Biometrics, Identity and Development

paraderollAI and Robotics

Nov 17, 2013 (3 years and 4 months ago)

44 views

Biometrics, Identity and Development

Research (very much) In Progress Seminar

September 16, 2010,

CGD

Alan Gelb and Caroline Decker

Biometrics Identity and Development (
abbr
) 3a 091510

1

Biometrics is not uncontroversial


The wasteful, bureaucratic and intrusive ID card scheme
represents everything that has been wrong with government
in recent years
.”

~ UK Deputy Prime Minister Nick Clegg (2010)


2


Privacy is dead, get over it
.”

~Scott McNealy, Sun Microsystems
(1999)


Leapfrog technology, very relevant
to development


Introduction


Original motivation : cash transfers of resource rents to
citizens
Identification is essential


But:


Identity
is a much wider
issue than transfers. Part of
development process.


Huge Externalities to Identity


Growing number of bio programs in developing countries +
new BIG player


Range of policy and coordination



issues: governments, donors,



private sector


No overall summary…


Objectives? Beyond a paper?

3

Outline of presentation


What is biometric identification?


Some system options


Five +1 case studies


Issues and debates on


biometrics


The future of biometrics


and development


Next steps for research?

4

What is Biometric Identification?



Identifiers


Fingerprints


Palm
prints


Footprints


Vein mapping


Hand geometry


Face prints


Earprints


Retina


Iris scans


Tongue
prints


DNA


Voice prints


Gait


Dynamic Signature


Possible uses


Security


Criminology


Anti
-
terrorism


Crowd recognition


Immigration


General
identification for
economic purposes



Different levels of precision
and intrusiveness





Identifying someone based on unique physical or behavioral trait

5

Fingerprints


Ancient: 650 BC speculation unique


150 years used for criminal forensics;

digitization in 1990s 40 different points


Simple readers $30


Accuracy: 99.9%
-
95% (latent prints, less)


Early capture age about 12


Susceptible to wear, damage


Potential to be fooled in lab conditions


6

Iris


1994 John
Daugmann

Iriscode

algorithm


266 digital feature points, still rapid comparison rate


Very accurate:
Prob

2 identical 1 in 10e78; 99% +


Stable by 8 months. Eyes self cleaning


Fooled by pictures in labs newer readers also check
for living eye


Does not involve physical contact


Readers $4,000 but quickly becoming cheaper

7

Universal ID Project


Typical national
ID
project


Malaysia, Costa Rica,
India,


Many applications


Economies
of
scale


C
oncern for security


Big Brother State?


Upfront cost before benefits




Programmatic ID


Individual projects


DRC, Malawi, KZN,
Gujarat


Smaller budgets;
cost
vs.
savings


Potential externalities


Need for planning and
coordination




Some System Options


1 Universal vs. Programmatic Adoption

8

2 Payment Options for Cash Transfers


Pull



Payment at specific time and
location


Allows dissemination of
information and services at
the same time


Less convenient for staff and
recipients


Example: DRC (final method:
mobile pay stations)










Push



Financial account necessary,
possibly e
-
banking


Convenient
-

cash accessible
anytime in
numerous POS,
merchants etc


More expensive to set up, but
cost effective long term


Examples: Namibia, Botswana
(ATMs), South Africa
(
Sekukula

account)



9

Methods of Payment



Cash


Risky: concentrated cash


No saving/banking options


Requires pull mechanism


Vouchers


Allow for saving, but not banking


Requires pick up, pull payment


Safer: no cash on hand at distribution


Smart Card


Can use pull or push payments


Allows for saving, not necessarily banking


Bank Direct Deposit (mobile banking)


Offers externality of banking services


Push payment service

10

3 Identity Validation


Pure Biometrics


C
onstant through life


C
annot be shared


D
oes not require literacy


Nothing
for recipient to
lose


Cheaper
, no smart
card


Doesn’t
support offline

push
system


Smart Card


Photo ID

11

3 Identity Validation


Pure Biometrics


Smart Card


Information on
card,
offline system. Validation with either
bio ID and:


Pin


Readers cheaper


Can be used remotely


Can be reprogrammed when compromised


Can
be
shared or stolen


Harder for illiterate
populations


Photo ID

12

3 Identity Validation


Pure Biometrics


Smart Card


Photo ID


Requires human interface, more susceptible to error


Easier to forge or fake


May appear cheaper than biometrics or smart card

13

Five +1 Cases:

Afghanistan/Pakistan (UNHCR)


Objective: Resettlement payments


Over 200,000 refugees processed


Began 2002


Iris scans before payment


Prevented round
-
tripping


Most basic use of biometric ID






It has proven very successful, meeting not only cultural needs of
the refugees but the operational requirements of UNHCR.”






14

DRC (World Bank)


Objective: demobilization payments


102,000 soldiers


13 payments over one year


Iris scans for registration


Push system later Pull system


Banking system not set up in time for full
use


Sparse rural areas

15

Malawi (DFID/Concern)


Objective: cash relief after drought 2006 (DECT)


11,000 households, only
lasted five
months


Fingerprinting for both registration and payment


Smartcard bank based system but bank not fully geared up so
manual payment
.


Biometrics not immediately cost
-
effective due to careful selection
process, small scale,

short term.

16

4 South Africa KZN (Government)


Objective: pensions and other social
payments


1990: first such known use of
biometrics


5 million grants delivered per year


Fingerprints for both registration and
payment


Modified Pull payment, fixed and
mobile centers. Developed into Push
system


Extended nationwide; also to Namibia


Apartheid


fingerprinting


grants to
poor


17

5 a India (Andhra Pradesh Gov’t)


Objective: Employment Guarantee and Pensions


Replaced traditional well
-
run system. 5 million people


Fingerprints, smartcards, Push payment system,


pure bio


Convenient, little to no leakage

18

5b India Unique ID (UIDAI)


Nationwide, multifunction program, Largest in world. ID only.


Registration underway, fingerprint and iris, not yet
compulsory….


Using international standards, potential to bring down costs
sharply for all countries


Payment systems will use Push system with POS





19

Debates on Biometrics

1. Too Costly?


Basic registration:


hardware 10c/ person..


Sao Tome 160,000 ;


5 stations cover in 2 years


Haiti 10 million;


200 stations cover in 3 years



National
ID

Costa Rica $1.11/per person; India $5; South Africa $10



Overall costs of applications need to include technology; single
smartcard can cost $2 +



20

Enrolment station


Are There Potential Savings?


Africa: PETS leakages 30%
-

60%, wages 20% Chad 98%


Southern Africa: Pension theft, ghost workers in KZN…reported savings up
to 20% in Namibia (NET1)


India: Andhra Pradesh bio showed prior losses of 12% ; other states 30%;
one state 70%


India: Estimated loss of 40% social payments ($110 billion). Two thirds
corruption, rest misdirection. Only 15% reaches poor.


USA programs lose $4 billion to welfare double dippers each year



Biometrics cannot solve all problems including those related to targeting,
but can reduce leakage. Losses very low.



“Technological approaches are more expensive to set up, but more cost
-
effective in the long run, not least because they have greater potential for
multiple applications.” Innovations Report, Malawi

21

Potential Savings

(India)

22

0
500
1000
1500
2000
2500
3000
3500
Food
NREGA
Kerosene
Midday Meal
Petroleum (ex
Kerosene)
Fertiliser
Expected leakage of government's funds likely to be spent over next five
years

Reaching the Deserved
Leakage
Rsbn

58%

30%

38%

25%

50%

48%

Can Bio be Used in Difficult Conditions?


Technology will be difficult for rural populations? No.


One hour of training needed in DRC for operators


“Clients adapted well to technology” ~Malawi report


May be easier than traditional ID methods


Different spellings of

names


PINs higher
-
risk


Tokens or books are

often lost


Some applications do

need mobile

communications but

spreading


23

Civil Liberties or Public Accountability?


Major concern in developed world, and in views of some in
developing countries


But opposing view: less liberties to erode, strengthens
citizens against rapacious officials


Some NGOs note the increased transparency of programs
strengthens accountability


What privacy and civil liberties are low
-
income citizens likely
to lose?


24

Undermines “Human” Approaches?


Direct transfers weaken community structures?


Offset: possibility of local taxation strengthening representation


See
Saving Ghana from Its Oil
(Moss and Young)


Weakens bureaucracy and less face
-
time for citizens?


Offset: depends on quality of governance


Political economy of Bio
-
ID including for transfers ?


Bypasses major chunks of

bureaucracy.


But potentially closer

connectedness to the state

because programs work?

25

Data Integrity for Future?



Will data still be secure in 20, 30, 40 years?


26

Questions for the Future

BIOMETRICS IS COMING.


How do we maximize its development potential?


Programmatic introduction in LICs often associated with
donor programs. Potential for huge saving in efficiency,
especially in less functional states.


Opens the door to better approaches. Haiti? Pakistan?


How to maximize externalities ,

towards a more coordinated

approach and international

standards
?


Survey paper + ? CGD?


Encourage research.


Little rigorous exists….


27

END

28