Credit application fraud management solution in OTP Bank
Attila Balaton
OTP Bank - Hungary
Fraud and Approval System Analysis Department
Banking Fraud Roundtable
4th June 2015, Athens
Summary
• Introduction of OTP Group
• Business need and anti fraud project scope
• System design and current application
• Usage of Visual Analytics
OTP Group is the biggest independent banking
group in Central Eastern Europe
OTP Bank Russia (2008) Russia
JSC OTP Bank (2006) Ukraine
OTP Bank Romania(2004) Romania
DSK Bank (2003) Bulgaria
CKB(2006) Montenegro
OTP banka Serbija(2007) Serbia
OTP banka Hrvatska(2005) Croatia
OTP Banka Slovensko(2002) Slov akia
OTP Bank
Hungary
In 2013 the OTP Group achieved
146 billion HUF (~0,5 billion EUR)
corrected consolidated profit after
tax. The profitability, liquidity and
the capital adequacy of the
Group is still outstanding in
international comparison.
OTP Group is offering universal banking services to more than 13 million customers in
9 countries via 1400 branches and more than 4000 ATMs.
OTP Group highlights• OTP is a dominant banking player in Hungary, founded in 1949 (privatization in 1995 –
introduced to Budapest Stock Exchange).
• Currently the bank is characterized by dispersed ownership of mostly private and institutional
(financial) investors.
• OTP Bank has completed several successful acquisitions in the past years, becoming a key
player in the region. Besides Hungary, OTP Bank currently operates in 8 countries of the region.
• Around 43.000 employees in the region, more than 10.000 billion HUF (around 33 billion EUR,
1/3 of Hungarian GDP) total assets.
• Consolidated net loan-to-deposit ratio is 89,0%, NPL coverage is 84,4% and consolidated core
Tier 1 capital ratio is 16,0% at the end of 2013.
• Despite the intense competition OTP Bank market position is stable in several segments, as well
as in terms of profitability and stability belongs to the European frontline.
Anti-fraud project scope – Business needs
• Credit risk relevant fraud detection solution system, which is integrated to application system and
approval process.
• Decrease significantly credit risk fraud losses and better early vintages in OTP Group.
• Implement for all retail products and easily extending to subsidiaries.
• Special know-how and new technics have to be acquired, which is supported by analytics andstatistics.
Early Vintage - Fraud Indicator (%)Unsecured
Mortgage
• Initial high level break-even calculation:
• Approximately 150 billion HUF (0,5billion EUR) retail new annually,
• Assuming 0,8-1,0% fraud rate and
performing fraud prevention systemwith 30-50% expected hit rate,
• Approximately 400 million HUF(around 1-1,3 million EUR) prevented
high risk fraudulent cases.
2013 status quo and selected vendor
6
• Fraud KRI became part of business team’s target.
• Decision on establishment of dedicated centralized fraud investigators’ team which would work on
high risk customers.
• Fraud prevention system would generate alerts via automated way.
• SAS selected as provider:
• ‚High-tech’ solution (like advanced network analysis)
• Positive experience with previous projects in OTP
• Existing and working SAS operational risk solution – OTP Group is working under AMA
• Full Hungarian team is supporting the project with international experts
• Already used SAS tools and applications.
SAS Fraud Solution – Logical system architecture
7
SAS Fraud solution
8
1. Data integration: powerful, user configurable and comprehensive data integration component to
draw in data from all relevant sources and also provide a component for name and addressresolution and verification.
2. Alert generation: advanced profiling engine that applies business rules, anomaly detectionalgorithms, predictive models, and social network analytics on the to raise alerts on entities for
fraud.
3. Social Network Analysis: build links between entities and uncover the hidden relationships thatexist within a customer’s data.
4. Alert Management & BI Reporting
5. Case Management: provides a systematic means for facilitating the investigation and capturingand displaying all information to an investigation.
6. Intelligent fraud repository: when fraud referral/case completed, the results are stored within theIntelligent Fraud Repository as known outcomes and the models are using them for improving their
performance and efficiency.
7. Alert administration: possibility to access the business rules, models, and network analytics. Itprovides the ability to generate and test the validity of new fraud business rules and analytical
SAS solution – Hybrid analytical approach
9
Using a hybrid approach for fraud detection
- For each element different SAS software would be used, to fulfill data management, business intelligence and network analysis technologies.
- Used SAS Software: Dataflux, Data Integration Studio, Enterprise Miner, Enterprise Guide, Visual Analytics
Investigator opening screen
Investigator screens: Scenarios
11 out of the 100 accounts connected to the customer’s employer have arrears or
sold. The customer requested 2 loans in the past seven days. There are
different start_date_at_employer and maritial_status in these applications.
Investigator screens: Social network
Different colors for different relations between entities and applications.
• Yellow: Home address
• Orange: Work address
• (Black: unknown relation, data is not available)
Possibility of growing the network by clicking on (+) sign
Reports for system operation
Reports for BPMS
Visualisation of statistics
Reports for scenarios