Do Banks Always Need to Know as Much as Possible about Borrowers?

© iStock

Economists from HSE University have demonstrated that collecting as much information as possible about borrowers does not always decrease banks’ risks. Sometimes, more is not better: on the contrary, increasing the volume of data might increase the risks of loan defaults to a certain extent. The study was published in the WP BRP HSE University, Series: Financial Economics series of working papers.

When banks are making a decision on whether to issue a loan and on what terms, they try to collect as much information about the client as possible. It is assumed that the more data is available to the analysts or algorithm, the more precisely they can predict whether the client will be able to return the loan or not. One source of information for banks is credit history: information on previous loans, their amounts and the timeliness of repayment, and certain characteristics of the borrower (such as their financial reporting), etc. Banks can get this data from a credit bureau or a credit registry.

But some studies demonstrate that efforts to collect the maximum amount of information on borrowers can have side effects. For example, information on unreturned loans can get lost in a sea of information and play an insignificant role in stimulating borrowers to be honest. This means that data on unreturned loans tends to result in a weaker disciplinary effect, and borrowers’ motivation to pay on time decreases.

The relationship between banks’ credit risks and the volume of information collected depends on whichever effect is prevalent: credit risks may decrease thanks to higher-precision forecasting of credit defaults, or grow due to decreasing credit discipline.

The authors of the paper analysed data on the volume of information collected by banks and the level of credit risks in order to understand the correlation between these parameters. The economists also evaluated how this correlation is impacted by national institutional quality and the level of financial development. To do so, they used data on 80 countries for 2004–2015.

In order to find these answers, the authors built a mathematical model in which credit risks (the share of defaulted loans in the total number of issued loans) in a specific country and a specific year were the dependent variable, while the independent variables were the depth of credit information index, as well as a few control variables. The depth of credit information index is calculated by the World Bank as part of the Doing Business project, and its value depends on whether the information is collected on individuals and business entities, whether it includes only information on credit defaults or on timely repaid loans too, the duration of information storage in the database, the borrower’s access to their credit history, and so on. The highest values of this index are observed in the U.S., Canada, Argentina and Germany, while the lowest are in Luxembourg, Afghanistan and Iraq. Russia is close to the highest end.

The researchers also divided countries into groups depending on the efficiency of public administration (via the relevant World Bank index), property rights protection (Property Rights Alliance) and the level of development of the financial system (two metrics by the International Monetary Fund).

The calculations demonstrated that the correlation between the level of information disclosure and credit risks is truly non-linear, with the function graph looking like an upside-down ‘U’: as the volume of information collected grows, the risks grow initially before decreasing after a certain point. The risks at the maximum levels of credit information index are considerably lower than at the minimum ones.  In countries with effective administration, the risks start decreasing at lower values of the index (and in absolute terms, the risks are also lower), which means that in a worse institutional environment, regulators can decrease credit risks only by introducing a system of credit bureaus and extensive requirements on the collection of borrowers’ information. Similarly, the correlation between the volume of disclosed information and risks is influenced by respect for private property and the level of development of financial institutions.

Maria Semenova,  co-author of the paper, Senior Research Fellow at the HSE International Laboratory for Institutional Analysis of Economic Reforms and Associate Professor at the HSE Faculty of Economic Sciences School of Finance

‘The results of this study demonstrate that in developing economies, central banks should be cautious about introducing new requirements for information exchange on credit markets (such as expanding the range of borrowers or types of credit, information on which goes to credit bureaus). If such innovations are singular and are not accompanied, for example, by extended periods of credit history storage, providing online access to credit files and expanded lists of suppliers of data on borrowers, they may fail to produce the desired decrease in credit risks.’

The study also identified the countries in which the level of information disclosure is least useful in terms of credit risks. They include Malta, Mauritania, Mozambique, East Timor and others. The results show that these countries may need to change their regulations on credit information exchange in order to collect either much less or much more data than they do today.