was the Manhattan coop loft market dead in 1Q09? Miller nugget provokes arithmetic angst

once more into the breach
This may be my last nugget from the Manhattan real estate market reports from the first quarter of 2009. Honors go to Miller Samuel:
 

Wow! By The Miller’s count, only one of every four lofts sold in Manhattan in the quarter was a resale. Put another way, there was essentially no market for Manhattan lofts in coops to start 2009 (some of those resales were condos, of course, while none of the new developments were coops). With only 89 loft sales in the quarter counted by The Miller, no more than 23 loft coops traded. While there may have been a pulse, the coop loft market was essentially dead for the quarter.

Not to (over) indulge in sycophancy, but I trust the numbers from The Miller as much as I trust any Manhattan real estate data, as even the most casual reader of Manhattan Loft Guy would know. So this datum is highly credible to me — and troubling for what it means about an important part of the loft market, since a large majority of "classic lofts" are in coops, not condos. It is also troubling for the way it conflicts with the data I have been collecting (and reporting) about loft sales on a real time basis.

angst alert
In contrast to the no-more-than-23 resale lofts counted as sold in the first quarter by The Miller, I counted 21 from March 12 to February 11 and another 13 in the prior 30 day period, plus another 22 in the 30 days prior to January 11, at least some of which were in 2009 (see my April 1 post), for a likely projected total for the quarter of 55 or so loft resales. The fact that I count more than twice as many as The Miller … makes me nervous.

This is troubling because it is yet another instance of the difficulty of reconciling market data in our still-oh-so-secret market. It is troubling because it points up a potential conflict between my own reporting modes about real time Manhattan loft sales, one which I have been aware of for a while but have not explained before.

how much does a crappy source determine garbage out?
As I have explained about my weekly "as of Sunday night" counting of new loft Manhattan listings and loft sales (my last such report was posted last night), my source is the inter-firm data-base, which has several structural defects for tracking the loft niche in the Manhattan market. First, it depends on agents updating the status of a listings as Sold & Closed to come to my attention as a recent sale. As is often the case, agents may take a while to update the data-base, if they update it at all. So, while I have been careful about reporting the number of Manhattan lofts "reported as sold" in the last 7 days, that is a subtlety likely to be missed by many readers (I still think it is a useful proxy for ‘actual’ sales; but more on that later).

Second, my report counts only those "lofts" identified by agents as "lofts" (as opposed to being identified as, e.g., a "2 bedroom" or "3 bedroom" apartment). This means my numbers are both over-inclusive (when an agent calls an "apartment" a loft) and under-inclusive (when a true "loft" is not identified as such). I have been willing to tolerate this messiness because I see no other way of providing real time information and I think that directionality is useful, as the messiness should be equally distributed and should not obscure trends.

But when I see such conflict with hard numbers from Miller Samuel it makes me wonder.

MLG 7 vs. MLG 30 / smackdown or love-fest?
In addition to my weekly "as of Sunday night" counting of new loft Manhattan listings and loft sales, I have now had four sets of reports of last-30-days-of-actual-loft-sales (my last such was as of March 12, posted April 1, as noted above). These reports start with the same inter-firm data-base, but I have supplemented those reported sales with other sales that I see publicly reported as having closed in the relevant 30 day period, and with lofts reported in the data-base as ‘recently’ sold that turn out to have closed in an earlier reporting period of mine. As I clean up and supplement my 30 day spreadsheets, they become more accurate as I add public information on deeds filed and closing prices and the extent to which there are lofts included that have not yet been publicly confirmed becomes clear.

The Manhattan Loft Guy dirty little secret is that I suspected after the second 30 day report that my 30 days reports might not match my 7 day reports. The reporting periods don’t line up exactly, but I was afraid I was getting different numbers.

So I tested the conflict from the periods that are in the first quarter of 2009 (including both new developments and resales, and using my updated 30 day totals):

total loft sales counted in 30 day reports, 30 days from

total loft sales counted in weekly reports, totals from weeks ending

too much inside baseball, or refreshing transparency?
It is my blog, so I can be as self-reflective (self indulgent?) as I wish…. This is the first time I have actually put the two reporting periods side by side, and they come out closer than I feared they would. There’s enough congruity to make me satisfied — if not ecstatic — about the internal consistency of the data (so far). The fact that my stuff is roughly compatible is less troubling than I feared, though the mis-match with the Miller Samuel numbers worries me.

Overall, I don’t think the data difficulties make my various counting efforts worthless, just rather imperfect. Provisionally, I will continue to go forward as before, in the belief that my counting identifies trends, and more valuably as the length of the reporting period gets longer. But I do wonder if this will prove to be distracting and irrelevant. I feel like I am between a rock and a hard place here.

Enough navel-gazing (for now).