Open Data and the Future of Banking

A hawker on the streets of Nairobi opens her cellphone and shares her most recent M-PESA and Equity Bank transactions with a budgeting app. She uses the app to schedule payments according to her cashflow, ensuring her bills are paid on time. An electrician working in the gig economy in Mumbai searches for an insurance policy to protect his family. He instructs Google Pay to share information about his earnings with an insurance company to set his coverage goals, and later automates premium payments matching his irregular income flow. In Medellin, a construction worker and his seamstress wife instruct an app to share information about their combined wages and income with prospective lenders, and find an affordable mortgage to buy a small apartment for their family.

Globally, a growing number of low-income people are entering the formal financial system but have not yet leveraged the full value of financial services. New systems for data sharing and payments flexibility could spur innovation by unlocking access to consumer data now held within payment companies, banks and other financial institutions  – as these examples illustrate. Combined with the ability to direct payments, these changes are enabling responsible players to design a range of new products and services that are more competitive, give customers more financial choices at lower price points and are better tailored to meet their specific needs. A more open use of data and payments can benefit millions of low-income people entering the formal financial system and improve their ability to engage with the real economy.

Yet the very same structures that hold out such promise for inclusion and growth also introduce new dangers. Sharing customer information among multiple players heightens the risk of misuse of their data leaving many millions vulnerable to being targeted with unsuitable offerings. Poor people have fewer assets and are more likely to be functionally or financially illiterate and therefore may be particularly vulnerable to exploitation. Building systems that include millions of poor will not happen through unfettered innovation alone but will require proactive action and dialog among public and private sector actors in shaping new regimes for data sharing and payments.

There are three design elements that we consider essential for maximizing the potential for delivering responsible and inclusive finance. These are:

• Wide and diverse data pools that draw information not just from banks but a range of players that interact with the poor.
• Data aggregation centers that enable consumers and data users to easily manage data permissioning.
• Third parties able to initiate payment requests directly from a bank on behalf of their customers.

Together these three elements form necessary (though not sufficient) building blocks for creating a successful open data and flexible payments regime that benefits people at every level of society.

If these principles sound familiar, it is because we are inspired by new regimes emerging in Europe, Mexico, and India, among other countries. To drive competition in retail banking in the United Kingdom, its Competition Markets Authority required the largest UK banks to open up and share their data. A recent study by the UK’s Financial Conduct Authority says the move could usher in more competition and innovative business models, delivering better customer services such as cheaper payment solutions, budgeting and money management tools based on customer data, and the ability for customers to easily switch to new providers. More widely, the European Union has recently passed Payments Systems Directive 2 (PSD2), which requires banks to share data. This sharing is enabled by APIs – software applications that enable disparate computer systems to communicate with each other, to exchange data and collectively perform a task. This would allow consumers to share their bank data with other parties and to initiate payments from their bank via third-party apps. Similar efforts to open up banking and payments are under way in Canada, Australia, and Brazil.

In Mexico, the 2018 fintech law establishes data-sharing requirements across all financial sector entities, and forthcoming changes in the national payments system will allow third parties to initiate payments. India has gone one step further by building an open payments system under National Payments Corp. of India (NPCI), called the Unified Payments Interface (UPI). More recently the Indian government has created a new type of institution for sharing customers’ financial data, called an Account Aggregator.

These countries provide early insights into how open data and payments regimes might work. Drawing upon their experience, we see immense potential for emerging economies that incorporate into their design these three principles for dynamic financial markets that foster inclusive growth. We will examine each one further.

Open Data: Financial data supplied from a wide spectrum of sources

In lightly banked, developing economies, the provision of customer data will need to extend well beyond banks. In these countries, banks have relatively few customers, many of whom have limited bank balances and lack deep financial relationships. Open data regimes will need to draw on a wider range of firms who reach a much broader customer base. These should include electronic money issuers (EMIs), which typically have significantly wider reach into poorer segments thanks to their lower cost structures. For example, bKash in Bangladesh has opened nearly 30 million accounts in a market where the largest banks have fewer than 1 million customer accounts. Payment service providers, such as Stripe or Visa, also should contribute to the pool because they process such large volumes of data. Mexico’s new fintech law is a good example of an effort to widen the data supply. It places a data-sharing obligation upon banks, payment service providers, credit cooperatives, insurance providers, money remitters, and licensed fintech firms.

Drawing on the data generated by a wide spectrum of firms will provide individual customers more data to draw upon and leverage towards their access of yet more services. If many financial firms are obligated to share data, it could generate the necessary critical mass to enable innovation. This has other benefits. Firstly, it should quell the concerns of incumbent financial service providers (FSPs) that third parties would unfairly benefit, free-riding the system by accessing the traditional banks’ data. Secondly, it could partially address monopolistic concerns about bigtech firms such as Google, Alibaba, Facebook and others gaining market dominance through their exclusive control of customer data since those players would also have to contribute financial services data. Importantly, the suppliers of the data could also be the beneficiaries by developing new financial services based on more data about the customers.

There is even a logic for expanding the data supply beyond finance to include telecoms data or utility payments. This would bring further value, potentially, but would also bring data sharing beyond the realm of financial sector policy alone. It may therefore be a longer term aspiration rather than practical near term possibility in many countries.

Data Access: Single point for consent across varied sources of data

Sharing data across multiple providers could be chaotic and overwhelming if consumers had to provide consent every time some of their account information is used, or if they had to submit data histories to multiple providers to get the best quotes on financial products and services. Or even more overwhelming would be efforts to end consent or recall data from multiple providers. Instead, consumers should be able to provide their data permissions in a single place and through a single permission channel allowing transfer of relevant financial information from different sources in a single package to an app, FSP or third party as needed. For example, the couple in Medellin wanting to buy a home would permission the mortgage app to access their financial information via a single request. This would be highly convenient for consumers, and a major cost saving for the data users.

An early example of this function is India’s Account Aggregators. Customers allow these new entities to link to multiple sources of their financial information from a range of different institutions (bank, broker, insurer, etc.), and then to share it upon their request. Account Aggregators act like pipes that move data easily but cannot either store or use the data themselves. The added benefit is for the users of the data, they need only make a single request to access the pipe to receive a wide spectrum of data.

Open Payments: Many third parties can initiate payment requests

All too often developed and under-developed markets have incumbent banks with legacy control of payments. This often introduces costs, delays and blockages that inhibit a consumer’s ability to make use of their own funds and therefore blocks access to better services outside their own bank. We have written on interoperability and the importance of moving money across a wide spectrum of bank and mobile accounts. However, payments regimes should go beyond interoperability and ensure that an even a wider spectrum of well-qualified third parties who are not direct participants in interoperable schemes can also help consumers initiate payments.

An example of this is Facebook Messenger in the Philippines, which has the ability to make payments from accounts held in a handful of electronic money providers. Customers can instruct Facebook Messenger to make a payment from their EMI account. Through an integration with Messenger, the EMI makes the transfer. Similarly in India, the NPCI has opened up its system to non-banks through a single integration known as the UPI. It enables Google Pay and BHIM, a local third-party app, to initiate payments from more than 100 banks, and since its launch two years ago, Google Pay (formerly Tez) has signed up more than 60 million monthly active digital payments users across India.

In Mexico, a new scheme called CoDi that is going live in the fall of 2019 will enable non-bank fintechs to connect directly to the low-value, real-time payment system (SPEI) and initiate payments across bank accounts. Though payments are made from the consumer’s bank account, neither the fintech nor the consumer will have to rely on the bank to start the transaction.

In more developed markets, opening up the payments infrastructure also is gaining momentum with the Bank of England allowing non-banks to participate directly in RTGS, its real-time, bank-to-bank payments and settlements system. By expanding access, the cost of moving money should reduce dramatically. Indeed, the aspiration of UPI in India and CoDi in Mexico is to dramatically drop the cost of payments perhaps even to zero, making the payments an integral part of a utility infrastructure available to nearly all in the country.

Not only should transaction costs drop, the changes should spur digital innovation in the real economy by enabling many more third parties to offer more compelling services. Imagine for example, a company organizing maid services via an online platform. Contract workers could be paid much more quickly and efficiently if, once the maid completes a job, the platform automatically requests payment to be deposited directly into the gig worker’s account. Such fast payments not only make it easier for the gig worker to manage her cashflow but might also create the data trail that later allows her to access a line of credit for her burgeoning business.

Open Systems Favor Inclusive Finance

It helps to think about how a new regime like this attacks many of the structural barriers in the financial system that continue to exclude many poor people in emerging markets. One persistent barrier is the high cost of traditional banking. Low-income consumers carry very small balances in their accounts and therefore generate very little revenue for the banking business, often less than $10 per year. When acquiring new customers typically costs banks between $20 to $30 for marketing, customer due diligence, account opening etc., the poorest segments become unsustainable to serve. Providers also face expensive service delivery costs when populations are dispersed across large distances. Not knowing much about the livelihoods of these customers makes it difficult to engage them or to offer compelling services.

With open data and payments, finding clients and turning them into customers can be far more efficient.  For example, a customer wanting to purchase a motorcycle and repay in weekly installments can give a lender permission to access data in his or her transactional account. The lender looks at the frequency of utility payments to assess the customer’s credit risk and combines that data with information supplied by the dealer – such as whether the motorcycle is a sturdy, low-cost entry model, whether it can transport light cargo, and where the dealership is located – to set suitable loan terms for a person who has no previous credit history. In this way, the lender is able to acquire a customer without high costs in finding him or her. Moreover, serving consumers this way in the real economy can be a conduit toward offering them a variety of complementary financial services.

Transactional data could also be used to customize loan offerings. An entrepreneur running a street food stall, for example, uses his mobile money account to purchase meat and vegetables. This generates a rich transactional data trail. He would like to expand his hours of business but needs a loan to buy more foodstuffs. He might benefit from a loan that allows him to repay in small increments as a percentage of actual sales, a method used by Kopo Kopo Grow loan in Kenya. Scoring algorithms can use a customer’s transactional history and mash up different types of data enabling the lender to more efficiently assess risk and price such a loan. And if the customer pays back on time, the lender could reward him in the future by offering lower interest rates. An analysis of the cost structure of banking in Africa undertaken by CGAP and McKinsey in 2015 revealed that incorporating data analysis and automation into existing financial services providers lending could reduce operating costs at least by 25 percent.

These examples illustrate how open data regimes can benefit the poor people through lower costs and more relevant products and services. Financial services that enable low-income customers to invest in their businesses, acquire more inputs, protect against risks, and easily manage their financial resources expand the extent to which those customers participate productively in the real economy. This is the kind of bottom-up economic inclusion that an open data and payments regime may activate.

Prudent Policymaking Required

For all the many advantages new regimes might bring, implementation can present significant challenges and introduce new risks. New data and payment regimes will require forethought, and design cannot be left solely to fintechs or the innovators alone to solve. These issues are being tackled in different ways by different countries. The challenges include:

• Should APIS be standardized? If every FSP designs a different API interface and protocols demanding multiple complex integrations, the benefits of open systems will be significantly reduced. It would be like the early days of railroad when companies were laying different gauge train tracks. In the UK and Mexico, efforts are underway to standardize data sharing to ensure a homogenous approach. India has gone a step further with one central system, the UPI, which gives access to more than 100 banks through a single integration.
• Should governments mandate open data and open payments? While policy makers in Europe and Mexico are driving change through legislation, in other countries private companies perceive business value in opening their APIs and are proceeding on their own. (see CGAP's work on Open APIs). In Japan a range of banks, encouraged by a June 2018 law promoting open banking, have adopted open APIs to provide account information services to third parties. By mid 2020, roughly 130 chartered banks of the largest 140 are expected to join. This level of private participation cannot be expected everywhere as incumbent banks, which have historically gathered more data on consumers, will likely be reluctant to share. In many countries it is possible that a policy nudge, though not necessarily a full mandate, will be required to push developments forward.
• How should cyber-security risks be managed? To have multiple sources of customers’ data flowing through a pipe and through a wide range of service providers creates new vulnerabilities. Providers need to agree on standards for data security and protection. Common encryption and security protocols must govern the data flow across multiple parties using the infrastructure.
• How to protect consumers against data abuses? Misuse of data may be partly addressed through laws and regulations. Consider a provider who offers a simple check-balancing service and gains a consumer’s consent to see his or her current financial transactions and positions. That service provider then monetizes the data in some other way to help an insurance provider develop and sell a product the consumer never intended to obtain. Such misuse risks undermining consumer confidence, to protect against which providers should be restricted to using the data solely for the purposes agreed to by the consumer. In a forthcoming paper, our colleagues David Medine and Gayatri Murthy articulate how this “legitimate purpose” limitation could work in a way that helps consumers while allowing growth in data usage.

For all the power that rests in data to potentially lower the costs of financial services and introduce new models for doing business that open up economic opportunities for poor people, we must also be highly cognizant that the same access can be used to systematically exclude. We already know in developed countries how low income or prior illness or pre-existing conditions can be used to exclude people from the credit or insurance they need. In countries where gender stereotypes prevent women from taking active roles in managing resources, data can be used to reinforce exclusion rather than overcome it.

We need to have a conversation on how to be intentional about what we aim to build, and we need to incorporate into the design of open data and payments systems principles that reflect the values of inclusion and empowerment . CGAP is committed to broad-based and equitable growth. For this to happen, we will need to double down on our efforts to ensure these values and priorities are embedded into how these new regimes are built. If we can do this, they can become major drivers of human progress. By thinking ahead, we can ensure that the introduction of technologies that change data and payments are channeled to build the kind of future where low-income people are not victims of a new economic order, but they stand to gain opportunities and protections that enable them and their families to prosper.