Using Data for Financial Inclusion: The Case of Credit Bureaus in Brazil
Less than a month after a partial lockdown in Brazil began on March 24, 91 million citizens had already experienced a late payment on a credit installment or a utility bill. This compares with 59 million in early March, and represents a jump from 39 percent to 58 percent of the adult population. Will this type of accident meaningfully impact the credit profiles of citizens following this current crisis? Moreover, how should late payments resulting from the pandemic impact scores? And will this change accelerate the use of positive data for credit scores in Brazil?
In April 2019, Complementary Law 166/2019 was signed by Brazilian president Jair Bolsonaro (complementing Law 12.414/2011), enhancing the roles and responsibilities of credit bureaus under an opt-out system. This came on the heels of years of discussions between banks, regulators, and consumer protection groups. Bureaus had been using positive data since 2011, though adhesion remained low due to an opt-in system. Presidential Decree 9.936 was then signed in July 2019 and banks began to contribute their data to bureaus in October 2019, alongside datasets from retailers and utilities, such as gas, electricity and telecom companies. With this addition, credit bureaus in Brazil leapfrogged from a binary to a continuous stage.
Bureaus can play an essential role in financial inclusion, as credit bureaus and the scores they produce reduce asymmetry of information between lenders and borrowers. Moreover, deeper financial systems are linked to higher and more stable levels of development. Therefore, accurate and complete bureaus bear the promise of higher levels of growth via lower levels of credit rationing and lower spreads for borrowers. From a policy standpoint, more data allow for the incorporation of insights from information economics, as opposed to an excessive dependence on monetary policy alone.
As the financial sector is continually transformed by the rise of FinTech and BigTech companies, data is becoming even more essential in assessing risk and supporting growth for lenders. Bureaus are adapting to a triple challenge: digitalization, influxes of new information sources, and new customer-centric regulations. Bureaus collect information on consumers’ payment history. They verify, compile, and analyze the data to assess customers’ payment ability, closing the information gap between lenders, utility providers, and users, while also empowering customers. Historically, bureaus were often developed by retail associations and then expanded to include data from utilities providers and financial institutions.
Scores represent the probability that a customer does not pay. They typically range from 0-1000 in Brazil. Four credit bureaus exist in Brazil and they are at the center of the implementation of the new regulation: Serviço de Proteção ao Crédito (SPC), Serasa, Boa Vista and Quod. The inclusion of positive data into scores meaningfully enriches the traditional use of the Credit Risk Center managed by the Central Bank of Brazil, which has provided credit-related information since 1997. Quod is a newcomer in the industry and was formed by the five largest Brazilian banks, each maintaining a 20 percent stake, having received conditional approval by antitrust authorities in 2016.
Smarter bureaus bear the promise of more efficient banking systems. The expectation is that complete and accurate credit scores can lead to lower banking spreads (the difference between rates paid by borrowers and the funding cost of banks) and increase financial depth. Credit rationing is particularly problematic for small and medium enterprises and lower-income individuals. This is true globally as well as in Brazil. A reduction in credit rationing would translate into a higher level of credit concession for any level of overnight rate.
How are the four existing credit bureaus using this opportunity? Are we seeing early evidence of an increase in financial depth in Brazil following the implementation of the new credit bureaus? The average corporate lending rate by regulated banks reached 15.7 percent in 2019. Within the large set of corporate loans, rates for the smallest category of companies (MEI) reached 58.8 percent. An increase in financial depth in Brazil is based on the premise that lending is mostly constrained by information. In fact, loan delinquency is the top component of banking spread, accounting for 37 percent, ahead of administrative expense (25 percent), taxes (23 percent) and banks’ profits (15 percent). Several policy measures can reduce the impact of delinquencies, including the use of collateral as guarantees, central registries, and enhanced credit bureaus.
The implementation of positive data will impact 120 million citizens. Estimates emphasize that 20 million more citizens could be included in a positive bureau, corresponding to individuals who do not have a traditional credit history with banks but could leverage other payment data. Calculations suggest that positive data could lead to an additional R$600 billion (about $115 billion) in new loan concessions, leading, in turn, to higher fiscal revenues. Credit penetration could also increase from 47 to 66 percent of domestic GDP.
The second promise of positive credit bureaus has to do with competition. Better scores will be available to a wider group of financial institutions, including, in particular, Fintech companies. While dominant banks have maintained a built-in advantage in terms of information gathered on their clients, positive bureaus will create a leveled playing field for lenders. This is consistent with the worldwide trend of customer-centric regulations, where citizens can dispose freely of their data.
However, Brazilian bureaus do face some limitations. The new rules implemented in 2019 incorporate only 12 months of payment history, and the full benefits of positive data may take some time to materialize––as long as two years, according to the Central Bank. Contributors, meanwhile, are adding their datasets gradually.
Moreover, bureaus have raised fundamental questions regarding consumer protection and data privacy. Negative data typically consist of late payments at stores and banks. But most recent trends for bureaus expand beyond the limit of credit. Payment history, bank account records, behavioral data, and psychometric assessments have been the most popular new avenues of research to assess creditworthiness. Consumers may worry that their data are being used in ways they do not anticipate or fully understand. They may also fear an increased risk of identity theft, especially in the case of data breaches. This is a timely concern, as a new law on personal data protection (LGPD) is set to enter into force in January 2021.
That said, safeguards are in place to protect customers: the scope of data gathered is limited, and information related to health or the types of goods and services acquired by consumers is not included in the Brazilian bureaus. Users can also choose to be removed from a given bureau at any time, and if they do, their data must be removed from all bureau databases within two business days.
Generally speaking, the scores produced by FinTech and data usage bear a question of social acceptability: is it socially desirable to deny a loan to a citizen based on a score, without further explanation? Can biases be embedded into those scores? How should these scores be used in exceptional times, such as the current COVID-19 crisis? In the US, a variety of measures have been proposed, on both bank-by-bank and state-wide bases. These include grace periods, forbearance, or a non-accrual of interest for a given period of time. In Brazil, the House of Representatives drafted a proposal in order to prevent bureaus from registering late payments taking place after March 20, that is, during the time of the pandemic. This law is to be analyzed by the Senate.
Removing data from bureaus is unlikely to create a favorable environment for lending and reduce credit rationing. As more people miss loan or bill payments due to the crisis, credit scores will need to capture the overall credit profile of individuals, not the impact of a particular accident alone.