Jonathan Miller and Questionable Inventory Statistics Part II

in the past, Miller has re-adjusted his inventory numbers so that might explain it. In Manhattan, it is rather difficult to accurately track inventory because there is no formal MLS.

i hope that since you like statistics so much you like to delve into them? have you paid your $10 to become an insider? if so you'll notice in recorded sales how many of the sales do not have a listing corresponding to them. sometimes this is merely an error, and when you pull up the building the recorded sale matches with a listing, but many, many times it does not. These units that were never listed but which traded hands (often new development, divorces, estates, sales between family members, etc.) are nonetheless inventory and must be added to the inventory numbers after closing. If you have followed the practices of new developments over the past couple of years you shouldn't be surprised in the slightest by a fairly robust readjustment of inventory numbers after sales are recorded.

AR
Huh? Once sales are recorded the units aren't in inventory. They're sold. Would be very surprised if the Miller Samuel numbers are adjusted as you describe.

For example, an estate sale done with family friends (never listed) closes today, June 20. For what period is that apartment part of inventory. Certainly not Q2 2009 because, again, it's been sold. For Q1 then? Every quarter since the owner died? How would Miller Samuel know this? Read the Times' obits?

Similar argument for new developments. Sales office sells unit X without ever having formally listed it. Sale closes today. For what historical period(s) is that unit added to inventory? This is different from Miller's estimates of new development shadow inventory that are not in his quarter-end surveys.

sidelinesitter, they were inventory. shadow inventory. you can't have a sale without it once being inventory, even if it wasn't listed. they have to be accounted for somehow, i don't know how it's done, but it has to be somehow if the accounting balance sheet is to even out.

"i don't know how it's done," - evidently, since you are now disclaiming your own description of how it is done

"but it has to be somehow if the accounting balance sheet is to even out." - no, in order for it to balance the difference between the inflow and outflow of inventory during a period just has to equal the change in inventory over the period. Example:
- Inventory at the end of period 1 is 10
- In period 2, 3 of those units sell but 5 new units are added
- Balance at the end of period is 12. The change in inventory (2) is equal to the difference between inflow (5) and outflow (3)
- if another unit becomes available during period 2 and sells, inflow is 6 and outflow is 4 but the balance sheet is still correct (up 2 units) at the end of the period. I suppose that the sixth unit was theoretically inventory for some brief period from when it was offered for sale until it sold (for example, from when the heirs sat down with the family friend to talk about selling mom's place until they finished their cup of coffee and shook hands on price), but no one tracks this. It's under the radar. No way does Miller make up numbers to adjust observable inventory for some hypothetical under the radar volume. He does estimate shadow inventory and occasionally comments on what he thinks it is, but, again, it's not in his quarterly inventory numbers.

From the condo section of the Miller Samuel Q1 report: "Inventory results do not include "shadow inventory" which are completed or nearly completed units that are ready to enter the market, but have not been formally offered for sale." He goes on to say that shadow inventory is very high, but the definition of inventory for the data series is unambiguous.

sidelinesitter, the inventory numbers don't include shadow inventory that HAS NOT been sold. they don't know what it is as it is not available.

and don't be snotty. i'm not miller samuel. obviously they have been retroactively increasing the inventory numbers. i'm offering an explanation why.

how do you account for units that never became a part of listing inventory? Hypothetically, SE shows that 100 units were available and went to contract, all closed, but 120 units are listed as recorded sales? were they never inventory?

"and don't be snotty." Your original, pedentic (and inaccurate - a great combination) post on this thread wasn't snotty? It's your snotty tone talking down (while talking nonsense) to StatsChecker that set me off in the first place.

For your hypothetical, see my previous post. Your 20 units are the 6th unit in my example.

did you read the last thread started by StatsChecker?

it's pedantic, btw.

Didn't read, or don't recall anyway. This one is a bit pedantic too (e.g., paragraph beginning "There are only...") but I think his questions are interesting and I'd like to know the answers for future reference, so I thought I'd give him a pass on the pedantry.

perhaps i came off a bit too strong, but i think the inventory issue is a thorny one. i'd love to hear an explanation for it, but i think in Miller Samuel's defense it might be difficult to formulate. maybe i'm wrong. i do think that JM has been doing his best to straddle livelihood and ethics, with ethics not being on the losing side.

I was rather disappointed when Mr.Miller came on here a few weeks back and made the claim that shadow inventory is greater than the sun of re-sales plus listed sponsor inventory together, and when I pointed out that "that is some claim" and asked him about it, he promptly disappeared (if he ended up answering it and I missed it , I apologize in advance).

But I also agree that a system calling a divorce sale where one party sells to the other as "inventory" con not possibly be giving an accurate count of what any "normal" concept of inventory should be. Same with estate sales going to kids of the decedent, etc.

To be fair to Miller, (as I mentioned above) Manahttan does not have a formal MLS like 99% of the country, so keeping track of inventory is quite difficult. Now, if Manhattan actually had an MLS and Miller got his inventory numbers wrong, then there would be no excuse for that and he would likely lose all credibility in the blink of an eye.

In this article from the NYT, Jonathan Miller is quoted as saying, "..numbers for the second quarter were not yet available. But, he added, “I can tell you that they’ve slipped a bit.” Can those of you who actually work in real estate explain what that means for the overall market? Isn't the 2nd quarter traditionally the busiest?

http://www.nytimes.com/2009/06/21/realestate/21cov.html?_r=1&ref=realestate

30_yrs. Of course that's inventory. Not a huge percentage of it, but it's a part of how the system works. Those numbers show up in the sales averages, no, and they always do? When the numbers come out showing the average sales price increase, the median, etc., those skewed numbers are included. It would be lovely to have a sophisticated system that could tell you what percentage that is, as it is illuminating not only for the price issue, but as a broader demographic as well.

LuchiasDream - I don't work in real estate, but suspect this is what Jonathan Miller was hinting at:

http://www.streeteasy.com/nyc/talk/discussion/11999-fed-beige-book-june-2009-nyc-real-estate

Thank you angler7 this was definitely insightful.

"30_yrs. Of course that's inventory. Not a huge percentage of it, but it's a part of how the system works. Those numbers show up in the sales averages, no, and they always do? When the numbers come out showing the average sales price increase, the median, etc., those skewed numbers are included. It would be lovely to have a sophisticated system that could tell you what percentage that is, as it is illuminating not only for the price issue, but as a broader demographic as well."

I think we're in "agree to disagree" mode here. And it's one of the reasons I ALWAYS yell how worthless "averages" are. But I disagree that they have to be included. There are all sorts of statistical methodologies to remove them, if one feels like using them. I think one of the problems is that most people using statistics don't really know enough about statistics. I may have told this story before, but my favorite stats prof at university 9actually, my favorite prof of all time who just happened to be my grad level stats prof) gave what i consider to be the best stats final ever: he gave us a set of data and a hypothesis, and the final was to both prove and disprove the hypothesis using the data and methodologies we'd learned.

they don't HAVE to be included, but they ARE included in the numbers we're given. and since they have been for years, you'd skew your analysis if you wanted to compare trends unless you went back and parsed out the categories over time.

not saying that i disagree that more meaningful analysis could be done with the numbers.

I long since discounted the overall inventory numbers, just like I have median price numbers. Pick your neighborhood and your price range and download a daily xls file from SE. Because the number is highly filtered it's small enough that you can give the file a sanity check. You can then manipulate the numbers and create your own stats specifically designed for you. You may be surprised with what you learn.

I shall take the allegation of being pedantic as a compliment. In both threads I wrote, I have only raised one core set of questions - which, by the way, have not been addressed by JM in spite of the fact that he commented on the first posting.

Questions: what is included in JM's inventory number; what is the statistical accuracy of it (if not known,that's fine, just say it's unknown); what is "back testing"; why back testing not disclosed in the "methodology" section of the website or in the report; why on earth does JM not footnote the inventory figures: in the current report, it should say that it could be adjusted following back testing and in the future reports - where the previously printed numbers have been adjusted upwards they should be footnoted to indicate that they have been adjusted.

Lastly, if the figures are that innacurate and require "back testing" why do they publish quarterly "stats" so quickly??? "Stats" that are so innacurate are not in fact "stats" - they are WAGs based on some base level of information. Why "back test" at all?? Why doesn"t JM publish his report ONLY WHEN HE IS CONFIDENT THAT THE FIGURES ARE FINAL.

""Stats" that are so innacurate are not in fact "stats" - they are WAGs based on some base level of information. Why "back test" at all?? Why doesn"t JM publish his report ONLY WHEN HE IS CONFIDENT THAT THE FIGURES ARE FINAL"

Probably because the audience prefers a WAG on some base level of data sooner rather than a more refined number later, especially if later is months later. I know that I would prefer this. It's much like GDP numbers that come out as a "flash" estimate about a month after the quarter end and are then updated about a month later and two months later (called preliminary and final, I believe). The market values the flash product even if it is explicitly incomplete.

This doesn't take anything away from StatsCehcker's concern about stealth updating of the numbers. I think the best approach would be to go with the WAG (and admit that it's a WAG) right after quarter end for quarter X and then footnote the updated quarter X data when it appears in the reports for quarters X+1 and later. Also agree that a description of the back testing methodology would be in order.

Mr Miller has produced a report which has the good fortune of being broadly quoted in the media. When the quarterly Manhattan Real Estate nunbers come out in a couple of weeks you'll note the number of references that are made to his report. I expect by virtue of being early to the table in terms of this type of reporting, his report has achieved some kind of "Gold Standard". Indeed the Fed beige book appears to make reference to his data when they issue their reports

Inventory numbers are extremely important in determining the state of any real estate market. They are an excellent leading indicatorr. When the NAR existing home sales report comes out each month, the data point I pay most attention to is in the number of mos of inventory on the market. Following that number over the past few years would have enabled a fairly unsophisticated investor to identify the peak of the RE market shortly after it occurred.

If JM wishes to publish inventory numbers using year over year trends I would suggest that the numbers should be challenged if they appear to be as inconsistent as StatsChecker has pointed out. I will be scrutinizing Mr Miller's next report with interest and be more then interested to see a detailed explainination about the variances rom JM

If i understand the root of the stats, current inventory numbers do not include shadow inventory, prior inventory numbers do include shadow inventory that have since traded, then clearly there will always be a bias towards higher past numbers. Statistics are fine, as long as you understand where they are coming from. So the question is, is there any reason to believe shadow inventory is significantly lower today than 12 months ago? That should tell you whether the real inventory number is down closer to 13% or 50%...I know where I stand on the issue

Correction: It should tell you whether real inventory numbers are UP 13% or 50%

"in the past, Miller has re-adjusted his inventory numbers so that might explain it"

Never. Go back and check. Not until the advent of REAL inventory numbers (StreetEasy) that make his numbers look like poo-poo did Jonathan Miller start "adjusting" the inventory figures.

And since he's tacitly admitted that he was wrong in the past (tacitly b/c no mention was ever made of it - it took sleuthing) all of his sales figures are now doubtful.

His role as a shill for the real-estate industry is now made manifest.

"prior inventory numbers do include shadow inventory that have since traded". dcorreale - pls provide basis for this statement. As StatsChecker points out, Miller Samuel does not make its methodology public, so is there another source for the definition?

I have no inside knowledge sidelinesitter, My basis is this thread. My understanding is Miller revised up prior inventory numbers based on actual sales of places that were not originally in his inventory numbers, which I guess is a combination of preconstruction, new buildings that have not released all of their units, or merely sales that Miller did not know were up for sale to begin with. How else do you explain him recalculating his prior inventory numbers up...

This leads me to believe his current inventory numbers are understated, and 6 months from now he will be recalculating them up as well based on actual sales of places not originally in his numbers.

dcorreale - It seems to me that re-setting the inventory figures at later dates to include sold shadow inventory makes no sense - and if that's what JM is doing, I'd be alarmed. My understanding from reading press about "shadow inventory" is that it's only a guess as to how many units comprise it. If you restate inventory to include ONLY the units that sold and you DONT INCLUDE the unsold shadow inventory, then the inventory figure is not an apples to apples number. If JM is willing to ballpark shadow inventory at the time the quarterly report is issued, that would make more sense - but because it's so unreliable, he would have to break it out.

Having said all of this, JM has only very recently re-stated inventory figures after the fact. I cannot determine that restating inventory has been a long standing practice. JM has not responded to the questions raised - which I think is in order.

I am not a conspiracy theorist - there must be a logical explanation - but JM is no longer the "independent appraiser" he once was. JM works for Douglas Elliman so far as I'm concerned.

"I cannot determine that restating inventory has been a long standing practice." It has not.

See below for the listing inventory shown in the Elliman reports from Q206 to present (Source: Elliman web site). To check for (in)consistency with prior published numbers, 1) start at prior Q (e.g., 9,081 in the Q109 report) and look one row down and one column to the left to see how the number was previously reported and 2) start at prior yr Q (e.g., 7,778) and look four rows down and two columns to the left.

The change happened in the Q408 report, when prior Q of 8,811 was reported 26% higher than the number in the Q3 report and the prior yr Q of 6,518 was reported 27% higher than the number in the Q407 report. In Q109, the prior Q comparison works properly again while the prior yr Q is again different. I think that he made a one time change in his methodology starting with Q408 and that we should see prior yr Q differ from the previously reported figure for the Q2 and Q3 reports this year; that is, until the impact of the change washes through a full year of numbers just as it has already washed through the sequential quarter numbers in Q109.

Current Q Prior Q Prior Yr Q
Q109 10,445 9,081 7,778
Q408 9,081 8,811 6,518
Q308 7,003 6,869 5,204
Q208 6,869 6,194 5,237
Q108 6,194 5,133 5,923
Q407 5,133 5,204 5,934
Q307 5,204 5,237 7,623
Q207 5,237 5,923 7,640
Q107 5,923 5,934 6,904
Q406 5,934 7,623 5,964
Q306 7,623 7,640 5,764
Q206 7,640 6,904 4,965

I won't spam the site with all the numbers, but in the Q408 adjustments about 80% of the impact is in condos and the balance in coops. In the Q109 adjustment (i.e., Q108 figures as reported at the time vs. as now reported as prior year quarter) the difference is split almost evenly.

Remember, the guy does get a ton of income from the brokerages....

A few other thoughts:

"My understanding is Miller revised up prior inventory numbers based on actual sales of places that were not originally in his inventory numbers" - dcorreale, please reread the thread. This is a kooky theory that was disclaimed by its author in her next post. No evidence has been offered to suggest that Miller actually did this.

"I am not a conspiracy theorist " - I'm not usually one either, but let me take a shot.

Mild version: Miller wakes up one day in late 2008 and realizes that his inventory tracking has got away from him with the cascade of new supply on the market. For whatever reason, his methodology missed (at least in part), the shift and he has to recalibrate his system to the new market dynamics. Nothing wrong with that, except that instead of 'fessing up he just changes the historical numbers in his reports without commenting on or correcting them and does no more to publicize the error than mumbling something about backtesting in a couple of interviews that get reported in the press.

Less mild version: First, let us recall where the inventory data come from.

"Listing Inventory In Manhattan, listing inventory has been collected by surveying the exclusive listings of top Manhattan real estate brokerage firms. The list of firms was developed based on our experience and Crain's New York Business Magazine's list of Largest Real Estate Brokerages which ranks them by annual direct property sales. In addition, data is collected through the the REBNY RLS system." (Source: http://www.millersamuel.com/reports/methodology.php).

In this version, the market has been going to pot and Miller has, to put it charitably, not been going out of his way to keep up with the escalation of inventory becuase he's tight with the brokerage firms (and needs access to them for data) and the truth would hurt their business. Eventually it becomes clear that he's so far behind reality that continuing to put his thumb on the scale is going to damage his carefully crafted reputation as a market guru while not even helping the brokerages anymore because the numbers are so manifestly out of line with the facts on the street. At this point, the implementation of the fix proceeds much as above.

This being NY real estate, there is probably an Oliver Stone version also, but my mind is not twisted enough to figure out what it is.

^ becuase = because

its pathetic that people have to correct their own spelling mistakes in subsequent posts. Pathetic because people are so shrill on online that an argument can be nullified, in some posters minds, because of a typo or grammatical error.

"How can we believe you when you say the Earth is round, not flat, when you misspell "spherical"!!!! Learn how to write English, you troll!!!!!!!!!!!!!!'

Hot off the Miller Samuel presses: JM complains about "black box" models in a Deutsche Bank RE report!!!!!!!

"This study seems to be another “black box” model where we don’t fully understand the methodology even though it was disclosed in the prior report."

If it were not so pathetic, it would be funny. Link to his blog on the DB report below:

http://matrix.millersamuel.com/?p=4797

The inventory charts on the Miller Samuel web site show the recast historical data series. They are in bar chart and line chart format, so we can't see the exact figures, but we can get very close by referring to the scale on the y axis of the charts. This first link shows monthly coop/condo listings since 2002.
http://www.millersamuel.com/charts/gallery-view.php?ViewNode=1168399259Jmtnq&Record=12
Comparing the quarter-end levels in this chart against the reported numbers from the Miller Samuel/Elliman reports since 2006 (see table a few posts above), there is a large gap at every quarter end from June 2006 until September 2008. Just eyeballing the chart, it looks as if the gap (i.e., the additional inventory in the new version of the data) ranges from about 1,000 to 1,500 units compared to what was originally reported. The chart and the Elliman report agree for the December 2008 and March 2009 quarters, reflecting Miller's adoption of the new series from last December forward.

We can see monthly detail for the same period in this chart (although the totals are a couple of hundred or so higher here because they include townhouse listings in addition to coops and condos).
http://www.millersamuel.com/charts/gallery-view.php?ViewNode=1212523305ebhwU&Record=13

Thanks for outing Miller Samuel
It seems like their numbers show a consistent 25% under-reporting bias
The recent numbers on SE are particularly interesting in this regard since inventory is declining despite the fact that New listings >> contracts signed.
However, even many of the new listings in the categories I monitor are properties that were on sale, were taken off and re-appear. Sometimes SE shows them as new prices but often as recently listed
So the question of what is shadow inventory comes up
I appreciate that many people do not NEED to sell, and play with the market. However, if they bought recently and are paying 6% in interest in the hope of flipping, they stand the genuine risk of being flipped.
Hence, either foreclosures or reductions have to emerge as the phenomenan of interest at some point.

jason, as to your spelling comment...could not agree more!

Sideline, this underreporting, is it a function of the way data is released? Would Miller Samuel have anything to gain by missreporting? Can we assume that the current data will change by the same delta?

joed - "It seems like their numbers show a consistent 25% under-reporting bias"
To split hairs slightly, I think that Miller Samuel's numbers SHOWED an under-reporting bias, rather than "show". Based on the numbers they are now using in their charts, the numbers that they previously reported were way low - big under-reporting bias, as you say. Their Dec 2008 and March 2009 numbers are pretty close to StreetEasy/Urbandigs numbers, so unless their methodology goes wandering off into the wilderness again, the bias (now corrected) should not be a problem going foward.

On the point about inventory declining despite new listings being greater than contracts signed, listings are taken off the market every day too, reducing reported inventory (although in many cases not shadow inventory, which is a whole different topic). If you have a day with 70 new listings and 45 contracts, inventory will still go down if more than 25 listings are taken off.

falco
"this underreporting, is it a function of the way data is released"
"Can we assume that the current data will change by the same delta?"
See exchange with beatyerputz and NWT near the bottom of the link below. One time major change in historical data series and going forward reporting methdology. Current data reflects new methodology so should not change in the same way.
http://www.streeteasy.com/nyc/talk/discussion/12468-inventory-dropping-to-2006-levels-says-millers-numbers

"Would Miller Samuel have anything to gain by missreporting?"
See conspiracy theories (somewhat tongue in cheek) further up this thread. On the other hand, he has credibility and stature in the market to lose by misreporting (at least if he gets busted doing it). The best thing for credibility would be for him to lay out the before and after methodologies so that people don't think he was corrupt and/or incompetent to under-report by as much as he did for as long as he did.

NEW FIGURES FOR YOU TO ATTACK!!!!

http://curbed.com/archives/2009/06/26/three_cents_worth_inventory_as_twohump_camel.php#more

However you slice it, shiznet is WACK, yo.

On cured, miller is all "@Squeaky Pete: Not sure where you got that explanation - amazingly strange theory. You might want to go back and re-read the StreetEast thread."

Did Miller EVER answer my question about his statement that there's shadow inventory greater than the sum of new offerings plus resales?