1. What are biometrics?
Biometrics refers to identifying individuals based on distinguishing physical or behavioral characteristics. This includes fingerprints, irises, face and hand geometry, gait, voice, signatures, DNA, and other traits. Fingerprints (inked and now digital) have historically been the most commonly used, but iris scans and facial recognition are becoming more prevalent. There is also increasing interest in DNA for identification following advancements in rapid analysis capabilities.
2. How is biometric technology used for identification?
Biometrics can identify individuals in two ways. First, biometric data can be used for authentication to verify a person’s identity. In this case, an individual’s biometric is compared 1:1 against a stored record (for example, on an ATM card). Secondly, they can ensure statistical uniqueness. Comparing one person’s biometric data against the larger population ensures that she is unique. In the context of elections, for example, biometrics can prevent people from registering to vote multiple times if each enrollee’s iris scans are compared with those of all other voters. This is a process of deduplication or 1:N matching. (For a primer on biometric identification, see Appendix 1 of CGD Working Paper 315.)
3. What does this have to do with development? Aren’t biometrics mostly used for law enforcement and access control?
In rich countries, almost everyone has a reliable form of official identification, and biometric technology has traditionally been employed mainly for security and forensics. Conversely, many developing countries suffer from an identity gap. Millions of people lack the official forms of identification—birth certificates, national ID cards, voter cards, etc.—that would allow them to access basic rights and services. Closing this gap has been increasingly recognized as both an instrument and goal of development. In the way that mobile phones have allowed poorer countries to leapfrog past landlines, biometrics have the potential to help solve their identification woes while bypassing the paper-based systems often found in the rich world.
4. What are the advantages of biometric identification?
Providing secure, inclusive identification to citizens and residents does not require biometrics. Other options include some combination of unique numbers (like the US Social Security number), cards or PINs. However, digital biometric technology offers certain advantages, particularly in a developmental context.
Biometrics:
- Are unique to every individual
- Cannot be misplaced or forgotten, and are very difficult to fake or steal
- Do not require literacy
- Can help to create an auditable trail for transactions
- Increase anonymity when used in place of personal details (names, addresses, etc.)
5. Who is using the technology and for what?
The particular benefits of biometric identification (see above) have prompted a number of low-middle income countries—and the donors that support them—to use the technology in their development programming. A CGD survey has identified over 160 such cases in more than 70 low-middle income countries. These cases span a number of sectors and include both “ID-driven” or “foundational” applications (such as national ID cards and civil registries) and “demand-driven” or “functional” applications (such as voter IDs, social transfers, emergency relief, banking, etc.).
6. But isn’t the technology too expensive or advanced for poor countries?
Yes and no. Biometric technology—which was prohibitively expensive not long ago—is falling in cost, due in part to the innovations of India’s UID program. Still, price tags can be hefty, and millions of dollars have been spent on projects that were never completed or didn’t work as intended. Studies have shown that, even in where biometrics have a positive impact, they may not be cost effective for smaller programs. For large-scale projects, however, the technology cost is likely to be much less than the cost of mass-enrollment campaigns, and it is not necessarily more expensive than rolling out a paper-based ID from scratch. Furthermore, when biometrics are used to rationalize spending programs (like the civil service payroll) they may pay for themselves.
Infrastructure can also be an issue, though problems with connectivity and rough terrain can be overcome with planning. Angola, for example, deployed helicopter teams to enroll voters ahead of its 2008 elections. Many countries opt to use smartcards in their ID system because they can be used offline. Though emerging economies like India have certain advantages when adopting biometrics (like local tech firms), poorer and more fragile states have also had success. Examples include demobilization payments in the Democratic Republic of Congo and civil service administration in Liberia.
7. How advanced is the technology (does it really work)?
Historically, very little data has been available on the performance of biometric technology for either identification or authentication. In 2012, however, the Unique Identification Authority of India (UIDAI) published a series of papers documenting the performance of its system. Based on these numbers, it is now possible to conclude that biometrics can work well (i.e., have relatively few errors) even in large populations. Performance, however, depends largely on the types of biometrics used and other technology choices. By collecting two irises with a dual eye camera and two labeled fingers, the error rates can be reduced to less than a percent. Still advances in technology cannot compensate for poor planning and administration.
8. Doesn’t biometric identification violate privacy?
A common concern is that biometrics will violate privacy. Technology such as facial recognition, which can be used for surveillance without the knowledge of the individual, raises a number of red flags. In addition, many countries that are implementing biometric programs lack legal frameworks to protect privacy rights. However, many of the commonly cited privacy concerns are not specific to biometrics, but relate to identification more generally. This includes the ability to link information across a number of databases, which can also be accomplished with a unique ID number (like Social Security).
Conversely, some cases indicate that the technology has actually improved privacy. For example, participants in a vaccine trial in Vietnam preferred being authenticated by their fingerprints, rather than by substantive personal information. There has also been concern over taking women’s biometrics in conservative populations, particularly photos and iris scans of women who wear a veil. However, in most cases (such as voter registration in Bangladesh and refugee identification in Pakistan) this has reportedly been a minor or non-issue.
9. Is there a risk of excluding people?
Like privacy, another concern is that biometric identification is inherently exclusionary, primarily because there are some people that cannot give biometrics, due to injury or age. Errors in a biometric-based identity system could also exclude people, for example by mistakenly determining that they have already enrolled when they have not. In both cases, exclusion remains a risk, but one that can be mitigated by smart procedures. This includes allowing multiple biometrics (those with damaged fingerprints may have perfectly healthy eyes) and creating manual processes for redress and comparison. India’s UID program provides one example of such measures.
10. Where can I find more information?
CGD Working Paper 315, “Identification for Development: The Biometrics Revolution,” looks at over 160 cases where biometric identification has been used for development and draws some general lessons. An earlier paper “Cash at your Fingertips: Biometric Technology for Transfers in Developing and Resource-Rich Countries” gives a summary of how the technology is used specifically for cash and in-kind transfers. CGD has published additional work specifically on the Unique Identification project in India, including a policy paper and brief. Comments and questions are welcome and can be sent to Alan Gelb (agelb@cgdev.org) and Sneha Raghavan (sraghavan@cgdev.org).