Fingerprint Imaging of Welfare Applicants and Recipients
The use of a biometric is the only way to positively eliminate recipient identification fraud. Five requirements are fundamental to the use of any biometric technology in a welfare application. The following is a list of these primary requirements for positive identification:
- 1. Biometric Definition: The "biometric" technique for positive identification must be based on measurement of a feature or characteristic that is provably unique to every individual, and that is invariant, repeatable, and capable of automated measurement and comparison ("matching").
- 2. Identity Verification: The biometric must be capable of rapid, automatic one-to-one Identity Verification matching, with a low error rate (near 0%) for both False Rejection (Type 1) and False Acceptance (Type 2) errors. Identity Verification is required for proof of identity for benefits access.
- 3. Identification Matching: The biometric must be capable of automatic one-to-many Identification Matching, with a high positive identification accuracy (near 100%) for identification matching when an acceptable-quality biometric sample is searched against a biometric registration database containing a large number of individual records, and when a corresponding biometric record from the same individual is contained in the database. Identification Matching is required for detection and elimination of multiple registration fraud ("double-dippers").
- 4. Technological Maturity: The biometric technology must be developed to the point that it is in productive operational use, it is routinely used in both Identity Verification and Identification modes, and it performs successfully for large, non-specific populations. This requirement ensures that: (1) the biometric and the necessary related technology are sufficiently well developed to provide a stable base for development of a large scale application without substantial additional research and development cost; and (2) systemic problems common to biometric technologies have been identified and resolved.
- 5. Legal Acceptance: The biometric technology must be accepted as a legal proof of identity. In denial of public assistance benefits or prosecution of multiple registration fraud, the biometric must be capable of use in a court of law to establish positive proof of identity and to support response to legal challenges of decisions based on the biometric identification process.
With these fundamental requirements as a basis, fingerprints represent the only viable current biometric technology for welfare applications. Consider the following:
Led by the development of automated fingerprint identification technologies in the mid-1970’s, a large number of automated biometric techniques have been invented, and the introduction of new techniques has increased in the last few years. These biometrics fall into two classifications: those based on physiological characteristics (e.g.: fingerprints, facial characteristics, eye retina and iris patterns, hand geometry, hand blood vessel patterns and bone structure, DNA, etc.); and the biometrics based upon behavioral characteristics (e.g.: voice recognition, signature patterns and dynamics, and keystroke dynamics). Of these, the majority have been developed and tested only for Identity Verification applications. Only fingerprint biometrics and hand geometry have experienced any level of success in the marketplace (Requirement 4).
With the exception of the well-established fingerprint technologies, the developers of only two other technologies, the eye-retina and iris pattern biometrics, claim sufficient uniqueness to allow their use in Identification applications (Requirement 3), although little or no statistically-significant testing has been performed to substantiate the claims. Identification matching using hand geometry and facial recognition biometrics has been attempted but both techniques have failed to perform adequately.
Based on twenty years of continuing technology development and successful operational experience, fingerprint biometrics remain the only scientifically and operationally proven biometric technology, the only biometric technology with a mature infrastructure of supporting technologies, and the only technology supported by national and international standards. Technology advances made over the past few years have resulted in off-the-shelf fingerprint ID systems with the capability to complete duplicate registration searches of very large databases in seconds.
Fingerprint identification is the one and only internationally accepted legal method of positive biometric determination of a unique identification.
This is not to say that some other biometric technology, either a current one or one that has not yet been invented, might eventually be proven to be cost effective and to work as well as fingerprints for large-population identification applications. However, the following major considerations suggest that a very lengthy and potentially costly process will be required to establish a second proven and accepted biometric identification method:
- the amount of development required to establish a large-scale operational infrastructure;
- the rigorous scientific and statistical testing required to prove the uniqueness of the identification technique;
- the technology development time to allow introduction of competing products to eliminate the problems associated with single-supplier technologies; and
- the formal process required to establish legal precedents through the courts.
Finally, although user acceptance is not one of the fundamental biometric requirements, negative connotations of traditional fingerprinting are often cited by advocacy groups as a major factor in arguing against the use of fingerprints. However, an independent study of the AFIRM system in Los Angeles County by Ernst & Young reports that ninety-seven percent of the roughly 500,000 welfare applicants had a positive or ambivalent reaction to the use of fingerprints; only three percent voiced a negative reaction during the registration process. With positive education of welfare applicants regarding the benefits to them of a fingerprint security program, the negative response rate will effectively vanish. The fact is that many, if not most, Americans have had their fingerprints taken for employment or licensing and view the use of fingerprints as commonplace. Two-thirds of the 30,000 fingerprint checks processed daily by the FBI are for civil (non-criminal) purposes. At least three studies h ave shown that law abiding citizens do not object to the use of fingerprints.
The fact that fingerprint biometrics are the only current biometrics which satisfy the requirements of the welfare application, combined with the rapidly-expanding implementation and acceptance of fingerprint biometrics in civil applications across the country, leads to the conclusion that fingerprint identification is the biometric process which should be selected for use in welfare administration.
The following paragraphs provide a brief description of some of the major implementation considerations associated with the development of a fingerprint system for fraud prevention. These are not intended to be comprehensive solutions, but to serve instead to highlight some of the important issues that must be addressed in proceeding from a system concept to a successful operational system.
1. Eliminating "Double-Dipper" Fraud
The implementation of a fingerprint identification system to eliminate welfare fraud is clearly a technically feasible undertaking, and the payoff in terms of real dollar savings can be immense. Fingerprint identification is a mature technology that has been used successfully by the law enforcement community for decades, and has been proven to be effective in eliminating welfare "double-dippers" by the Los Angeles AFIRM system. The primary issues in implementing a fingerprint identification system are the typical systems design considerations of system topology (numbers of terminals, hosts, nodes, interfaces, etc.), system capacity (number of current records, growth, archiving requirements, etc.), system performance (identification speed and accuracy requirements, query response timing, etc.), and operational considerations (initial enrollment processes, records management, update procedures, etc.). These considerations, which are sometimes very complex, require the development of a detailed set of specifications which define the functional requirements, performance requirements and operations concepts for the system. Without a detailed statement of the specifications, implementation of a complex system is unlikely to be accomplished successfully.
2. Eliminating Fraud at Benefits Access Points
Also technically feasible is the expansion of the fingerprint identification system to provide identity verification at benefit delivery points, which will accomplish the elimination of many of the forms of recipient fraud. Here, decentralized verification is the preferred approach, using a counterfeit-resistant ID card with encoded biometric information. New low-cost products are on the horizon which integrate a fingerprint verification device with secure data storage media such as smart cards and memory cards for storing the fingerprint data. Another exciting, potentially low-cost approach uses two-dimensional bar codes for biometric data storage. But because these are new products, they have little or no operational track record. Even if the cost of the devices is low enough to permit wide-spread installations, a careful assessment of one or more approaches (and possibly operational testing in a pilot environment) is required to determine if these products are robust enough to perform effectively in the real world.
3. Interfaces to Existing Public Assistance Management Systems
Integration of the fingerprint identification system with existing public assistance management and data systems is another key requirement which must be carefully considered. Interoperability of the systems will require not only interfaces between the new and existing systems for data interchange, but also an integration of fingerprint identification functions into the overall operations of the existing systems infrastructure. To successfully accomplish this integration, the design and operation of all existing systems must be fully understood by the design team that develops the specifications for the new fingerprint systems.
4. Interfaces to External Systems
In addition to integration of the fingerprint system with public assistance management systems, interfaces with other systems will be required. For example, the current direction of national policy suggests the creation of three classes of persons within the U.S.: citizens; legal resident immigrants; and illegal aliens; and that the provision of public assistance benefits may likely differ among the classes of recipients. The INS is in the planning stages for a non-citizen registry which would support employer verifications of work authorizations for legal immigrants, and state agency verifications of immigration status for welfare and public assistance applicants.
Consideration of interfacing with these systems will be required, as states come under increasing public pressure to control non-citizen consumption of public assistance benefits.