Bank customers generate large quantities of useful data when they access financial services. Periodic deposits, overdrafts, and late payments help predict a potential borrower’s riskiness. Account balances and transactions reveal a customer’s financial needs, while a small business’ transaction data can inform lenders about its health or promote the cross sale of financial management services.
Traditionally, customers’ data has, for the most part, been under the bank’s control. As a result, incumbent banks were assumed to have an informational advantage over entrants providing financial services (Hauswald and Marquez 2006). This advantage can lead to rent extraction and dampened competition (Rajan 1992). Barriers to data access also have the potential to stifle financial innovation and inclusion, stymieing the development of useful products that could provide advice to consumers.
However, data has become increasingly easy to process, store, and transport. This technological change has upended banks’ exclusive access to customer data through a movement known as open banking (OB), which grants bank customers the ability to share their financial transactions data with other financial service providers. Governments have been keen to enable bank customers to share their data freely with the aim of boosting innovative entry, competition, and financial inclusion.
In a recent working paper, we consider four aspects of open banking (Babina et al. 2024). First, we assemble the first comprehensive database on government initiatives to establish open banking and conclude that the policy is becoming ubiquitous internationally. Second, we use data from the UK – an early adopter – to show how open banking has affected individual consumers and small businesses. Third, we examine the global impact of open banking by showing that at the country level, adopting OB is associated with new venture capital flows in Fintech entrants. Fourth, we provide a quantitative model of customer data sharing, which measures the overall and distributional effects of OB.
Our dataset shows that many governments are actively debating mandating OB. As of October 2021, regulators in 80 countries out of the 168 we examined have taken at least some steps to encourage the adoption of OB (see Figure 1), and 49 of the 80 have already adopted their key OB policies. A key driver of OB policies appears to be consumer trust in sharing data with fintechs: intuitively, willingness to share data increases the potential benefit of these policies.
Figure 1 International prevalence of government open banking policies
Note: Map shows the current implementation status of government-led open banking policies and the year in which the major open banking policy was passed. Panel (a) shows the implementation status of their government open banking policies. Fully implemented corresponds to countries that have implemented open banking government policies; Implementation to those that have determined the specifics of the open banking approach and are currently implementing it; Discussion to those either considering implementing open banking policies or discussing that implementation; None to those with no government open banking approach; and NA to those where we have not collected data.
To explore how these policies have affected bank customers, we consider microdata from the UK. As an early adopter of open banking, the UK implemented a series of policies in the middle of the 2010’s. The most notable, covering consumers, was the Payment Services Regulation of 2017 – a UK-specific regulation that, in turn, grew out of the EU’s wider 2015 revised Payment Services Directive, one of the earliest examples of OB regulation (Navaretti et al. 2023).
We present specific evidence from a UK policy targeting small businesses. The UK’s commercial OB-related policy was introduced slightly earlier in 2016, and applied only to SMEs with annual sales below £25 million. This cut-off allows us to compare outcomes for eligible and non-eligible SMEs following the policy’s implementation. To illustrate how the policy promotes competition and potentially limits adverse selection in credit markets, Figure 2 shows the relative rate of new secured credit relationships formed by UK SMEs below the threshold compared to those above it. The policy raised the relative likelihood of an SME obtaining a new loan from a new creditor in a given year by approximately 1.5 percentage points, or around 25% of the mean relationship formation rate. These new relationships are concentrated among non-bank lenders (e.g. fintechs). In terms of distributional effects, we also find that treated firms with prior lending relationships are more likely to get new loans, and SMEs that form new lending relationships with non-banks pay less interest.
Figure 2 Event-study of SME data sharing and new lending relationships
Note: Figure shows changes in new lending relationship formation for SMEs treated by the UK Commercial Credit Data Sharing (CCDS) policy using a panel event-study analysis. The underlying data is company-year data on secured loans for UK firms with 2016 sales between $10 million and $40 million from Companies House via Bureau van Dijk for the 2014–2019 period. Firms are classified as treated if their 2016 sales are below the CCDS’s $25 million eligibility threshold, with firms above the threshold serving as the control group. The event-study specification is estimated using one period lagged firm-level control variables of the log of total assets, a low credit risk dummy, cash to total assets, and leverage ratio, as well as firm, sector-by-year, region-by-year, and relationship stage-by-year fixed effects. Low credit risk is destined as a QuiScore above 80, sectors are defined based on one-digit 2003 UK SIC codes, regions are the 124 postcode areas, and relationship stage is the decile of the average relationship length the firm has with its lenders. Shaded regions denote 95% confidence intervals calculated using standard errors clustered at the firm level.
We leverage survey data on UK households to inform discussion of the consumer side of open banking. The UK Financial Conduct Authority asks consumers directly for their attitudes surrounding and uptake of financial applications that require them to share their financial data, alongside information on demographics and other financial product use. A key finding is that consumers seem to make use of OB for two distinct reasons. The first is financial planning and advice – consumers share their financial data with applications that allow them to track spending, manage accounts at multiple banks, or advise on useful products. The second reason is to obtain credit – consumers share data with potential lenders in order to improve their outcomes in credit markets. There is little overlap between these two use cases: most consumers who engage in OB do it for a single reason only. Nonetheless, both use cases are positively associated with consumers being employed, having a greater willingness to share data, and having missed prior payments. We also find suggestive evidence that OB use improves consumer outcomes: using OB for advice is associated with greater financial knowledge, whereas using it for borrowing is associated with greater credit access.
Given the uptake of OB at the customer level, it is important to ask whether the policy also spurs entry and investment in financial services – a key motivation behind OB. To answer this, we go global, measuring entry at the country level using data on venture capital investment in fintech startups. Using the staggered implementation of OB policies across countries, we show that venture capital investment in fintechs surges following OB policy adoption (see Figure 3). We also find that countries whose residents are more willing to share data see greater post-OB investment, suggesting that consumer preferences play an important role in OB’s impact.
Figure 3 Event-study of fintech investment after open banking government policies
Note: Figure shows changes in fintech venture capital (VC) activity around the passage of open banking government policies using a panel event-study analysis. We perform this analysis on a Pitchbook panel of 2011–2021 data for the 21 countries with at least five fintech VC deals in the 2000–2010 period. Figure shows an event study on the log of one plus the millions of US dollars invested in fintech VC deals. Year 0 is the passage year of each country’s major open banking initiative. The coefficient for year 0 is set to zero and other coefficients are presented net of country fixed effects and region-by-year fixed effects. Regions are 1) Africa, Middle East and North Africa; 2) Europe and Central Asia; 3) Latin America and the Caribbean; 4) North America; 5) South Asia, East Asia, and the Pacific, following World Bank geographic terms. EU member states are weighted to count as a single country for estimates and standard errors. Shaded regions denote 95% confidence intervals calculated using standard errors clustered at the country level.
While our empirical results offer valuable evidence regarding OB in practice, they are silent on the mechanisms by which access to OB data increases entry, and have little to say on welfare, general equilibrium effects, or distributional consequences. Moreover, while the consumer and SME microdata are informative about the UK case, our cross-country results highlight the importance of customer privacy. How might OB look in countries with different social attitudes?
We develop a quantitative model of OB to tackle these questions. This model incorporates data sharing into a standard model of consumer choice with heterogeneous consumers. A bank customer’s data reveal her preferences (allowing the creation of better products) and costliness to serve (allowing better screening). An incumbent relationship bank always sees the data, while entrants see it only if the customer shares it via OB. We calibrate the model based on a credit use case and an advice use case. In both calibrations, OB spurs competition but through different channels (see Figure 4). When OB is used for credit, unequal data access discourages entry by giving relationship banks an advantage, creating adverse selection for entrants. OB reduces this adverse selection, raising entrants’ profits. In the advice case, unequal data access impairs entrants’ ability to offer customised products. OB corrects this, raising customer demand, entrant profitability, and entry.
Figure 4 Aggregate and distributional outcomes of open banking
Note: Figure presents model-implied aggregate changes after open banking (OB), with each bar showing the percentage change in the relevant outcome caused by moving from the status-quo relationship banking regime to the OB regime. Magenta and cyan bars show outcomes for the financial advice and non-GSE residential mortgage calibrations, respectively; # entrants is the number of new entrants. Quantities (all) is the population fraction obtaining the financial service, which we further split into Quantities (relationship), i.e. relationship bank, and Quantities (outsiders), i.e. fintechs. Price (average) is the average fee or rate charged. Relationship profit is relationship banks’ profits.
While OB unambiguously increases competition, this sometimes comes at the cost of financial inclusion. No customers lose when OB is used for advice; product quality improves for those who share their data. In contrast, there can be negative distributional consequences from credit OB. Entrants can better exclude higher-risk customers. Users who share unfavourable data lose directly. Privacy-conscious consumers who opt out of sharing are inferred to have hidden unfavourable data. Thus, consistent with the evidence from UK businesses, the customers who benefit the most may be those who already have credit access. Customers who opt out of sharing still gain from increased entry and competition, but lose because they are now inferred to be higher risk.
Our model shows that societal preferences for privacy also play an important role in explaining these distributional effects. Strong societal preferences for privacy blunt the impact of OB, as few customers opt in to data sharing and so few firms enter. However, strong societal preferences for privacy have a silver lining, as opting out sends only a weak signal about one’s riskiness. In fact, under reasonable parameters – including those obtained in our UK calibration – OB is welfare-improving for all customers even when data is used for screening. The negative inference lenders draw against opt-outs is more than offset by the benefits these customers derive from increased entry and innovation.
To summarise, we document that government policies to promote OB are prevalent: about half of countries have some OB effort. Our empirical analyses and quantitative model show that OB data can have beneficial economic effects. By giving customers the ability to share their transaction data, OB can shift organisation of the financial sector and the market for retail financial products. The welfare and distributional effects of this, however, depend on specific uses of customers’ data and their willingness to share it.
References
Hauswald, R and R Marquez (2006), “Competition and Strategic Information Acquisition in Credit Markets”, Review of Financial Studies 19(3): 967–1000.
Rajan, R (1992), “Insiders and Outsiders: The Choice between Informed and Arm’s-Length Debt”, Journal of Finance 47(4): 1367–400.
Babina, B, S A Bahaj, G Buchak, F De Marco, A K Foulis, W Gornall, F Mazzola and T Yu (2024), “Customer Data Access and Fintech Entry: Early Evidence from Open Banking”, NBER Working Paper 32089, January.
Navaretti, G, G Calzolari and A Pozzolo (2023), “Open banking’s far-fetched promise of a financial revolution”, VoxEU.org 23 May.
