Technology in the Tropics: a Visit to Bangalore

This is a joint post with Julia ClarkBangalore in September. Beautiful weather, luscious gardens, and the din of metro construction. But most importantly (for our purposes, at least), Bangalore is headquarters to the world’s largest biometric identification project. Every 24 hours, the Unique Identification Authority of India’s data center performs 100 trillion matches to ensure that each of the day’s 1 million new enrollees is distinct from the 200 million people already identified. This number crunching will only increase as the program scales to cover India’s 1.2 billion people.It’s fitting, then, that we traveled to the Garden City to present a paper on biometric identification in developing countries (forthcoming, download PowerPoint) at a conference on technology innovations for social program delivery.The conference—ably organized by the World Bank and Bangalore’s IIIT—was timely and salient. Governments in developing countries are rapidly creating new public programs, revising old ones, and using new technologies (GPS, mobile phones, biometric identification, etc.) to improve targeting, reduce leakage and increase access. There is much ado about these cases (exhibit a), and development practitioners are eager to replicate successful innovations and policies.Choosing the “right” social protection models and policies is important (to PMT or not to PMT?). But as the conference highlighted, gaps in beneficiary identification, targeting, enrolment and payments are more often caused by poorly conceived or implemented processes than by specific policy choices. (For example, one presentation cited data from 84 programs showing that 80 percent of the variation in certain performance outcomes occurred within programs that used the same targeting methods.)Process is critical, but can technology help?Representatives from 18 (mostly Asian) countries attended the conference to discuss this question and present diverse cases from India, Pakistan, Chile, Mexico, Fiji, Kenya, and many more. Though each program and country context differs, a number general themes and conclusions emerged that—fortunately for us—mirrored those from our own research:

• Leapfrogging is possible! Technology can improve program delivery when used smartly. It can shift the cost curve of program administration, giving implementers better results for their money. In one example, Dataprev estimates that Brazil’s move to electronic social payments saved some $788 million. Technological innovations offer developing countries a way to leapfrog past the often fragmented and inefficient social protection systems of many richer nations. For example, traveling to a bank to collect NREGA payments in the traditional way can easily cost 20 percent of a beneficiary’s weekly wage.
• The customer is always right. With technology, a citizen-centered model for service delivery is emerging. Too often, social programs exist in silos, designed to suit the administrative needs of government and with little planning from below. Some countries are moving away from this fragmentation by A) using general registration drives to enroll people in several programs at once (“backbone” enrollment), B) delivering multiple services via “one-stop shops,” and C) creating unified social protection ministries and registries (e.g., the Philippines, Pakistan and Chile). These measures facilitate cross-checking and are often more cost-effective for governments and users (reduced travel time and fees, and fewer occasions for bribe paying). New identification and payment technologies are also more accepted by beneficiaries than policymakers often think—especially when they offer clear benefits. [Those most fearful of new technologies are frequently the powerful stakeholders concerned with their loss of discretionary power.]
• But if it sounds too good to be true ... There is no perfect technology or program design (shocker!). This is hardly an epiphany, but the fervor surrounding new innovations can cloud our vision and let us forget that technology is rarely a silver bullet. It cannot undo bad processes or replace the need to develop human capacity. When applied poorly, high-tech systems can complicate service delivery, inflate budgets, and turn into embarrassing boondoggles for governments and donors alike. Conference presentations emphasized the high cost of needing to resurvey populations because of inadequate data quality. Examples from our own research also include some cases of expensive, poorly executed biometric projects that are barely used or discarded.
• Lions and tigers and vendors, oh my! Many technology-based identification and social protection programs have run aground in the muck of bad procurement processes and costly contracts for proprietary systems. In order to avoid vendor lock-in—which is costly and often stymies system growth and adaptation—countries should seek advice from independent technical advisors and consider adopting open systems (like India’s UID). Each country must define its own needs and requirements, rather than relying on pre-packaged solutions.
• South-South learning is a reality (and a necessity). Pakistan’s identification authority, NADRA, competes successfully for contracts in other countries; it is currently implementing Kenya’s passport program. Indonesia’s national ID program is benefiting from India’s technology. Many countries have useful (and sometimes surprising) data and expertise to offer their peers. For example, a survey in Bangladesh reported that 80 percent of women found it easy to remember PINs whereas 80 percent of men found it difficult! (A fingerprint could fix this…)

Beyond technological innovations, additional messages about general social program implementation became clear:

• Updating information—it’s complicated. Enrolling beneficiaries (finding people, capturing data, issuing program identification) is a formidable task. Large-scale population censes are labor- and resource-intensive and can’t be done every day—or even every year. Yet the challenge of how to update the initial list of beneficiaries remains. What about people who become eligible after enrollment? What about those who become ineligible (e.g., by increasing their family’s income level)? How do you find people who have moved or migrated? Some suggest that frequent updates should focus on exclusion errors, allowing for longer periods between full-scale re-enrollment campaigns. India’s RSBY has a goal of touching beneficiaries once a year; Cambodia’s IDPoor census will update records every three years.
• Politics will interfere. Project ownership and ministerial coordination is tricky business. In order to ensure efficient, effective programs, government agencies must clearly delineate responsibilities and incentives for cooperating. The case of Chile’s social protection regime suggests the need for non-sectorial administration of centralized or coordinated systems (e.g. the Ministry of Planning); Mexico has created an Intersectoral Committee to oversee the implementation of a common identifier for its social programs.
• Process, process, process. Above all else, implementation will make or break the success of individual projects and social protection regimes at large. For example, a grievance system for program beneficiaries is a necessity, but too often an afterthought. A well-functioning system should have both administrative and independent channels for redress. Programs that rely on community accountability systems (like scorecards and social audits) must link back to formal grievance processes in order to be effective.

The conference showed that technology can certainly change the way that many social programs are implemented.  But it was also clear that we are at only at the beginning of this (r)evolutionary process.