What’s With the December Bump in Closed Sales?

As promised, I’d like to take a little time to go into the odd December bump in closed sales according to the numbers released this week by the NWMLS. Here are the three points I’ll be covering:

  • There definitely was a bump in sales in December.
  • The NWMLS numbers overstate the size of the bump.
  • Much of the real bump came at January’s expense.

We already know that both the warranty deeds filed with King County and the NWMLS data show a seasonally-uncharacteristic increase in sales between November and December. There is also a third data point that confirms a move in that direction: Redfin sale records.

If you do a search of all SFH sale records in King County for the last three months, then split it up small enough to download the results to a spreadsheet, you will find that according to Redfin (which pulls data from the NWMLS as well as public records), there were 1,120 sales in November and 1,346 in December. That represents an increase of about 20% month-over-month. That’s still quite high for this time of year, but does not stretch the boundaries of believability the way a 33% increase does.

So why would the NWMLS data be overstating December’s volume by nearly 10%? My theory is that the 110 or so extra sales in the NWMLS data are probably just another instance of the problem we discussed at length a few months ago with the way they collect and report their data.

The short version is that the number of “closed sales” the NWMLS puts in their monthly reports is not really the number of sales that actually closed during that month. Instead, it is the number of closed sales that NWMLS member agents entered into the system during that month, regardless of what month the sale actually closed. Note the carefully selected wording in their press releases: “MLS members reported 4,430 closings…”

The NWMLS defended this misleading practice in yesterday’s Seattle Times article by stating that “it has reported sales the same way for two decades, and its statistics are accurate.” As if somehow doing the same wrong thing for two decades makes it not wrong anymore.

Here’s why I think that much of the December bump came at January’s expense:

King County SFH Pending & Closed Sales

Note that after dropping off a cliff following the expiration of the tax credit (down 40% in a single month), pending sales were basically flat for six months, then dropped nearly 25% in the last two months. Closed sales had a slower descent (down 38% from June to September), but instead of flattening out, have see-sawed a bit in October, November, and December.

While far fewer pending sales eventually end up closing than used to (~75% today vs. 95% a few years ago), every sale that the NWMLS reports as closed has to at some point have been counted as a pending sale. A big boost in December closed sales when there was no simliar boost in pendings during the months prior indicates that sales are just shifting around between months.

My bet is that closed sales drop back down in January, to around 1,100 or so (a 25% MOM drop). Since sales lost 35% MOM between December and January last year, we’ll likely see a double-digit YOY percent gain, but we won’t really know how sales are shaping up for the 2011 “spring buying season” until we see the March numbers.

Full disclosure: The Tim is employed by Redfin.

0.00 avg. rating (0% score) - 0 votes

About The Tim

Tim Ellis is the founder of Seattle Bubble. His background in engineering and computer / internet technology, a fondness of data-based analysis of problems, and an addiction to spreadsheets all influence his perspective on the Seattle-area real estate market. Tim also hosts the weekly improv comedy sci-fi podcast Dispatches from the Multiverse.


  1. 1

    Using Regression Analysis, Tim

    The “eye ball” kind, your closed sales YOY chart is approximately “flat” sales for 2010 [approx 1250 units/mo].

    Certainly no reason to pop out the champaign if you’ve loaded yourself up with RE debt.

    I’d add too, when YOY sales are approximately flat from a severe 2009 recession year; where’s the bottoming of the Recession? Looks like a flat table at the bottom of a recession sinkhole to me.

  2. 2
    David S says:

    RE: softwarengineer @ 1 – Within my search parameters on Redfin I noticed that there are a few Short Sales that closed remarkably fast and on 12/30/10. When I say fast I mean sold in two months, pending around the first of November. That is just my little micro search area but if that projected to the region, it has the effect of a definite bump in sales do to rapid processing of short sales before 2010’s end. And, what ever else may be pulled in from January as is suggested.

  3. 3
    Pegasus says:

    RE: David S @ 2 – I saw the same on some of the short sales I track closing during the holidays. With the foreclosure pipeline in a temporary slowdown perhaps the banks decided to take the bird in the hand(short sales that normally take eons if ever to close) knowing that when they restart flooding the market with foreclosures they will devastate existing real estate prices…….

  4. 4
    ARDELL says:

    RE: David S @ 2

    It’s more likely the sales were pushed forward from November than “pulled” from January, given the 4th quarter numbers as a whole are not out of proportion, and November was as out of whack on the low side as December was on the high side. Also January is most always lower than December, so lower in January won’t prove much, unless you first calculate the standard deviation.

    Short sales December 2010 were lower than December 2009 and the % of short sales in December was not higher than the % in November at between 9% and 10% and closer to 9%. There was more increase in closings of bank owned transactions than short sales.

    Single Family King County – no mobile homes or townhomes, etc…

    Looking at the full 4th quarter volume, 2010 was higher than 2008 by 24% and lower than 2009 by 20%. Seems about right.

    The volume for the quarter is down 20% YOY. More likely you had more 45 day closings written in October that closed in December vs November, as loan processing times during the holiday period from Thanksgiving to Christmas tend to be longer.

    I don’t think it’s a huge deal, and looking at full 4th quarter makes the most sense. That puts us 20% down YOY and 24% up compared to the year prior. Sounds about right.

  5. 5
    David S says:

    RE: ARDELL @ 4 – Well good then. Thanks.

  6. 6
    ARDELL says:

    RE: David S @ 5

    I think you focus on some “south” markets. If you look at Kent, Burien, Federal Way and Auburn, the percentage of short sales was higher at roughly 12% vs 9% and the total number of SS did increase from November to December, by about the same proportion as total closings increased.

    So you are correct, you did see and there were more short sales closings. That “more” was the same % of total sales. Both increased to the same degree.

  7. 7
    Julie Lyda says:

    Here are some interesting statistics for you for Kirkland.

    October Sales 45
    November Sale 45
    December Sale 72
    (REO’s 10 – Shortsales 13 – Market sales 49)

    I researched each sale looking for anomalies such as long pending dates and double checked the closing dates that were recorded at the County and found only 1 that had actually closed in November. There were no anomalies to report.

    Another interesting fact:

    $1Million+ Sales
    October 3
    November 4
    December 12

    Of those 10 closed in the last week of December.
    All sales were reported as of the closing date NOT the status change date made by the agent.

    So there you go – no funny business. Looks like December was a great month for sales in Kirkland. My guess is it’s more a tax related issue.

    I’m still gathering numbers on other areas and will let you know if I find any anomalies.

    *Sales not compiled nor published by the NWMLS

  8. 8
    David S says:

    RE: ARDELL @ 6 – You are right but 98038 and 98027 specifically.

  9. 9
    ARDELL says:

    RE: David S @ 8

    Those two zip codes combined, 5 of 49 were short sales in December. Another 10 were Bank Owned. Last December there were 5 of 60 and 11 bank owned. November 2010 – 5 of 37 were short sales and 4 were bank owned.

    Required Disclosure: Statistics not compiled or published by The Northwest Multiple Listing Service.

    Note to Julie, the rule regarding the disclosure changed (in July of 2010 I think) to must say “The Northwest Multiple Listing Service” and not abbreviated to NWMLS. Rule 188 I think. Just a heads up

  10. 10
    deejayoh says:

    Instead, it is the number of closed sales that NWMLS member agents entered into the system during that month

    I wonder if it might reflect the number of closed sales that were entered at the time they pulled the reports – e.g. the system probably looks at the date of the transaction, not the date of entry. So whatever is in there on the 4th or 5th of the month when they publish their stats is what they go with. When you pull the same data from your Redfin system later, you get the benefit of seeing everything that was input with a transaction date in the actual month. The NWMLS never goes back to restate the numbers because – I am guessing – they probably don’t care so much about what was published. Internally they can still look at the trends and they aren’t primarily in the business of giving hyper accurate numbers to the press. Most people just don’t care about the decimal point accuracy as much as participants on this blog.

  11. 11
    ARDELL says:

    RE: deejayoh @ 10

    A “closing” is typically a phone call saying “we have recording numbers” at 4 to 5 p.m. on the day of closing. The most typical “inaccuracy” would be allowable, which is to post it on the first business day after it closes. That would put all last biz day of month closings into the next month. But to “correct” that in one month without correcting it in all months would still not create an “accurate” picture.

    I just checked on some closings from month end and the mls is more accurate than the county on some. The county is still showing the owner as the bank on some bank owned sales recorded as December closings. On those I would tend to believe the mls, as the agents would have gotten that call from the County, and also know if the new buyer has both recorded and been granted possession rights.

    There will never be 100% accuracy in any system at the time of a press release. That is why I post my numbers afterward and update them as new info comes in. Even then there are many inaccuracies unless you check every mls closing date with The County records, and as I noted, that doesn’t always work either.

    Maybe we can agree on an acceptable margin for error?

    The system likely relies on that margin for error being a constant on a % basis.

  12. 12
    ARDELL says:

    “When you pull the same data from your Redfin system later, you get the benefit of seeing everything that was input with a transaction date in the actual month.”

    That is not likely an accurate statement unless they are cross checking the County Record data base.

  13. 13
    Julie Lyda says:

    RE: ARDELL @ 9

    Thanks with the heads up – I did clarify that NWMLS is acceptable.

    Also – I should have noted that when pulling my statistics I do that same as you by excluding mobiles and townhouses.

  14. 14

    From main article: “The NWMLS numbers overstate the size of the bump.”

    I think you’re starting to get a bit over-obsessed with this issue of the reporting, and it’s showing in your analysis. Specifically, you won’t be able to state the above until we know how many sales will be late-reported for December. Given how early they reported (before agents even technically had to report sales per the 3 day rule), it’s very likely there were more than 110 (your number, but mine is close to that) December sales that will be late reported in January on.

    This late reporting was a big deal the month it was discovered because of the number of them due to the Matrix changeover. Since then, not so much, except that it did push down the median in November.

    Again though, I’m not crazy about the current system and wish that the prior month’s stats only included that month’s sales, with the late reported simply being added to the YTD. Under that system though, I don’t know what you’d do with December late reported, and others that went past a year end.

  15. 15

    RE: deejayoh @ 10 – When Tim first discovered this it became pretty clear they were including sales from the month of entry. Thus, December sales include all of the December sales reported, plus any earlier sales that were late reported after the prior cutoff.

    You’re right not too many people care, and to a great extent it doesn’t matter. If every month had 100 late reported sales the volume numbers wouldn’t be affected at all, assuming the distribution of the late reports was also evenly distributed. The mean and median would likely be affected though. But again that really only matters if you want to try to draw a straight line from the prior month and extend it out indefinitely.

  16. 16
    speaka says:

    RE: ARDELL @ 12
    [Deleted: read the rules before posting.]

  17. 17
    drshort says:

    I’m struggling to understand why this possible 10% overstatement of closed sales matters at all. Doesn’t this mean other months were understated? Your argument isn’t that there are fake closed sales. It’s only that they’re attributed to the wrong month.

    Where are all the posts titled “closed sales under reported in June!”

  18. 18
    drshort says:

    By ARDELL @ 12:

    “When you pull the same data from your Redfin system later, you get the benefit of seeing everything that was input with a transaction date in the actual month.”

    That is not likely an accurate statement unless they are cross checking the County Record data base.

    Redfin provides both the MLS and county sale information.

  19. 19
    ARDELL says:

    RE: drshort @ 18

    I spot checked several sales and many show a completely different date closed in the mls vs the County Records. You would have to pull the data as of the County Record date vs the mls feed. It’s possible to do, but I wouldn’t assume someone who has both data points is integrating them in that manner. The mls has access to both as well, as do agents, but that does not mean we can integrate them into one valid list. In fact at the moment we can’t, as there are December closings not posted in the County record yet that are showing in the mls, based on my brief tests of data today.

    I’ve tried running the numbers by County Record date vs mls, but it picks up banks buying back their properties at foreclosure, which is not really a valid “sale”. It also picks up nominal value transfers from husband to wife, etc…administrative name change “sales”.

    We do have a “general query” function to search the County database, but I don’t think any method is 100% accurate. It would be interesting to see how many homes are sold without agents at all, as they won’t appear in the mls data ever. If that % starts to climb, or has already started to climb, that would skewy the multi year tracking.

    I can think of one fairly foolproof method, but we would have to decide how many months and years we would want to go back and correct. Where do you draw the line? It would be a massively tedious project to go back and correct years worth of data. Someone would have to be committed to purifying the data pool.

  20. 20
    David Losh says:

    RE: drshort @ 18

    Does that mean redfin is providing better sales data than the Northwest Multiple Listing Service? Is that the point here? Otherwise this wrangling makes absolutely no sense.

  21. 21
    Ahau says:

    Tim, let me know if I’m on to something here–

    It looks like your 20% bump in Redfin is based on taking an early report for December and comparing it to a late report for November. Looking at Redfin’s spreadsheet from early December, they showed 1052 SFR sales in King County. Your more recent search showed 1150. The 98 sales that make up the difference were found and added by Redfin between about Dec 10 and today. This happens every month. October’s sale count went from 1207 to 1289 between the initial report and the next months’ report, and September’s went from 1109 to 1249.

    If you want to compare apples to apples, you either need to wait until early February to compare January to December, or you need to compare your early January numbers to the early December number. That was 1052. This shows a 28% MOM increase, which still doesn’t match the MLS’s 33%, but makes it a little bit more plausible, with the difference perhaps coming from November’s late reports and/or the percentage of sales that Redfin will find late this month.

    All that said, I agree that January won’t be holding at this level, as there aren’t enough pending sales in December to sustain it. The information about short sales is interesting–perhaps the banks were more willing to close short sales at a loss, once they realized that they were making huge profits for 2010, and were looking to reduce their tax liability.

Leave a Reply

Use your email address to sign up with Gravatar for a custom avatar.
Your email address will not be published.

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Please read the rules before posting a comment.