During the last year, the release of two smartphones with fingerprint readers by two different manufacturers was met with a lot of excitement. People in the payments industry were keen on the ability of the new phones to better authenticate mobile payments. Fingerprints are one of several biometric methods used today to supplement passwords.

Biometrics refers to techniques that use measurable physical characteristics that lend themselves to automated checking techniques. In addition to fingerprints and vein recognition, biometrics can include voice, facial, and iris recognition, and even DNA matching, among others.

As the Federal Reserve’s report Consumers and Mobile Financial Services 2014 noted, consumers’ security concerns are a big barrier to the adoption of mobile banking. Mobile proponents believe this barrier can be reduced with the additional security features that mobile phones can provide, along with consumer education. There is no question that the mobile phone offers a number of ways to authenticate the user more positively, using both overt and covert methods. One well-known covert option is the smartphone’s geolocation function, which allows verification that the phone is in the location it’s supposed to be. Another covert method is “device fingerprinting,” whereby a number of digital characteristics about the consumer’s phone can be captured and used to verify that the phone being used is the one originally registered.

The most common overt biometric methods being tested today are fingerprint and facial recognition. While only a small number of mobile phones in use today in the United States have fingerprint readers, the vast majority have a camera that could support a facial recognition application. Both of these biometric methods are minimally invasive.

The key difference between biometric verification and user ID and password verification creates the greatest challenge for implementing biometrics authentication: with passwords, unless there is a 100 percent match between the data on file and the data the user enters in trying to gain access, the request is automatically rejected. It may be the legitimate user trying to gain access but maybe he or she forgot the password. Nevertheless, the system rules block access until the user’s identity can be authenticated through some other means. On the other hand, the nature of biometrics is such that a 100 percent match between the stored template value and the live template value is rare—possibly because of differences in lighting conditions or angles when biometric measurements are made, or differences between readers, or some other reason. To deal with this gap, the manager of each application has to determine an acceptable accuracy level for both false-positives (whereby a party incorrectly matched is authorized) and false-negatives (whereby the authentic party is denied access). Naturally, false-positives pose the greater threat. False-negatives generally just involve some level of inconvenience until the individual can be authenticated and provided access.

No matter what biometric authentication methodology a system uses, the most important step is validating each customer’s biometrics upon enrollment in the program. We will discuss this issue and other challenges for biometric programs in future issues of Portals and Rails.

By Dave Lott, payments risk expert in the Retail Payments Risk Forum at the Atlanta Fed