Globally, one in seven people (1.1 billion) are unable to prove their identity through any legally recognized means. As a result, these individuals lack access to critical social services and economic opportunities, amongst countless other rights and services.
It is imperative that we solve this challenge; however, hurried interventions can result in unintended consequences, including underutilized resources or even inadvertent harm to vulnerable individuals. At ID2020, we believe that piloting is critical for identifying risks, optimizing operations, assessing cost-effectiveness, and tailoring programs to local requirements. Therefore, while recognizing the need for immediate action, ID2020 believes in an expeditious but cautious approach to implementing digital identity programs.
The National Team for the Acceleration of Poverty Reduction (TNP2K) under the Government of Indonesia leads evidence-based efforts to improve the effectiveness and efficiency of poverty reduction programs. To that effect, TNP2K regularly conducts pilots and randomized controlled trials.
In 2012, J-PAL and TNP2K conducted an evaluation of an identification card system to improve access to rice subsidies for poor households, which produced significant improvements in household access and leakage reduction. These findings encouraged further pilots for identification systems that support other poverty alleviation programs, including some which leverage digital technologies.
With Everest, TNP2K and ID2020 implemented a digital identity pilot with the aim of increasing access to liquefied propane gas (LPG) subsidies for low-income families in Indonesia. The Everest platform used biometrics, blockchain and digital wallets to enable households to securely and conveniently access their allocated subsidies. This digital identity program could further improve operational efficiencies, minimize corruption and wastage and, ultimately, reduce government expenditure. If successful, the chosen platform is rapidly and easily extended to other social service programs.
This pilot was offered to approximately 6,000 households across three dispersed sites in Indonesia: Bukittingi (Sumatra), Tangerang (Java), and Tomohon (Sulawesi). Each of the chosen locations differs in its cultural context, geographical landscape, infrastructure capabilities, and population needs - all of which affect the generalizability of future findings and the implementation of the digital identity program. With these insights, the Government of Indonesia is equipped to make a more informed decision about scaling the digital identity program to other locations and other services.
The Bukittingi, Tangerang, and Tomohon operations have independent mechanisms to solicit feedback from all participants - including consumers, local agents, and partner organizations. Additionally, the program has a lateral feedback system, through which each location shares insights and exchanges best practices with the other sites.
For instance, we found that the functionality of both the software and mobile devices was affected by local infrastructure and cultural contexts. Despite some offline capabilities, weaker connections caused delays in enrollment and, subsequently, authentication for some residents. These delays meant higher operational costs, and greater frustration for agents and participants alike. The connectivity problems were pronounced in rural areas and for individuals located near an airport in Tangerang which deploys network jammers. During the pilot, Everest responded quickly to improve processing times by simplifying the data transfers.
The biometric authentication process, which uses facial recognition and fingerprints, was also found to be influenced by the local environment. Certain participants used headscarves, veils, or turbans, which hindered facial recognition. Fingerprint scans were affected by worn fingers and by dust, dirt, or other contaminants. Some of these issues could be easily addressed with simple operational procedures, including using powder to enhance fingerprint patterns or more frequent cleaning of equipment. Regardless, the pilot was designed for each participant to use multiple biometric methods to ensure everyone could be enrolled and authenticated, and to assess their relative performance in these conditions for future iterations.
Similarly, in response to local conditions, the platform and program were modified to accommodate linked accounts to increase access while still limiting fraud. For elderly participants, this feature allowed trusted family members to assist with the enrollment process, or even redeem subsidies on their behalf.
In accordance with TNP2K policies, the digital identity program was offered principally to women; they were prioritized because of their role in managing the household, and because existing research suggests that women maximize the benefits of government services. Within that group, the pilot focused on the elder woman in each household which typically included mothers, aunts, and grandmothers. In households without older women, the program was offered to the elder daughters or the fathers.
Traditional gender roles in the pilot locations had the potential to cause discomfort for women who were required to interact with male service agents. Anticipating this problem, the pilot increased the number of female service agents to ensure the program remained inclusive. In Bukittinggi and Tangerang, the populations are predominantly Muslim and hence a majority of the participants wore a hijab, which affected the accuracy of enrollment and authentication. If scaled, the program design including the choice of a biometric technology would need to account for such cultural considerations.
Throughout the pilot, the active participation and support of local governments was a critical factor for success. For instance, in Tomohon, the local governments took a much more active role, spearheading outreach and onboarding efforts, and ensuring that participants clearly understood the pilot benefits, risks, and processes. This resulted in a significantly higher number of individuals who were willing to participate, as compared to other locations.
Strong performance of the pilot in each location provides evidence for the viability of a larger scaled program for this specific use case. As compared to a program at scale, this pilot allowed a wider scope for participant engagement, and thereby more nimble responses to risks, challenges, and opportunities. It informed future program design around the inclusion of contextual needs. Hence, this program, if scaled, has the potential to be both a robust and effective tool for addressing leakage and improving transparency and individual access to vital services.
Responsibly implementing digital identity programs at scale requires reasonable certainty that these initiatives will positively impact the lives of the participants, while not exposing them to excessive risk. Information gained through the LPG Digital ID pilot provides evidence of the potential impact and will enable improvements to future program designs. While the pilot itself is complete, detailed monitoring and evaluation efforts remain underway and are likely to yield additional insights for future digital identity programs.
Additional details and data on this specific pilot will be made available in an upcoming report.