BEFORE THE CALIFORNIA STATE BOARD OF EQUALIZATION
5901 Green Valley Circle, Room 207
Culver City, California
REPORTER'S TRANSCRIPT
JUNE 23, 2011
ITEM C7
SALES AND USE TAX APPEALS HEARINGS
PETITION FOR REDETERMINATION
filed by
CITY OF LOS ANGELES - CONTROLLER
(Case No. 489155, 509958 AA)
Reported by: Beverly D. Toms
CSR No. 1662
1
1
2 P R E S E N T
3
4 For the Board Jerome E. Horton
of Equalization: Chairman
5
Michelle Steel
6 Vice-Chairwoman
7 Betty T. Yee
Member
8
George Runner
9 Member
10 Marcy Jo Mandel
Appearing for John Chiang
11 State Controller
(per Government Code
12 Section 7.9)
13 Diane Olson, Chief
Board Proceedings Division
14
15 For Board of David Levine
Equalization Staff: Appeals Division
16
Scott Lambert
17 Sales and Use Tax
Administration
18
Robert Tucker
19 Legal Department
20 Kevin Hanks
Sales and Use Tax Department
21
22 For Petitioner: Daniel Whitley
Representative
23
Ricky Deguchi
24 Representative
25 Bruce Truitt
Witness
26
27 ---OOO---
28
2
1 Culver City, California
2 June 23, 2011
3 ---oOo---
4 MR. HORTON: Ms. Olson, what is our next
5 matter?
6 MS. OLSON: Our next item is C7, City of Los
7 Angeles - Controller. Please come forward.
8 MR. HORTON: Mr. Levine, as they come would you
9 please introduce the issues in this case.
10 MR. LEVINE: Yes. The issue in this appeal is
11 whether the Department's audit, although conforming with
12 the requirements of the Board's Audit Manual, should be
13 rejected.
14 MR. HORTON: Would the Appellants pleases
15 introduce themselves for the record.
16 MR. WHITLEY: Good afternoon. Deputy City
17 Attorney Dan Whitley of the City Attorney's office of
18 Los Angeles.
19 MR. DEGUCHI: Rick Deguchi, City Controller's
20 office.
21 MR. TRUITT: Bruce Truitt, a Applied Statistics
22 and Government Auditing expert with a focus on criminal
23 and civil improper payments.
24 MR. HORTON: Thank you.
25 MS. YEE: I didn't -- I didn't hear that last
26 part, Mr. Horton.
27 MR. HORTON: Would you repeat that, please.
28 MR. TRUITT: Focusing on criminal and civil
3
1 improper payments.
2 MS. YEE: Thank you.
3 MR. HORTON: Okay. Ms. Olson, is this one that
4 we allowed additional time?
5 MS. OLSON: Yes, we did. I'm sorry, I didn't
6 announce that they get 15 minutes for their
7 presentation.
8 MR. HORTON: Okay. Gentlemen, you have 15
9 minutes for your presentation. We will return to you
10 after the Department uses their 15 minutes for rebuttal.
11 MR. WHITLEY: Good afternoon again. My name is
12 Dan Whitley, as I said earlier. As an attorney I know
13 you don't need to hear me speak about statistical
14 methodology so I'll do what's hard for attorneys and try
15 and keep this brief.
16 What I'm going to do is explain our view of the
17 case and where we are from the City's perspective.
18 The -- this case, as was mentioned earlier,
19 involves a statistical methodology that was used by the
20 Board of Equalization to make the determination that --
21 of taxes that the City owes to -- to the State.
22 And as I said earlier, this doesn't involve
23 taxable versus non-taxable transactions or proper
24 reporting, it involves whether a proper methodology was
25 used to determine using a statistical sample how much
26 the City owes.
27 But this doesn't mean that although there
28 aren't specific regulations or statutes that apply to
4
1 our issue that this is free of a legal analysis.
2 There's a basic standard that applies to all
3 determinations and decisions made by government, which
4 is that they be rational and non-arbitrary. And as will
5 be explained by Mr. Truitt and Mr. Deguchi later the
6 methodology that was used by the Board to determine the
7 City's liabilities resulted in an arbitrary
8 determination.
9 Because the determination was arbitrary it
10 shouldn't be used to determine the City's liability.
11 Now, I don't want to make it appear that we're
12 saying that this -- the Board's manual is incorrect as
13 applied to every taxpayer. The purpose of statistics is
14 to give you a result that has some reasonable basis to
15 give you a reliable estimate. It's never going to give
16 you an exact number, it's going to give you an estimate
17 of what the exact -- of what the proper result should
18 be.
19 And for some taxpayers in some circumstances it
20 might be more effective and efficient for them just to
21 get a ballpark number. And so even though the
22 methodology that -- that the Board uses doesn't give you
23 you a very reliable estimate, these are relative terms.
24 And so although as Mr. Truitt and Mr. Deguchi
25 will explain, this is not a very reliable method of --
26 of achieving a determination. Some people may just want
27 to get a quick and dirty number that's good enough for
28 them and they just want to be done with it.
5
1 The City's not in this position. We need to
2 justify every expenditure to our taxpayers. We need to
3 know what our liability is and why we're paying it. And
4 we can't rely on a determination that we can't go to our
5 taxpayers and say with specificity this is a reasonable
6 determination; we know why this number is the right
7 number and that's why we're paying it.
8 I also want to explain that we're not asking
9 for a complete refund of the determination that was
10 made. What -- what this statistical methodology
11 resulted in, as will be explained in a second, was a --
12 a point estimate, which was what the determination was
13 and then a -- a range within that point estimate will be
14 found. And we're willing to pay the lower limit of the
15 range.
16 Although the methodology doesn't give you a
17 very good reason to think that the point estimate is
18 reasonable and reliable we do think that the range
19 that -- that the methodology gave us is reasonable
20 enough for us to say that we at least owe what the lower
21 limit of the range is and we're willing to pay that
22 amount.
23 And so with that I'm going to pass this off to
24 people who know more about statistics and can give you
25 more relevant information.
26 MR. DEGUCHI: Good afternoon. Just very
27 briefly on my background, I've been an auditor for over
28 30 years for two of the largest organizations in
6
1 California; the County of Los Angeles and the City of
2 Los Angeles.
3 Over the last 20 years I was designated as the
4 statistical sampling guru for those offices, which meant
5 pretty much that any sample that was done -- statistical
6 sample that was done, they had to come to me before the
7 sample and after the sample.
8 And I've trained probably well over a hundred
9 people in statistical sampling methodology.
10 As Dan had indicated, Mr. Whitley, we have
11 revised our request for refund from 1.4 million down to
12 $450,000. This is irrespective of any interest that
13 would be due to us in terms of interest from the date of
14 the transaction through the date we made the refund plus
15 any interest due from the time we made the refund until
16 we receive a refund.
17 With regards to the 1.4 million we have no
18 issue with $763,000 of this amount. That amount was
19 determined by going through a hundred percent review of
20 all transactions within a certain subset within the
21 population.
22 With regard to the other remaining 639,000 that
23 was an amount that is an estimate based on a flawed
24 sampling which our statistical expert will cover.
25 In a nut shell, I want to just summarize that
26 the auditors are 90 percent confident that the 639,000
27 is accurate within plus or minus 71 percent. Stated
28 another way, they are 19 percent -- percent -- 19
7
1 percent confident that the estimate is within plus or
2 minus ten percent.
3 We believe that the 639,000 should be reduced
4 to 138,000, which is the lower limit of that range.
5 To put this in perspective, I don't think any
6 of us would be happy here if the IRS came to us and
7 said, you know, "We did an audit of your taxes over the
8 last several years. We determined that based on our
9 sample you owe us $650,000 and we're 19 percent
10 confident that's within plus or minus ten percent, or
11 we're 80 percent confident that that amount is within
12 plus or minus 71 percent."
13 MR. TRUITT: Just a little bit more information
14 on my background. Like I said, I've been in applied
15 statistics and government auditing for 25 years. My
16 analytical technologies and methods are used in all 50
17 states and 33 foreign countries.
18 I teach statistical sampling at the U. S.
19 Government Audit Training Institute in Washington, D.C.
20 and at the National Medicaid Integrity Institute. And
21 I've also taught at the University level, the graduate
22 level in quantitative methods in applied statistics.
23 Now, since sampling figures so prominently in
24 the entire discussion here, I think a very short
25 definition and some clarification of terms is in order.
26 Sampling is certainly a well-established audit
27 procedure, and it relates to examining less than a
28 hundred percent of the items in a population in order to
8
1 come to some sort of conclusion about that population,
2 to estimate a population value.
3 If you look at everything in the population
4 you've done a census. You have no margin of error,
5 whatsoever, on your estimate. You know exactly what's
6 going on in the population. And you can assert a single
7 number.
8 If you do not look at everything in the
9 population you have not done a census by definition; you
10 have a sample and also by definition you're going to
11 have a margin of error.
12 We have two basic types of sampling.
13 Non-statistical sampling, in which the auditor controls
14 the probability of being included in the sample, and as
15 a result of this control you cannot calculate the margin
16 of error nor can you make an assertion about the
17 population. That's for non-statistical sampling.
18 In this case statistical sampling was done and
19 the primary differences are statistical sampling lets
20 you do two things that non-statistical sampling does
21 not; measure the precision or the margin of error, and
22 make an estimate of what's going on in the population.
23 That's the so-called point estimate.
24 However, a sample that does not measure
25 precision cannot be used to estimate a population value
26 and automatically becomes non-statistical. The AICPA
27 sampling standards are extremely clear on this point.
28 It says, and I quote Standard 2.20, "Any sampling
9
1 procedure that does not permit the numerical measurement
2 of the sampling risk," in other words the margin of
3 error, "is a non-statistical sampling procedure." And
4 we've already established that you cannot measure margin
5 of error or make an assertion about the population from
6 a non-statistical procedure.
7 AICPA 2.20 adds that, quote, "Even though the
8 auditor rigorously selects a random sample the sampling
9 procedure is a non-statistical application if the
10 auditor does not make a statistical evaluation of the
11 sample results."
12 Statistical evaluation means you measure the
13 margin of error. You measure your precision, your
14 sampling risk. It comes with all sorts of different
15 names.
16 Now, another issue that has figured prominently
17 in this discussion has to do with confidence level and
18 precision. And I absolutely do not want to put
19 everybody to sleep with a long discussion of that but
20 there are a couple of things to understand.
21 Confidence level is actually very simple. It's
22 really just the probability of getting statistically the
23 same results if you continue sampling.
24 So let's say you're running for office and you
25 have done an exit poll and it comes back and it says
26 we're 90 percent confident that you're going to get 55
27 percent of the vote, plus or minus three percent.
28 Okay. That means that you can be 90 percent
10
1 assured of receiving 52 percent all the way up to 58
2 percent, right? 55 plus and minus three.
3 90 percent confident, though, means that
4 there's a ten percent probability of your being wrong
5 when you make that assertion. Okay.
6 Now, in this instance the Board chose to work
7 at 88 percent confidence, according to its audit manual.
8 What this translates into realistically is that the
9 Board is willing to accept a 20 percent probability of
10 incorrectly estimating taxes owed.
11 Now, if the IRS came to me and asked me to
12 write them a check based upon a number that had a 20
13 percent probability of being wrong, I'm not real sure
14 how I'd feel about that.
15 Also, in the present case the margin of error
16 per policy is allowed to be plus or minus 75 percent.
17 That gives you a window that is 150 percent of the value
18 of the tax in width. That's extremely wide. Extremely
19 wide.
20 Again, if the IRS came to me and said, well, we
21 want you -- we want you to write a check to us and the
22 range of the number could be 150 percent wide, that's
23 not very precise. It's actually not very useful as
24 management information, unfortunately.
25 Now, as I understand it, the City has no
26 dispute with the point estimate, itself. Is -- that's
27 correct, okay. Nor do we necessarily have a dispute
28 with the amount of precision that would be calculated if
11
1 it were used, and I believe that's about 57 percent --
2 MR. DEGUCHI: 71 percent at the 90 percent
3 confidence level.
4 MR. TRUITT: 71 percent at 90 percent. But at
5 80 percent it's about 57 percent.
6 So under their confidence level their window is
7 still plus or minus 57 percent. That is -- if I'm doing
8 the math right, 114 percent wide. I can do statistics
9 but I can't do arithmetic. I think that's right.
10 Okay. That's a very wide window. And, again,
11 that really provides you with marginal information for
12 decision-making. You get a very wide and imprecise
13 range, which works out to be a range of numbers from
14 about 188,000 -- oops, sorry about that -- 188,000 up to
15 just over a million bucks. That's a very wide range of
16 values inside of which to try to assess tax.
17 And, again, I remind you, any sample that does
18 not measure the precision, and I'm quoting here, is a
19 non-statistical sample and therefore cannot be used to
20 estimate the population parameter.
21 In other words, the Board must take the margin
22 of error into account when it's make -- it makes its
23 decision. Otherwise, the sample is by definition
24 non-statistical and it cannot be extrapolated at all.
25 And to insist that the amount of tax owed is
26 $640,000 without considering the precision also makes
27 the sample non-statistical and it assumes that the
28 auditors know what the amount of tax owed is when in
12
1 fact they do not. The only way they could know what it
2 is is to look at everything in the population. They
3 have not done that.
4 So to insist that it is $640,000 is incorrect.
5 In fact, and every statistician on the plant will tell
6 you this, that's my gross -- my candidate for the gross
7 generalization of the year award, but I'm -- I'm pretty
8 sure that's right -- you have zero confidence in the
9 point estimate, because the probability of getting the
10 point estimate again in another sample is zero. It's
11 not going to happen.
12 So you've got to take the margin of error into
13 account. If you don't you're now saying, "I have zero
14 percent confidence that my point estimate is $640,000."
15 Now, there was also an issue about the use of
16 the data from a stratum that examined 100 percent. That
17 information should not even come to the table into a
18 discussion of sampling. AICPA again makes this very
19 clear.
20 Standard 1.09. "100 percent examination is not
21 a procedure that involves audit sampling." And the
22 Audit Standards Board adds that, quote, "Any items that
23 the auditor has decided to examine one hundred percent
24 are not part of the items subject to sampling." So that
25 census stratum, as it's called, should not have anything
26 to do with any of the calculations that have to do with
27 the sample. That needs to be taken off the table.
28 Finally, we want to bring your attention to a
13
1 crucial standard that has been promulgated by the United
2 States Government Accountability Office, GAO, the
3 pinnacle agency for government auditing, and they have
4 set a nationwide legally recognized standard called
5 relative error -- relative error, to assess the
6 precision of estimates of dollars expended in error.
7 MS. OLSON: Time has expired.
8 MR. TRUITT: Pardon me?
9 MR. WHITLEY: It's 15.
10 MR. HORTON: Time --
11 MR. TRUITT: Okay, I'll --
12 MR. HORTON: -- has expired.
13 MR. TRUITT: -- be very fast.
14 Relative error for GAO needs to be 15 percent
15 or less. The relative error in this case is 42 percent
16 or 178 percent less precise than is necessary.
17 And I'm going to briefly call your attention to
18 Exhibit 1 and I'll wrap this up very fast.
19 In Exhibit 1 please look down the middle
20 column. You'll notice that the confidence levels used
21 in financial tax and recovery audit sampling goes 90,
22 95, 90, 90, 90, 95, 90, 90, 90, 95, 90, 90, 95, et
23 cetera, et cetera.
24 The message there is if you're going to have
25 money on the table you must be confident and you must
26 have good precision. Neither of these has occurred in
27 this case. Unfortunately, I don't think you have useful
28 information upon which to base a decision as yet.
14
1 MR. HORTON: Thank you very much. We will now
2 go to the Department. Would you please introduce
3 yourself for the record. You have 15 minutes to make
4 your presentation.
5 MR. LAMBERT: Good afternoon, Chairman Horton
6 and Members. My name is Scott Lambert. To my right is
7 Kevin Hanks. And to his right is Robert Tucker and
8 we'll be representing staff.
9 Also available is John Calzada, who is the
10 Computer Audit Specialist Supervisor for the Board of
11 Equalization in -- in case it becomes a detailed
12 statistical sample, he's probably a better person to --
13 to answer the questions as he's -- he's considered an
14 expert in -- in our office.
15 Just to --
16 MR. HORTON: Given that the -- that's the
17 premise of the debate I mean you might want to start
18 there. But it's entirely up to you.
19 MR. LAMBERT: I -- I think I'll present the
20 information the way we have it and then we'll -- we'll
21 get to that point. Because I think we have to set a
22 background of -- of where --
23 MR. HORTON: No, no. No, no, please, just --
24 MR. LAMBERT: Yeah, where we're at and I'm --
25 I'm sure John will --
26 MR. HORTON: Commence.
27 MR. LAMBERT: -- get his chance to --
28 MR. HORTON: Thank you.
15
1 MR. LAMBERT: -- talk. So, we basically did a
2 statistical sample and in this particular type of sample
3 we prepared stratums. In other words, we took a look at
4 each individual -- or actually looked at the sample
5 population -- I'm sorry, we took a look at the
6 population as a whole and then divided -- or decided to
7 divide the population into different stratums. And --
8 and there were four different ones. We looked at -- and
9 these are all purchases of various pieces of equipment
10 that the City of Los Angeles is purchasing ranging
11 anywhere from $100 in this particular case to three and
12 a half million dollars.
13 So there's a wide range of purchases here.
14 But, anyway, our -- our first strata is we
15 broke it down between $100 and $700. The second one was
16 $700 to $20,000. The third stratum was $20,000 to
17 $300,000. And then the fourth stratum we took a look at
18 everything above $300,000.
19 And -- SO, one of the things that -- we
20 basically went by the Chapter XIII policy statistical
21 sample in the Field Audit Manual. And what that says is
22 every time you have a stratum you have to look at at
23 least 300 items. And if you're going to project that
24 sample you have to find at least three errors in each
25 one of those samples.
26 If you don't you have to either decide to
27 increase the -- increase your sample or just pick the
28 two items up, include them in the audit on an actual
16
1 basis or maybe just the -- the suppliers from -- from
2 that -- from that one vendor or what not.
3 So, when -- also what you have to do is the
4 variables that come into play for the confidence we use
5 an 80 percent confidence and it has to be plus or minus
6 75 percent.
7 This went through policy when the Chapter XIII
8 was rewritten in the 2000-2001 era, went through the
9 B -- BTC and we came up with those standards.
10 The lower limit was -- was argued during the
11 BTC and it was decided not to use the lower bound.
12 And -- and what I'll say is that of all the states that
13 are doing this there's only four states that use the
14 lower bound. And -- and John later on can kind of
15 explain in statistical terms where the lower bound
16 falls. It -- it's probably easier for him to do that.
17 But in terms of the lower bound the Board
18 Members took a look at that and decided that we're not
19 going to use that. And the -- the entities that
20 supported the staff included the cities, the counties
21 and the Department of Finance. They opposed the use of
22 the lower limit.
23 There's -- there's two types of sampling.
24 And -- and what the Board does is variable sampling.
25 And it's different than attribute sampling. Attribute
26 sampling is basically yes or no. Variable sampling is a
27 wide range and generally what you're going to find is is
28 when you have a wide range of purchases you're going to
17
1 have a larger interval.
2 So, when you -- when you hear an exit poll
3 you're going to be more precise on that. You're not
4 going to have plus or minus 75 percent. That wouldn't
5 be the standard for that. It's going to be a lot more
6 narrow. This is -- we're auditing different type
7 information.
8 So, we -- the -- the Board does attribute
9 sampling. The -- in terms of using that -- that 80
10 percent confidence and the 75 percent interval, that was
11 also argued during the BTC and there was one group that
12 wanted a higher standard.
13 And -- and actually the Department position was
14 they wanted a 90 percent confidence with a 50 percent
15 interval. There was one party that wanted it even
16 higher than that. All the rest of the parties, the
17 interested parties, wanted to maintain it, which was
18 the -- the staff's original position was 80 percent
19 confidence with a 75 percent interval.
20 And so that's what the Board Members decided
21 and that's what the policy and that's what this audit
22 follows, is the 80-75.
23 And what I'll say is that the -- when you take
24 a look at the sample you should include all items in the
25 population. And we -- we followed, which was used as an
26 example, the Ehrens and Lebeck book that's written on
27 stratified sampling, and it basically states that you
28 need to include all items of the population in the
18
1 calculation of your confidence interval.
2 And based on that the Board comes up with a 24
3 percent interval at a 80 percent confidence, which is a
4 fairly tight interval, especially in this type of
5 area -- or especially in this audit.
6 When you take a look at the number -- the --
7 the dollar amount of items selected you have almost 50
8 percent of the purchases were looked at. So, there were
9 $670 million in purchases and I believe we looked at
10 somewhere around $330,000 and -- $330 million in
11 purchases, which is a substantial amount.
12 The whole idea for doing statistical sampling
13 is the fact that we -- if -- if you're able to do this
14 stratified sampling you can look at a lot more of the
15 dollar items. If we would have conducted a block
16 test -- in other words we would have taken one-quarter
17 out of each individual year theoretically we'd have only
18 looked at 25 percent of the purchases. And so, if
19 you -- if we obtained a percentage of error from that
20 and applied it, that type of test has been supported by
21 the Court.
22 In this particular case we've looked at almost
23 50 percent of the dollar value.
24 Also, the Department was willing to lower the
25 level down to $100,000 to take a look at those on an
26 actual basis and that would have include -- that would
27 have increased the amount of dollar purchases that we
28 looked at to 62 percent of the total population, which
19
1 is a significant amount.
2 The -- the Department -- even though taxpayers
3 might want us to look at items on an actual basis or say
4 a 90 percent confidence with a ten percent interval it's
5 not practical for us to do that. It's -- if you did a
6 90 percent confidence with a ten percent interval you --
7 and this is just an estimation, you would probably
8 increase your sample size about seven times. And -- and
9 that's a significant amount of work to theoretically get
10 to the same place that we're at, although admittedly if
11 you pull a bigger sample you're going to be more
12 confident on the number that you come up with.
13 So it really comes down to how much time do you
14 want to expend to get a better number. And what the
15 Board has decided when they approved Chapter XIII is
16 that if you look at -- or have an 80 percent confidence
17 with a 75 percent interval, also with the 300 items and
18 the three errors, which other states don't have and so
19 it's really -- you can't really compare the two -- and
20 there's a lot of differences between what the other
21 states do and what the Board of Equalization does --
22 that this is really the most efficient way of -- of
23 conducting a -- a sample of the taxpayer's records.
24 So, with this I'll leave it to -- to John to --
25 MR. CALZADA: I just want -- good afternoon
26 Chairman Horton and Board Members. I just wanted to say
27 a few things.
28 The confidence interval and the confidence
20
1 level that they're computing in the computed sample at,
2 those factors were not in the design of the test. Our
3 test was designed to meet a criteria of an 80 percent
4 confidence level with no more than a plus-minus 75
5 percent confidence interval. The standard to which
6 we're being held in this example is a much higher
7 standard.
8 In order to achieve that standard at the onset
9 we would have had to have selected up to seven times the
10 amount of sampling units. The formula for correct --
11 for computing the sample size required is somewhat
12 complicated when stratum are involved.
13 The second thing I'd like to say is the fact
14 that there is a huge difference in attribute versus
15 variable sampling. To put it as basic as possible and
16 not to try to make this too elementary, an attribute
17 sample would be how many times do you walk into a room
18 with the light on, or off. It's answered distinctly,
19 yes or no. There's no variability, partially on,
20 partially off.
21 In a world of transactional auditing there
22 could be minimal errors, there could be partial errors,
23 there could be a transaction where there's $100,000 in
24 measure of which $75,000 deals with tangible personal
25 property and than maybe $20,000 deals with an exempt
26 freight portion, and $5,000 with an expedite charge,
27 which would be subject to sales tax.
28 If we selected that as a sample unit and the
21
1 whole $100,000 was selected the error would be the
2 $5,000. That unit of itself has a lot of variability.
3 It's denoted in the $100,000 items but the error,
4 itself, is only $5,000.
5 This is what causes a lot of difficulty with
6 transactional-based variable sampling. Several states
7 that are on the list in front of you, specifically
8 Tennessee, where it indicates a ten percent precision,
9 this is not the case. I am rather friends with
10 Dr. Keith Culp and Rosie McClarkin who runs their
11 program. This is the criteria for designing the sample
12 test. They have no criteria for projecting based on a
13 precision interval.
14 I looked carefully over this and I don't want
15 to get in a "he said"/"she said" point of argument, but
16 I think there's a lot of misinformation being put forth
17 because we are doing variable sampling where the sample
18 unit itself could be $100,000 but the actual error could
19 be a very small portion.
20 And when you do variable sampling there is
21 going to be a wide variance. You can have $100,000 and
22 above ranges with a $5,000 error, which would appear to
23 be odd, and then you could have a $5,000 to $10,000
24 range with a $10,000 error.
25 And on the appearance of it when you compute
26 the confidence interval you're going to have much
27 variability.
28 The reason for computing the confidence
22
1 interval isn't to indicate whether a sample is a good
2 sample or whether it's a bad sample. It's simply to
3 measure the variability of the sample, itself.
4 I'd also like to indicate that over the last 12
5 years I've been involved with the Computer Audit
6 Specialist in the Stat. Sampling program. We've had
7 much input from Will Yancey, Dr. Pfaffenberger, Dr. Culp
8 and a number of individuals designated as experts in
9 transactional based auditing.
10 Dr. Pfaffenberger and I have worked extensively
11 together and he has yet to indicate to me at any level
12 that our standards are under par for transactional-based
13 variable sampling.
14 Thank you.
15 MR. HORTON: Thank you very much. On rebuttal,
16 please.
17 MS. STEEL: Mr. Chairman, can I just ask them
18 that -- to -- I mean, this is getting really complicated
19 about the stats.
20 Can you just give us that you are asking for
21 two things right now. 90 percent confidence level. And
22 second thing is exclude $300,000 -- over $300,000 strata
23 number four.
24 So, can you explain to us about those? You
25 know, not really going into complicated -- the stat. I
26 mean, you know, when you start going in I -- there's
27 just no end. So just to make it a little more simple.
28 MR. TRUITT: I think it would probably be most
23
1 appropriate --
2 MR. HORTON: During --
3 MR. TRUITT: -- for --
4 MR. HORTON: During your --
5 MR. TRUITT: -- the City to answer those
6 questions.
7 MR. HORTON: -- rebuttal, please.
8 MR. WHITLEY: Okay. Go ahead. Just --
9 MR. DEGUCHI: Just on that question, the actual
10 sample over 300,000 that just has to do with calculating
11 how precise the sample was. And they're saying that
12 their 32 percent -- the precision was plus or minus 32
13 percent. We're saying it's plus or minus 71 percent.
14 It's undisputable that the estimate is plus or minus 71
15 percent. We both agree on all the numbers. I don't see
16 what the problem is. We both agree that 639,000 is the
17 point estimate. We both agree that the precision is
18 450,000. So you just divide 450,000 divided by 639,000,
19 it works out to 71 percent.
20 So I don't quite understand what -- how they
21 can say that -- that their sample achieved plus or minus
22 32 percent.
23 If terms of what we're asking for, we're asking
24 for the lower limit of this -- of this wide range. As
25 we indicated the range is from 188,000 to a million.
26 They indicated that some states use the lower -- only a
27 few states use the lower precision limit. Most of them
28 use the midpoint.
24
1 We would have no problems using the midpoint if
2 they had a 90 percent confidence level and a plus or
3 minus ten percent precision. And we would have no
4 problems about paying the -- the midpoint. But when the
5 range is this wide where they have no idea of what the
6 tax really is, we have -- as Mr. Whitley said, we can't
7 accept that.
8 MR. TRUITT: May I, Mr. Chairman, momentarily?
9 MR. HORTON: Yes, please continue and --
10 MR. TRUITT: Did you get your question
11 answered?
12 MS. STEEL: No.
13 MR. TRUITT: I didn't think so.
14 MR. DEGUCHI: Well, I don't know what the
15 question is.
16 MS. STEEL: You know what, just go your
17 rebuttal and then I'll come back with the questions.
18 MR. TRUITT: Okay.
19 MR. HORTON: And try not to argue among you.
20 So --
21 MR. TRUITT: Okay. First of all, the mention
22 of an attribute sample is totally immaterial to this
23 discussion. It was offered merely as an example to
24 demonstrate how a confidence interval is constructed.
25 That should be completely stricken from the record.
26 Also, the percentage of the purposes --
27 purchases or the percentage of the dollars that were
28 looked at is also immaterial. It doesn't matter. It
25
1 didn't figure into the calculation of the sample size,
2 nor does it have anything to do with the extrapolation
3 unless you've done a dollar unit sample, which all
4 indications are you have not done in this case.
5 There was mention of, I think, there in one
6 case being no criteria for the assessment of the sample.
7 That's a violation of the statement on auditing
8 standards number 39 and the AICPA sampling standards
9 because the auditor is required to compare the actual
10 precision to the target precision.
11 That is not a legitimate pre -- procedure at
12 all.
13 In terms of using all of the strata, what this
14 book in fact says is, "Stratified sampling is an
15 extension of variable sampling in which all of the
16 elements in the total population are divided into two or
17 more sub-populations. After the results of the
18 individual parts have been completed," -- notice that
19 they're talking about sampling here, variable sampling.
20 The title of the chapter is "Sampling". Okay.
21 "After the results and the parts have been
22 completed they're combin -- combined." That's right.
23 That's only from the sample. You cannot use a census
24 stratum in this calculation. It is illegitimate to do
25 that.
26 As far as the efficiency issue goes, whether or
27 not sampling is efficient, I certainly would have to see
28 the calculations to support the assertion of a seven
26
1 times greater sample size. There are many other things
2 I'd like to see with respect to how the sample size was
3 calculated, but those are relatively -- I guess
4 relatively immaterial here. I would posit that one of
5 the main reasons that your sampling is so inefficient is
6 because you're inadequately stratifying. It's because
7 you have so much variance in your strata. You have
8 one -- was the last -- or the -- what's stratum 3 --
9 what -- what was that, 300,000 -- that's okay. Every --
10 there's a case in there I think where you had $100,000
11 or maybe 10,000 to 300,000. That's an enormous dollar
12 range of values to look at. So I think you might want
13 to revisit the efficiency of your stratification
14 process.
15 90-10 is practical. 90 percent confidence; 10
16 percent precision is practical. I also point you back
17 to Exhibit 1 momentarily. The agencies that are listed
18 here, the great majority of these agencies are doing
19 recovery audits. Their objective in the audits is to
20 recover dollars. That's what they do.
21 And the great majority of the Federal agencies
22 here use the lower limit. And one reason for that is to
23 avoid exactly these kinds of discussions. It makes
24 things very simple.
25 It's also equitable to the person who has made
26 the improper payments. It doesn't shift an
27 inappropriate part of the compliance burden to the
28 person who is owing the tax or, you know, is responsible
27
1 for the improper payments.
2 MS. OLSON: Time has expired.
3 MR. TRUITT: Thank you.
4 MR. HORTON: Sir, you -- would you -- I'm sure
5 there will be several questions.
6 Discussion, Members?
7 MS. STEEL: Just question.
8 MR. HORTON: Member Steel.
9 MS. STEEL: You -- you were asking 90 percent
10 confidence level then it's going to come out to
11 confidence interval is going to be from 57 -- 24 percent
12 to 57 percent. So, that number is coming higher. So,
13 what amount that you are asking us to figure it out at
14 90 percent confidence level because the bigger number of
15 a confidence interval is actually not really good --
16 MR. TRUITT: I understand what you're saying.
17 MS. STEEL: -- at this point. So the 90
18 percent is not really good. So how -- what you are
19 asking at this -- because Mr. Deguchi just said there's
20 four strata and he said if we're going to choose one
21 strata then he's going to be happy with it, if I heard
22 it right. Maybe I'm wrong.
23 So can you just explain to us that what amount
24 that we were talking about here?
25 MR. DEGUCHI: Well, again, the initial
26 assessment was 1.4 million.
27 MS. STEEL: Right.
28 MR. DEGUCHI: And --
28
1 MS. STEEL: That's the 80 percent.
2 MR. DEGUCHI: That's when they first started
3 the sample, they -- they came up with the 1.4 million
4 based on their sample. Of that amount about 763 were
5 actual errors found in a population where they looked at
6 100 percent of the items. So we obviously have no issue
7 about paying that money back.
8 So if you take the 1.4 million minus the
9 763,000 that equals 639,000. So that's the amount
10 that's in dispute.
11 To get that 639,000 they only found $60,000 in
12 actual errors. So they found $60,000 in errors and they
13 said, well, we want you to pay us back 639,000.
14 So the 639,000 is, again, their best estimate,
15 which is not really an estimate at all because of the
16 wide precision and the -- and the low confidence.
17 MS. STEEL: So what you are saying is BOE
18 method is not really reliable, so you are asking us to
19 give you the lower number?
20 MR. DEGUCHI: Just as Mr. Whitley said --
21 MS. STEEL: That's the bottom line at this
22 point?
23 MR. DEGUCHI: -- it's kind of an arbitrary --
24 it's kind of an arbitrary figure because of -- because
25 it is so wide of a precision. That's --
26 MS. STEEL: So you are asking for the lower
27 number?
28 MR. DEGUCHI: Well, we're asking for -- that
29
1 the 649,000 be reduced to 188,000. Because the range
2 that they're saying that they're 80 percent confident
3 that the -- that the amount is between 188,000 and a
4 million dollars.
5 And, again, we do this all the time in our
6 auditing. We -- we know what the lower limit if -- if
7 it's a real wide precision.
8 MS. STEEL: So, yes -- yes --
9 MR. DEGUCHI: Yes.
10 MS. STEEL: -- you are asking for the lower
11 number.
12 Mr. Lambert --
13 MR. LAMBERT: Yes.
14 MS. STEEL: -- so what's wrong with that
15 figure, because 80 percent of confidence level is not
16 really reliable right now, even our manual says that 80
17 percent is good enough. So why we are arguing here?
18 MR. LAMBERT: Why we argue -- well, why aren't
19 we agreeing with them, is that --
20 MS. STEEL: Yeah. Because -- yeah, they say
21 it's not really reliable. I see that it's not really
22 reliable.
23 MR. LAMBERT: Yeah.
24 MS. STEEL: But --
25 MR. LAMBERT: Well, let -- maybe I'll -- I'll
26 just try and explain it because I know it's com -- it's
27 complicated as soon as they start doing numbers. But
28 it --
30
1 MS. STEEL: Oh, I did Stat for a couple of
2 years ago so I -- I know exactly --
3 MR. LAMBERT: Yeah, me --
4 MS. STEEL: -- what you guys are --
5 MR. LAMBERT: I know, but I -- it gets -- it
6 does get confusing --
7 MS. STEEL: Right.
8 MR. LAMBERT: -- you know, when you start
9 mixing numbers.
10 But if you just look at it like doing a block
11 sample, right -- if you did a block sample and you
12 looked at three months you would take the errors and
13 divide it by the base and come up with a percentage of
14 error, right? And then you would apply that percentage
15 of error to whatever is in the population, right? And
16 that's the way that we've handled it all the time.
17 In effect, that's the same thing that we're
18 doing here. What we do is we break out the different
19 stratifications and then -- and I'll use an example,
20 stratum 2. There was 3412 items. We decided to look at
21 ten percent of those items, which was 350 purchases.
22 So, what it -- it came out was a three percent
23 of error. So, in other words, we took the 350 items, we
24 found out -- I believe there was -- I'm going to say
25 eight errors -- I believe that there was eight -- and we
26 came up with a three percent of error.
27 MS. STEEL: Right.
28 MR. LAMBERT: And so what we did is we applied
31
1 that three percent to all the population in that
2 stratum.
3 MS. STEEL: Okay.
4 MR. LAMBERT: Right?
5 MS. STEEL: Let me --
6 MR. LAMBERT: And --
7 MS. STEEL: You know what?
8 MR. LAMBERT: -- and so when -- yeah.
9 MS. STEEL: That's --
10 MR. LAMBERT: Maybe I can do one --
11 MS. STEEL: Go for it.
12 MR. LAMBERT: -- just one more -- yeah.
13 MS. STEEL: Go for it. Okay, sorry.
14 MR. LAMBERT: So, when we did the 80 percent
15 basically what that does is it -- it's just telling us
16 how many items that we're selecting for -- for sample.
17 MS. STEEL: Got that.
18 MR. LAMBERT: And then based on that when we
19 get those items and we -- we calculate it, it's going to
20 tell us what range it falls in so we're -- you know, 80
21 percent confident it falls in this range.
22 MS. STEEL: If --
23 MR. LAMBERT: If --
24 MS. STEEL: If we did 90 percent --
25 MR. LAMBERT: You would --
26 MS. STEEL: -- then what amount comes out?
27 MR. LAMBERT: Well, if we did the -- if we --
28 see, originally what we -- if you're going to do 90
32
1 percent --
2 MS. STEEL: Right.
3 MR. LAMBERT: -- you should have picked a
4 larger sample.
5 MS. STEEL: Right. Right.
6 MR. LAMBERT: Right? And we didn't do that
7 because we're following the Board's audit method. This
8 is the approved method that industry wanted --
9 MS. STEEL: But it was ten years ago, though.
10 MR. LAMBERT: Well, it was -- it was ten years
11 ago. But I think if you go back to industry and ask
12 them the reason why they don't want it more is because
13 it's intrusive. When you're asking for -- say you
14 wanted to double or triple the sample to come up with a
15 -- a more confident sample you're looking at three times
16 more invoices, which means the taxpayer's pulling more
17 items, the -- the auditor is looking at -- at more items
18 and -- and industry -- in fact all the interested
19 parties there was only one party that wanted it higher.
20 Every other one wanted it exactly the way that we're
21 doing it right here.
22 And so, when you start taking a look at, oh, we
23 want this lower limit because that's really what you're
24 looking at is they basically want to be 99 percent
25 confident that they're not overpaying their tax.
26 That -- that's what -- that's what the argument is, is I
27 want to be 90 percent confident and, you know,
28 the Department of Finance just came back and said, you
33
1 know what, we think it should be the middle. We're --
2 that the taxpayer and the State should take equal risk
3 that it's either higher or lower. And that's why we're
4 going with the midpoint. And that's why we use a lower
5 sample because we do the most efficient way of -- of
6 doing an audit.
7 MR. HANKS: So even outside of the statistical
8 samples that we're talking about today we used the
9 midpoint in calculating any errors, if we're doing block
10 samples, so that the calculation of -- of the error
11 factors and the application of those error factors to
12 reported amounts is identical in -- in a block sample as
13 it is with the statistical sample. We're using a
14 midpoint.
15 So we're -- we're not disadvantaging anyone.
16 When we calculate a -- a factor of error of ten percent
17 in a block sample we can't say we're plus or minus any
18 factor. We're going to use the ten percent.
19 MR. HORTON: Okay.
20 MR. TRUITT: Would it be possible to make two
21 very brief -- very brief comments?
22 MR. HORTON: It's up to Mrs. Steel.
23 MS. STEEL: Yeah.
24 MR. TRUITT: Okay. One of the reasons that the
25 sample sizes would be large is because they are actually
26 sampling on the wrong mathematical distribution. Since
27 the error rates are so low, one percent, three percent,
28 et cetera, et cetera, they should be using -- not the
34
1 normal curve, they should be using either the Poisson
2 distribution or the binomial distribution.
3 MR. HORTON: Oh.
4 MR. TRUITT: So they're say -- they're using
5 the wrong mathematical distribution to calculate their
6 sample sizes and they should be doing dollar unit
7 samples. It would be much more efficient, they would
8 get much better coverage of the dollars.
9 MS. STEEL: So you are asking 90 percent
10 confidence level with excluding three -- over 300,000
11 strata number 4. That's what you are asking at this
12 point?
13 MR. TRUITT: Oh, yes.
14 MS. STEEL: That number comes out to the
15 lower.
16 MR. TRUITT: Correct.
17 MS. STEEL: That's --
18 MR. TRUITT: That stratum must be excluded
19 because it wasn't part of the sample. It was looked at
20 completely.
21 MS. STEEL: But it -- it was looked at
22 completely so that was the right number?
23 MR. TRUITT: Exactly. But you don't use that
24 number in calculating the precision for the sample. It
25 has nothing to do with the sample.
26 UNIDENTIFIED SPEAKER: Ms. Steel --
27 MR. TRUITT: We're -- added all together it has
28 no margin of error. It should not figure into any
35
1 calculation related to the sample.
2 MR. HORTON: You know, that --
3 MR. TRUITT: We're -- we're --
4 MR. HORTON: -- seems to argue against your
5 point but I'm not defending your case, but I certainly
6 would want to caution you that discrepancy in your
7 arguments are working against you.
8 But, anyway, the other thing is it sort of
9 reminds me of -- of when I took my first Statistics
10 class, just I think a little levity here, and that I was
11 confused as to why I was in the class until the teacher
12 asked, he said, what is the probability of a die. And
13 then from that point on I fell in love with statistics.
14 And the one thing that I realized that is the closer you
15 get to an actual test the larger the universe. And so
16 the larger the universe of your sample. And that all of
17 these things are interrelated.
18 And so it makes me sort of ponder who -- who
19 agreed to this test from the onset because you could
20 have actually started off from the premise and said we
21 disagree with the test and we want a 90 percent
22 interval, and the Department could have said, yes, of
23 course, but in order to accomplish that these other
24 variables have to be this, thus and that.
25 And then the City could have said, well, yes or
26 no, and we wouldn't be here today.
27 But at any rate --
28 MS. STEEL: Okay, I have just one more
36
1 question --
2 MR. HORTON: Ms. Steel.
3 MS. STEEL: -- to the Department. So when we
4 do confidence level to 90 percent then we have to have
5 much more sample. So why don't we do 80 percent that we
6 already have and exclude strata number 4, that over
7 300,000, that what amount comes out?
8 MR. HANKS: In the evaluation?
9 MS. STEEL: Yeah.
10 MR. CALZADA: We had spent quite a bit of time
11 dealing with recognized experts, Dr. Don Roberts, Ph.D.
12 from Stanford in Statistics and the authority with the
13 Internal Revenue Service, and also Drs. Roger
14 Pfaffenberger and Dr. Yancey and Dr. Culp and Dr.
15 Cavaney, (phonetic) and our methodology that we're using
16 is based upon authoritative sources.
17 Variable sampling creates this variability in
18 results simply from the fact that it's not a distinctive
19 yes or no answer. It's not entirely correct or taxable
20 or entirely non-taxable.
21 So the very nature of variable sampling to --
22 to have this high standard requires a massive amount of
23 looking at the population in order to have that
24 confidence interval brought down.
25 And we have -- I can't speak on a tax issue but
26 I think Kevin can -- can talk about what was offered.
27 MS. STEEL: I'm asking just number, that 80
28 percent of confidence level --
37
1 MR. CALZADA: We have -- yeah --
2 MS. STEEL: -- excluding strata 4 over $300,000
3 then what amount comes out? It's really simple
4 question.
5 MR. CALZADA: Yes. It -- we've analyzed that
6 using their confidence interval computation. It still
7 meets our criteria at an 80 percent. It -- it's very
8 high, it's like the 60s.
9 But it meets the confidence interval
10 achieved is 61 percent.
11 MS. STEEL: The amount, itself. I'm talking
12 about tax -- taxable amount that how much it comes out.
13 MR. LAMBERT: I -- I'll try. It depends --
14 it's coming out the same. The only difference is if you
15 want to use the lower limit.
16 MS. YEE: Right.
17 MR. LAMBERT: That's the only thing that that
18 has to do with. So the -- the average -- the mean is --
19 the midpoint is coming out --
20 MS. STEEL: It's the same. So we are using
21 midpoints so it's --
22 MR. LAMBERT: Right.
23 MS. STEEL: -- going to come out same --
24 MR. LAMBERT: So if you --
25 MS. STEEL: -- even we don't use -- we exclude
26 strata number 4? So like $300,000 ones because it was a
27 hundred percent that completely that we did. So
28 excluding that so we just use from stratum 1 to 3 --
38
1 MR. LAMBERT: Yeah.
2 MS. STEEL: -- with 80 percent confidence level
3 then amount supposed to be changed. You cannot have
4 same amount when it comes with exclude and include, that
5 has to be two different numbers.
6 Do you have those numbers here?
7 MR. DEGUCHI: Yes, we do. The number would go
8 up from 188 to -- to 288 and it would be --
9 MS. STEEL: That's the lower -- lower. But
10 where is the midpoint? That's what I'm asking.
11 MR. DEGUCHI: The mid -- the midpoint will
12 never change on this based on the sample. So it's --
13 MS. STEEL: So midpoint is same?
14 MR. DEGUCHI: Yes.
15 MS. STEEL: So lower level is going to go up?
16 MR. DEGUCHI: Right. At -- because you're only
17 80 percent confident --
18 MS. STEEL: Right.
19 MR. DEGUCHI: -- which is going to make the --
20 the --
21 MS. STEEL: So how much to how much?
22 MR. DEGUCHI: So again it would go up from 188
23 to 288, and we would be willing to accept that amount.
24 MS. STEEL: So 288 is a lower?
25 MR. DEGUCHI: Yes.
26 MS. STEEL: How much is the higher?
27 MR. DEGUCHI: The higher is -- like what's
28 (inaudible) --
39
1 MR. CALZADA: That's 639,000 plus or minus
2 350,000. So --
3 MS. STEEL: Okay. Instead of 450,000 then?
4 MR. CALZADA: Yes.
5 MS. STEEL: Okay. 300 how much?
6 MR. DEGUCHI: 639,000 plus or minus 350,000
7 MS. STEEL: 350,000.
8 MR. DEGUCHI: Just kind of rounding. I mean
9 it's 358 -- 350,857.
10 MR. LEVINE: That sounds like 900,000 to me.
11 MS. YEE: Yup.
12 MS. STEEL: Thank you.
13 MR. LEVINE: The plus.
14 MS. YEE: Exactly.
15 MR. HORTON: Just for clarification, is -- is
16 the testimony on the table that the Board used the
17 medium rate?
18 MR. LEVINE: Can I comment --
19 MR. LAMBERT: The midpoint --
20 MR. HORTON: One second, Mr. Lambert.
21 MR. LEVINE: -- in standard terms?
22 MR. HORTON: Yeah, I -- one second.
23 MR. LAMBERT: We used -- we used the midpoint.
24 MR. HORTON: Midpoint.
25 MR. LAMBERT: In each one of the stratums.
26 MS. YEE: All right.
27 MR. HORTON: Oh, God.
28 MS. YEE: Right.
40
1 MS. STEEL: Can I ask just one more? What's
2 the lower bound?
3 MR. LAMBERT: What's the lower bound?
4 MS. STEEL: Uh-huh.
5 MR. LAMBERT: That -- well, that's the -- the
6 standard deviation that you -- you have from the -- the
7 mean and -- and we're coming up with different
8 calculations just based on how you calculate it.
9 And -- but really in -- in any case the way we
10 apply this is using the midpoint. And if you're using
11 something else than that you're basically discounting
12 the audit, and that's why the Department of Finance
13 didn't like it and -- and I'll go one step further,
14 that's why the cities didn't like it. Because the
15 cities get less revenue when this is incorporated.
16 And in fact in this particular case our
17 calculations show that the city -- if -- if we change
18 the policy and allow the lower bound and it was 50
19 percent on average lower that the City of L.A. would
20 receive $500,000 less every year in tax revenue.
21 If everything else stayed constant.
22 MS. STEEL: Okay.
23 MR. HORTON: And he's arguing against that.
24 Mr. Levine.
25 MR. LEVINE: Yeah, I was -- I think we're
26 getting a lot of technical stuff that's above my head.
27 I did ace my statistics class 30 years ago. But this
28 midpoint I think is a misnomer. It's -- may be the
41
1 underlying concept behind what we do, but we don't do
2 any points. We do this to evaluate the test, and if the
3 test evaluates we take the error and we multiply it. We
4 project it.
5 And that happens to be the -- if I get this
6 correctly, the midpoint. But we don't calculate a
7 midpoint because we take the dollar error from the test
8 that has evaluated well, and we multiply it.
9 MR. HORTON: Yeah, that --
10 MR. LEVINE: That's it. So that's all -- and
11 that's why --
12 MR. HORTON: Yeah.
13 MR. LEVINE: -- the actual part of the test can
14 never affect the -- the projected parts of the test
15 because we've just done a straight line projection --
16 MS. YEE: Right.
17 MR. LEVINE: -- because the test evaluated.
18 The solution to a bad test is not to take the midpoint.
19 I've never heard that done at the Board. The solution
20 is throw out the test.
21 MR. HORTON: Okay.
22 MR. LEVINE: Start over.
23 MR. RUNNER: I have a question.
24 MR. HORTON: I think the Department is pretty
25 clear in that regard.
26 Mr. Runner.
27 MR. RUNNER: Yeah, real quick. First, for the
28 record, I did not ace my statistics class. In fact, I
42
1 do have this reoccurring dream that I still have. And
2 that is I find myself sitting in my Statistics class
3 totally lost. And that just came back again to me.
4 (Multiple voices.)
5 MR. RUNNER: But, yeah. But let me ask you
6 this, and this goes to the issue of the sample. And --
7 and just, again, I'm just again trying to see what --
8 if -- if I understand at least kind of the -- trying to
9 see if we came up -- come up with something that does
10 make sense when you're trying to come up with a
11 statistical average here.
12 Okay. We've already established that the above
13 300,000 -- it was actual numbers so there's no debate in
14 that discussion, okay.
15 So the second -- that second level, the 20,000
16 to the 200,000 -- is that right? Yeah. Or up to 3 --
17 299, up to 300,000 the sample at that point was -- it
18 was just over ten percent of that particular strata,
19 right?
20 MR. LAMBERT: That's correct.
21 MR. RUNNER: It's a pretty hefty strata, a
22 sample of that particular strata, right?
23 I mean, it seems to me that's a -- that's a --
24 that's -- an end, and just looking at the numbers of
25 the -- of the numbers that's like -- what, about 40
26 percent or so of the -- of the -- of the -- of the total
27 dollar population outside of the -- outside the -- the
28 upper strata?
43
1 MR. LAMBERT: I think it would be ten -- ten
2 percent of that strata. We're looking at ten percent of
3 the invoices and the total of those --
4 MR. RUNNER: The dollar value?
5 MR. LAMBERT: The dollar value of those 350 --
6 MR. RUNNER: Of that population?
7 MR. LAMBERT: -- is ten percent of that total
8 stratum.
9 MR. RUNNER: Right, but I'm just talking about
10 the total of that stratum -- strata in terms of the
11 dollar value -- because maybe I'm missing that.
12 MR. LAMBERT: It's 2 -- well, the total is 223
13 million --
14 MR. RUNNER: Uh-huh.
15 MR. LAMBERT: -- is the total.
16 MR. RUNNER: Right.
17 MR. LAMBERT: And we looked at 23 million --
18 MR. RUNNER: So -- right.
19 MR. LAMBERT: -- as a sample.
20 MR. RUNNER: But -- but my point is $223
21 million in that strata --
22 MR. LAMBERT: Yes.
23 MR. RUNNER: -- and -- and the -- and if you
24 take -- take the other -- total of the other three
25 stratas -- or the other two stratas, right -- or I guess
26 there's -- one, two, three stratas.
27 MR. LAMBERT: We've projected --
28 MR. RUNNER: There's four -- there's two.
44
1 There's --
2 MS. STEEL: Four.
3 MR. RUNNER: -- four stratas -- the other two
4 stratas, the total is close to 130 million?
5 MR. LAMBERT: Yes.
6 MR. RUNNER: Okay.
7 MR. LAMBERT: 138.
8 MR. RUNNER: That's -- that's what I -- when I
9 said 40 percent I meant of the total population that --
10 in terms of total dollars.
11 MR. LAMBERT: Uh-huh.
12 MR. RUNNER: So, I'm -- what I'm saying is
13 of -- of the -- of what our test was, we tested ten
14 percent of -- of what is -- in terms of the dollar
15 population of about 225 --
16 MR. LAMBERT: Yeah.
17 MR. RUNNER: -- well, okay, 225 million, right
18 of that -- of that total.
19 MR. LAMBERT: Uh-huh.
20 MR. RUNNER: Of ten percent of that.
21 MR. LAMBERT: Correct.
22 MR. RUNNER: And so it seems to me that's a
23 pretty healthy chunk of that particular group.
24 Now, my question I have is tell me about
25 whether or not that's -- having a 20,000 to a 300,000
26 strata is problematic. Just in terms of its sheer
27 size -- in terms of its variation.
28 MR. LAMBERT: We evaluated it based on the
45
1 results of it.
2 MR. RUNNER: Uh-huh.
3 MR. LAMBERT: And -- and it came out within our
4 parameters.
5 MR. RUNNER: Okay.
6 MR. LAMBERT: So -- and -- and I'll --
7 MR. RUNNER: So it was considered?
8 MR. LAMBERT: Pardon me?
9 MR. RUNNER: It was considered whether or not
10 that was to -- for instance, my thought is for a -- I
11 mean why -- the actual is -- if you took the actuals
12 down to the 100,000 and took it -- you know -- you know,
13 you used -- and 3 -- and did the actuals starting at
14 100,000 on up then you shrunk that -- that -- that
15 variable plus you got a much cleaner number -- bigger
16 number that you were actually doing the actuals on.
17 MR. LAMBERT: Right. And I believe the number
18 in that -- that range between 100 to $300,000 that
19 you're talking about --
20 MR. RUNNER: Uh-huh.
21 MR. LAMBERT: --- there was 609 -- if I -- if I
22 get it right.
23 MR. RUNNER: Uh-huh, yeah, it's real close to
24 that.
25 MR. LAMBERT: -- 609 sample items in there.
26 And we did offer to take a look at those on an actual
27 basis.
28 But, you know, the one thing you do take a
46
1 risk -- well, I guess the taxpayer takes a risk when
2 they do that because there is about $100 million in
3 purchases in those 6 -- 609 items. And when you -- when
4 we do something like that it's basically you've made the
5 decision, we're going forward with it. If it's good for
6 you, great. If it's not that's the way it is. So --
7 MR. RUNNER: Okay. Thank you.
8 MR. HORTON: Member Yee.
9 MS. YEE: Thank -- thank you very much,
10 Mr. Chairman.
11 Let's see, where do I start?
12 I guess first I -- I wanted to just establish
13 how the Board approached this. We have guidelines that
14 have been adopted that are reflected in Chapter XIII of
15 the audit manual. Yes, they're ten years old but they
16 are guidelines and that's what we're using.
17 I think my first concern with looking at a
18 different approach as suggested by the Petitioners that
19 we are going to look at disparate treatment of taxpayers
20 in terms of how we're approaching this type of sampling,
21 so let me just say that at the out -- outset. I think
22 the impression I'm left with from this discussion is
23 that there are probably a variety of methods that can be
24 used when we look at evaluating a stratified random
25 sample.
26 I'm troubled by a couple of things that the
27 Petitioners have said and I -- let me just start with
28 seeing if I can dispense with the first concern about
47
1 the treatment of strata 4. You keep calling it a
2 census. And -- but it's not a hundred percent of the
3 population.
4 MR. TRUITT: It is in that stratum.
5 MS. YEE: Okay. But that's not what the census
6 generally is -- that's not generally called the census.
7 MR. TRUITT: When you do stratified sampling
8 you are dividing the population up into sub-populations,
9 four separate groups.
10 MS. YEE: It's a hundred percent of the
11 sub-population --
12 MR. TRUITT: Exactly.
13 MS. YEE: -- I agree with that.
14 MR. TRUITT: But they look -- they pulled a
15 sample, pulled a sample, pulled a sample, looked at
16 everything in stratum 4. It was not part of the sample.
17 They looked at everything.
18 MS. YEE: Well -- and we had everything to look
19 at from the City in that strata.
20 MR. TRUITT: So by looking --
21 MS. YEE: -- strata.
22 MR. TRUITT: -- at everything the results of
23 this stratum should not figure into calculating the
24 precision. All you can do with the dollars from this
25 stratum that you looked at everything is simply add it
26 to whatever you find in error. That's all you can do.
27 MS. YEE: Okay. Let me ask the Department, if
28 we were -- I'm sure you've looked at this because we've
48
1 posed the question, if we were to take away strata 4 --
2 stratum 4, does it still fall within the parameters of
3 the interval?
4 MR. LAMBERT: It does.
5 MS. YEE: Okay. All right. Let me also ask
6 the Department, the statement about if we were to
7 consider the Petitioner's approach and the additional
8 workload that that would encompass, can you speak to us
9 about how you get to seven times more transactions? I
10 mean, what's --
11 MR. CALZADA: The formula for computing a
12 sample size is dependent upon the confidence interval
13 and the confidence level, what you plan to achieve.
14 The reason I said that is I did not have the
15 time to take out and compute each one separately. There
16 are generalized tables and formulas for achieving a
17 confidence -- with a given confidence level and inter --
18 and determined interval of how many units you'll need to
19 do. But that's on a very simple sample with no
20 stratification.
21 Once you stratify the formulas change. And I
22 did the quick cursory count and that's where I came up
23 with the number.
24 MS. YEE: All right.
25 MR. CALZADA: But even if you were to use a
26 simple formula I do believe it would be in excess of
27 three times the sample size.
28 MS. YEE: Okay. Petitioners agree?
49
1 MR. DEGUCHI: Do we agree that it's going to be
2 three to seven times as large?
3 MS. YEE: Uh-huh.
4 MR. DEGUCHI: It's very difficult to determine
5 because it's going to depend on as they take more items,
6 if they find more errors --
7 MS. YEE: But it will be larger?
8 MR. DEGUCHI: Oh, yes.
9 MS. YEE: Okay.
10 MR. DEGUCHI: Most definitely.
11 I would estimate it's going to be at least --
12 it would be at least five times as large.
13 MS. YEE: The City can afford the time and the
14 resources to devote to that?
15 MR. DEGUCHI: Yes, we've -- we've checked with
16 our people in Accounting and they -- they have agreed to
17 make the documents available.
18 MS. YEE: Wow, you're much better off than the
19 Board is, resource-wise.
20 MR. WHITLEY: We -- we need to know.
21 MS. YEE: Okay. I mean, I think we're both
22 after the same goal, and -- and -- and this is really
23 about, you know, kind of what's -- what's the best
24 approach. We have guidelines that we've been using.
25 I'm not -- I think it would be inappropriate to change
26 our approach at this time because it would be treating
27 you unlike we've treated other taxpayers that we've put
28 in this situation.
50
1 We also have to balance, obviously, the
2 resources that we have available to go in and do these
3 audits utilizing this method.
4 I think we're all after the correct measure of
5 tax and so I don't -- I mean, I'm not open to another
6 test but the City of Los Angeles appears to be open to
7 another test --
8 MR. DEGUCHI: Correct.
9 MS. YEE: -- if your approach is embraced by
10 this Board.
11 MR. DEGUCHI: Well, let me clarify it, that
12 test is going to have to be a 90 percent confidence
13 level and plus or minus ten percent precision. So --
14 that's what we're talking about in terms of another --
15 MS. YEE: Okay, yeah, which -- yeah.
16 MR. WHITLEY: It requires a larger sample.
17 MS. STEEL: But --
18 MS. YEE: Well, requires a larger sample and
19 it -- and it requires us to move away from our already
20 established guidelines.
21 MR. RUNNER: I'm -- I'm sorry --
22 MS. MANDEL: No, no.
23 MR. RUNNER: I was going to say, it -- it would
24 but it would -- but it would mean that the taxpayer then
25 on their dime is coming up with -- with the answer at
26 that point. So, yeah, it's outside of our norm but
27 they're coming back with what they feel more confident
28 in and -- and they're paying for it.
51
1 MR. LEVINE: Well, so are we.
2 MR. RUNNER: That --
3 MR. LEVINE: Which is part of the reason why
4 this was adopted because it takes a lot more of our
5 staff time, meaning less audits and less money back to
6 the City on the local tax portion.
7 MR. RUNNER: Well, let me ask you, when you say
8 "so are we," I mean I -- help me understand that a
9 little bit. Because I mean we are -- we are not running
10 this -- we are not doing that -- that at that point,
11 right? We are just reviewing -- aren't we going to be
12 just reviewing what it is that -- they're --
13 MR. LEVINE: Correct. Presumably they'll pull
14 everything and I have no idea how long it takes the
15 Department, but the Department has to review each one
16 and depending on the nature of the errors we're
17 examining, sometimes it's quick, just look at the
18 documents, and sometimes it's research.
19 But it takes time. If you're reviewing a
20 thousand rather than 300, that's more than three times
21 as long to do the review of that portion -- that portion
22 of the audit. That's considerable.
23 MS. YEE: Yeah.
24 MR. LEVINE: And my understanding of
25 statistical theory it's as -- as likely that a further
26 review will show more taxes due --
27 MS. YEE: Right.
28 MR. LEVINE: -- as it will show less taxes due.
52
1 MR. RUNNER: Less taxes.
2 MS. YEE: Uh-huh.
3 MS. STEEL: Sure.
4 MR. HORTON: One -- one second. Members -- I
5 believe Ms Yee has the --
6 MS. YEE: Yeah. I mean, just --
7 MR. HORTON: And then we'll come back to you,
8 Ms. Steel.
9 MS. YEE: I really feel strongly about the
10 point that we are going to be treating this particular
11 taxpayer differently. But that's not to say I'm not
12 open to revisiting the guidelines that we currently have
13 in place through another formal process. Maybe it's
14 time to do that, ten years later.
15 But I think given what Petitioner's seeking and
16 the guidelines that we have in place I'm not so sure
17 that the additional expenditure of resources, time
18 and -- human and otherwise, is actually going to get us
19 any less comfortable with where we are currently for the
20 Depart -- for the Board and the Department.
21 And for the City, I agree with Mr. Levine. I
22 think once you open that up you probably are going to be
23 on the hook for more tax --
24 MS. STEEL: Uh-huh.
25 MS. YEE: -- than not. So it's kind one of
26 those be careful what you ask for.
27 MR. HORTON: Member Steel.
28 MS. STEEL: I have a solution. We --
53
1 MR. RUNNER: Okay.
2 MS. STEEL: We --
3 MS. YEE: A 30-30-30?
4 MS. STEEL: Even that manual -- no. No, no,
5 no, no. Even that manual is over ten years old, still
6 manual said that 80 percent compton -- confidence in --
7 level, so we already have that number. So we just
8 exclude strata 4 over the highest one, $300,000 over
9 because that was a hundred percent already done. So why
10 don't we exclude that, you know, we just put strata 1
11 through 3 and then we can ask them to pay for that.
12 MR. RUNNER: I --
13 MS. YEE: Are you --
14 MS. MANDEL: They're not paying for the actual
15 errors in --
16 MS. STEEL: Just exclude only strata 4 because
17 that's the -- strata 4 is the one that it's not samples,
18 but it's hundred percent calculation were there.
19 MS. YEE: And -- and what are you suggesting
20 for stratas 1 through 3?
21 MS. STEEL: Yeah, 1 -- 1 through 3. 1 -- 1 to
22 3.
23 MS. YEE: But it doesn't matter.
24 MR. LEVINE: Yeah, I think -- I think that the
25 complicated discussion confused me, and I think, to be
26 honest, you're in the same position.
27 The -- taking out -- that relates to evaluating
28 the test, not change for -- the way we do it, not
54
1 changing the bottom line number. When you take out the
2 actual basis portion, stratum 4, it -- if that was
3 included in the evaluation then the evaluation will
4 change. But then you're still looking at the remaining
5 samples 1 through 3 strata and deciding if it comes
6 within our parameters or not. And if so, it's good to
7 go. And if not you reject it.
8 And here what the Department said is it comes
9 within our parameters whether 4 is included in the
10 analysis or not and therefore it doesn't -- doesn't
11 matter, because either way under your current guidelines
12 it just --
13 MS. STEEL: Mr. Levine, can I ask that our
14 expert --
15 MR. CALZADA: Yes.
16 MS. STEEL: -- about my idea? Solution.
17 MR. CALZADA: Well, I just want to confirm that
18 the -- removing census stratum 4 and computing --
19 MS. YEE: Don't call it a census.
20 MR. CALZADA: Okay, computing stratum 4, the
21 interval, it meets our criteria.
22 If we were to remove 4 it would have to be
23 added to the interval. But the manner in which we
24 project we use the addition of -- of the determined
25 ratio, but we use a ratio, we find the total errors, and
26 in this case we would have the errors for the sam --
27 sample plus the actual errors and we would be at the
28 same point.
55
1 The only difference would be is if somehow we
2 assessed at the lower bound. The lower bound would then
3 result in a net -- a tighter interval, a less amount
4 of -- of tax and we would then add the census -- or, I'm
5 sorry, the actual amounts. And those two numbers would
6 differ. I don't know by how much -- how much they would
7 differ by, but it would only apply if we were projecting
8 at the lower bound. The mid -- the midpoint or -- or
9 the ratio would still be the same.
10 MR. HORTON: Okay. Further discussion,
11 Members? Did the City of --
12 MR. RUNNER: Please, no.
13 MR. TRUITT: Are you sure?
14 MR. HORTON: Did the City of Los Angeles agree
15 to the test from the onset?
16 MR. DEGUCHI: It's my understanding that we
17 signed a document. I wasn't involved because again the
18 Accounting Division -- I only got involved because they
19 -- because they wanted a statistical expert to become
20 involved. I'm not sure what exactly that document said
21 in terms of whether it was specific in terms of the --
22 the confidence levels and the intervals. MR. HORTON:
23 You know, it seems to me that the -- the current
24 requirement is acceptable even ten years later to
25 determine whether or not we are -- the test, itself, is
26 reasonable. I mean, it -- it sort of leans in the
27 taxpayer's favor, and that if there's less than three
28 errors we throw it out. If it's between this margin we
56
1 throw it out. Meaning that we identify that errors
2 exist but yet still we don't -- we say that your
3 liability is zero, and pick it up on an actual basis.
4 And if -- however, if a taxpayer wants to
5 conduct an audit -- I mean a test on an actual basis,
6 far be it for me to say no, you can't do that, to
7 determine their liability. But that doesn't change the
8 criteria in which we -- we accept the test.
9 So now you're going from 90 percent plus or
10 minus ten percent to a hundred percent, basically,
11 almost and I don't -- but far be it for me to say no,
12 that you can't go out and do that test and come back and
13 use that test to dispute the test that the Department
14 did. I would certainly -- being from L. A., I would
15 certainly caution you that the request would be to allow
16 you an opportunity to go and examine it and then if
17 it's -- if it doesn't work out the way you think it
18 might work out statistically then you just pass, if you
19 will.
20 But far be it for me to deny you that.
21 The Department mentioned that if we were to use
22 this methodology it would impact the prospective dollars
23 to the City of Los Angeles. Can you elaborate on that?
24 MR. LAMBERT: I -- I can't because that -- that
25 was the first thought that crossed my mind when I looked
26 at this, is that if we're going to use the lower bound
27 that -- what I wanted to do was to see what percentage
28 of the revenue the Department produces from audits --
57
1 what percentage was from statistical samples, and what I
2 came up with is that 22 percent of the revenue that we
3 derive from the -- from our audits is from statistical
4 sampling.
5 And then I took a look at about how much
6 revenue the -- the audit agency brings in, and it's
7 about $400 million a year.
8 And so, what I was figuring was is probably 50
9 million of that 400,000 is going to go to the -- the
10 local. And then I took a look at what percentage we pay
11 out to the City of L.A., which was eight percent.
12 And so I came up with 8.22 percent. And so I
13 came up to about $4 million or so and then -- then I --
14 I figured out, you know, using the lower limit it's --
15 it's going to be a portion of that. And that's assuming
16 that everything stays the same.
17 MR. HORTON: Okay.
18 MR. LAMBERT: So --
19 MR. HORTON: Is it the City of Los Angeles'
20 position that the methodology that you're proposing
21 should be applied consistently throughout -- is it a
22 policy that the City of Los Angeles is willing to adopt
23 for themselves when it comes to evaluating taxes due to
24 the City? Is this a policy that -- that the City is
25 advocating that should be the norm even for the City of
26 Los Angeles?
27 MR. WHITLEY: Well, speak -- speaking as much
28 as I can on behalf of the City, what -- what we require
58
1 from our taxing agencies that they act reasonably and
2 non-arbitrarily.
3 MR. HORTON: That's not the question.
4 MR. WHITLEY: But -- but that is -- that is the
5 answer to your question, though. And so a statistical
6 sample --
7 MR. HORTON: But legal answer.
8 MR. WHITLEY: -- needs to give you a -- a
9 result that you can rely on, that gives you a reasonable
10 basis for thinking your determination gives you a reason
11 to make --
12 MR. HORTON: Let me change it.
13 MR. WHITLEY: -- to decide different than the
14 taxpayer.
15 MR. HORTON: Is a 90 percent with a plus or
16 minus ten percent reliable in every case?
17 MR. WHITLEY: 90 percent and ten percent would
18 be. But that doesn't mean there might not be other
19 measures that would also be reliable.
20 MS. YEE: Right.
21 MR. WHITLEY: 90 percent and ten percent
22 certainly is.
23 MR. HORTON: No, I mean it is. So is 80. So
24 is 95. So is a hundred.
25 MR. WHITLEY: Well, it -- can I -- can I
26 interrupt you a second? 80 percent would be depending
27 on what --
28 MR. HORTON: That's not a good idea, but --
59
1 MR. WHITLEY: -- on what your --
2 MR. HORTON: -- I will --
3 MR. WHITLEY: But -- but it did --
4 MR. HORTON: -- I -- I will -- I will allow
5 you -- allow you to -- to respond.
6 MR. WHITLEY: Well, but -- but it -- these ---
7 these are linked measurements (inaudible).
8 MR. HORTON: I know, and that's my point.
9 MR. WHITLEY: So --
10 MR. HORTON: I mean, that's my point, that it's
11 linked together. And all these variables are -- are
12 tied into each other. And so one argument begets
13 another. And therein is the challenge.
14 And it's a double-edged sword. And so we have
15 to be prepared to let it cut both ways.
16 And as I said earlier, I mean as a citizen I'm
17 not necessarily prepared to let it cut both ways but I
18 understand you don't -- you don't argue a case that way.
19 You win the case one case at a time. And I get that.
20 But --
21 MS. STEEL: Chairman Horton.
22 MR. HORTON: Okay. Well, let me -- is there --
23 you've heard all the testimony. Is there anything else
24 you'd like to share, at least that can bring some
25 clarity to the debate that's been here on the floor?
26 I'm going to ask you and then I'm going to go to the
27 Department, as well.
28 MR. DEGUCHI: There's something you want to
60
1 add, Bruce?
2 The only thing I want to add is that the --
3 MR. HORTON: Why don't I ask you this, again as
4 a citizen. I'm going to go to the Department first and
5 then I want you guys to spend a little time with each
6 other, giving an advantage, if you will.
7 MR. LAMBERT: Well, I'll -- I'll argue the --
8 the point and I -- I believe all the issues that they've
9 brought up have been vetted when Chapter XIII was
10 prepared. When it came time to actually have people
11 step up and say, "I want the 90-10," they didn't want
12 it. They wanted to keep the 80 percent with the 75
13 percent interval. And -- and that's the bottom line,
14 there was only one company that said they wanted
15 tighter.
16 And so --
17 MR. HORTON: Hindsight is 20/20 but let's say
18 you're in there doing an audit and the taxpayer says,
19 "You developed these four different stratas. I want you
20 to raise the -- the -- the strata number 4. Instead of
21 300,000 I want you to look at everything over a half a
22 million on an actual basis."
23 And the taxpayer says, "Well, you know what?
24 And I'm prepared to prepare -- to give you all the
25 documents, we'll do a preliminary review. I'll even pay
26 for someone to come in objectively to take a look at
27 this."
28 Are we going to say no to that?
61
1 MR. LAMBERT: Absolutely not. And in fact to
2 prove that point in -- in the subsequent audit that's
3 exactly what we did. When we went in to do the audit we
4 started the actual basis at $100,000 and -- and did the
5 audit on that basis.
6 So, if you come to -- if you're asking the
7 Department would we be willing -- if -- if you want us
8 to do it on an actual basis the entire audit --
9 MR. HORTON: No, no, no, no on that.
10 MR. LAMBERT: -- we're -- we're not going to do
11 it. We're not going to do it --
12 MR. HORTON: Yeah.
13 MR. LAMBERT: We're not going to do a 90-10.
14 But if you're saying be reasonable or -- well, maybe not
15 reasonable, but maybe move it down to $100,000 and we
16 take a look it it and we say, oh, okay, we -- we could
17 take a look at enough of those and -- and we'd be okay
18 with that.
19 Yeah, we would do that. And that's --
20 MR. HORTON: I mean, if you -- if you change
21 strata 4 that very well -- may very well change your
22 confidence intervals and everything else. The higher
23 you go up the more you examine, the higher number you
24 examine on an actual basis.
25 So, I mean there are some in betweens here that
26 we could -- we keep -- I mean, we could go to. I would
27 certainly encourage a conversation with the Controller
28 and the Mayor relative to what resources you guys are
62
1 prepared to allocate, but that's just one person's
2 opinion, yeah.
3 MR. RUNNER: Just to clarify, Mr. Chair, real
4 quick --
5 MR. HORTON: Well, I'm still rolling here.
6 MR. RUNNER: Okay.
7 MR. HORTON: So let me understand what it is
8 L. A. is prepared to do relative to a subsequent test.
9 And what -- even though I believe that you -- this
10 should have occurred from the onset, but I'm just
11 curious.
12 I mean, Member Steel brought up what I thought
13 was a -- a good idea, as well, that might get you
14 somewhere. Mindful, we just need to be in the ballpark.
15 And we're somewhere in the middle now. And so, you can
16 go to the right, you go to the left, we need something
17 in order to be able to do that.
18 L. A., you're up.
19 MR. DEGUCHI: Okay. Again, our preference
20 would be to pay the amount based on the lower limit of
21 the 90 percent confidence level. However, we are
22 willing to have additional samples pulled whereby they
23 achieve a 90 percent confidence level plus or minus ten
24 percent.
25 I guess another option would be to take any
26 transactions on an actual basis over whatever dollar
27 cutoff they want and any tax that -- any errors they
28 find we would repay that tax if we underpaid the tax.
63
1 MR. HORTON: Member Steel, then Ms. Mandel.
2 MS. MANDEL: No, I -- I have nothing.
3 MR. RUNNER: I just -- I just want to go back
4 and clarify what -- the question that you had -- the
5 Chair asked, real quick, before we go too far.
6 Were you suggesting that maybe the way to get
7 through this was to go ahead and -- and lower the --
8 include the actuals and go down to the $100,000?
9 MR. HORTON: No, that's not my job to --
10 MR. RUNNER: Well, I thought -- I'm sorry, I
11 misunderstood. I -- I guess --
12 MR. HORTON: I may put something out there
13 hypothetically.
14 MR. RUNNER: Right. Well, that -- that's --
15 MR. HORTON: But I --
16 MR. RUNNER: -- that's what I want to clarify.
17 MR. HORTON: -- I went back to the City of Los
18 Angeles to try to hopefully get them to --
19 MS. STEEL: Can I speak now?
20 MR. HORTON: -- come to some conclusion as to
21 what they would like to do.
22 MS. STEEL: Okay.
23 MR. HORTON: Ms. Steel.
24 MS. STEEL: Okay.
25 MR. TRUITT: Mr. Chairman, may I ask a
26 question?
27 MR. HORTON: One second, sir.
28 MS. STEEL: Can -- can I just ask what --
64
1 MR. HORTON: We will come back to you.
2 MS. STEEL: Well, if City of L. A. wants to go
3 there and to have 90 percent interval confidence -- I
4 mean confidence level then that's fine with me, I can
5 give 30-30-30 that you can go back and you can figure it
6 out another number.
7 But I have another idea. Why don't we just
8 take the strata number 4 plus the lower range of the
9 sample test using 80 percent confidence level?
10 Is Department can agree with -- with that?
11 MR. LAMBERT: I -- I wasn't sure -- I don't
12 know what that means.
13 MS. STEEL: Okay.
14 MR. HANKS: Ms. -- Ms. Steel --
15 MS. STEEL: Take --
16 MR. LAMBERT: What was that?
17 MR. HANKS: -- in terms of the time, though,
18 you had suggested a 30-30-30 time.
19 MS. STEEL: If they agree 90 percent, but you
20 know we are wasting time and money -- even City, I live
21 in L. A. County, not City, but you know that's going to
22 cost another bunch of money. It's a tax money we're
23 talking about.
24 So what I'm suggesting is that we take the
25 strata 4 because I -- I -- my -- the other one was just
26 take that out and just give 80 percent confidence level.
27 Now, take strata measure -- 4 measure plus the lower
28 range of the sample test using 80 percent of confidence
65
1 level, is that --
2 MR. LAMBERT: On 4?
3 MS. STEEL: -- agreeable? Yeah, 4 plus lower
4 range --
5 MR. LAMBERT: 4 --
6 MS. STEEL: -- of the sample test.
7 MR. LAMBERT: 4 plus the lower on 2 and 3?
8 MS. STEEL: 1, 2 -- yeah.
9 MR. LAMBERT: Well, not -- because 1 there's
10 no -- we didn't --
11 MS. STEEL: So 2 and --
12 MR. LAMBERT: -- we found just a couple errors
13 and we didn't project it, in strata 1.
14 MS. STEEL: So just in 2 and 3?
15 MR. LAMBERT: So it's just 2 and 3 and that's
16 what we're -- that's what the argument is right now on 2
17 and 3, is using the lower limit.
18 MR. DEGUCHI: Yes, we agree a hundred percent
19 with that approach --
20 MS. STEEL: Okay.
21 MR. DEGUCHI: -- exactly what you said.
22 MS. STEEL: So you want to do that? 2 and 3.
23 MR. LAMBERT: But that's what the whole
24 argument's been about then, the --
25 MS. STEEL: I try to give --
26 MR. LAMBERT: Be -- I mean other than it went
27 down to 80 --
28 MS. STEEL: -- something for both of you to,
66
1 you know, discuss about and then we can settle something
2 here.
3 MR. LAMBERT: Right.
4 MS. STEEL: But if you're not agreeing then --
5 MR. LAMBERT: I don't think I can agree and --
6 and the reason I can't is --
7 MS. STEEL: Mr. Lambert, you're not going to
8 agree with me anything, anyway.
9 MR. LAMBERT: I would -- I will --
10 MS. STEEL: We've been going through all these
11 stats and, you know, it seems like I'm in the class,
12 too, and I'm not happy with those classes, anyway.
13 MR. LAMBERT: Yeah. Yeah, I can't agree with
14 it because it's just contrary to --
15 MS. STEEL: So what can you agree here then --
16 MR. LAMBERT: Can I --
17 MS. STEEL: -- without going for 90 percent?
18 MR. LAMBERT: It -- what can I agree to?
19 If I agreed to anything it would be what we
20 agreed to originally, is that we would go down to look
21 at items $100,000 and -- between 300,000 and --
22 MS. STEEL: Oh, so you want to redo it then?
23 MR. LAMBERT: No -- well, redo it -- not -- I
24 wouldn't call it re --
25 MS. STEEL: Be -- because you have to put the
26 numbers for --
27 MR. LAMBERT: Well, you would -- you would
28 recalculate the numbers --
67
1 MS. STEEL: Right.
2 MR. LAMBERT: -- for just stratum 2 -- you
3 would recalculate it for that. But -- but remember, the
4 exposure here is for those 600 items or whatever -- or
5 the 609, is there's $100 million worth of purchases
6 there and that if you -- if you -- there's a possibility
7 it could be a higher amount but we were willing to do it
8 back then, we'd be willing to do it now.
9 So I would --
10 MR. HORTON: Okay.
11 MR. LAMBERT: If that's what you want to do
12 that's --
13 MR. HORTON: All right.
14 MR. HANKS: I think that's a point that -- that
15 we haven't really discussed, is that -- that the offer
16 to -- to expand that sample was made previously with --
17 with the City of Los Angeles but they rejected that.
18 And so we're mindful of --
19 MR. HORTON: Well, I -- I think that of the --
20 MR. HANKS: -- of that difficulty.
21 MR. HORTON: My -- my apologies for
22 interrupting but I -- I think that the City of Los
23 Angeles is just saying that if -- they believe that if
24 you increase the confidence intervals to 90 percent you
25 won't pick up anything. That your test will be
26 invalidated. Which I just don't see if all the other
27 variables change.
28 But possibly if the City of Los Angeles is
68
1 willing to change the strata somehow, if you've done
2 some analysis that you think that will reflect the -- I
3 mean the closer we get to examining more transactions on
4 an actual basis your theory would say that that would be
5 better. Because the premise of your theory is that you
6 need to be as close to a hundred percent as you possibly
7 can be.
8 And so theoretically if we get somewhere
9 closer -- because this whole conversation's about 80
10 percent plus or minus ten percent is really not the
11 basis of the audit liability, it's just the basis for
12 saying we accept or we reject the audit liability. In
13 this case the Department say we -- we accept the -- that
14 the errors occurred, and it seems we all occur, and
15 we've extrapolated those errors. And then at the end of
16 the day they say, well, does this make sense? And based
17 on the criteria of the Board of Equalization it does
18 make sense.
19 And if you want to change that criteria, then
20 you go back and you change the test, and you get
21 somewhere in between a hundred percent and where they
22 are now.
23 But, theoretically -- and this is just from --
24 it kind of reminds me, if you will, of -- what do you
25 call it, a recount? Every time I see a recount and the
26 numbers are close it -- it's like the odds is, is that
27 the numbers that were projected end up being the same.
28 They really don't go the other way. And if they do go
69
1 anywhere, they go up.
2 And so, I -- I kind of want to -- I don't want
3 to -- I don't want to defend the case on behalf of my
4 own city. I'm getting there, close to that, but I --
5 no, I'm not. I'm being very objective here, let me
6 clarify the record.
7 I don't really see the argument that the City
8 of Los Angeles is putting forth as having merit when you
9 look at it from a holistic perspective. And so -- but
10 somewhere in between there may be an adjustment that the
11 numbers may reflect that may change the outcome but is
12 not incumbent upon this body to make that decision.
13 It's sort of incumbent upon the appellant to listen to
14 the direction that the policy-makers are going and then
15 offer up some alternative.
16 One alternative might be 30-30-30, which allows
17 you 30 days to evaluate your situation, have a
18 discussion with the Department and possibly consider
19 what the Department has put forth, or debate whether or
20 not that adjustment -- it should be different.
21 The other is to allow it -- allow this body to
22 determine whether or not it should overturn ten years of
23 policy and procedure that would ultimately result in a
24 negative to the City of Los Angeles if we did so going
25 forward. According to their estimates, not mine.
26 So with that I sort of -- it's the appellant's
27 turn to respond to the original question.
28 MR. WHITLEY: I'm sorry, I -- I've kind of lost
70
1 track of the original question.
2 MR. HORTON: Oh. Well, you know, again that's
3 not my job.
4 All right. Further discussion, Members?
5 Is there a motion?
6 MS. YEE: Move to take the matter under
7 submission.
8 MR. HORTON: It's been moved to take the matter
9 under submission by Ms. Yee, second by --
10 MS. STEEL: Second.
11 MR. HORTON: -- Ms. Steel.
12 Without objection, Members, such will be the
13 order.
14 Thank you so very much for the discussion
15 today. It's been very interesting.
16 We will take the matter under consideration
17 later on and send you a written report of our decision.
18 ---oOo---
19
20
21
22
23
24
25
26
27
28
71
1 REPORTER'S CERTIFICATE.
2
3 State of California )
4 ) ss
5 County of Sacramento )
6
7 I, BEVERLY D. TOMS, Hearing Reporter for the
8 California State Board of Equalization certify that on
9 June 23, 2011 I recorded verbatim, in shorthand, to the
10 best of my ability, the proceedings in the
11 above-entitled hearing; that I transcribed the shorthand
12 writing into typewriting; and that the preceding 71
13 pages constitute a complete and accurate transcription
14 of the shorthand writing.
15
16 Dated: July 21, 2011.
17
18
19 ____________________________
20 BEVERLY D. TOMS
21 Hearing Reporter
22
23
24
25
26
27
28
72