses in Ladera Ranch have17 bedrooms?
No_Such_Reality said, “I would expect multicollinearity. Housing is bed/baths/space dependent.”
The symptoms of a multicollinearity problem would be regression coefficients that change greatly in value when independent variables are dropped from the equation (or added for that matter), ridiculously large or small coefficients, flipping signs, and of course, a High R2. I don’t see evidence of these things in the results reported here, unless IR found his coefficients changing drastically or signs flipping. Further, IR regressed each of the independent variables on all others, through the use of the VIFs. That should put to rest the concern over multicollinearity.
former_irvine_resident , “Agreed. I think a plot of the data points and the regression line would be an appropriate visual.”
It would only be possible to show a regression line for bivariate relationships. Multiple regression is not two-dimensional.
Very nice number crunching, IrvineRenter.
For several years
Very nice number crunching, IrvineRenter.
For several years I have been using a simple formula/ratio to find prices of houses that are either larger or smaller than house of known value.
As with your method, this works when all else is equal. Similar neighborhoods, houses not extremely different in age, etc.
I usually use my house as a known-price. I feel I am relatively knowledgeable about the market value of my own house. Really.
I have found that as the size of the house increases, the land does not increase at the same rate. There is (as you mentioned) an intrinsic value in a small house.
As the size of a house increases, each additional square foot is only worth half as much as the smaller house. That is; a house that is twice as large, is worth 150% as much. In other words, a 100% increase in size equates to a 50% increase in value.
This holds true, very nicely, with up to a 50% change in size. It does pretty well with a 100% change in size.
Of course, a decrease in size is the inverse. Just remember to use the known house-value as the denominator of the fraction (bottom of the ratio) so you get numbers like 2/3 instead of 1.5.
Here’s an example of finding the value of a smaller house.
Let’s say you know your house is worth $500,000. and is 1500 square feet, and you are trying to find the value of a similar house that is 1200 square feet. It would go like this; 1200/1500 = 0.8 so this is a -20% change (0.8-1 = -0.2). The 1200 foot house would be 10% less than the 1500 square foot house (20% divided by 2), or 90% of $500,000 (or $450,000).
When finding the price of a condo, you need a known condo price for comparison. You need a known town home price for a town home, and a known house to find the price of a house.
This method works very well, but it also points out the fact that a noticeably smaller house will only be a little less money, and vice-versa.
When using the approach for houses on the same street, that were built at the same time (with different square footage), the numbers are dead-on.
I would love to take credit for this, but it was the reader named Zileas who did all the work. I am just the messenger.
Okay Prof, you said it much better than I did!
I am not nearly as sophisticated as Zileas, but I have been tracking the population of active MLS homes in my DC neighborhood for a few months. I include all homes in a narrow geography having more than 2 beds, and at least 1.5 baths, up to 600k in asking value.
As of today, I have watched 466 homes enter the listings since April this year. I began to compile statistics on them when I thought I had enough of a sample–200 or so in July. Over the past 120 days, active listing ask prices have fallen over $7.5mm. There have been, in that time frame, an average number of active listings of 170, for an average decrease of $44,344 each.
In the months of October and November, the number of active listings has increased to 196 (avg), and the ask decreases have accelerated to just short of $500/day (Oct $477, Nov $479). The avg ask per square foot is $364.
These are Capitol Hill numbers–in city, plentiful access to public transportation, and relatively crime free neighborhoods. Last spring I had realtors telling me how immune to decreases it would be. I’m still renting.
I think it is very difficult right now to determine a true value in Orange County. Someone might think that their house is worth $500,000 but can they really sell their house for that? Even if your neighbor, who’s house is similar just sold for $500,00, that doesn’t mean you are going to find another buyer for your home at $500,000. There aren’t very many buyers who could qualify and would want to buy a home for that amount. We are going to have this issue with determining value for a few years. I see it often in MLS descriptions “listed at $100k under appraisal just done 6 months ago.” Well abviously appraised value means nothing right now if the property has been on the market for 4 months at $100k less….and it’s still not selling! That should tell you something about what so called “value” means right now!
The $134 per square foot means.. if you take a house, hold EVERYTHING constant — meaning, # of beds/baths, day it was sold, etc, you add $134 for every square foot. As you increase square feet you eventaully add bedrooms. The “$ per square foot” we usualyl has also includes bedrooms, bathrooms etc,. this analysis breaks out the component parts and lets you know what contributes to what and how much… so $134 is not an underlying resistance on the price, its just how the valuation seems to work these days — when we say $300 per square foot, thats not counting 2 vs 3 bedrooms at the same square footage — im saying a 1000 sqft house that is 3 bedroom will sell for maybe 25k more than a 2 bedroom, thus giving that $134 number.. .sortof 🙂 I can explain in more detail if anyone is interested.
I use it mostly because it is a meaningful part of the prediction 🙂
– Zileas
I trimmed the data to only be 1/2/3 bedroom condos under 600k i believe to deal with a lot of these issues — its focused on a very narrow band of the market for these reasons.
The rental value of these places is well above $125 per SF. Probably at least $200 to $250 per foot.
Multicolinearity was strangely weak nonehtless. There was enough predictive power in those variables to jutsify all of them being included, look at the significance levels
I know we all hate VIFs, but they were all below 2 as well, which is way low IMO.
The statistical probability you can find with the listed confidence intervals on each variable at the bottom
For example, the coeff for $ decrease per day is $334 but the 95% confidence interval is liek 187 to 481. That means that there is a 95% chance that the true decreases falls between $187 and $481 per day, with $334 being the likeliest value, GIVEN the data I’ve used.
Yes, this model only predicts what sales are RIGHT NOW, it doesn’t predict what the bottom will be.
see my comment about $ per sqft above your comment. Houses are valued by a lot of things, and when we just say “$ per square feet” we rae ignoring all those other factors. My model doesn’t have a complete picture, but it has more pictures, and its basically showing that other factors besides $ per square foot impact price…
What the $ per squrae foot measure in the model gives is how much extra people will pay for a slightly larger living room or bedroom, but no additional house rooms/features beyond that. Fior example: How much more would you pay for your living room to be $200 sqft more? In this model, apparently its around 200 * $134 🙂
you’ll also notice at the bottom of my anaylsis there is a variable called “const” or something like that. That is the value that the house is worth at 0 sqft 0 bedrooms etc. It seems non-sensical, and it sortof is since you cant have a house that big, but small houses always haev a higher $ per square feet, and that’s a reflection of that.
Zileas,
Got it. That makes sense to me. I think some other posters jumped on the $134 per foot number as being the point that the market was heading towards. That is what I was commenting on as being unrealistic.
Who knows. I may be wrong. There may end up being so much pessimism that we overshoot rental value.
61 points is pretty good for that # of predictors, and if you look at the output, I was getting good enough p values for something like this. Maybe not good enough for say a drug test, but plenty good for economic analysis.
The Stata output log is there, read as you please. I tested some alternate models to verify robustness within the data set, and got largely similar numbers chopping off bits of data, and using different sets of predictors.
I avoided hetegeneity mostly because I’m using ladera ranch which is quite homogenous, and I trimmed the data to a narrow band of similarity with less other factors at play. I ran this same regression on Aliso Viejo and got an R^2 of like 50% instead of the 75% you are seeing here, and the AV data had wayyyy more points. Ladera is homogenous, ever been?
I would say this model is most appropriate for looking at 2 bed/2bath condos, and less accurate for 1s and 3s. But the confidence intervals of prediction I was getting on specific points int hat band were pretty good.
I didn’t, if you want to hook me up with the data I will. It wont be as accurate though, SFRs have more variety in their characteristics.
California leads the 50 states in almost everything.
I am from Wisconsin, so I am not inhibited by most other than taxes.
I suggest that the California market put their assets on-line in real-time to those not just in Califronia but global.
Since I do this in Wisconsin, can California do this too?
Selling in a market where reality is an issue, I suggest we do this in real-time.
Steven Nowak
I just used an ordinary least squares regression, then whitewashed it to get rid of any heteroskedasticity. You can see the regression output at the bottom, the P values are pretty good. I believe that OLS is also called BLUE as you describe it, the method i use just tries to minimize variance of predictions vs actual reality, no weighting on the various points or any stuff like that.
I tried adding in some more predictors. Some were significant, but a lot of those were not robust as i cut out swaths of data or added/removed variables (maybe multicolinearity, maybe other issues), so I only kept the ones that were producing fairly consistent types of coefficients and also had clear economic justification, trying to capture major economic factors in each variable. My comment about overpricing = lower sale price was from another model that looked really good, but got a p value of like 10% when I put it into my final model, so I didn’t keep it.
Was hard to catch more subtle valuation trends with the data set — lack of tandem vs wide garage, HOA dues, upgraded level of property, etc. Maybe with more points it would be possible.
the $134/sqft (vs the “apparent” low end sqft value of around 300-400) mirrors your comment perfectly. The “constant” value at the low end is a little over 100k of value for a “no space” house 🙂 so your intuitition matches my analysis.
I’m interested in getting better data to do more of these, and/or focus on specific questions of interest to people.
If you have direct MLS access, that would help me a ton. I can ask my Realtor for stuff relevant to properties I’m seriously considering, but I can’t really ask for wide data dumps of multiple zip codes 🙂
This may be useful for Realtors too if you want further evidence to convince your sellers to price reasonably.
I can be contacted by my name on the internet: Tom ( at ) Zileas [Dot} COM
sorry.. have to spam email disguise.
Speakig of Ladera Ranch, can anyone with MLS or other resources enlighten me on what happened at 19 Dennis Ln, Ladera Ranch, CA 92694 (4,578 sqft)? This property was posted about here at IHB on March 28,2007. Here is the lowdown:
–10/2005: $1.6M financed with $1.4M fom Bear Stearns Residential Mortgage.
–12/2005: HELOC of $250,000 with American First Credit Union.
–03/2006: HELOC of $500,000 with Mortgage Inc.
–07/2006: refi of $2 million from AllState Home Loans.
–08/2006: refi of $215,000 from AllState Home Loans.
–01/2007: listed on MLS for $2.9 million.
–02/2007: reduced to $2,498,900.
–02/2007: NOD filed 6 months after latest refi, another 1st payment mortgage default on $2.215M note.
This is the subject of an ongoing bet!
Oooops….it was actually posted at bubbletracking.blogspot.com. Same difference. Anyone know how it shook out.
Prices at the bottom?
My intuition says:
National average will be $85 per square foot,
Riverside will be $110 per SqFt,
O.C. will be $220 per SqFt.
San Diego? Who knows. I think it will have the biggest “undershoot” in bottom price.
That’s just another of the 800# gorillas in a roomful of them.
Ten-year yield opens at 4.08% this morning, and trendlines only point straight south. The last time we tested the water below 4.0%, a war was beginnning in Iraq.
Yields snapped back within a month, that time.