.svg)
The US Federal Reserve has introduced a new data collection model known as ScamClassifier that aims to promote consistent reporting of scams and improve mitigation strategies.
Payments firms can use ScamClassifier, which launched last week, to standardise how they identify, classify, report and analyse scams and related trends.
The voluntary model is the result of a partnership between the Federal Reserve and two working groups of US payments and fraud experts.
In March 2023, the Fed set up a working group for defining scams and improving fraud mitigation. Participants included representatives from Citi, Mastercard, FIS Global, the American Bankers Association (ABA) and the US Department of Justice (DOJ).
A second working group, set up in the summer of 2023, focused on optimising information sharing related to scams. Participants included representatives of Amazon, Google, J.P. Morgan Chase, Feedzai, American Express and BNY Mellon.
Mike Timoney, vice president of payments improvement at Federal Reserve Financial Services, has led the ScamClassifier project from its design stages to its completion.
Timoney said that by standardising how firms collect data on scams, the payments industry as a whole can respond more effectively and can protect against losses.
“Fraudsters are using similar tactics across organisations to conduct scams, and lack of information sharing helps fraudsters successfully repeat the same tactics,” he said.
“Timely access to fraud information can help improve fraud management strategies, detect scams more quickly and avoid potentially devastating losses within an organisation, as well as within the industry as a whole.”
In 2023, US consumers reported more than $10bn in scam losses to the Federal Trade Commission (FTC) — an increase of 14 percent compared with 2022.
How does it work?
ScamClassifier uses a flowchart structure similar to that of an existing model, FraudClassifier, which was introduced by the Federal Reserve in 2020.
Like its predecessor, ScamClassifier uses a series of questions to help firms classify scams and attempted scams by category and type.
Source: Federal Reserve Financial Services
The model begins by asking whether the incident in question meets the regulator’s definition of a scam, namely: “the use of deception or manipulation intended to achieve financial gain”.
If the answer is yes, the user is then asked what action resulted from the scam: was an authorised party tricked into making a payment, or was an authorised party tricked into allowing the scammer to gain access to an account?
Next, users are asked how the authorised party was deceived or manipulated. The answers to this question are used to categorise the incident as either a goods or services scam or a relationship or trust scam.
In a goods or services scam, a product or service ostensibly offered by a seller is not delivered to the buyer following payment. In a relationship scam, the scammer poses as a trusted third party to manipulate the victim into sending an authorised payment.
In the final step, the user of the model is asked to categorise the incident as one of nine scam types that make up the ScamClassifier taxonomy:
- Merchandise scam.
- Investment scam.
- Property sale or rental scam.
- Romance impostor scam.
- Government impostor scam.
- Bank impostor scam.
- Business impostor scam.
- Relative/family/friend scam.
- Other trusted party scam.
The regulator said that consistent use of the model will allow firms to “speak the same language” when it comes to scam reporting.
“Financial institutions and other organisations should evaluate their processes for fraud and scams to determine how the ScamClassifier model could provide consistency and value in combating scams,” the Federal Reserve said.
“In addition, organisations should assess potential integration of the ScamClassifier model into their existing scam classification case management tools.”
Firms are invited to register to access the full ScamClassifier model, including its supporting terms and definitions.
After signing up, they will also receive notifications of model updates and will be able to provide feedback about the model’s performance and potential enhancements.
How does the model differ from FraudClassifier?
Whereas ScamClassifier focuses on scams that have arisen with new forms of digital payments and communications channels, FraudClassifier focuses on more traditional types of fraud.
FraudClassifier begins by asking who made the potentially fraudulent payment: an authorised or unauthorised party?
It then asks how this party made the payment in question. Depending on the user’s answers to the question, the incident may eventually be categorised as a type of fraud that is not catalogued by ScamClassifier.
For example, embezzlement, false claims, synthetic ID and physical forgery are included in the FraudClassifier taxonomy but not in the ScamClassifier taxonomy.
However, the Federal Reserve said that both models can be used sequentially, and encouraged firms to do so in order to achieve the most accurate results.
Trends in scam fighting
The decision of the Federal Reserve to standardise scam reporting mirrors that of regulators in other jurisdictions.
In Australia, as covered by Vixio, the National Anti-Scam Centre was launched in July last year under the authority of the Australian Competition and Consumer Competition (ACCC).
Similar to ScamClassifier, one of its roles is to offer a centralised database for the reporting of scams from across the payments industry.
The centre also shares this intelligence with other government and law enforcement agencies and with the private sector to disrupt scammers.
In the six months following the launch of the centre, reported losses to scams fell 29 percent compared with the same period in 2022.
