How Reuters analysed the data FINRA doesn't want you to see

By Benjamin Lesser and Elizabeth Dilts

NEW YORK (Reuters) - FINRA requires brokers to disclose 23 types of incidents that might give investors concern, such as regulatory sanctions, lawsuit judgements and bankruptcies. The regulator then publishes those disclosures on its Brokercheck website, allowing investors to search the backgrounds of individual investment advisors.

But FINRA refuses to release that data in bulk form, which would allow for database analysis to find patterns at brokerages that might also trouble investors - such as a firm's propensity to hire brokers with a history of client complaints or regulatory run-ins.

Reuters obtained that bulk FINRA data from researchers at Columbia University Law School Datalab, who wrote code to extract it from FINRA's Brokercheck website.

To analyse it, reporters Benjamin Lesser and Elizabeth Dilts chose to analyse only 12 of the 23 required FINRA, isolating only those incidents that might cause investors the most concern.

The reporters examined firms with at least 20 brokers, and FINRA-mandated disclosures between the years 2000 and 2015.

They found 48 firms where at least 30 percent of brokers have such FINRA flags on their records. [L1N1J50RV]

(For full coverage, see: http://www.reuters.com/investigates/special-report/usa-finra-brokers/)

The FINRA flags included regulatory sanctions, customer complaints that resulted in a payment, criminal cases that resulted in a plea agreement or conviction, and bankruptcies.

The incidents analysed by Reuters varied in severity. Some involve final legal determinations of misconduct, such as criminal convictions, civil judgements or regulatory fines. A very small percentage of those cases remained under appeal by the broker or firm involved, according to the FINRA data.

The analysis also included incidents with no finding of misconduct, but that did come at a cost to the firm or employee involved - such as broker terminations after allegations of misconduct, or settlements in which firms paid restitution to customers but admitted no wrongdoing.

Reuters used the data provided by Columbia Law School researchers to identify firms meeting the above criteria. Reporters then manually updated the data, using the Brokercheck site, to include FINRA-mandated disclosures by brokers at those firms through the beginning of 2017.

Below are the definitions - written by FINRA - of the 12 incidents used in the Reuters analysis.

(Editing by Janet Roberts and Brian Thevenot)