USING A SAMPLE OF FEDERAL ESTATE TAX RETURNS TO EXAMINE

USING A SAMPLE OF FEDERAL ESTATE TAX RETURNS TO EXAMINE

THE EFFECTS OF AUDIT REVALUATION ON PRE-AUDIT ESTIMATES

Martha Britton Eller and Barry W. Johnson, Internal Revenue Service

Martha Britton Eller, P. O. Box 2608, Washington DC 20002

Key Words: Audit, Stratified Sample, Non-response, Imputation

Introduction

Federal estate tax returns are filed for America’s

wealthiest decedents, and these returns contain detailed

information on decedents’ assets, liabilities and

personal characteristics. Data from Federal estate tax

returns provide a unique opportunity to study the

demographic and portfolio characteristics of America’s

wealthiest individuals. In addition, these data provide

valuable insight into the transfer of wealth between and

among generations. The Statistics of Income Division

(SOI) of IRS collects and publishes data from annual

samples of Federal estate tax returns. During the

collection process, data are subjected to extensive math

tests and are edited to remove obvious errors. Similar

to other studies conducted at SOI, data for the annual

Estate Tax Study are collected prior to any systematic

audit of returns, since the audit process can take several

years to complete. Published pre-audit data are,

therefore, likely to contain an understandable bias

resulting from taxpayers’ desire to avoid tax (Harriss,

1994), as well as taxpayers’ misunderstanding of estate

tax law. Because estate tax data are an important

research tool in both the public and private sectors, the

merits and deficiencies of such data warrant careful

analysis.

After IRS processes Federal estate tax returns for

revenue purposes and after SOI extracts data for a

sample of these returns, IRS service center tax

examiners, those with expertise in estate tax law,

review the returns to determine whether they should be

examined by estate tax attorneys in IRS district offices.

Finally, estate tax attorneys in the district offices survey

the candidates for audit and select a subsample of

returns for complete audit processing. Federal law in

effect for returns filed in 1992, the year examined in

this paper, allowed the IRS up to three years after a

return is filed or after the date it was due, whichever

date is later, to complete the audit process. However,

this statute of limitations may be suspended in special

cases.

The IRS grants estate tax attorneys, or auditors, wide

discretion in examining records and in interviewing

fiduciaries, survivors, heirs and others who may be

acquainted with decedents’ affairs. If an auditor

changes values reported on the tax return,

representatives of the estate can appeal to the IRS

Appeals Office, the U.S. Tax Court, the United States

Claims Court or U.S. District Court.

In order to understand the effects of audit revaluation

on pre-audit estimates, SOI undertook a study of postaudit estate tax returns. The 1992 Estate Post-Audit

Study, initiated in 1996, is a sample of 4,433 returns

filed in 1992. Returns were previously selected for

SOI’s annual Estate Tax Study. Filing year 1992 was

selected as the focus year for SOI’s post-audit study in

order to allow time for returns to complete the entire

audit process. For filing year 1992, a decedent’s estate

was required to file a Federal estate tax return if the

value of gross assets, at death, exceeded $600,000.

Federal estate tax returns were filed for 59,178 wealthy

estate tax decedents during that year.

This paper will discuss the sample design and

development of final weights for the 1992 Estate PostAudit Study. Estimates of audit changes in tax

assessments and asset values will be presented.

Estate Tax Study Sample Design

The sample for SOI’s Estate Tax Study, the sample

frame for the 1992 Estate Post-Audit Study, is a

stratified random sample, executed over a 3-year

period, with three stratifying variables. In every year

since 1982, the stratifying variables have been year of

death (focus year, non-focus year), total gross estate

and age at death. Gross estate is divided into 5

categories: $600,000 < $1 million, $1 million < $2.5

million, $2.5 million < $5 million, $5 million < $10

million, and $10 million or more. Age at death is

divided into age < 40, 40 < 50, 50 < 65, 65 < 75, and 75

and older. Sample rates vary from 3 percent to 100

percent, with over half the strata selected with certainty.

Returns are selected for the sample as they are

processed for IRS revenue purposes.

Weights for the Estate Tax Study sample are calculated

in several steps. First, population and sample counts

are adjusted for returns that were selected into the

sample but, upon close examination, do not satisfy the

parameters of the study. This occurs, for example,

when an incomplete return is filed because the estate

was not able to compile all the necessary

documentation by the filing deadline. In such cases, a

second and final return is filed by the estate once the

required information is obtained. Second, counts are

adjusted for the small number of returns that were

unavailable for SOI processing because they were

under review by other areas of IRS. Third, adjustments

are made for misclassified returns, which typically arise

from taxpayer or IRS processing errors that cause

returns to be assigned to an incorrect sampling strata.

Finally, data are poststratified using auxiliary data from

the IRS Master File. These auxiliary data have been

examined and corrected in an attempt to adjust for large

returns not originally available for sampling due to data

transcription errors.

Figure 1: SOI Sample Selection Criteria

(1) fees (including executors’ commissions, and

attorneys’ fees) > 5 percent of total gross estate

(2) adjusted taxable gifts > 5 percent of total gross

estate

(3) other non-corporate business assets > 5 percent

of total gross estate

(4) other stock > 30 percent of total gross estate

(5) credit for tax on prior transfers > 0

(6) art > 0

(7) decedent had a power of appointment over trust

property

1992 Estate Post-Audit Study Sample Design

The sample frame for the 1992 Estate Post-Audit Study

consisted of returns originally selected for SOI’s study

of Federal estate tax returns filed during calendar year

1992. The 1992 Estate Tax Study sample consisted of

7,559 returns out of a population of 59,178. The 1992

Estate Post-Audit Study sample was a stratified random

sample of these 7,559 returns. The sample was

stratified by a variable that predicted a return’s

likelihood of being audited, an “audit likelihood

indicator.” Because the process of determining whether

or not a return will be audited is highly subjective,

development of this post-audit project included meeting

with IRS auditors in several regions of the country to

discuss the criteria they use in selecting audit cases.

Based on these interviews, SOI developed the sample

selection critera listed in Figure 1 below. In addition to

these criteria, returns for married decedents are usually

not audited, since the unlimited marital deduction for

bequests to a decedent’s surviving spouse typically

offsets any increase in the taxable estate that might

result from audit revaluation. With this information, an

audit likelihood indicator that served as the sample

stratifier was developed. Returns with a high likelihood

of audit, those which met one or more of the criteria in

Figure 1 and were filed for non-married decedents,

were selected into the sample with certainty.

A

random sample of the remaining returns was selected at

a sampling rate of about 0.4. Overall, the final sample

included 4,433 returns, 2,433 of which were selected

because they conformed to the marital status (nonmarried) criteria and the criteria in Figure 1; the

remaining 2,000 were selected randomly.

While filing year 1992 was chosen as the focus year for

the Estate Post-Audit Study, a significant portion of the

sample, 229 returns, was still unavailable to SOI at the

study’s close. After SOI economists tested and cleaned

the data, as well as performed some limited imputation

for item non-response, weights for non-missing returns

were developed in two stages.

(8) closely-held stock > 5 percent of total gross estate

(9) real estate > 40 percent of total gross estate

(10) lifetime transfers > 30 percent of total gross estate

(11) total gross estate > $5,000,000

First, an initial base weight was calculated for nonmissing returns, taking into account both the probability

of selection in the underlying sample of estate tax

returns and the probability of selection in the post-audit

subsample. In the initial post-audit sample, there were

a small number of cases for which only a limited

amount of data on the magnitude of audit revaluations,

but no underlying detail, existed. Since researchers are

most interested in studying this detail, incomplete cases

were eliminated from the final data set. Therefore, in

the

second

stage

of

weight

development,

poststratification was used to adjust the weights of the

remaining cases. Adjustment cells were constructed

based on the size of the change in taxable estate. In

each adjustment cell, a ratio adjustment was computed

that utilized data available from both complete and

incomplete cases. Using these ratios to adjust base

weights of the cases for which complete data were

available, final weights were then constructed. The

final data file included complete records for 4,182

returns, of which 1,357 had been audited. A careful

analysis of the 229 missing returns was conducted using

the IRS Integrated Data Retrieval System (IDRS). It

was possible to determine whether each of the missing

returns had been audited and, if so, the outcome of

audit. The distribution of missing returns on these key

characteristics was almost identical to that of the nonmissing returns, so no additional “non-response”

adjustment was necessary.

Revaluation at a Glance

Raw audit data for the 1992 Estate Post-Audit Study

were collected from Forms 1273 and 3228 prepared by

IRS auditors during the formal audit process. These

forms, in most instances, capture both pre- and postaudit estate tax return values. Revaluation data were

extracted from the forms by IRS personnel in the

Ogden Service Center. Therefore, the revaluation data

$12.0 billion, an aggregate net difference of $117.0

million. Therefore, the post-audit value of 1992 total

allowable deductions reached $43.6 billion.

described in this paper are estimates of the audited

population based on actual IRS audit results.

IRS tax examiners audited an estimated 11,338 Federal

estate tax returns filed in 1992, representing 19.2

percent of the 59,178 returns filed during the year.

Figure 2 shows that, as the size of gross estate

increases, the rate of audit increases. Returns filed with

gross estates less than $1.0 million were audited at a

rate of 11.1 percent. However, almost 50.0 percent of

returns filed with gross estates over $5.0 million were

audited, even though the audited returns in that

category represented only 9.7 percent of the entire

audited population.

The original net estate tax liability for filing year 1992

was an estimated $10.2 billion. Almost two-thirds, 63.2

percent, of the original liability, or $6.4 billion, was

subject to audit. Overall, net estate tax liability

increased an estimated $559.8 million as a result of

audit, a change that represented 5.5 percent of the

original liability reported in 1992. Of audited cases,

60.0 percent were closed with additional net estate tax

owed, 19.0 percent were closed with no change in tax

assessment, and 21.0 percent were closed with a

reduction in the original net estate tax liability. The

value of additional estate tax owed was $676.6 million,

while the reduction in estate tax liability totaled $116.8

million. Post-audit estate tax revenue for filing year

1992, the sum of pre-audit estate tax revenue and

revenue derived from audit, reached $10.8 billion.

Figure 3: Change in Value of Total Gross Estate,

Allowable Deductions and Net Tax Due to Audit

Figure 2: Number and Percentage of Returns

Audited, by Size of Total Gross Estate

Size of Total

Returns

Returns

Percent

Gross Estate

Filed

Audited

Audited

Under $1 million

31,376 (30.6)

3,475

11.1%

$1 million under $5 Million

25,542 (59.6)

6,760

26.5%

$5 million or more

2,260

1,098

48.6%

Total

59,178 (100.0) 11,338

19.2%

(9.7)

Numbers are in millions of dollars.

Combined total gross estate for 1992, the accumulated

wealth of estate tax decedents whose returns were filed

during 1992, exceeded $100.0 billion, and 34.9 percent

of that value was subject to audit by tax examiners.

Total gross estate increased $1.2 billion, from $34.9

billion to $36.1 billion. The post-audit value of 1992

combined gross estate was $101.2 billion (see Figure

3).

Total

gross

estate

Allowable

Net estate

Deductions

tax

Pre-audit value

100,017

43,530

10,199

Value subject to audit

34,880

11,905

6,443

Percent subject to audit

34.9%

27.3%

63.2%

Audit revaluation amount

1,222

117

560

Percent change due to audit

1.2%

0.2%

5.5%

Post-audit value

101,239

43,647

10,759

Demographic Data for the Audited Population

Total allowable deductions, available against gross

estate, reduce a decedent’s taxable estate and include

deductions

for

marital

transfers,

charitable

contributions, administrative expenses, indebtedness,

taxes and casualty loss. The original, pre-audit value of

total allowable deductions for 1992 exceeded $43.5

billion, and 27.3 percent of that value, or $11.9 billion,

was subject to audit. An unexpected result, total

allowable deductions increased from $11.9 billion to

Revaluation Data by Sex of Decedent

The filing population in 1992 included more male

decedents than female decedents, with 32,843 males, or

55.5 percent of the total population, and 26,335

females, or 44.5 percent of the total population (see

Figure 4). However, females were more prevalent than

males in the audited population. Of the 11,338

decedents in the audited population, 52.9 percent, or

Figure 4: Audited Estate Tax Returns, Change in Assessment as a Percentage

of the Number of Returns Audited, by Sex

[Money amounts are in thousands dollars]

Males

Number

Females

Percent

Number

Total

Percent

Number

Percent

Filing population

32,843

55.5%

26,335

44.5%

59,178

100%

Audited population

5,340

47.1%

5,998

52.9%

11,338

100%

Additional tax assessed

2,951

55.3%

3,856

64.3%

6,807

60.1%

Tax reduction assessed

1,289

24.1%

1,095

18.3%

2,384

21.0%

No tax assessment change

1,100

20.6%

1,047

17.4%

2,147

18.9%

5,998 decedents, were female and 47.1 percent, or

5,340 decedents, were male. This last finding may be

explained by the prevalence of widowed decedents,

most often female, in the audited population. That is,

61.2 percent of the audited population were widowed

decedents, and, more than two-thirds, 68.7 percent, of

widowed decedents were female. Again, the overriding

presence of widowed decedents, and therefore female

decedents, in the audited population is a result of the

audit selection process, which favors non-married

decedents.

The estates of females in the audited population owed

an additional $292.9 million in combined net estate tax,

or 52.3 percent of the total change in estate tax liability

for the audited population, while the estates of male

decedents owed an additional $266.8 million in

combined net estate tax, 47.7 percent of the total

change in estate tax liability. This difference in the

change in tax liability by sex reflects the original

difference in pre-audit liability by sex. For, in the

original 1992 estate tax filing population, female

decedents owed 52.1 percent of the combined net estate

tax liability, or $5.3 billion, while male decedents owed

47.9 percent of the combined liability, or $4.9 billion.

Again, the prevalence of widowed decedents, mostly

female, in the audited population may explain the larger

share of additional estate tax liability assumed by

female decedents’estates.

unknown, 4.1 percent of all decedents. The audited

population looked quite different, with a larger

percentage of widowed decedents and a smaller

percentage of married decedents than in the filing

population.

Widowed decedents comprised 61.2

percent of the audited population, while married

decedents comprised only 23.1 percent of the audited

population. The remaining decedents in the audited

population were divided almost equally between the

last marital status categories, single and legally

separated, divorced or unknown (the “Other” category),

8.8 percent and 6.8 percent, respectively.

The estates of married decedents and the estates of

widowed decedents saw similar increases in net estate

tax as a result of audit. Widowed decedents’ estates

assumed 42.5 percent of the total increase in net estate

tax, or $237.8 million, while married decedents’ estates

assumed 41.2 percent of the total increase, or $230.9

million. Estates of decedents in the “Other” category

experienced a $59.3 million increase in estate tax

liability, and the estates of single decedents experienced

a $31.8 million increase in estate tax liability.

Figure 5: Estate Tax Returns Filed in

1992 and Audited Returns, by Marital

Percent

Status

80

61.2

60

In terms of the type of change in tax assessment, Figure

4 shows that the estates of males and females exhibited

similar tendencies. For both males and females, returns

closed with additional estate tax owed were dominant,

with 55.3 percent of estates for male decedents owing

additional tax and 64.3 percent of estates for female

decedents owing additional tax. Overpayment of the

original tax was more prevalent than no change in tax

assessment for both sexes. However, compared to the

estates of female decedents, the estates of male

decedents were more likely to have overpaid. While

estates of males overpaid in 24.1 percent of cases and

had no change in tax liability in 20.6 percent of cases,

estates of females overpaid in only 18.3 percent of

cases and had no change in tax liability in 17.4 percent

of cases.

Revaluation Data by Marital Status of Decedent

Married decedents comprised the largest percentage of

decedents in the 1992 estate tax filing population, with

46.6 percent of all decedents married at death (see

Figure 5). The second largest marital status category

was widowed decedents, with 40.6 percent of all

decedents widowed at death. The remaining decedents

were single at death, 8.6 percent of all decedents, or

were legally separated, divorced or marital status

46.6

40.6

40

23.1

20

8.6

8.8

4.1

6.8

0

Married

Widowed

Estate Tax Filing Population

Single

Other

Audited Population

Revaluation Data by Age of Decedent

In terms of decedent age, estate tax examiners selected

audit cases that fairly represented the 1992 estate tax

filing population, since the distribution of age in the

audited population was quite similar to the distribution

of age in the 1992 filing population as a whole (see

Figure 6). The largest group of decedents in the 1992

filing population were “80 under 90,” 32.5 percent of

the filing population, while the second and third largest

groups were “70 under 80” and “90 and older,” 26.4

percent and 16.0 percent of the filing population,

respectively. In the audited population, those estates

selected for audit, the largest group of decedents were

“80 under 90,” 34.6 percent of the audited population,

while the second and third largest groups were, as

above, “70 under 80” and “90 and older,” 24.3 percent

and 20.9 percent of the audited population, respectively.

Percent

40

35

30

25

20

15

10

5

0

Figure 6: Estate Tax Returns Filed in 1992 and

Audited Returns, by Age

34.6

32.5

26.4

24.3

15.7

11.9

20.9

16

6 5.7

3.4 2.6

Under 50

50 under 60 under 70 under 80 under

90 and

60

70

80

90

older

Estate Tax Filing Population Audited Population

Revaluation of Assets

In the course of the annual Estate Tax Study conducted

by SOI, detailed asset data are extracted from each

Federal estate tax return included in the study. Assets,

the building blocks of total gross estate, are assigned to

one of several asset categories, which allows SOI to

produce estimates of total asset holdings, by asset type,

for the filing population. The Estate Post-Audit Study

utilized these same asset categories, permitting

comparisons between the 1992 filing population, as

reported on the original, 1992 estate tax return, and the

audited population, as reported on Forms 1273 and

3228, audit forms completed by estate tax examiners.

Asset data available from the 1992 Estate Post-Audit

Study include original (pre-audit) and corrected (postaudit) values for individual assets that were revalued

during audit.

Audited returns most frequently included cash assets,

present on 99.0 percent of all audited returns, followed

by other corporate stock, primarily publicly traded

stock, present on 78.3 percent of audited returns (see

Figure 7, Column 1). However, looking at the relative

frequency of audit revaluation among audited returns

with a particular asset, the findings are quite different

(see Figure 7, Column 3). Real estate assets, excluding

the personal residence, were most frequently revalued

among those audited returns that included real estate

holdings; 38.0 percent of audited returns with real

estate included revalued real estate assets. The stock of

closely held corporations was second in the relative

frequency of changes. Of audited returns with closely

held stock, 37.0 percent included revalued closely held

stock. Cash assets followed closely held stock, with

34.0 percent of audited returns that included cash

encountering revaluation of cash assets.

Again, if audit results for asset holdings are examined

in conjunction with asset-holding data from SOI’s

annual Estate Tax Study for filing year 1992, it is

possible to examine the degree to which revaluation

affects pre-audit estimates of assets. Comparing the

original value of revalued assets with the total, preaudit value for assets in a particular category among all

audited returns reveals that 41.3 percent of depletable

and intangible assets, such as copyrights, were

examined for revaluation purposes, making these assets

most subject to audit (see Figure 7, Column 4).

Ignoring unclassified mutual funds, an asset category

that includes funds that could not be classified by their

component assets, second among asset values subject to

audit was the total value of real estate, with 34.3

percent of real estate in the audited population

examined and revalued. Closely held stock holdings

were third in this ranking, with a little less than 27.0

percent of the value of closely held stock examined and

revalued. Farm assets as a percentage of the total value

of audited returns with farm assets exceeded 100.0

percent. This finding suggests that the original value of

revalued farm assets, as derived from the Estate PostAudit Study, is larger than the total value of farm assets

in the audited population, derived from the Estate Tax

Study. The difference between studies points out the

difficulty in identifying farm assets that are used to run

a farm or agricultural business. Because of this

difficulty, farm assets are excluded from the following

analysis.

Asset revaluation that occurred during the audit process

may be examined in a number of different ways. First,

the net aggregate change in asset value may be

examined (see Figure 7, Column 5). Closely held stock

had the largest net aggregate change in asset value, a

$319.7 million increase. The net aggregate revaluation

of mortgages, notes and claims, the second largest

revaluation, totaled $268.9 million. The third largest

increase was in other real estate, a net increase of

$140.1 million.

The average net change per revalued return is another

variable to use in analyzing asset revaluation (see

Figure 7, Column 6). The largest average net change

per revalued return was almost $500,800, the average

change for mortgages, notes and claims. The average

change for closely held stock was the second largest

change, with the value of that stock increasing, after

audit, more than $387,000 per revalued return.

Ignoring the change per revalued return for farm assets,

third in this ranking was depletable and intangible

assets, with an average change per revalued return of

$193,600.

The revaluation of assets may also be examined in

relation to the original value of revalued assets (see

Figure 7, Column 7). The post-audit increase in the

value of mortgages, notes and claims was the largest

increase as a percentage of that revalued asset, 170.2

percent. Life insurance on the life of the decedent was

second in this ranking. The net increase in the value of

life insurance was 116.9 percent of the value as

originally filed. Depletable and intangible assets

sustained the third largest increase, an increase

equivalent

to

49.0

percent

of

revalued

depletables/intangibles as originally filed. Nine of the

15 asset categories increased by 10 percent or more,

while three categories increased between 5 and 10

percent and only 2 categories increased between 0 and

5 percent. Only one asset category, unclassified mutual

funds, sustained a net decrease, -3.6 percent.

Finally, it is useful to examine asset revaluation in

relation to the original value reported in a particular

asset category across all audited returns (see Figure 7,

Column 8). With some exceptions, the revaluation of

assets represented between less than 1.0 percent and 4.0

percent of the original value on all audited returns.

Once again ignoring farm assets, the largest increase in

asset value, as a percent of total asset value, was in

mortgages, notes and claims, 28.9 percent, followed by

depletable and intangible assets, a 20.3 percent

increase. Closely held stock was next in this ranking,

although it saw a much smaller increase than

depletables/intangibles; closely held stock increased 7.7

percent of its original value on audited returns.

Comparing Columns 7 and 8 reveals that, while

revalued assets were often changed substantially during

audit (Figure 7, Column 7), the changes were, in most

cases, still a small portion of the original asset value for

all audited returns (Figure 7,Column 8). For example,

look at other non-corporate business assets. While the

value of the revalued asset increased 44.5 percent, this

change represented only 7.0 percent of the total value of

other non-corporate business assets on all audited

returns. The same is true for other assets, although in

varying degrees: limited partnerships, for which there

was a 33.0 percent increase in the revalued asset

compared to a 3.8 percent increase in the total value;

cash (26.9 percent compared to 2.1 percent); closely

held stock (28.8 compared to 7.7 percent); and bond

holdings (17.2 percent compared to 0.5 percent). Of

course, there are a couple of exceptions. The increase

in depletable and intangible assets, for example, was

large, both in relation to the original value of the

revalued asset, 49.0 percent, and in relation to the total

value of depletables/intangibles, 20.3 percent.

Mortgages, notes and claims are similar, 170.2 percent

compared to 28.9 percent. Overall, however, the

changes that estate tax attorneys made to revalued

assets during audit did not introduce large, overall

increases to total asset holdings, regardless of the

magnitude of change in the revalued asset.

References:

Erard, B. (1998) “Estate Tax Underreporting Gap

Study: A Report Prepared for the Internal Revenue

Service Economic Analysis and Modeling Group.”

(Order Number TIRNO-98-P-00406) Internal Revenue

Service.

Harriss, C. L. (1949) “Wealth Estimates as Affected

by Audit of Estate Tax Returns.” National Tax

Journal, vol. 2, number 4, pp. 316-333.

McCubbin, J. (1994) “Improving Wealth Estimates

Derived From Estate Tax Data.” Compendium of

Federal Estate Tax Data and Personal Wealth Studies,

Dept. of Treasury, IRS Pub. 1773, pp. 363-390.

The Authors would like to thank Gerald Auten for

helpful comments. Thanks also to Ogden Service

Center Personnel, Catherine Gullickson, Tamara Rib,

Dorothy Wallace, and Patricia Slaughter for their work

on this project.

Figure 7: Audit Revaluations For Audited Returns Filed in 1992, by Asset Type

Asset

Type

Personal Residence

Other Real Estate

Closely Held Stock

Other Stock

Bonds

Unclassified Mutual Funds

Cash

Insurance

Farm Assets

Limited Partnerships

Non-Corp. Bus. Assets

Mortgages & Notes

Annuities

Depletables/Intangibles

Art

All audited returns

Number

Amount

(1)

6,251

7,611

2,234

8,879

8,295

2,242

11,234

5,569

1,042

1,498

1,943

4,219

3,421

1,160

425

(2)

1,717,199,287

4,335,051,498

4,155,033,484

8,687,221,836

6,301,791,305

301,332,782

3,717,282,372

568,572,557

102,287,640

520,072,214

696,284,538

931,794,868

773,034,691

156,666,051

663,691,995

Returns

with change

(3)

15.9%

38.0%

37.0%

21.8%

11.6%

10.4%

34.0%

8.5%

22.9%

14.0%

16.7%

12.7%

6.0%

14.1%

23.3%

Value

audited

(4)

17.7%

34.3%

26.7%

24.2%

2.7%

39.7%

7.8%

6.2%

113.4%

11.5%

15.7%

17.0%

6.3%

41.3%

6.7%

Change in

asset value

Average

change

(5)

13,720,129

140,142,057

319,689,764

113,406,228

28,991,250

-4,362,642

77,693,621

41,170,651

45,794,993

19,733,013

48,536,881

268,927,346

3,241,130

31,746,604

1,464,218

(6)

13,789

48,509

387,034

58,487

30,168

-18,724

20,371

86,675

191,611

94,416

149,344

500,796

15,810

193,577

14,790

Percent

Change

(7)

4.5%

9.4%

28.8%

5.4%

17.2%

-3.6%

26.9%

116.9%

39.5%

33.0%

44.5%

170.2%

6.6%

49.0%

3.3%

Change,

audited

returns

(8)

0.8%

3.2%

7.7%

1.3%

0.5%

-1.4%

2.1%

7.2%

44.8%

3.8%

7.0%

28.9%

0.4%

20.3%

0.2%

SOURCE: Turning Administrative Systems Into Information Systems, Statistics

of Income Division, Internal Revenue Service, as Presented at the 1999 Joint

Statistical Meetings of the American Statistical Association, Baltimore, MD.,

August, 1999.