Financial Inclusion in Indonesia: What Can We Learn from a Data Dive?

It was a full house at the Lab this week as we welcomed a new batch of participants for our final Research Dive of the year, this time diving into alternative datasets to answer policy and development questions on how to accelerate financial inclusion across Indonesia. The usual three-day activities that make our Research Dives what they are — a set of intensive collaborative research sprints — remained intact, yet ideas about new data sources and approaches to consider coupled with the participants’ enthusiasm gave this Research Dive a certain dynamism. Coming full circle on the last day for final presentations, four teams took to the floor to share their experimental approaches, preliminary findings as well as receive feedback from a number of domain experts, academics, government officials and other researchers in attendance.

Contextualising the Research

In 2017, Indonesia scored around 50 per cent on the Global Financial Inclusion Index, which means that half of the Indonesian adult population does not access formal financial services. As discussed by guest domain experts from the Indonesian Financial Services Authority (OJK) and MicroSave during the introductory sessions of the Research Dive, several known factors such as geographic coverage and lagging regional infrastructure inhibit the pace at which cohorts within the population become financially included, but knowledge gaps still remain. Across the country, new data are being generated that provide opportunities for financial institutions and those in the policy-making domain to understand the needs of different communities in order to increase access to financial services and products.

What better way to dive into this critical research but with a mixed group of passionate participants, bringing together skills and experiences across academia, government and the private sector combined. Examining the various dimensions of financial inclusion, what financial inclusion means for Indonesian society in particular, as well as the progress made and challenges that are ahead, four research areas were outlined to help answer pressing policy and development questions:

  1. Measuring financial awareness and financial literacy through social media
  2. Measuring financial access through formal financial and non-financial institutions
  3. Modelling gender-based differences in financial inclusion
  4. Assessing the digital opportunity impact on financial inclusion

Insights and Takeaways from the Research

Guided by the expertise of four advisors who have conducted related research in areas of spatial econometrics (Rahma Fitriani, Ph.D. — Universitas Brawijaya); topic modelling (Edi Winarko, Ph.D. — Universitas Gadjah Mada); geographic information systems (Adityo Dwijananto, Humanitarian OpenStreetMap); and financial technology (Chaikal Nuryakin, Ph.D. — Universitas Indonesia), the teams spent three days hacking away at a mix of both new and traditional datasets.

In a spirit of the event, all the teams worked hard and later took the opportunity to share some preliminary findings from their research:

Team 1 — Measuring financial awareness through social media

Social media data contains a wealth of information on measures of financial inclusion. The team developed a model to process public tweets that contained finance- and money-related keywords to predict the level of financial awareness in seven cities, namely Medan, Jakarta, Bandung, Yogyakarta, Surabaya, Banten and Makassar. Initial analysis showed that Jakarta had the highest financial tweet ratio (the number of tweets containing financial attributes divided by all tweets), and there was a steady increasing trend throughout 2014 observed for six of the seven cities. Yogyakarta, however, was relatively stagnant and had the lowest financial tweet ratio.

Datasets — anonymised 2014 Twitter data from seven major Indonesian cities

Team 2 — Supply Side measurement of Financial Inclusion

The second team explored the possibility of extending the calculation of the Financial Inclusion Index to the village level (in Pontianak) using non-financial channels such as post offices, convenience stores (Alfamart and Indomaret) as well as mapping their geographical proximity. A new Financial Inclusion Index at the village level covering both formal and non-formal channels was produced for the city, but could be extended to other areas as a proxy for Financial Inclusion metrics calculated from surveys. The team also found that, empirically, the proposed Financial Inclusion Index is in accordance with other related indicators such as GDP per capita and the Human Development Index.

Datasets — (i) Location data of Banks, ATMs, Alfamart, Indomaret, and post offices in Pontianak obtained from Google Maps, (ii) 2016 National Survey on Financial Literacy and Inclusion (SNLIK), (iii) 2017 Population Data from the Indonesian National Board for Disaster Management and (iv) The Village Potential Statistics (PODES) 2014.

Team 3 — Modelling Gender-Based Differences in Financial Inclusion: Rural Area Analysis

This team pursued a pair of objectives: a) understanding the extent of financial inclusion disaggregated by gender in rural areas and b) quantifying the probability of being financially included based on gender in rural areas. Using explanatory variables such as gender, income, education, age, marital status and household size as well as response variables (from a 2016 Financial Services Authority Survey) such as savings, credit, insurance and investments, the team’s preliminary findings indicate that women (especially belonging to the higher income brackets) have the highest level of financial inclusiveness in rural areas. The team plans to expand their model to include urban areas to see whether gender-based differences may vary based on differences in settlement type.

Team 4 — Measuring the Effects of Digital Opportunity on Financial Inclusion

The fourth team looked at whether the growing adoption of digital technology can accelerate financial inclusion. Using 2017 Susenas data and about 90 million anonymised tweets from 2014, the team worked to model Financial Inclusion from this perspective. They developed three visualisations to show internet ratio, savings ratio and phone ratio at the regional level. The team found that digital opportunity is correlated with Financial Inclusion, but that high-density use of Twitter is not correlated with Financial Inclusion. The team suggested ways to reduce the digital divide between the western and eastern parts of the country and based on a literature review recommended the use of FinTech to deliver some social assistance.

Datasets — (i) Susenas, 2017 (466 cities, 297,276 households) and (ii) anonymised 2014 Twitter data from seven cities

The research continues…

Being able to receive on-the-spot peer review from fellow researcher divers and experts is one of the many benefits of participating in these events. Furthermore, the dives are designed with the awareness that three days of research are not sufficient to generate polished models and approaches that can be immediately replicated or applied, and so participants are encouraged to use the feedback received to inform and refine their research. During the final presentations, we were happy to be joined by representatives from the Indonesian Government (the Financial Services Authority, Secretariat for the National Council for Financial Inclusion and Indonesian National Statistics Bureau), Indonesian Fintech Association and research think tanks who offered comments on how these research projects can be improved and possibly used to fast track financial inclusion efforts in the country.

We were pleased to host all the participants, advisors and guest domain experts in this Research Dive on financial inclusion, and we look forward to reading the forthcoming technical papers being produced by the teams.


Pulse Lab Jakarta is grateful for the generous support from the Government of Australia.