© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d . E x p e r ia n a n d th e m a r k s u s e d h e r e in a r e s e r v ic e m a r k s o r r e g is te r e d t r a d e m a r k s o f E x p e r ia n In fo r m a t io n S o lu t io n s , In c . O th e r p r o d u c t a n d c o m p a n y n a m e s m e n t io n e d h e r e in a r e th e
t r a d e m a r k s o f th e i r r e s p e c t iv e o w n e r s . N o p a r t o f th is c o p y r ig h te d w o r k m a y b e r e p r o d u c e d , m o d i f ie d , o r d is t r ib u te d in a n y fo r m o r m a n n e r w i th o u t th e p r io r w r i t te n p e r m is s io n o f E x p e r ia n .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
Optimizing Income Tax Return Identity Fraud Prevention
2© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
§ Common Fraud Definitions that drive Tax Refund Fraud Detection
► Why fraud definitions are critical
§ Tax Refund population
► Who’s asking for refunds?
► Who are we contacting for verification?
► More importantly– where do agencies look for fraud dollars?
§ Putting it all together- definitions, verifications and outcomes
§ Research indicates the common definition may not be accurate
Content
3© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
§ Generally accepted, two-part definition of “ID fraud” used in Tax Refund Fraud.
► Confirmed fraud– where a risky return generates a notification to which a real taxpayer responds and claims to have not filed the refund request
► Non-responders- who receive a notification and do not respond
§ Definitions are critical
► Dictate who is impacted by fraud prevention efforts and who is not
► They become a self fulfilling prophecy
► Dramatically impact perceived false positives
Common Definitions
4© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
Majority of Requested Refund Amounts are low
5© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
… and people who request large refunds are much more likely to be flagged as suspicious
6© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
But when we convert “percentage of returns” to dollars, returns with lower refund amounts get plenty of attention in terms of suspected fraud
7© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
Using the common definition of “fraud”
8© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
Separating “fraud” into confirmed fraud vs non-responders
9© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
An abrupt decline in non-responders
correlated to a dollar range
Low volume of confirmed frauds
in low dollar ranges
10© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
1. Notifications sent to low-dollar cases don’t reach a “real” taxpayer that contacts the agency to confirm fraud. This seems unlikely.
2. The notifications do reach the “real” taxpayer, who is willing to walk away from the requested refund
► These are actual taxpayers using their own identities that choose not to engage with the agency
● Possible misrepresentation of tax information to increase refund
● Possible mistrust of the validity of the notification
Two possible explanations
11© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
Evidence that non responders may not be fraud
12© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
Evidence that non responders may not be fraud (2)
13© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
… so where do we focus our effort to stop fraud?(light gray overlay on right axis shows the total dollar amounts of returns that get a notification for each refund account)
Non Responders Confirmed Fraud
14© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
There is significant emphasis on low-dollar refunds where benefit comes from non-responders and not fraud
28% of all dollars associated with a notification occur in dollar ranges (<$2,200) where non response is very high and confirmed fraud is low
15© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
§ Using the common definition of fraud has two pitfalls
► Liability -- Non Responders may surface and verify
► Integrity– Based on the volume and dollars, analytics built to find fraud that includes non-responders, will drift away from “confirmed” fraud
Impacts
16© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
§ Experian advocates the exclusive use of “confirmed” fraud for building analytics that prevent tax refund fraud
► Generates lower overall false positives
► Less Ambiguity- Better segments results into either confirmed fraud or
verified good
► Stops the analytics from drifting toward the wrong target
§ More details on the analytical approach available
A better analytical approach
17© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .
§ If the dollars you are saving come from non-responders, you should be concerned
► Evidence indicates a significant number of these are real taxpayers
► The dollars to be saved by chasing good taxpayers away from refund requests are simply larger and easier to create than finding confirmed fraud
► Over time, the analytics will focus more on generating no response- -leaving ID fraud room and incentive to grow
§ Focusing on non responders
► May create liability for the agency
► Will erode savings over time if/when taxpayers get more comfortable with engaging with the agency
► Takes the focus off of confirmed fraud and will corrupt the analytics over time
Wrap Up
18© 2 0 1 6 E x p e r ia n In fo r m a t io n S o lu t io n s , In c . A l l r ig h ts r e s e r v e d .
E x p e r ia n P r o p r ie t a r y a n d C o n f id e n t ia l .