Fraud Analytics and Recovery

by Jul 27, 2016Financial Services, Success Stories

Background and Challenges

Risk Intelligence and Control (RIC) department of a leading bank was facing challenges in controlling the fraud at application level for unsecured lending products. RIC department of the bank is responsible for document processing and identifying frauds at application level itself. Due to limited resources and huge amount of applications, the bank realized it was practically impossible to scan each and every application form and do advanced verification processes like CPV etc.

Our challenge was to identify potential frauds at pre RIC stage and to help RIC in more scientific sampling to increase hit rate within limited resources. The scorecard was to be used at the acquisition stage to detect possible fraudulent cases as early as possible.

Our Approach

To develop a predictive model that would predict the future propensity of a Consumer committing fraud at application level in the pre RIC stage using scorecards on unsecured loan portfolios (Personal Loan, Business loan and Small Business loan).

As per the requirement of the bank, we set out to predict the propensity of Consumers to turn fraud within first three months of the relationship and label the Consumers according to their likelihood of a specific behavior as defined below:

  • Good: Any Consumer whose application is approved by the bank
  • Indeterminate: An account will be termed as indeterminate if the Status = ‘WIP’ (Work in Progress) or ‘Rejected’
  • Bad: All accounts provided as Fraud

The bank wanted to integrate the scorecard on its system such that it could automatically throw alert for the cases that required high attention of RIC. We developed a linear model so that it could be easily integrated with the client’s internal system.

Results and Implementation

Each Consumer is given a score of likelihood to turn fraud in next three months. All applications were scored once and the scores were usable for entire lifetime. Using these scores, alerts are raised for most suspicious cases and these cases are picked by RIC for a thorough scanning. Using this scorecard, RIC was able to increase its fraud detection significantly.

Related Posts

[et_pb_dfbm_blog dfbm_fullwidth=”off” article_distance=”20″ item_animation=”faded” custom_posttypes=”post” include_post_categories=”248″ posts_number=”3″ post_content_background_color=”#f9f9f9″ content_absolute=”on” content_absolute_event=”off” thumb_same_height=”on” define_thumb_height=”300″ header_tag=”h3″ show_content=”excerpt” show_limit_words=”on” show_comments=”off” show_limit_categories=”on” show_author=”off” show_date=”off” read_more=”off” show_pagination=”off” use_inner_shadow=”off” disabled_on=”off|off|off” _builder_version=”3.19.1″ post_body_line_height=”1.5em” post_header_font=”Montserrat||||||||” post_header_font_size=”17px” post_header_line_height=”1em” post_meta_font=”Montserrat||||||||” post_meta_text_color=”#000000″ post_meta_font_size=”11px” post_meta_line_height=”1.2em” post_meta_link_font=”Montserrat|300|||||||” post_meta_link_text_color=”#000000″ post_meta_link_font_size=”11px” post_meta_link_line_height=”1.2em” post_content_line_height=”1em” text_orientation=”left” custom_css_filt_blogs_post_meta=”font-weight: 200;” custom_css_filt_blogs_post_meta_a=”font-weight: 200;” saved_tabs=”all” /]