more counting wars / Manhattan coop + condo inventory
(In which, Manhattan Loft Guy gets into the weeds — again — with numbers and takes StreetEasy to the woodshed.)
Let’s beat that old horse of Different Firms, Different Numbers again … as a follow-up to my February 23, inventory numbers: different firms, different issues, and as a lead-in to a report on my conversation with StreetEasy’s Director of Research, Sofia Song and my dissatisfaction with an important way they report Manhattan coop and condo real estate inventory.
The inventory numbers that PruDE reports are quarterly, defined as "active listings at the end of the quarter". Corcoran publishes monthly figures, which I take to mean, as of the last day of the month. Among other interesting data, Streeteasy reports on "available listings" weekly. So I believe that we can get the best apples-to-apples by taking Corcoran’s December 2009 number, The Miller’s 4Q09 number, and StreetEasy’s number for the last week of 2009:
The Miller: 6,851, down 24.6% YoY; off 18.3% quarter to quarter [page 1 of 4]
Corcoran: 7,942, down 29.3% from 11,231 December 08, and 18.3% from 9,718 in September [p6 of 15]
StreetEasy 9,188 [p2 of 15], down 5.4% from 9,708 for the week beginning December 22, 2008 (from the 4Q08 report) and down 13.6% from 10,640 for the first full week of the quarter (October 5) (Prediction: Ms. Song will read that and say "this is wrong", which I explain below.)
(Here comes that horse…) You’d think that the easiest number to find would be how many coops and condos are for sale, but — of course — you’d be wrong. Corcoran found nearly 1,100 more listings than The Miller, and StreetEasy found nearly 1,200 more than Corcoran. I find no explanation in the reports of precisely what each firm is counting, but I did get an explanation from Streeteasy’s Director of Research in a phone conversation (thanks for sharing, Sofia!).
StreetEasy scans "all" firm websites for listings with identifiable addresses (for firms like Brown Harris, which make something of a fetish of not giving an address on their public site, the addresses are confirmed in the inter-firm data base [ROLEX, or OLR]) and gets a direct feed from some small firms. Ms. Song reports that they are learning to make adjustments based on the very strong suspicion that some firms put specific addresses on the web for bogus listings , so they do filter out some firms from their scan; let’s not name names here, OK?
So one possibility for the "over count" on StreetEasy is that they pick up non-REBNY firm listings (as noted in Ms. Song’s comment on my February 21 post, quoted below), but I seriously doubt that non-REBNY firms account for 10% of available inventory. Ms. Song said in her gracious comment on that February 21 post,babbling in the Real Estate Industrial Complex / NY Magazine edition:
Miller Samuel/Elliman and Corcoran inventory numbers will always be lower than ours because we include listings from independent brokers who are not members of REBNY. We have many small brokerages who manually upload their listings onto our site.
With no explanation from Corcoran or The Miller there is simply no way to account for the difference between these two REBNY stalwarts. I assume that (a) each sincerely believes that its number is the most accurate, and (b) considers it something of a trade secret to reveal the adjustments it makes.
StreetEasy’s "inventory" has an odd definition
The StreetEasy Market Snapshot page about Inventory [p2 of 15 in the report] says this about inventory trends: "Inventory levels this quarter were 7.0% lower than last quarter but were 11.2% higher than this time last year." Take a deep breath and follow me ….
Note that these percentage changes (down 7% from the prior quarter and up 11.2% from the prior year) are very different from the percentages I calculated above, when I compared StreetEasy’s December 29, 2009 weekly total to October 5, 2009 and to December 22, 2008, when I got a 13.6% decline from the prior quarter and a 5.4% decline from the prior year. My math is correct, but I am not comparing the same things that StreetEasy is, which is odd, and something you wouldn‘t realize if you only looked at the Market Snapshot page.
The Significant Findings on the first page of StreetEasy’s report explains the difference. The paragraph on inventory says: "There were a total of 13,922 listings that were available at some point in this quarter, a 7.0% decline since last quarter but an 11.3% increase compare to the prior year quarter." What on the page 2 Market Snapshot is described as "inventory levels" is defined on page 1 as "listings that were available at some point in this quarter", a subtle but significant difference. How significant?
what a difference an odd definition makes, mixing apples and oranges
Significant enough that the trend line is reversed. When StreetEasy concluded that inventory was 11.3% higher year-over-year, they did so based on a cumulative figure for listings available at any point in a 90 day cycle. When I compared the StreetEasy December 29, 2009 number of "available listings" to the December 22, 2008 number (i.e., comparing two snapshots, interestingly enough), I read StreetEasy as reporting that inventory was 5.4% lower year-over-year, a (huge!) swing of 16.7% from the StreetEasy prose.
As I told Ms. Song when we spoke, I offered my opinion that the "available at some point in this quarter" usage is a very weird duck and not — to me — particularly interesting. Now that I have had to struggle to explain all this and confront what I see as StreetEasy’s mistake about inventory trends, I find that this very weird duck is pretty dangerous because it is — to me — misleading.
That "available at some point in this quarter" number strikes me as counter-intuitive, and I would be surprised if anyone following Manhattan real estate data would think that a preferable way to count "Inventory" than simply looking at Available Listings on Day X and on Day Y. I realize that I have never before wondered exactly what the other reports mean by "listed available inventory" (in the Corcoran phrasing) or "listing inventory" (per The Miller). The Miller offers a definition ("active listings at the end of the quarter"), confirming that it is a one-day snapshot, but Corcoran does not provide a definition.
paging Will Robinson
Getting back to trend analysis, I don’t think that there is any question that the better measure of an inventory trend is The Miller’s formulation, which I assume to be the same as what Corcoran uses. As we see, applying that measure to StreetEasy’s raw data produces the opposite conclusion as reported by StreetEasy. That’s pretty alarming.
Obviously, StreetEasy is new to this game, and Ms. Song told me that they are interested in improving what they track and how they report. I happen to believe that the "available at some point in this quarter" number is simply not as interesting or as powerful as the "available in the last day of the quarter", but they are free to report and believe otherwise. But when they do so, I suggest they make it clear on the Market Snapshot page (which provides Available Listings for each week of the quarter) that the prose about "Inventory levels" make it clear that it is not (ironically enough) a snapshot of a day, but an accumulation of listings that were "available at some point in this quarter". (Per Ms. Song’s comment on my February 21 post explaining what this number includes, "This 13,922 number [for 4Q09] includes all the listings that were available for at ANY point during that quarter, which includes listings that came onto to market, were de-listed, went into contract, or taken temporarily off the market during the same quarter.")
I hope you have had the patience to walk through this morass with me, and that I have been clear about what the problem (as I see it) is, and why it is important.
Recall that I got focused on StreetEasy’s inventory reporting in my February 21 rant, babbling in the Real Estate Industrial Complex / NY Magazine edition, where I went off an a NY Magazine piece that that relied on a better-than-overall-market Flatiron inventory trend.
You’d think that the higher prices found in Flatiron are related to the shrinking inventory, and that this is something that accounts for Flatiron having stronger pricing than elsewhere in Manhattan. After all, that is the point of the article, right?
Trends are important. As I said in that February post about StreetEasy’s inventory reporting: "To repeat: I just don’t trust that number. What’s up with StreetEasy’s inventory counts??" Now that I better understand it, I can ‘trust’ it, but I hope i have proved here that it is misleading vis-a-vis 2009 vs. 2008 inventory trends.
Let’s move on to two small points about the various quarterly reports.
2 numbers for new listings
Both Corcoran and StreetEasy report new listings to market; The Miller does not. Evidently, The Miller made a judgment years ago that this data point is not very interesting.
Adding the three months of October, November and December 2009 together, I get 4,369 Corcoran new listings for the quarter. Adding the 13 weeks totals from StreetEasy, I count 3,560 StreetEasy new listings for the quarter.
Is this data point interesting?
Moderately, so I wish The Miller started counting it. New listings is some indication of the attitude of sellers … showing some confidence that they can find the right price in today’s market. What is odd (to me) about the Corcoran vs. StreetEasy data is that Corcoran has a much higher proportion of new listings in the market than StreetEasy (4,369 of 7,942 vs. 3,560 of 9,188).
I found absorption! (you may not have realized it was missing)
In doing the heavy listing of again parsing the major Manhattan real estate market quarterly reports I found what I believe is the answer to a small but annoying MLG mystery. Very close readers will recall that on February 23 I lamented that Corcoran does not define "absorption" as used in describing inventory trends, but that I thought that its reporting partner, Property Shark, did, though I was unable to find it on the Shark’s website or reports. It seems that I mis-remembered, as I see "absorption" defined by StreetEasy (on p2 of 15): "Absorbed listings refer to listings that have either entered into contract or have been taken off the market". I have to assume that Corcoran uses the same definition, but this usage strikes me (again) as odd.
As I discussed with Ms. Song on the phone, I just don’t think it at all useful to lump together two very different market reactions to a listing in this way. In one case, The Market loves a listing (those that go into contract); in the other case, the seller reacts to something (to the way The Market ignores the listing, or to changed circumstances, or to in-laws invading for a few weeks, or to something else) by withdrawing it from The Market. In both situations, you can’t buy it today, but so what? That mere similarity is not interesting to me.
I concede that there might be some interest in tracking the Off The Market listings over time, but only if that is separately broken out.
This is a small point that I probably would have suppressed — except that I previously stated that the term was defined in Property Shark. I am pretty certain that the only place I have seen it defined is StreetEasy.
Will stop here (you’re welcome!), as this has dragged on way too long.
© Sandy Mattingly 2010