Thanks for watching our fraud and financial crime update! For more around ethical concerns for payment service providers and acquirers, fraud risk management, and payments industry news, please subscribe: [ Ссылка ]
CHAPTERS:
0:00 Ethics & Risk Appetites Among PSPs & Acquirers
0:56 Crypto Wallets Servicing Unbanked Populations
1:33 The Downside of Consortium-Based Models
2:23 Real-Time Data, Feature Plans & Custom Models
TRANSCRIPT:
How can Acquirers and PSPs help remove unfairness and bias in their fraud prevention processes without losing accuracy?
This is James Hunt and this is the Feedzai Financial Crime News Weekly Update.
If we're looking at the ethical waters that PSPs and Acquirers have to navigate, you’ll typically find the majority of this will be based on their own internal risk appetite. So for example, you will have Acquirers and/or PSPs which will deal specifically with high-risk merchants, so a merchant that deals in the adult space, gambling. Their propensity for chargebacks is a lot higher. So therefore those Acquirers can often charge higher processing fees.
There's also conditions within both Visa and MasterCard which state, High-risk merchants need to be registered as such and therefore adopt kind of higher processing fees and different rules and regulations within the Visa and MasterCard environments.
If we look at some crypto wallet solutions, some of these crypto wallets were actually founded to try and help facilitate the unbanked populations of the world - certain places like Venezuela, where access to banking services may not be as forthcoming. So these companies kind of specialize in those areas. They're actually also very, very appealing to fraudsters.
So I think really when we're looking at kind of ethical issues, can fraud solution providers use data around particular behaviors in demographics and/or countries to influence their decision making process? Well, ultimately in some cases some fraud solution providers will use what is called consortium-based models to influence the outcome or a decision of a transaction where they will constantly look at patterns and trends over certain periods of time to determine whether or not the transaction is risky.
Consortium data is essentially a central data pool where decisions are made. So if I look at this from a modeling context, all of a particular fraud provider’s transactions will flow through their system, they’ll flow through a single model, and then that model will look at those transactions to determine what is good and what is bad. And it will use all of that data to decide an outcome on whether or not it believes that transaction to be fraudulent.
The downside to kind of using that approach is that they often require very extensive rule sets to make sure that false positives are kept to a minimum.
I think those fraud solutions which are good at being able to predict good and bad behavior don't necessarily base their data on consortium-based models. Yes, there is definitely information that can be used to help build models, but everyone's business is different. So therefore what's risky for one merchant may not be risky for another merchant.
So the ability to kind of take that information, then use it in a more real-time way, understanding actually how transactions, et cetera, are performing at the time of payment rather than what's happened in the past, is a much better way to do it and also I would say a little bit more ethical, but without losing accuracy. Just because a certain postcode is well known for fraud doesn't mean that every single resident is going to be facilitating fraudulent transactions.
We would create what we call feature plans, which is where we take known fraudulent features from other customers across the globe and use those features to create custom models. So being able to not just take in learnings from the past, but actually being able to base decisions on things such as real-time data analytics - what we're physically seeing from consumer behavior today - is incredibly important.
What are your concerns with treating customers fairly and without any bias? Why don't you drop us a note in the comments below? Thanks for watching, and that was your Feedzai Weekly Update.
