Urban Place Consulting Charged Palisades BID 62% Less For Establishment Than They Are Charging Fashion District For Renewal, $21K vs. $55K. The Resulting Linear Model Suggests That Each Additional Parcel Adds Around $18 To The Price Of BID Consultancy, But Comparison With San Pedro Casts Some Doubt On Accuracy

This chain of emails from December 2015 reveals that the Pacific Palisades Business Improvement District paid Urban Place Consulting $21,000 for guiding the establishment process and an additional $4,000 to the consulting engineer.1 This is yet another piece of the BID consultancy puzzle that I’ve been trying to decipher since it became clear that almost certainly BID consulting qualified as lobbying under the Municipal Lobbying Ordinance and that almost all of the qualified consultants were breaking the law by not being registered with the City Ethics Commission like, e.g., Tara Devine.2

And this small piece of evidence is especially valuable given the fact that by now it’s essentially impossible to coax records out of the Palisades BID. They’ve even hired a lawyer specifically to thwart my requests, as if the bred-in-the-bone intransigence3 of PPBID ED Laurie Sale, which presumably they’ve already paid for, weren’t enough in itself.

In particular, because we already knew that Urban Place was charging the Fashion District $55,000 for renewal consulting and because it’s the first time we’ve known the rates that a single consultant is charging two different BIDs, it’s possible for the first time to attempt to model UPC’s fee structure. The gory details are available after the break, but the upshot it’s possible to estimate that UPC’s baseline fee for establishing/renewing an ideal BID with zero parcels in it is about $19,583 and that each additional parcel adds a little more than $18 to the cost of establishing/renewing the BID.

First of all, because there are only two data points available, it’s hard to choose a kind of model. I’m going to make the following assumptions:

  1. The independent variable is the number of parcels in the BID — This is a huge assumption, and there’s not that much support for it. It’s not completely implausible, though. Part of the consultancy process is mailing out and collecting ballots and also doing outreach to reluctant property owners. Also consultants organize informational meetings with property owners and other such stuff. All of these are reasons why we might expect the consultant’s fees to be influenced strongly by the number of parcels. Note that there are almost certainly other factors, but I have to construct a model on the basis of what can be measured, and this can be.4
  2. The consultant’s fees increase as the number of parcels increases. — Obviously it’s not going down, right?
  3. A linear model will describe the data accurately. — Obviously there’s not much foundation for this assumption, but since there are only two data points, there’s really no empirical way to choose a model type. It’s plausible that each additional parcel causes the cost-to-consultant to increase by a fixed amount, though, which means a linear model is correct. It’s also plausible that the model is only piecewise linear, e.g. if there’s essentially no increase between 100 and 120 parcels, but at 121 an extra outreach meeting becomes necessary or something.5 There’s no way to discern things like this given our limited data, so it’s reasonable to use a simple linear model.

So at its last renewal, the Fashion District had 1990 parcels.6 Urban Place Consulting is charging them $55,712. And at its establishment, starting in 2015, the Palisades BID had 78 parcels.7 Also, as we just discovered, UPC charged the PPBID $21,000 for its services.8

Thus the two known points are (78,21000) and (1990,55712). This makes the slope of the linear model

m=\frac{55712-21000}{1990-78} = 18.16

So if this model is accurate, each additional parcel adds a little more than $18 to the consultant fees. Also, we can solve for the base price using the slope-intercept form of a linear equation9

y = mx + b

21000 = 18.16  \times 78 + b

21000 = 1416.48 +b

21000 - 1416.48 = b

19583.52 = b

So the resulting model is:

P = 18.16x + 19583.52

Where P is the BID consultant’s fee and x is the number of parcels in the BID. Now, there are very few things more unsatisfying than using all of one’s data on making a model and having none left over to test predictions against. So it occurred to me to check the model against the San Pedro Historic Waterfront BID, since we know that they’re paying Ed Henning $20,000 to handle their renewal. They have 804 parcels,10 so you can see already that the model’s going to go badly wrong. It predicts a cost of

18.16 \times 804 + 19583.52 = 34184.16

Which is 71% more than the actual figure of $20,000. But of course, the San Pedro BID is using a whole different BID consultant, so maybe it’s OK. Also, note that the SPBID has only 281 stakeholders, so a little less than 4 parcels per stakeholder. This is a significantly higher number of parcels per stakeholder than the two UPC BIDs I used for the model. This may be an argument for redoing the analysis using stakeholders rather than parcels as the independent variable. It doesn’t seem worth the time right now, though, until I collect enough data to allow for cross-checks.

Ideally, if we could obtain a plausible model, it could be useful in estimating the amount of time that the consultant spends on the renewal process. In turn, this could be used to make a case against consultants for unregistered lobbying even if it turns out not to be possible in the future to find out precisely how many hours they’re spending. That’s journey’s end, anyway, and today’s analysis is just another step along the road.

  1. Hat tip to a wide-awake reader of the blog who brought this email to my attention last week. I’ve had it all along, but I only started thinking about issues surrounding BID consulting recently, so hadn’t realized what I had. I couldn’t do most of what I do without you all, my dedicated readers! Thank you!!
  2. Lobbyist Larry Kosmont is a welcome exception to this general rule. He registered and scrupulously followed the reporting requirements for his renewal consultancy with the Gateway to LA BID in 2015.
  3. It’s hard to tell the intransigence from the bloody-mindedness from the cluelessness. They all have the same effect, though, so pragmatically there’s really no difference.
  4. Also, I strongly considered using the number of property owners rather than the number of parcels for the independent variable, since almost certainly property owners only need to be convinced once for all their parcels, which would make outreach costs more accurately figured per owner than per parcel. But I decided to stick with parcels because probably the number of property owners is essentially proportional to the number of parcels across BIDs, making the distinction irrelevant with respect to the predictive power of the model. There is some evidence that this assumption may be inaccurate, but it’s empirically decidable, at least, although I haven’t done the calculations.
  5. There are a bunch of other ways an accurate model could turn out to be piecewise linear, but I’m skipping over them. It’s not like this exercise has much chance of matching reality anyway, given the sparse data available.
  6. See page 2 of the City Clerk’s 2013 report to the Council at the time of the FDBID’s last renewal (careful, that’s a huge PDF there). The number may have changed for the current FDBID renewal process, but it probably didn’t change that much.
  7. See page 2 of the City Clerk’s 2015 report to the Council at the initiation of the BID establishment process.
  8. There are some differences between the BID renewal process and the BID establishment process, but they really don’t seem significant. Also, I have no choice but to ignore them at this point given the spotty state of my data.
  9. And substitution, which is how one solves these things!
  10. Or they did at their last renewal in 2012, according to the City Clerk’s initial report to Council.
Share

Leave a Reply

Your email address will not be published.