Okay. All right, good. Now, also a question on just a follow-up from my colleague, Councilman Green, regarding the money sheet and the way that process is, and, again, this goes back to technology. So, the money sheet information is provided by the assessor, and it is then 160 entered by someone else
The data is entered. So, what kind of quality controls do we have in place to make sure that that's done correctly?
Well, once again, this speaks to, you know, the process and why we're doing it like we are. You know, we're bringing the information in and we actually have a quality control unit that's responsible for auditing what it is that's been updated in the database.
Okay. And one last question was, I just wanted to follow-up on Councilman Jones's asked, he had requested an assessment of all government owned property and that there will be a breakdown by Council District, that's correct, that you will be providing?
Well, he had asked, if I am not mistaken, for PHA, some of the governmental entities, but he didn't specify Council District. 161
We can do it by ZIP code and we can do it by what is included in the Council District, but our data isn't exactly stratified by Council District, but we can do the best we can with it.
Thank you, Councilwoman. The Chair recognizes Council Member Green.
Thank you. So, the process is the evaluator, since you came to OPA, have been coming back from the field responding to emails and phone calls, sitting down where they have noticed something that they think is big enough to effect the value of the property potentially in 162 a significant way, they fill out a money sheet for all other data that they get that is different from I guess the binder they went out into the field with which has, you know, whether or not it has got a garage, how many bedrooms it has to the extent you can tell, the size --
Right, it's two story versus three story, things you can see
They would correct that themselves in Characteristics Database.
Right. Now, we in an effort to time to expedite things, we had brought in some temp individuals to help with the data entry because some of my evaluators aren't as swift data entry wise, so we had brought in some additional assistance to help speed the effort along, but most of them are putting it themselves because they are quite confident they can do that.
So, if we have 163 collected data on 75 percent of the properties in the city.
The evaluators have input into Characteristics that data, have they not?
I can't say that they have put in all 75 percent because some of them are a little less efficient at data entry. Some of them have focused more on the field collection. Because we were up against winter time, and we were really fortunate to have the type of winter that we had, but we didn't know we were going to have that type of winter, so the efforts moved towards data collection, data collection, data collection waiting for the terrible winter to finally hit and it never hit, so we are more along the process of collecting than we are at data entry.
Well, it's a simple -- where is the room with -- does each evaluator keep at their desk the binders of unentered data? Are the binders returned to a 164 data entry pool?
It's a highbred. Some of the evaluators have the binders at their desk and they enter them in themselves. Some of the evaluators are using an assistant and the assistant has the binders and they're sitting there all day constantly entering in information.
Okay. And so most days the evaluator is in the field and there is people doing data entry back in OPA --
-- combination of things happening. We have evaluators out in the field collecting data. We have evaluators in the office entering data. And then we have assistants that are entering data for some of the evaluators.
Well, there is property count, so it varies, so we try to --
We try not to -- we try to keep it, you know, it's property count and then it's the type of properties you have. If you have properties that are very homogeneous, you may be able to have a higher property count than someone who has properties that are all unattached, you know, large in size and scope, and more difficult to value, so it varies. Some may have 13,000, some of them may have 5 or 6,000, some of them have 20,000. We try to, as we get additional staff and get them trained up, we try to reduce the property count per parcel per evaluator down to something that's more easily acceptable with standards.
Thank you. As the -- I am going to go until they cut me off
Okay. So, as the 166 properties are -- I guess the question is somebody with 20,000 properties, they have personally gone and looked at 75 percent of those
No, the total office is at 75 percent, not each evaluator.
And that is somebody going into the field or are they using pictometry or are they using Google Earth?
That's going into the field and looking at the properties. Now, some properties we can't see because of logistics, you know, they may be fenced off or they may be behind a barrier, we'll use pictometry for those. Some things we can't see, even though we can see the front of the property, we can't see the back, so we use pictometry for those.
Is every property checked with pictometry by the evaluator?
Right, if they can't see because they typically would see the front and they can drive down an alley to see what is going on in the back.
So, who's auditing the data that's being handled by the data entry pool or verifying the data that's being put in by the evaluator himself or herself, the evaluator's assistant, or the data entry pool, does anybody audit that information that goes into Characteristics?
Another step in this process is to, once a GMA is completed and all the data is entered, then we call that a hundred percent complete, that's both field inspection and data entered, then we attempt to run a report on all of the changes and we can look at -- the supervisor can look at a couple of things to see, you know, inconsistencies. The evaluator looks at it, and, you know, so we have a quality control also that will look at, we will spot-check certain neighborhoods and 168 behind certain evaluators to ensure that everything is consistent.
Right now I'd have to go back and find that out. I know I have several that are a hundred percent field inspected, but I don't know how many are a hundred percent data entry, data, you know --
I would like the answer to both questions provided to the Chair, please, a hundred percent field inspected and a hundred percent --
We can go through that on call backs. Are evaluators taking photos of the properties they inspect while they're out in the field?
Only new construction or things where they feel as though a property has changed a great deal.
Okay. Let me see here. So, I guess this leads me to some 169 questions about the methodology -- I'm sorry. You said that the money sheet goes into a system that is separate from the Characteristic VSAM Software
It's, and Larry can explain this better, our IT guy, but it's entered into a front end part of the database that is designed just to pick up the Characteristics and then that's loaded into the primary database at the end of the day so that the next time we go to check and we run the report we can see what is touched and what changes were made.
Is the primary database Characteristics or is it something else?
No, it's Characteristics 170 is what we put the characteristics in, that goes into VSAM which currently is our primary database.
Okay. And the money sheet information that goes into the front end, is that going to go to VSAM or something else?
Okay. And that is the current property evaluation system in the City?
Okay. Well, let me just go through this before I get to that. So, my understanding is that every year there are only 15 to 16,000 valid Bona fide arm's length residential transactions; is that correct?
Please provide that to the Chair. How many years of sales are built into the software for comparison purposes with the data being currently entered?
My understanding is that it should be at least months, and if 10 it's been a stable market it can be 30 to 36 11 months, but when you go back too far you're 12 creating a data set that can be bad. 13
Well, we're 14 considering five, but we didn't say that we 15 necessarily going to calibrate everything with 16 five years. We looked at five years worth of 17 sale and then we are going to really hone in on 18 the most recent two years and then we'll do a 19 hold out sample of the most recent months. 20
That's statistical package, well along with Evalassist now. Evalassist is a database that was created in the office for the purposes of looking at sales and things like that, and recording information from sales, so when we get to the point of doing modeling we primarily use SPSS which is a statistical package and then that information -- some of the information loaded into SPSS comes from Evalassist with the help of Evalassist.
Well, the methodology is to, basically what you're doing is you're trying to estimate what a market is, so you're looking at, you have the City broken down into different GMAs and you're analyzing all the 173 sales data for all particular properties, be it single family, twins, rowhouses, fully detached properties, you're breaking all of that out individually and then you're trying to tie co-efficient to everything that you picked up that you think adds value from the characteristics.
You're comparing the characteristics of the property -- if I can restate --
So, you're taking the characteristics of the properties you've sold, whether it has one, two, three bedrooms, two or three bathrooms, a garage attached or separate, and you're taking the 15 or 16,000 such properties that sell a year, maybe less in the last couple of years because I know our transfer tax has gone down, but let's just call it 15 to 16 over the last two years, so it's a pool of roughly 32,000 properties with 174 characteristics assigned in different GMAs, well, obviously a smaller pool in each GMA.
And you're comparing that to the characteristics that are in a particular GMA?
So, you're taking sales data and comparing it to characteristics of properties that are not sold, but are every other property in the city?
And so that's what you're doing. So, really, you could do that today?
No, you can't do that today in my opinion because you don't have everything vetted out. You don't have all the characteristics for your city. You don't have all those parcels looked at and you don't have --
So, you would be, in my opinion professionally, you would be haphazardly placing value on properties that you don't know if that property has that characteristic or not.
Well, you know that in the database other similarly situated properties probably don't have that characteristic either. I recall as the OPA headed down this path, it was really driven by our former budget director of gathering every single piece of data, you reach a point of diminishing returns with respect to data, you either have a statistically significant sample of data or you don't.
But let me tell you why I wouldn't take that approach just specifically. If you were to value a lot of properties on a particular block and you said that they had garages and garages added, let's hypothetically say $30,000 to a particular single family property, then you're going to give those people an assessment notice that 176 represents something, an additional $30,000 that they don't have. Now, of course those people can appeal, and we have a first level process, but you're making them go through a process that you put on them because you did not vet the data, and then that's somewhat --
Well, that's true, but there was data in Characteristics before OPA was created.
Right, but, sir, it is bad, that's why we couldn't go with the last reassessment, if I'm not mistaken. And if you look at what we're finding, we're finding, you know, things that need to be documented like houses that have been knocked down that, you know, we're still carrying on the books and vice versa, new construction where we have vacant lots. So, without giving it the proper time and consideration to really look at that data, to really clean it up and vet it, in my opinion you would be acting somewhat irresponsibly.
Well, what percentage of the City's data has a house constructed that is not in your system? And, frankly, that is not the fault of anybody but L&I or whoever is supposed to be sending that information over to you.
It's a simple question. What percentage of the properties, residential properties, has such a big change like no house there, which is probably not going to be a taxpayer who is going to pay a bill, and the opposite which is there is a house that we didn't know was there, what percentage statistically --
I don't have a percentage. What I have is verbiage from my residential supervisors who informed -- we meet every week, every Friday, and I am told there are blocks, there are streets, that we don't 178 have values for where the houses have been removed. I am hearing from my supervisors, as well as some of my staff, that we have numerous properties where additions have been built and we don't have it on the books. And, so, these are the reasons in my opinion why I stated from day one our process has to involve a field inspection for at least every residential parcel and a great portion of the commercial ones as well. And, also, and not to cut you off, but that doesn't even begin to speak of the commercial piece that we haven't even --
So, really, wherever a house has been knocked down that your records show was there, the value of the property may not change all that much with respect to that, I imagine, but let's just say and where ever there is a new house that appears it's not in the Characteristics 179 database, that would produce a money sheet, would it not?
Well, 75 percent of the city is done so it's going to be a pretty good sample size. Statically that's way better than a poll.
So, it's a simple question. Can you please provide the Chair with the number of money sheets that are in the front end system that effect value at all, and break it down by 10,000, by 20,000, by 30,000, by 40,000, by 50,000 and just the total number of money sheets that went into the system.
We may not be able to do that because the money sheets I don't know if they're coded to pick up whether something is new construction or not.
Yeah, see, you would have to have that coded. And one of the things I want to do for next tax year is to have some sort of system where we can tally up new construction, but right now we don't have a system that will tell you whether something was new construction or exactly what the cause of the money sheet was like that.
So, what are the codes that are available for a money sheet?
We do have codes that are available to indicate a change, but they don't necessarily --
What codes are they? What codes are available for money sheets?
I think among those 12 codes we can put together a request that will -- 181
I am sorry. You don't code -- when you take the money sheet data and you put it in the front end system, you don't record the code that created the money sheet in the front end system?
Okay. So, I can ask for a run based on the various codes?
And based on the change in values in the front end system?
We'll provide you a list of the codes and then you can tell us what you want.
I already know what I want. I would like a run of the money sheets 182 by code and by value, so $10,000 change, $20,000 change in $10,000 dollar increments.
Yeah. So, if the front end system doesn't record the change in value that was created by the money sheet?
As I look at my IT director, he is saying he doesn't know if our system will be able to, but we will do the best we can and try to get you a report and we'll work with you if you're amenable to seeing what we can do.
Okay. I mean, certainly I will wait to hear back what you say you can do.
And I will reserve judgement on whether or not I am amenable to that. So, my understanding is that the following data points can lead you to 85 percent of the value of a residential property: Sale price, location, year built, condition, 183 poor, fair, good, excellent, and type of construction, brick, masonry, stone, wood, et cetera.
You're asking me is that true. I think that it depends on what market you're in. I mean, that might be a guideline, but that might not necessarily be true for every locality.
Don't we have far more data points on almost all properties in the City of Philadelphia currently in Characteristics?
Once again, the current data points might be erroneous. That's why we're out doing a --
They might be, but they might be -- in other words, when we gather this data going city wide, every year we're going to have to go out and see if there are changes again and again and again, and at a certain point you have enough data statistically that you say, okay, this is close enough.
Can I, if I follow 184 you, let me say this. I think year one and two, that might be -- we might be better off to say something like they'll be a small change in certain, but we have never ever as a city gone out and inspected everything. And I think until we do that and we do that and collect the data accurately and it reflects accurately in our system, I think we're -- I don't think you can say that.
Statistically, the data produced by the BRT in 2009 I think was presented and was presented on a disk to Council, which I have in my office, met IAAO standards and it met it based on the characteristics that were currently in the system. So, it was a statistically significant enough sample for us to meet those standards two years ago. You've added data for three years since that time that that run was produced, so statistically you have got a big enough sample size today to get it almost perfect. You're never going to get every property right, but at a certain point you have to say I have enough data, let's run it and see 185 how it looks. And that's kind of, you know, I don't think it will be unreasonable for Council to ask you to do that before we reach the end of this process. And I think you heard the President trying to make that point earlier
I think I would have to advise or recommend against that based on what we have seen. I can't speak for what was done in 2009 and what reports were run. I know that since I've been there, there have been numerous discussions amongst our staff about things that they are finding out in the field and I think there was some criticism from the media perspective of what that reassessment back then might have produced as far as errors. So, I professionally would advise against it without seeing everything.
So, what you're telling me is you have never even evaluated the BRT assessment yourself, it's just you're going based on what you have read in media reports and?
I have talked to the 186 people who ran that file and a lot of them are currently there, and so based on what I have been told, I trust in the methodology we're using now versus what they may have done back then.
Right, but it was just statistically -- basically they used statistics to correct for data that wasn't in the system and they got it basically right. If you go look at a -- do you have a disk of that?
When you're doing a re-assessment, you want to do it as accurately as possible --
So, I mean, without being too picky, so to speak, I think that if you can continue to go out and find errors in the data of substantiality, additional stories that aren't there, properties that we don't have listed as being vacant lots, I mean, I 187 think that you owe it to your profession and the taxpayer to take the steps to do it right as much as possible.
And I, based on the historical information that I have gotten, I have been told a lot of individuals are happier, have more confidence in what we're doing now versus what they did back then.
I don't know what that means sort of people having conversations about what happened in the past or not happened in the past. This is a simple area, and it is clear that the more data points we have, the more accurate the system is going to be. At a certain point, you have to fish or cut bait and say do we have enough data to produce a statistically significant sample that is going to be correct for 99 percent of the residential properties in the city or 97 percent of the residential properties in the city, and I am not an assessor, but since they could do that without all the extra data that was provided and get it right within IAAO standards, it's my 188 suggestion that we could --
Well, first thing's first. I have no knowledge of how those studies were run statistically. I'm not going to, you know, cause immense doubt in them, but I would be more confident in doing the steps that a typical jurisdiction would do recommended by IAAO that you would get the type of results that you could stand by.
I completely agree. I am not suggesting you stop having people --
I am not suggesting you stop having people gather data. I mean, that's constantly what us evaluators should be doing all the time, especially as we go to a mass appraisal system where based on that data it's computers that are going to be generating values and evaluators are going to be confirming what is produced based on their field experience rather than the other way around going into the computer system. So, I completely agree, that it's only a simple 189 point, and that point is at a certain point you have enough data to have it be statistically significantly correct such that you can go out and meet IAAO standards and be there. And I don't hear you saying that we would not meet standards if we didn't --
I am getting ready to say I don't know that you would. You know, without tossing -- without saying that, you know, everything that was done in 2009, I can't speak to it, you're speaking to me about it, but I don't know that you could produce the type of statistics that would be industry standards based on what we have in our system now. And my professional opinion is that you continue with the steps that are necessary to ensure that you would meet those standards.
Well, I guess I would like to know statistically of the 75 percent of the properties that you have reviewed, how many have had a data change that is significant enough to effect their value. Because if the answer to that question is not a large number, then we're just spinning our 190 wheels. We need to go ahead and produce values based on what we have and continue to update as every assessor's office should every year. Or we can wait, as you're suggesting, if this is the professional thing to do, then we should wait for all of the data to be in so that this body can make an informed decision about the measures we are going to put in place to protect low income, seniors, and other people, et cetera, that would be legal, and without knowing the data and the aggregate values, we can't come up with whether 15,000 or 90,000 is an appropriate Homestead exemption
But you cannot get a value tied to what we have collected until you have run, cleansed, and reviewed the models, and we are not at that point yet. And I think that's what we were discussing with Councilman Clarke, Council President Clarke, is that we're not at that point. So, all we have are characteristics that have been collected on whether a property has a garage, what condition it's in, things like that, but there is no value tied to any of 191 that yet.
Well, we have that data because we have the last months of 4 sales data, the last 36 months of sales data, 5 and we could at any time we wanted apply that 6 to the information in Characteristics to see 7 what results are produced and then do a 8 statistical analysis to see if that met IAAO 9 standards 10
I think when you're 11 modeling, and, you know, probably Dr. Bob 12 Gloudemans would be a better individual to tell 13 you since I am not convincing, but you can't -- 14 it's not recommended that you model based off 15 of information that you collected year end 16 analysis you've done years ago. If you're 17 going to have -- 18
I don't understand 19 that question, because isn't the pricing model 20 based on the sales over the last 24 to 36 21 months? 22
-- but for 2009 analysis you had to have had sales prior to 2009. So, you're looking at sales if you went back, if the team who did that analysis did three years back, they're back into 2006, and that's a whole different type of market.
I am not asking for values in 2009, I am asking for values in 2012, and you get that based on the value --
But you're using characteristics that haven't been finished.
But that's actual sales data. That's actual sales data which you have. You have the actual sales data, don't you?
We have sales data running up to as current as possible, but, you know, we don't have characteristics that have been collected and vetted for right now.
But you have actual sales data for every property that's sold over the last three years, don't you?
And isn't it that information that creates the pricing model?
That information is used, but that information is used with characteristics that are in that time period.
Sure. And you could use it with all of the characteristics you're going to continue to gather.
So, I guess what I'm saying is, okay, so you're asking to use current sales and you want to use old characteristics.
I want to use the characteristics that are currently in the characteristics database.
Right, but once again while 75 percent have been collected, not 75 percent have been put in, so you basically will be using a lot of old characteristics. 194
Sure you would. And most of the old characteristics is accurate. You're not changing -- in other words, you're not suggesting that you're changing most of the characteristics on every single property, are you, because we just determined that 85 percent of a value could be determined by just a few small number of characteristics, when you add the number of windows, the number of bedrooms that is already in your data.
I think that the characteristics that we are finding are enough to state that we need to collect them all. I mean, you got to also think about equity here. What happens if the characteristics that we have collected partially in a neighborhood, we use those and the other part of the neighborhood we haven't, we used the old ones, and so now you got, in my opinion, an equity problem, have the taxpayers been treated fairly if, in fact, they are -- some of them are using new data and some of their values are used for old data. It's just not a clean -- 195
And the more data you have, the more fair the system is, would you agree with that?
No. Not -- just in the general sense. Good data, I think the key here is good data. Yeah, you can have data, the data module is filled out, is it good, is it accurate, is it clean, not so in our case. I think that's why we had so many problems.
You say that, but you don't produce data to demonstrate that. I mean, you're saying, well, you know, we do change a lot, but I mean I think if we're going to be asked to do what we're being asked to do, we should expect you to be able to answer that question. What percentage of the data has been inaccurate with respect to a particular property on average.
Right, but I won't know whether it was accurate or not until I go see it all. 196