Ohio Tax Statistical Sampling Preparation for a Sample Audit - - PDF document

26th Annual

Tuesday & Wednesday, January 24‐25, 2017

Hya Regency Columbus, Columbus, Ohio

Ohio Tax

Workshop RR

Statistical Sampling … Preparation for a Sample Audit

Wednesday, January 25, 2017 2:00 p.m. to 3:00 p.m.

Biographical Information Brad W. Tomlinson, Senior Manager, Zaino Hall & Farrin LLC 41 South High Street, Suite 3600 Columbus, OH 43215 btomlinson@zhftaxlaw.com 614-349-4818 Brad has more than 33 years of experience in state and local taxation, specializing in statistical sampling techniques for multiple taxes. From 2007 until joining the firm, Brad was an assistant administrator for the Ohio Department of Taxation, Audit Division, while also serving as the manager of the Computer Assisted Audit Group for more than 11 years. In that position he was responsible for the formulation, implementation, and oversight of the Department's statistical auditing practices; including the approval of data populations, sample designs, and the review of all statistical sampling procedural agreements. Prior to that, Brad spent several years auditing fortune 500 companies in manufacturing, telecommunications, computer services, and retail for sales and use tax compliance. Brad was instrumental in the design and implementation of the Audit Division's computerized auditing program (OFAST) used by multiple divisions to audit personal property, corporate franchise, employer withholding, pass-through entity, and sales and use taxes; as well as managing quarterly updates and releases of the application. Brad's other responsibilities included the establishment of single rate reporting procedures for taxpayers as well as assisting in reviewing and approving the Department's PARSA (Previous Audit Representative Sampling Analysis) agreements. During his tenure, Brad was a frequent presenter on statistical sampling at the Ohio Tax Conference and the Federation of Tax Administrators national computer and technology

• workshops. He also was consulted by other states and Canadian provinces to assist in the

development of their respective statistical auditing programs. He is a graduate of Columbus State Community College (Computer Programming) and The Ohio State University (Business Administration). Jonathan Ross, Senior Tax Consultant , Deloitte Tax LLP 180 East Broad Street, Suite 1400, Columbus, OH 43215 (614)229-5932 joross@deloitte.com Jonathan is a Senior Tax Consultant in the sales and use tax practice with Deloitte Tax LLP in the Columbus, Ohio office and has over ten years of sales and use tax experience. In addition, he has a background in statistical sampling as it relates to sales and use tax audits. Jonathan was a sales and use tax auditor for the State of Ohio for three and a half years. During that time he conducted sales and use tax audits and refund reviews. In addition, he transitioned into the state’s computer audit support group which oversees all statistical samples done by the audit division as well as data manipulation support for the entire audit division. Oversight of the statistical sampling program required: refining populations, designing statistical samples based on certain parameters, generating samples, and evaluation of the samples upon conclusion of audits. Since leaving the state Jonathan has assisted clients with state sales and use tax audits, conducted reverse audits, voluntary disclosure negotiations, and provided sales and use tax and statistical sampling consulting services for a variety of clients in a multitude of industries. During his tenure with Deloitte Tax Jonathan has been involved in design and implementation teams for various bolt-on automated tax solutions. Jonathan is a graduate of the University of Kentucky with a B.S. in Accounting

Biographical Information Roger C. Pfaffenberger, Ph.D., Director, Audit Sampling, Ryan, LLC Three Galleria Tower 13155 Noel Road, Suite 100 Dallas, TX 75240 roger.pfaffenberger@ryan.com 972.934.0022, Ext. 101279 Fax: 972.960.0613

• Dr. Roger Pfaffenberger is a Director and Practice Leader for Ryan’s Sampling Analysis and

Evaluation practice and is based in the Firm’s Dallas Office. Dr. Pfaffenberger is responsible for the design and evaluation of audit sampling methodologies with an emphasis on the use of multiple audit sampling methods in statistical and non-statistical sampling for transaction taxes. Roger has provided statistical consultation and expert testimony for a variety of companies, government entities, and law firms. Prior to joining Ryan, Dr. Pfaffenberger was Professor of Decision Sciences for Texas Christian University and Founding Director for the Center for Teaching Excellence. Dr. Pfaffenberger has made numerous presentations on audit sampling issues to professional and academic

• rganizations. He has a Doctor of Philosophy Degree in Statistics from Texas A&M University.

www.ryan.com

00/00/0000

Presentation Title

January 25, 2017

26th Annual Ohio Tax Conference Workshop RR – Statistical Sampling

• Dr. Roger Pfaffenberger, Ryan LLC

Jonathan Ross, Deloitte Tax LLP Bradley Tomlinson, Zaino Hall & Farrin LLC

Manufacturers’ Education Council

2

Presenters:

Roger C. Pfaffenberger, Ph.D. | Director, Audit Sampling Practice Ryan Three Galleria Tower 13155 Noel Road , Suite 100, Dallas, Texas 75240 Direct: 972.934.0022 Ext. 10-1279 | Cell: 817.917.0095 Email: roger.pfaffenberger@ryan.com

January 25, 2017 26th Annual Ohio Tax Conference

Jonathan N. Ross | Senior Tax Consultant Deloitte Tax LLP 180 East Broad Street, Suite 1400, Columbus, OH 43215 Direct: 614.229.5932 Email: joross@deloitte.com

3

Learning Objectives

Apply best practices in planning and managing sample audits to insure they proceed efficiently and effectively Understand and avoid mistakes that lead to sample audits producing inaccurate results Learn taxability determination resolution techniques and steps required to close the audit

January 25, 2017 26th Annual Ohio Tax Conference

4

Session Topics

Planning for initial meeting with auditor Managing initial meeting with auditor Discussion topics at initial meeting Discussion topics in later meetings/contacts Resolution of tax determination issues Steps to close the audit

January 25, 2017 26th Annual Ohio Tax Conference

5

Planning for Initial Meeting with Auditor

Understand your data sources and how transactions are recorded in your system Know your data well – conduct preliminary investigation of documents before auditor arrives Know the jurisdiction’s sampling policies and procedures, and authority to sample Know the jurisdiction’s recent case law on sampling challenges

January 25, 2017 26th Annual Ohio Tax Conference

6

Planning for Initial Meeting with Auditor

Confer with IT, AP, AR on data extraction process and on availability of required data and documents If there was a prior audit with the jurisdiction, review audit results – what went well and not so well If no prior audit, discuss experiences other tax directors have had with jurisdiction’s sample audits Develop preliminary audit plan and audit sampling plan (timetable, grouping transactions, excluding transactions not in scope, treatment of special cases)

January 25, 2017 26th Annual Ohio Tax Conference

7

Managing Initial Meeting with Auditor

Take control of the meeting – be proactive in conducting meeting Present your plan for the audit to the auditor Do not provide documentation requested by auditor at initial meeting Do not permit access to firm’s databases Establish audit timetable Set up next meeting with auditor to address auditor questions and requests

January 25, 2017 26th Annual Ohio Tax Conference

8

Discussion Topics at Initial Meeting

Determination of audit period Setting the scope of the audit (sales, purchases, types of purchases within scope) Factors that may affect audit planning (mergers, acquisitions, cost centers, multiple plants) Process for extracting relevant data for audit Process of reconciling data (AP reconciled to GL; accruals reconciled to tax returns) Process of situsing data appropriate for jurisdiction

January 25, 2017 26th Annual Ohio Tax Conference

9

Discussion Topics at Initial Meeting

Treatment of special situations – Missing documentation – Bad debt – Progress payments – Negatively-valued transactions (reversals, voids, etc.) – Treatment of EDI/ERS transactions (no printed invoice) – Isolation of transactions with potential high error rate (e.g., p-cards) – Tax only transactions

January 25, 2017 26th Annual Ohio Tax Conference

10

Discussion Topics at Initial Meeting

Sampling agreement – Binding or not binding – Scope of agreement – Consequences of not signing agreement – Point in audit timeline when agreement must be signed – Option for adding provisions proposed by tax manager to agreement via addenda

January 25, 2017 26th Annual Ohio Tax Conference

11

Discussion Topics in Subsequent Auditor Meetings/Contacts

First and foremost, take control of audit - be proactive and on top of audit status throughout audit timeline Meeting following initial meeting with auditor – Items to agree upon before auditor develops sampling plan and pulls sample

• Agree on accounts of interest
• Selection of controlling date field
• Grouping of transactions for sampling (e.g., assets,

expenses, taxed, non-taxed, utilities, p-cards)

January 25, 2017 26th Annual Ohio Tax Conference

12

Discussion Topics in Subsequent Auditor Meetings/Contacts

Meeting following initial meeting with auditor – Items to agree upon before auditor develops sampling plan and pulls sample

• Data extraction procedures
• Sampling and estimation methods for each group of

transactions

• Sample size guidelines for the groups
• Minimum set of fields in database provided to auditor
• Method of conveying database to auditor

January 25, 2017 26th Annual Ohio Tax Conference

13

Discussion Topics in Subsequent Auditor Meetings/Contacts

Meeting following initial meeting with auditor – Key questions for auditor at this point in sample audit

• When should data files be provided to the auditor?
• Can auditor share taxpayer data with other

jurisdictions?

• Is the proposed audit timeline reasonable and

acceptable to the auditor?

• Who pulls the documents for the sampled transactions?
• Can auditor take electronic data to his/her office?

January 25, 2017 26th Annual Ohio Tax Conference

14

Discussion Topics in Subsequent Auditor Meetings/Contacts

The life cycle of the audit data

January 25, 2017 26th Annual Ohio Tax Conference

Reconcile Data Define Population Sample Design Results Get Data

15

Discussion Topics in Subsequent Auditor Meetings/Contacts

Next Meetings (or meetings!) – Development of sampling plan in concert with auditor – Characteristics of sampling plan for each group

• Type of sample (block, simple random, stratified

random)

• Sample size
• Type of estimator used to project sample results to

population

• Guidelines for acceptable estimates (precision,

confidence interval, confidence bound)

January 25, 2017 26th Annual Ohio Tax Conference

16

Discussion Topics in Subsequent Auditor Meetings/Contacts

Next Meetings (or meetings!) – Development of sampling plan in concert with auditor – Characteristics of stratified random sampling plan

• Determination of lower threshold (floor)
• Determination of upper threshold (ceiling)
• Choice of number of strata
• Choice of strata breaks
• Determination of strata sample sizes

January 25, 2017 26th Annual Ohio Tax Conference

17

Discussion Topics in Subsequent Auditor Meetings/Contacts

Example of sampling plan proposed by auditor

January 25, 2017 26th Annual Ohio Tax Conference

18

Resolution of Tax Determination Issues

Issues that may (will!) arise as auditor audits sampled or selected transactions – Missing documentation – EDI (Electronic Data Exchange) or ERS (Evaluated Receipt Settlement – SAP) transactions without printed invoices or other printed documents – Transactions qualifying for exemptions (e.g., manufacturing exemptions) – Transactions used for Enterprise Zone credits – Contesting items scheduled for assessment in general

January 25, 2017 26th Annual Ohio Tax Conference

19

Steps to Close the Audit

Importance of being proactive in audit – Constant and continuous communications with auditor throughout audit process – Addressing issues arising in audit in a timely manner – Developing professional rapport with auditor which engenders a mutual trusting relationship – Compromising with auditor on tax determination issues not worth fighting over from a cost/benefit perspective – Not letting differences of opinion become personal

January 25, 2017 26th Annual Ohio Tax Conference

20

Steps to Close the Audit

When all else fails: – After attempting to work with auditor, elevate discussion to the supervisor level – Request replacement of auditor (desperate move that rarely produces good outcomes) – Seek out professional advice and consultation – Move on to contesting the audit following the jurisdiction’s policies and procedures for contested audits

January 25, 2017 26th Annual Ohio Tax Conference

21

Steps to Close the Audit

Appealing audit on statistical sampling grounds – Two grounds for contesting audit:

• “Generally accepted sampling procedures” were not used in

the audit. This expression comes from the AICPA SAS 39 document.

• The sample is not representative of the population or typical

transactions associated with taxpayer’s business.

– Insufficient to merely claim that a sample is “bad” – you must provide concrete evidence in support of the claim – Difficult to win an appeal solely on statistical sampling grounds

January 25, 2017 26th Annual Ohio Tax Conference

22

Thank You!

January 25, 2017 26th Annual Ohio Tax Conference

23

Supplementary Topics

Audit Sampling Roles & Responsibilities – Taxpayer and State Audit Sampling Procedures Audit Sampling Estimation Methods Evaluating an Auditor’s Proposed Sampling Plan Sampling Concepts and Terminology Determination of Sample Size Treatment of Special Situations Projection of Overpayments in Audits and Refund Claims Frequently Asked Questions

January 25, 2017 26th Annual Ohio Tax Conference

24

Audit Sampling Roles & Responsibilities – Taxpayer and State

We begin with a section on planning for a sample audit. A successful sample audit is almost always due to extensive planning and preparation. Well, sometimes we get lucky, but not often. This section describes the roles and responsibilities for the taxpayer and for the state to insure a successful sample audit. Interesting and pertinent quotes about planning and preparation

“By failing to prepare, you are preparing to fail.” ― Benjamin Franklin “If you don't know where you are going, you'll end up someplace else.” ― Yogi Berra “Success depends upon previous preparation, and without such preparation there is sure to be failure.” ― Confucius

January 25, 2017 26th Annual Ohio Tax Conference

25

Audit Sampling Roles & Responsibilities – Taxpayer and State

Who is responsible for what? – Jurisdiction responsibilities

• Notification of Sample Audit/Sampling Agreement
• Sampling plan
• Audit results
• Confidentiality

– Taxpayer responsibilities

• Data for the sample audit
• Input during the entire process
• Results verification

January 25, 2017 26th Annual Ohio Tax Conference

26

Audit Sampling Roles & Responsibilities – Taxpayer and State

The Role of the State – Notification of Sample Audit/Sampling Agreement

• Sampling agreement may or may not be binding
• Sampling agreement should address the following:

» What is being estimated? » What is the sampling unit? » What is the main variable unit (transaction amount)? » What is the audit period and/or data period? » How will the sample be designed (sampling method)? » How will the results be extrapolated from the sample to the sampling population (estimation method)? » Will projections be allowed for tax overpayments (accrued and tax paid)?

January 25, 2017 26th Annual Ohio Tax Conference

27

Audit Sampling Roles & Responsibilities – Taxpayer and State

The Role of the State – Notification of Sample Audit/Sampling Agreement

• Treatment of special topics

» Missing records and/or replacement » Errors that are later corrected » Bad debt items » Tax only items » Progress payments, installments, joint interest billings » Extraordinary items

• California leads the way with the best sampling agreement (BOE-472)

“Strange events permit themselves the luxury of occurring.” Charlie Chan

“The little things are infinitely the most important.” Sherlock Holmes

January 25, 2017 26th Annual Ohio Tax Conference

28

Audit Sampling Roles & Responsibilities – Taxpayer and State

The Role of the State – Population definition

• State should provide a “roadmap” containing the criteria for deriving the

sampling population from the initial taxpayer data downloads

» A third party should be able to follow this roadmap and arrive at the same sampling population » If the derivation process cannot be replicated, the data are corrupt; the criteria are not well documented; or, the auditor did not adhere to the agreed upon criteria » The state should provide copies of the roadmap and sampling population files

• Examples of derivation criteria include:

» Controlling date » Accounts of interest » Transaction type

January 25, 2017 26th Annual Ohio Tax Conference

29

Audit Sampling Roles & Responsibilities – Taxpayer and State

The Role of the State – Sampling plan and results

• State should provide a summary of the sampling plan design with clear

explanations and reasons for selecting the design » Multistate Tax Commission (“MTC”) audit sampling software comment

• Software is increasingly employed by states
• Software produces confusing output format
• State should provide agreed upon schedules in a timely manner

» Throughout the review process, error schedules should be distributed indicating where (in the sampling plan) the errors occur.

January 25, 2017 26th Annual Ohio Tax Conference

30

Audit Sampling Roles & Responsibilities – Taxpayer and State

The Role of the Taxpayer – Provide good data

• Data should be reconciled to Trial Balance or some financial statement
• Produce control totals to ensure auditor uploads the data file correctly
• Perform thorough preliminary analyses to identify problem areas

– Provide meaningful input during entire sample design process

• Account/vendor selection
• Grouping transactions (e.g., assets, expenses, taxed, non-taxed, p-cards)
• Numerical stratification based on amounts

– Verification of results

• Agree on sample items that are in error
• Verify extrapolation (projection) of errors
• Verify penalty/interest calculations

January 25, 2017 26th Annual Ohio Tax Conference

31

Audit Sampling Procedures

In this section, we will discuss the distinction between statistical and non-statistical sampling and describe the basic types of statistical sampling methods

January 25, 2017 26th Annual Ohio Tax Conference

32

Statistical and Non-Statistical Sampling

Statistical sampling:

• Provides objective quantifiable estimates of sample risk based on

statistical theory (“I am 95% confident that the estimated population error is within 5% of its actual value were the population detailed and the exact population error determined.”).

• Statistical sampling requires explicit consideration of sampling risk and

probability theory during all phases of the sampling process, including planning, selection, evaluation and estimation (“projection”).

Non-statistical sampling:

• Provides no objective measure of sampling risk
• Evaluation of sampling risk depends on the auditor’s judgment and

subjective opinion

January 25, 2017 26th Annual Ohio Tax Conference

33

Statistical and Non-Statistical Sampling

Statistical sampling methods: – Simple random sampling

• Random sample is taken from a population

– Stratified random sampling

• Population is subdivided into groups called “strata”
• Transactions have similar properties within each group, but properties

vary across groups

• A simple random sample is taken from each stratum
• Most common random sampling method for sales and use tax audits

January 25, 2017 26th Annual Ohio Tax Conference

34

Statistical and Non-Statistical Sampling

Statistical sampling methods: – Cluster sampling

• Population is subdivided into groups called “clusters”
• The transactions in each cluster have approximately same

characteristics as transactions in population (i.e. each cluster is a replica

• f the population)
• A simple random sample of clusters is taken out of the set of all clusters

– Sequential (or Systematic) sampling

• Transactions in audit population are placed in a sequence, usually

chronologically

• Every kth item in the sequence is sampled (k > 1).

January 25, 2017 26th Annual Ohio Tax Conference

35

Statistical and Non-Statistical Sampling

Non-statistical sampling methods: – Block sampling

• Typical blocks are time periods such as months within audit period (e.g.,

6 months are selected in a 36-month audit period, and all transactions are detailed for these months)

• Store-days in sales audits
• Vendors (e.g., from list of all vendors, choose a subset to detail)

– Convenience (or Opportunity) sampling

• Select records that are “convenient” to audit

26th Annual Ohio Tax Conference January 25, 2017

36

Statistical and Non-Statistical Sampling

Statistical sampling – advantages and disadvantages – Advantages

• Appropriate when data are voluminous
• Efficient – smaller overall sample size compared to block sampling
• Produces representative samples in well-designed statistical sampling

plans

• Most states have moved to statistical sampling

– Disadvantages

• May take nearly as much time as block sampling, particularly in the

planning stage

• Mathematical complexity

January 25, 2017 26th Annual Ohio Tax Conference

37

Statistical and Non-Statistical Sampling

Block sampling – advantages and disadvantages – Advantages

• Potential for less time required for audit
• Possibility of controlling periods selected in the block sample
• Appealing data requirement (e.g., 3 months of data for 36 month audit

compared to 36 months required for statistical sample)

– Disadvantages

• Does not permit measurement of sampling risk
• Sample may not represent entire audit period well – high potential for

bias

• Auditor may short-test block sample by sorting by vendors, then auditing
• nly the first 10 to 25 transactions for each vendor, potentially leading

to biased results

January 25, 2017 26th Annual Ohio Tax Conference

38

Audit Sampling Estimation Methods

In this section, we consider the two basic types of estimation methods and the two estimators used most commonly in sample audits based on stratified random sampling.

January 25, 2017 26th Annual Ohio Tax Conference

39

Audit Sampling Estimation Methods

Types of estimators – Point estimator

• Single number (point on real line) that is estimate of population error amount
• Most U.S. states use point estimate

– Confidence bound estimate

• Lower bound for net assessments and upper bound for net refunds where underpayments are

positive and overpayments are negative

• Typical confidence levels: 75% (Tennessee), 80% (Washington and Utah), 95% (U.S. IRS)
• Example: “The estimated population error amount is an assessment of $100,000, and we are

80% confident that the true population error amount is at least $85,000. “ Assessment is made at lower confidence bound.

• Works in reverse for net refunds: Estimate is -$100,000 (net refund), and upper 80% confidence

bound is -$85,000. Refund is made at upper confidence bound.

January 25, 2017 26th Annual Ohio Tax Conference

40

Audit Sampling Estimation Methods

Statistical estimation methods – Mean per unit estimator

• Compute average sample error
• Multiply average sample error by population size
• Easy and straightforward to compute

– Separate ratio estimator

• Form ratio of sample dollars in error to total sample dollars
• Multiply the ratio by total population dollar amount
• Known widely as “percentage method” (percentage of sample dollars in

error times total population dollar amount)

• Easy and straightforward to compute
• Method capitalizes on linear relationship between sample errors and

sample amounts in each stratum

January 25, 2017 26th Annual Ohio Tax Conference

41

Evaluating an Auditor’s Proposed Sampling Plan

The evaluation of an auditor’s proposed sampling plan is a crucial step in managing a state-initiated sample audit. Most states use stratified random sampling in their sample audits. Accordingly, we consider a proposed sampling plan developed by a Virginia audit as an illustration

• f the evaluation methods.

January 25, 2017 26th Annual Ohio Tax Conference

42

Evaluating an Auditor’s Proposed Sampling Plan

A Virginia sales and use tax sampling plan example – The proposed auditor’s sampling plan for expenses

January 25, 2017 26th Annual Ohio Tax Conference

43

Evaluating an Auditor’s Proposed Sampling Plan

Check list for evaluating a proposed sampling plan

– The plan should satisfy the guidelines for attributes of a good plan

• The tax projection on an average item measure should ideally have a maximum
• f $50,000 for the sample strata in the plan

» The maximum value of this measure occurs in Stratum 1D, and its value is ($570,000/475) x ($600,000/$570,000) x 0.07 (the assumed tax rate) » $1,200 x 1.05263 x (0.07) = $88.42 » (sample average x projection factor x tax rate) » This plan easily meets the $50,000 maximum value for this measure.

• The detail stratum should contain at least 20% of the population base dollars

» The percentage in this plan is $750,000/$2,550,000 x 100% = 29.41%

• Stratum sample sizes should be at least 200 for a commingled population of

taxed and non-taxed transactions

» All the sample sizes easily meet this guideline

January 25, 2017 26th Annual Ohio Tax Conference

44

Evaluating an Auditor’s Proposed Sampling Plan

Check list for evaluating a proposed sampling plan – The plan should satisfy the guidelines for attributes of a good plan

• A lower exclusion threshold of $10 has been used, but we do not know

how many items have been excluded by looking at this plan

• The plan should contain three more strata:

» A stratum with the count and amount of the negatively-valued transactions » A stratum with the count of the $0 transactions » A stratum showing the count and amount for the transactions with amounts greater than $0 and less than or equal to $10

• The plan should also report on the number of positive and negative

matches the auditor made in the process of deriving the sampling population

January 25, 2017 26th Annual Ohio Tax Conference

45

Sampling Concepts and Terminology

Audit sampling has a language of its own which is Greek to most people (literally and figuratively!) But, understanding key audit sampling concepts and terms is very important when negotiating an audit plan and a sampling plan with an auditor. This section introduces, explains and describes the terminology commonly used in audit sampling.

January 25, 2017 26th Annual Ohio Tax Conference

46

Sampling Concepts and Terminology

Sampling risk

– Risk that projected error (assessment or refund) is different from the amount that would have been determined by detailing all items in population – Sampling risk can be reduced by using efficient sampling designs and by increasing the sample size

Non-sampling risk

– Risk that auditor reaches conclusion different from true population error for any other reason than sampling procedures – Non-sampling risk exist even if auditor details population – Potential causes:

• Data file does not represent population
• Tax errors incorrectly determined (measurement error)

January 25, 2017 26th Annual Ohio Tax Conference

47

Sampling Concepts and Terminology

Precision – Absolute – “We want the estimate of the total population error to be within $25,000 of the true value." – Relative – “We want the estimate of the total population error to be within 5% of the true value." Confidence Level – Example: “We want to be 95% confident that the estimate of the population error is within 5% of its true value.” – Interpretation: If all possible samples of a fixed size are taken from the population, 95% of the samples will produce an estimate within 5% of the true population error.

January 25, 2017 26th Annual Ohio Tax Conference

48

Sampling Concepts and Terminology

Confidence interval estimate – Instead of a point estimate, a range of possible estimates is given

• Example: The point estimate via the ratio method of the tax assessment

for a taxpayer is $200,000

• A 95% confidence interval estimate might turn out to be from $160,000

to $240,000

• The 95% point estimate and confidence interval statement:

» “The estimate of the tax assessment is $200,000 and we are 95% confidence that the true population tax assessment amount is between $160,000 and $240,000." » The width of the confidence interval gives an idea of the precision of the estimate

January 25, 2017 26th Annual Ohio Tax Conference

49

Sampling Concepts and Terminology

Confidence interval estimate – Comments about the confidence interval estimate

• 95% confidence is equivalent to 95% probability before the sample is
• drawn. Once the sample is drawn, the probability that the confidence

interval contains the true population tax assessment amount is either 0

• r 1.
• Interpretation: If all possible samples of size n were taken from the

population, 95% of the samples will produce a confidence interval that contains the true population tax assessment amount; 5% of the samples will not.

• The higher the level of confidence (e.g., 99%), the wider the confidence

interval

January 25, 2017 26th Annual Ohio Tax Conference

50

Sampling Concepts and Terminology

Margin of error – One-half the width of the confidence interval Relative precision – Margin of error divided by point estimate Example – Point estimate is $200,000 and 95% confidence interval estimate is $160,000 to $240,000

• Margin of error = ($240,000 - $160,000)/2 = $40,000
• Relative precision = $40,000/$200,000 = 0.20, usually expressed as a

percentage: 20%. » Interpretation: At 95% confidence, we expect the point estimate to be within plus or minus 20% of true population tax assessment amount

January 25, 2017 26th Annual Ohio Tax Conference

51

Sampling Concepts and Terminology

Confidence bound estimate – Instead of a point estimate, an upper or lower confidence bound estimate is used

• Example: The point estimate via the ratio method of the tax assessment for a

taxpayer is $200,000

• A 95% lower confidence bound estimate might turn out to be $175,000
• The 95% point estimate and confidence bound statement:

» “The estimate of the tax assessment is $200,000 and we are 95% confidence that the true population tax assessment amount is at least $175,000. » Relative precision = ($200,000 - $175,000)/$200,000 = 0.125, or as a percentage, 12.5%.

• Used by IRS, State of Washington and Utah (80% confidence bound), Tennessee

(75% confidence bound), and by other states

January 25, 2017 26th Annual Ohio Tax Conference

52

Sampling Concepts and Terminology

Percent difference between sample and population mean taxable amounts – (sample mean – population mean)/(population mean) – Usually expressed as a percentage – Example:

• Population mean = 400 and sample mean = 350
• (350 – 400)/(400) x 100% = -12.5%
• Interpretation: Sample mean is 12.5% smaller than population mean
• Provides idea of how well the sample represents the population in mean

value

• Texas: percent difference must be between -8% and +8% for stratified

random sample

January 25, 2017 26th Annual Ohio Tax Conference

53

Determining Sample Size

The determination of sample size is a crucial step in the development of a sampling plan. If the sample size is too small, the precision of the estimate may be so bad as to render the estimate of the total tax error

• useless. If the sample size is too large, valuable and scarce auditing

resources may be unnecessarily expended.

January 25, 2017 26th Annual Ohio Tax Conference

54

Determining Sample Size

Two approaches for setting sample size: – Fixed minimum sample size/auditor judgment – Use of statistical formulas for setting the sample size – British Columbia: Set maximum sample size at 1,000 – Virginia: Total sample size of 1,000 split among 3, 4 or 5 strata Block sampling – Most jurisdictions use fixed minimum sample sizes (e.g., Texas: minimum of two randomly selected months for each year in audit period) – Some jurisdictions use auditor judgment (e.g., Virginia may use only

• ne month for 36 month audit period!)

January 25, 2017 26th Annual Ohio Tax Conference

55

Determining Sample Size

Simple random sampling – Sample size determined by formula or by setting minimum sample size (e.g., Texas requires a minimum sample size of 250 transactions for simple random samples, which it calls “manual samples”) Stratified random sampling – Use of fixed minimum stratum sample size (e.g., Texas = 100, California = 300, though now more like 400!) – Determination of sample size by formula (e.g., Tennessee with minimum of 70; Florida with minimum of 100)

January 25, 2017 26th Annual Ohio Tax Conference

56

Treatment of Special Situations

Inevitably, special situations arise in sample audits due to the complexity

• f typical audit data. These situations include missing documentation,

bad debt, installment payments and ERS transactions. The treatment of these special situations should be discussed with the auditor and the treatments agreed upon in the planning stage of the sample audit.

January 25, 2017 26th Annual Ohio Tax Conference

57

Treatment of Special Situations

Typical special situations arising in audit sampling include: – Missing documentation

• Treat as 100% taxable
• Use alternate information (e.g., similar invoices from same vendor)
• Do not use spares or replacement items

– Negatively-valued transactions

• Match as many negatively-valued and positively-valued items as

possible before sampling

• Exclude remaining negatively-valued transactions and sample only the

positively-valued transactions

• Include remaining negatively-valued transactions and stratify on

absolute value

January 25, 2017 26th Annual Ohio Tax Conference

58

Treatment of Special Situations

Typical special situations arising in audit sampling include:

– Progress/installment payments

• Know how jurisdiction treats these payments and at what payment tax is

considered due

– Bad debt

• Discuss with auditor what to do when transaction that later results in bad debt

hits the sample

– Tax only items

• Should be treated as non-error if item hits sample

– ERS transactions

• What to do if the auditor demands hard copy (e.g., invoice) for transaction
• Screen shots
• Sitting with auditor at computer to display ERS transactions

January 25, 2017 26th Annual Ohio Tax Conference

59

Projection of Overpayments in Audits and Refund Claims

Not all states permit the projection of overpayments in state-initiated audits. And, very few states permit refund claims to be based on a sample audit conducted by the taxpayer. In this section, we suggest appropriate courses of action to take to overcome these limitations.

January 25, 2017 26th Annual Ohio Tax Conference

60

Projection of Overpayments in Audits and Refund Claims

Approximate statistics on states permitting projection of

• verpayments in state-initiated audits

– Projection of tax paid to vendor

• Twenty-two states permit projection of tax paid to vendor

– Projection of use tax paid

• Thirty-seven states permit projection of use tax paid

Refunds based on samples – taxpayer initiated refund claims – Tax paid to vendor

• Thirteen states allow refund claims for tax paid to vendor based on samples

– Use tax paid

• Eighteen states allow refund claims for use tax paid based on samples

January 25, 2017 26th Annual Ohio Tax Conference

61

Projection of Overpayments in Audits and Refund Claims

The projection policy issues and what to do about them

– State audit sampling policies and procedures are difficult to change

• Takes a considerable amount of time
• Requires a “top-down” policy change process

– Convince auditor to allow sampling for select and specific subsets of transactions

• California will sample accruals if asked as a part of a state-initiated audit

– States that do not allow refunds based on samples will often sample detailed refund claims – Why not include these items in the state-initiated audit or base a refund claim on them based on a sample?

• This is a fairness and equity issue!

– Consider a managed audit that will encompass all transactions with potential errors – overpayments and underpayments.

January 25, 2017 26th Annual Ohio Tax Conference

62

Frequently Asked Questions on Audit Sampling

And, maybe some answers!

January 25, 2017 26th Annual Ohio Tax Conference

63

Frequently Asked Questions on Audit Sampling

Can we participate in the sampling process, and in particular, in the development of the sampling plan? We know we have serious missing documentation problems with our procurement card transactions and have just been notified of an audit. How can we deal with this issue in the planning of the sample audit? Much to our surprise, we were hit with a very large assessment that is due in large measure to missing documentation. The errors were taxed at 100%. Frankly, we were stunned by the size of the assessment and the consequences that may follow for the tax department and the firm. What can we do to try to reduce the part of the assessment due to missing documentation?

January 25, 2017 26th Annual Ohio Tax Conference

64

Frequently Asked Questions on Audit Sampling

The auditor and the jurisdiction take the position that overpayments will not be projected within the audit. So, how is this fair and equitable? I thought the

• bjective of the audit was to determine the correct amount of tax due whether

that leads to an assessment or a refund. Is there anything I can do about this for my audit? The auditor wants to use a stratified random sample for my sales audit, but I am far more comfortable with using a block sample for the audit. Do I have any recourse in this matter? We have reached an impasse with the auditor over several contentious taxability and sampling issues. The auditor is pressing hard to close the audit, so it looks like we are headed toward an objection or appeal. What advice can you give us about an economic analysis or a cost/benefit analysis as we decide between pressing for a settlement or committing to the courts for resolution of the impasse?

January 25, 2017 26th Annual Ohio Tax Conference

65

Frequently Asked Questions on Audit Sampling

It was determined after the sample was drawn that some portions of the population should have been removed before sampling. Can you accurately remove transactions if you replace them with transactions from a replacement sample? Is it a best practice to always include replacement samples in the sampling process? If the audit has been completed through stratified random sampling and the sample has been extrapolated over the population, should vendors be allowed to seek reimbursement of sales tax they paid during the audit period to us? If a small grouping of our vendors occupy the majority of our purchases, would you suggest completing samples on certain vendors to obtain taxability percentages per vendor?

January 25, 2017 26th Annual Ohio Tax Conference

66

Frequently Asked Questions on Audit Sampling

During audit sample reviews, we have encountered situations where taxes are paid on the entire PO as a whole and not necessarily on each invoice. Occasionally, these taxes are not paid appropriately on the PO as a whole and it becomes difficult to determine the amount of error that should be applied to the sample invoice. How should that determination be made? What have you found to be a beneficial way to identify negative matches and remove the matches from the population prior to sampling?

January 25, 2017 26th Annual Ohio Tax Conference

67

Frequently Asked Questions on Audit Sampling

We have a significant amount of progress payments for equipment where taxes are paid on subsequent invoices. If a progress payment

• ccurs for taxable equipment and the tax payment occurs outside the

audit period, or has not occurred yet, how would you suggest treating this situation in the sample audit? If taxes have not been calculated correctly on the entire equipment purchase, how would you suggest treating the transaction in the sample?

January 25, 2017 26th Annual Ohio Tax Conference

68

Reference Materials

California Board of Equalization Form 472. This form provides an excellent vehicle for planning the audit and the audit sample with the

• auditor. It also covers virtually all the special cases that should be

addressed in the planning phase of the audit such as missing documentation, progress payments, bad debt, tax only items, etc. Texas Notification of Sampling Procedures form, while not as effective as the California BOE Form 472 for planning purposes, is a very useful document. In addition, Texas does an excellent job in providing the taxpayer with very detailed information on development of the sampling plan and sample.

January 25, 2017 26th Annual Ohio Tax Conference

69

Reference Materials

Audit sampling software contacts:

– IDEA (www.audimation.com) – ACL (www.acl.com) – TsepWin (www.mtc.gov): TsepWin was written by Richard Kulp for use in Tennessee sales and use tax audit. Statistically, it is excellent, but its use can be challenging for those with limited statistical knowledge. The program can be downloaded for free from the Multi-State Tax Commission’s website. – Multistate Tax Commission – Take the MTC four-day “Statistical Sampling for Sales and Use Tax Audits” training class and get a copy of the MTC audit sampling software with the class.

The FTA “White Paper” written in 2002 is a very approachable description of audit sampling. It is available at the FTA’s website (www.taxadmin.org)

January 25, 2017 26th Annual Ohio Tax Conference

70

Reference Materials

Organizations – COST (Council on State Taxation) www.cost.org

• Excellent publications of materials, updates and references on audit

sampling

• Very active in working successfully with states in changing their audit

sampling policies and practices

• Great networking – a best source for getting assistance on sample

audit issues are your fellow tax directors, and COST provides an effective forum for that.

– IPT (Institute for Professionals in Taxation) www.ipt.org

• Has the CMI certification program for persons conducting tax audits
• Also, a great organization for networking with your colleagues in tax

January 25, 2017 26th Annual Ohio Tax Conference