In 2005 I was the AML Officer for Riggs Bank overseeing an AML team and hordes of consultants.  Riggs was caught up in a whirlwind of regulatory, law enforcement and Congressional investigations.  Selling the bank was the only option, and before approving any sale, the regulators wanted a bunch of projects completed.

One of the projects involved negative news screening.

We “batch ran” a list of about 2,000 high risk client names against the leading negative news database.  That database, which I will not name, remains the industry’s “leading” source of negative news screening.

After nearly 4 months of work and hundreds of thousands of dollars in consulting costs here is what we found:

2,000 high risk customers were run against the negative news database generating over 30,000 “hits.”  From those “hits” there were zero (not one) investigation that uncovered information we needed to report to FinCEN.

Every so called “hit” (aka “alert”) was in fact either not related to the Riggs customer, the information in the database was incomplete, or the underlying information in the database was no longer available for review by the consultants.

Ultimately, the entire exercise was a waste of time and money.

What Exactly Is Negative News Batch Screening

Negative News batch screening has become a nearly universal process for financial institutions of all types and sizes.  Why?

There is a perception among AML professionals and regulators that the database products used to batch screen names contain vast amounts of information about people and entities that have engaged in, or alleged to have engaged in financial crime activity. If this were true, then negative news batch screening would in fact be a good use of resources.  But the reality is much different.

Negative news databases actually include only a very small amount of publicly available financial crime risk information.  How small?  Consider this:  the leading negative news database has about 2.3 million of what are often called “profiles” of individuals and entities who, according to the database company, present some sort of financial crime or regulatory risk.

Two million names may sound like a large number.  It is not. This year alone, more than 5 million new pieces of negative news data will be published.

This means that over the 15 years since AML groups first began using these negative news applications, the leading databases have not yet accumulated the equivalent of 5 months of publicly available negative news information!

Shouldn’t this be a problem for anyone relying on these datasets to actually find negative news information? This includes AML officers, AML investigators, and the regulators.

The Problems with Negative News Databases

They Miss More Information Than They Find

The fundamental challenge for any database is data. “Batching” or automating the screening of a list of customer names requires that there be another list against which your customer list is compared.  Such lists must be “structured.”  Think of a structured database as an Excel sheet with columns and rows.

To create these negative news databases, employees of the database companies read news articles and then select limited information to enter into the database. This is a slow, tedious and highly subjective process.  This is a problem of all databases that rely on human curation.  No wonder the datasets are so small.

And perhaps they are even smaller than many realize.

When these databases say they have 2 million records or “profiles” how many of these profiles are actually in fact just sanctioned entities listed by OFAC and other similar government agencies from around the world?  Including long lists of Politically Exposed Persons is another way databases increase their number of profiles.

No doubt every AML group needs to know if they are dealing with a sanctioned party or a PEP, but this begs a very important question that users of these databases should be asking:  Of the 2.3 million names how many are actually names harvested from true negative news?

Endless False Positives

One would think that with so little data in the databases, it wouldn’t be much work to run a list of customer names and find the very few (if any) that turn up in the database.  Wrong.

Instead of actually finding negative news that is about your customers, the AML department is deluged with “false positives” which is another way of saying, “unrelated results.”  This happens because names are not unique identifiers and trying to “match” one name against a batch of similar names creates all sorts of problems.

Today’s Reality

After more than 20 years in AML, I realize simply doing away with a process like negative news batch screening would raise eyebrows.  However, it is important that those paying for the direct and indirect costs of such batch screening negative news (and the regulators) realize what it is exactly that is being accomplished, what is not, and at what price.

What is Needed for Negative News Screening

Many parts of AML compliance remain untouched by the technological advances that have changed nearly every aspect of our lives.  In 2016, AML professionals and the regulators should expect technology to enable much more from negative news applications, including:

• Finding as much publicly available negative news as possible – in real time.
• Drastically cut down on false positives or unrelated information.
• Simplify search techniques so entire AML departments work consistently.
• Easy and fast ways to fully document all negative news search results.

Negative news searching remains a fundamental element of identifying, reporting and reducing risk.  It is time to improve how it is done.

TransparINT solves the problems caused by the existing crop of outdated negative news applications.  Contact me at david@transparint.com or (202) 641-7485 to see how