Market Data Conflicts
By Charles TurnerYesterday, commenter Squeezed supplied us with a link to this chart on housingwatch:
It shows a median price drop from $369,000 to $332,900 or a 10% drop this year. Heading out to the homepage of the site, I found the following disclaimer/methodology:
These are asking prices and not final sale prices of homes in various cities and their surrounding regions. Final sales prices can be higher or lower than asking prices depending on factors such as market conditions, or the needs of the buyers and sellers involved in the real estate transaction. Data is taken from MLS data sources. We are not responsible for any errors and/or erroneous data.
They are looking at listing prices, not sales prices. RMLS looks at sold prices. Two sources of “prices” with different results. RMLS data reports:
Average Sales Price for February, 2007 – $326,700
Average Sales Price for February, 2008 – $342,600
Median Sales Price for February, 2007 – $279,000
Median Sales Price for February, 2008 – $280,000
Average Sales Price Year to date for 2007 – $323,800
Average Sales Price Year to date for 2008 – $342,800
Adding another data source for the same market:Case Shiller index data for Portland (current reporting period is only through December):
March 2007 181.72
April 2007 183.55
May 2007 185.21
June 2007 185.76
July 2007 186.51
August 2007 186.00
September 2007 185.67
October 2007 185.10
November 2007 183.65
December 2007 182.47
How do we put it all together? It doesnt all make sense. Housingwatches’ median listing price was below RMLS’s median sales price in Feb 2007??? A year later the median listing price dropped 10% and there was a slight increase in median sales price? Any thoughts?
Just had a thought that needs to be taken into account: Days on Market (DOM). If Housingtracker reports a listing price on day X, RMLS reports the sale of that house on day X + DOM + escrow. The data needs to be shifted to be compared. Right?
Charles, DOM would be a good # to know if it was a true #. Houses get listed, expired and listed again. Each ‘new’ listing starts a new DOM.
Very true, Kevbo. It looks like we have a garbage in, garbage out problem. One property can be listed at any price, canceled, sold or relisted at any price.
The relisting of a million dollar house just a few times would change housingtracker data but not RMLS data. You could list it a dozen times but only sell it once. Days on Market would be skewed for RMLS but that is not part of the equation until you get to inventory.
I think we need to take all data with a grain of salt. The metrics for real estate are clearly imperfect. Looking at one might not give an accuate picture. Lookin at two or more may not give an accuate picture either but at least it highlights that the data doesn’t define a rule.
Thanks for posting the data and the interesting analysis. Are you certain that MLS asking price data include each re-listing? If so, asking price data probably overestimates price drops. I think its still a barometer of market sentiment.
Case-Shiller is time shifted to the left by several months (and also measure the closing price). I expect Case-Shiller to start showing significant decline in the summer.
The comments regarding the DOM issues realted to Case-Shiller vs. MLS listings is good work.
However, the issue of repeated relisting of a house at lower prices somehow skews the numbers is in my opinion an accurate measure of the market. (More accurate than “median sales” since that number can be skewed upwards by expensive homes $750+ which are arguable in a separate market than starter or median priced homes)
Looking back in time and ask how often houses get relisted a new lower prices. As I understand, this is considered a big no no in the market and most RE agents would counsel that the listing should be at the market. The fact that homes are being relisted at lower prices is the canary in the coal mine:10% down has already happened.
Once spring sales start coming in even further below asking only then will the second relisting downward push occur in late summer and along with that more HELOC’s get shut down.
In fact: found some comments that were discussing this housing watch Asking Price index as compared to Price Shiller in the Pheoniz Arizon market in summer 2006
“Remember that the median sales price numbers everyone are keying on are houses sold, when we all know stuff ain’t sellin’, and the number is bogus because it doesn’t take into account incentives and freebies. I like asking prices as a more accurate reflection of today’s conditions on the ground, and what’s to come …
Meanwhile another commenter was talking about Portland in 2006 (where the bubble was still running hot) and the housingwatch numbers had Portland median at 365K while Case Shiller had it only at $280K. I think that $365K is much closer to the right number for August 2006 in Portland (bubble peak)
Might be a good test to look back and see which index was more accurate during the bubble in 2005-2006… my guess asking prices (as they are the inside market)
This would almost be funny if it wasn’t the email I just received:
The 60% Loan To Value isn’t scary but the rest of it is. Oh so professional.
readyandable: yes I would tend to agree with that Phoenix post. Since I doubt very many houses right now are selling for over asking price (it may be happening in a few hot neighborhoods, but I suspect it’s well limited) I would tend to think that looking at asking prices makes some kind of sense here and certainly gives you a data point that you won’t get from CaseShiller for several months. I wish housingwatch would also track how many homes sold in the period (usually it’s a 3 or 4 day period for them and a one week period for housingtracker).
FWIW the housing watch numbers represent the entire bell curve of what’s on sale in PDX today – that’s a pretty good metric of what’s going on – although it can be skewed when condo’s flood the market.
The sales price numbers reflect an unknown mix of the small percentage of houses that sold this month. This is highly susceptible to swings due to the changes in the lending landscape amongst other things – so it may not be the best datapoint.
The case schiller is a backward looking datapoint that because of lag time in getting sales into the MLS and collating the info and publishing it probably lags what’s going on right now in the market.
From what I’ve seen/heard sales are going through at around 10% below list – although I’m sure Charles has a better view of what’s going on from his seat than I do from mine.
The truth is probably somewhere in between, 5% to 10% down is my wild guess.
The data they’re showing on Housing Watch doesn’t match the RMLS data that I’ve seen. Personally I have also seen a number of houses pulled off the market before the sold.
I”m still trying to figure out the spread between median listing and median sold prices, which runs anywhere from 15% to 30%.
Anecdotally, I’ve seen alot of pendings that didn’t. IMO, the mortgage mess is now starting to really impact PDX.