2 Sep 2008 by Chetan Shah.

While sourcing watch list consider a watch list provider like www.world-check.com. Rather than manually collecting watch lists from different agencies like OFAC or Bank of England, the software solution which automatically fetches watch lists (via ftp or some other method) is much more reliable and scalable. Keep the manual uploads to your watch list scan application as minimal as possible.

Posted in AML | No Comments »

25 Aug 2008 by Chetan Shah.

Watchlist Scanning applications which have the capability of reporting “alerts” should consolidate alerts arising from the same “Entity” from different watch list as 1 alert rather that reporting the same entity multiple times.

This will eliminate duplicate work as multiple analysts can work on the same entity if she is reported more than once due to the fact that she is on different watchlists.

Posted in AML | No Comments »

17 Jun 2008 by Chetan Shah.

While parsing last names from SDN watchlist (or any other watchlist for that matter. BOE , OSFI etc.) , make sure that the parsing logic takes care of multiple names in last names.

For example :

Last Name : Giraldo Sarria

In this case the financial entity may have registered this person as having Sarria as their last name, in which case the parsing logic may completely ignore this SDN entry as a potential match because the last name from SDN side is Giraldo Sarria and on the financial entity side it will be Sarria.

On the same lines is the scenario of hyphenated last names like Smith-Johnson

Posted in AML | 1 Comment »

3 Mar 2008 by Chetan Shah.

One of the key components of an efficient watchlist matching software application is its “matching” matrix. This matrix defines on which fields should the application match the enterprise data with the watchlist data. This matrix also defines the priority of each match. I.e. : If there is a customer in the company whose last name, first name, complete address and nationality perfectly match with what is on the watch list side then the risk analysts need to look at this customer asap. This can be done by assigning a priority for each match pattern which the software application looks for as part of the matching process between the enterprise and watchlist data.

Axiom :
The Matching Matrix should be reviewed and updated on a regular basis.

Reason #1 :
As the enterprise grows, its data characteristics change. The match patterns created last year might not be efficient this year due to data changes. Old/Outdated patterns can create a lot of false positives for Risk Analysts.

Reason #2 :
The watchlist data supplied by different agencies also undergo cleansing. Data which was harder to extract in the past might be easier to extract now. For example : prior to the SDN’s XML Format, the date of birth was in the comments field of each record and hence was not guaranteed to be in a proper format.

Posted in AML | No Comments »

31 Jan 2008 by Chetan Shah.

While parsing the OFAC SDN list provided by US Treasury, the alerts should be generated at the main entity level.

SDN list consists of a parent entity. The child/ren of this parent entity are the “AKAs” by which the organization or an individual is also known.

For example :

6365
Usama bin Muhammad bin Awad
BIN LADIN
Individual

SDT
SDGT

4227
a.k.a.
strong
BIN LADIN
Usama

4757
a.k.a.
strong
BIN LADEN
Usama

4771
a.k.a.
strong
BIN LADEN
Osama

4772
a.k.a.
strong
BIN LADIN
Osama

4773
a.k.a.
strong
BIN LADIN
Osama bin Muhammad bin Awad

If alerts are generated at a child level, possibly the system will generate more alerts or duplicate alerts. For large financial institutions this can create a lot of unnecessary work loads for Risk Analysts. This in turn creates a backlog of alerts to be looked at. Having backlogs of alerts creates a major financial/reputation risk for the institution as there may be a bad guy on their books but they could not take action in time because of them being severely backlogged on their alerts processing.

Posted in AML | No Comments »

28 Jan 2008 by Chetan Shah.

Never try to find out whether a company has a bad guy on their books by just matching a SDN list entry with the company customer data on basis of just first name and last name. You will be surprised how many “Jose Luis” you will find in your company database.

A better approach is to use date of birth in addition to the first name and last name. This rule will generate more accurate matches and your risk analysts will thank you for not generating “False Positive”

Posted in AML | 1 Comment »