Collections, Predictive Analytics and Taxpayer Compliance Management

1
Collections, Predictive Analytics and Taxpayer
Compliance Management

• John McCalden
• McCalden Consulting
• john_at_mccalden.biz

2
Agenda

• Some Collection Theory
• Decision Analytics
• Taxpayer Compliance Management
• Q A

3
Percent of Cases Aging per Month (SC)
Source South Carolina ARMS
4
Percent of Cases Aging per Month (SC)
100
80
60
Percent
40
-0.6291
y 1.0489x
2
R
0.9988
20
0
0
12
24
36
48
60
Age in Months
97 - 99
Power (97 - 99)
5
Collection Rates,Based on Different Levels of
Performance
100
80
60
Percent Remaining
40
20
0
0
12
24
36
48
60
Age in Months
Forecast -0.2
Forecast -0.5
Forecast -0.6291
Forecast -1.0
Forecast -2.0
Forecast -0.7
6
Effect of Raising the Level of Performance From
-0.6291 to -0.7
100
80
60
Percent Remaining
40
-0.6291
y 1x
2
R
1
20
-0.7
y 1x
2
R
1
0
0
12
24
36
48
60
Age in Months
97 - 99
Forecast -0.7
7
Average Balance by Age of Case
2,500
y 390.49Ln(x) 529.3
y 390.49Ln(x) 529.3
2
R
0.9334
2
R
0.9334
2,000
1,500
Average Balance ()
1,000
500
0
0
12
24
36
48
60
Months
Average Per Case
Log. (Average Per Case)
Source South Carolina ARMS Summary Receivables Report 12/31/2004
8
Application of Aging Curve (-0.6291) and Average
Balance Curve to a Hypothetical Cohort of 10,000
Cases
9
Percent of and Collected, per Month of Aging
10
Comparison of Percent of and Collected, per
Month of Aging
11
Hypothetical Improvement in Collections When
Exponent Increases From -0.6291 to -0.7
Aging
Aging
Cases
Total
Cases
Total




Month
Model 1
Model 1
Model 2
Model 2
Difference
Difference
Improvement
Improvement
1
10000
5,763,100
10000
5,763,100
0
0
0.00
0.00
3
5010
4,963,770
4635
4,592,230
375
371,540
3.75
6.45
6
3239
4,056,100
2853
3,572,724
386
483,376
3.86
8.39
12
2095
3,171,340
1756
2,658,173
339
513,167
3.39
8.90
18
1623
2,705,105
1322
2,203,419
301
501,686
3.01
8.71
24
1354
2,403,706
1081
1,919,059
273
484,647
2.73
8.41
36
1049
2,022,712
814
1,569,578
235
453,134
2.35
7.86
48
876
1,784,201
665
1,354,445
211
429,756
2.11
7.46
60
761
1,614,037
569
1,206,816
192
407,221
1.92
7.07
12
How Do We Transition to a Higher Level of
Performance?
100
80
60
Percent Remaining
40
-0.6291
y 1x
2
R
1
20
-0.7
y 1x
2
R
1
0
0
12
24
36
48
60
Age in Months
97 - 99
Forecast -0.7
13
Use Decision Analytics!!

• Use Information Intelligently to Make Business
  Decisions
• Optimize Collection Activity
• Prioritize Audit Candidates
• Supply Education to the Needy!
• And Repeat
• (Taxpayer Compliance Management Program!)

14
How Do We Use Information Intelligently?

• Forecast Performance (models)
• Appropriate Actions (decision strategies/treatmen
  t scenarios)
• Controlled Experiments (champion/challenger)
• Performance Reporting

15
Actual and Forecast 'Good' Probabilities for
Repeat Filers (SC)
16
(No Transcript)
17
Treatment Scenarios

• Allow low-risk cases to self-cure
• Accelerate high-risk cases to enforced
  collection actions
• Focus collector resource on medium-risk cases
• All scenarios end with enforced collection actions

18
Low-Risk Treatment Scenarios (SC)
19
Medium-Risk Treatment Scenarios (SC)
20
High-Risk Treatment Scenarios (SC)
21
Treatment Scenarios in MA(Initial Design)
Treatment A
Field
Yes
Phone Auto
-
Call

RP
FN

Call

Open or
NOA
NOD
Auto
-
Levy

Med
.
Risk

Research
(
trustee
)
RP Deem
Assets
(
High Balance
)
No
FR
Auto
-
Levy
OCA
Case
Assigned
LIEN
Treatment B
Assign
Yes
FN
Phone Auto
-
Open or
NOA
NOD
NIL
Call

High Risk

Auto
-
Levy
Research
Assets
(
Low Balance
)
RP
Bus
.
No
FR
Auto
-
Levy
OCA
Treatment C
Wage Levy
LIEN
Yes
Phone Auto
-
Wage

High Risk

NOA
NOD
NIL
Auto
-
Levy
Research
Levy
(
High Balance
)
(
Low Balance
)
Case
No
Ind
.
FR
Auto
-
Levy
OCA
LIEN
Treatment D
LIEN
Field
Phone Auto
-
Open or
Call
NOA
NOD

High Risk

FN
Yes
Research
Assets
RP
-
Propose
(
High Balance
)
Call
Bus
.
Deem RP
No
Low
FR
Auto
-
Levy
OCA
Treatment E
NOA
(
Very High
Call
Med
Balance
)
Ind
.
Assign
High
Low
Treatment F
NOA
(
Very High
Call
Med
Balance
)
Bus
.
Assign
High
Day
Day
Day
Day
Day
Day
Day
Day
Day
Day
1
30
2
14
45
61
90
97
105
111
22
Champion-Challenger Evaluation
Primary Primary
Challenger 1 Challenger 1 90 Grossed Up
10 Grossed Up
Available 450 500 50 500
Collected 90 100 11
110 Collection 20 20 22 22

23
2004 State Tax Revenue
Source FTA Web Site - U.S. Bureau of the Census and Bureau of Economic Anaylsis.
24
Probability of Making an Assessment PA Data
Actual
Forecast
25
Sort Candidates by Cell and Probability (PA)
cum_ Obs hours cum_yield
myrank . . . 3507 676851 196963641 3507
3508 677044 197019804 3508 3509
677237 197075967 3509 3510 677430
197132130 3510 3511 677623 197188293
3511 3512 677816 197244456 3512
3513 678009 197300619 3513 3514
678202 197356782 3514 3515 678395
197412945 3515 . . .
26
Collection Action Transition Probabilities
(Markov-Chain Analysis)
To
FTF
Assessment
Payment
Levy
Lien
Field Visit
Revoke
Seize
Responsible
Cure
Notice
Plan
Party
New
0.55
0.35
0.00
FTF Notice
0.00
0.70
0.10
0.20
Notice of Assessment
0.20
Assessment
0.20
0.20
0.05
0.25
0.30
Payment Plan
0.05
0.05
0.20
0.70
Levy
0.60
From
Lien
0.70
Field Visit
0.40
Revoke
0.10
Seize
0.50
Responsible Party
0.80
27
Probability of Curing by Age of CE and Type of
Collection Action
28
(No Transcript)