The theoretical case for credit registries is strong. When lenders share information, defaulting borrowers lose their reputation in the credit market as a whole and not just with their current lender. This can reduce moral hazard. Centralised credit data may also ease adverse selection (Pagano and Jappelli 1993) and prevent borrowers from taking on too much debt by borrowing from multiple banks (‘double dipping’).
Notwithstanding these clear priors, the empirical evidence on credit registries remains scarce and is mainly based on cross-country comparisons. This evidence indicates that information sharing is associated with more lending to the private sector and fewer defaults (Jappelli and Pagano 2002, Djankov et al. 2007). More recent evidence, using micro data, qualifies these macro impacts somewhat. For instance, Giannetti et al. (2015) show that when banks are forced to share borrower information, they may manipulate credit ratings before making them public, thus reducing their usefulness. Recent anecdotal evidence also raises questions. When a new registry was introduced in the United Arab Emirates in 2014, it was widely expected to increase banks’ appetite to lend but instead led a sharp decline in loan approvals.
In a new paper (Bos et al. 2016), we present micro evidence of what happens when a new credit registry obliges lenders to start sharing borrower information. Evaluating the impact of such a regime change is challenging for two reasons. First, borrower information is typically only available after a registry is introduced. Second, even if pre-registry data exist, it remains difficult to identify the impact of information sharing if all lenders and borrowers are similarly affected by the new regime.
To surmount these challenges, we use data on the complete loan portfolio of a small-business lender in Bosnia and Herzegovina. We exploit information on the terms – amount, maturity, interest rate, collateral, and performance – of all small-business loans of this lender. We also have information on all loan applications that the lender rejected and why they were rejected. Crucially, these contract-level data refer to both the period before and after the introduction of a credit registry in Bosnia and Herzegovina in July 2009. We therefore observe lending decisions by the same loan officers under different information-sharing regimes. A difference-in-differences framework combines the time variation in information sharing with geographical variation in competition and borrower variation in lending history. Our analysis yields four main results:
- The registry is actively used
Our analysis first of all shows that the lender whose data we analyse actively uses the new credit registry. The registry requires lenders to submit a report for each loan to a firm or private individual that is disbursed, repaid in full, late or written off. It thus contains both ‘negative’ information on past defaults and ‘positive’ information on pre-existing debt. We find that after the registry introduction, loan rejections are based increasingly on hard information – especially registry information about applicants’ outstanding debt. The use of negative information increases too and this is especially the case in high-competition areas where adverse selection problems are typically more severe. In contrast, the probability that a loan gets rejected due to soft information (such as a bad recommendation) declines.
- The registry leads to fewer and smaller loans
Our analysis shows that loan officers tighten lending at the extensive margin. The rejection rate almost doubles from 8 to 15% after the introduction of the registry. Using loan-officer fixed effects, we find that the introduction of the registry also tightens lending on the intensive margin – first-time borrowers receive smaller, shorter and more expensive loans that require more collateral. Average loan size drops by 19% (and in high-competition areas even by 25%) while the average loan maturity shortens by 13% in low-competition areas (equivalent to 90 days) and by 16% in high-competition areas (almost 120 days).
Interestingly, however, repeat borrowers can now signal their quality to competing lenders and this appears to force the incumbent lender to offer better loan terms. Using borrower fixed effects, we show that – in line with a decline in switching costs (Ioannidou and Ongena 2010) – repeat borrowers receive progressively larger, longer and cheaper loans once the registry is in place. This effect is driven by high-competition areas. In areas where more lenders compete, information sharing opens up more outside options to formerly captive borrowers and the impact of information sharing on repeat borrowers is consequently larger.
- The registry improves loan quality
The tightening of lending standards at the extensive and intensive margins also improves loan quality, in particular in high-competition areas and for first-time borrowers. Figure 1 provides a non-parametric look at our loan-quality data in the form of a Kaplan-Meier survival analysis for the period June 2002 to December 2012. Both panels focus on high-competition areas, where our results are most pronounced. Each panel follows the survival rate of loans after disbursement, both for the period before and after the introduction of the registry.
Figure 1. Survival rate of loans disbursed with and without information sharing
Both panels reveal a substantial difference in repayment behavior once the registry is in place. After the registry is introduced, loans have a significantly higher survival probability compared with loans approved without information sharing. A comparison of the panel on the left (new borrowers) and right (repeat borrowers) shows that the registry impact is larger for new borrowers. The panel on the left indicates that in high-competition areas the survival probability for new borrowers after 12 months increased from 92.5 to 97.5%.
- The registry improves lender profitability
To assess the impact of the registry on the lender, we evaluate its profitability in the year before (June 2008–June 2009) and after (July 2009–July 2010) the registry introduction. After the introduction, the present value of total lending goes down by 49.7% due to the combined effect of more rejections and smaller loans. At the same time the registry leads to a strong increase in repayment performance. As a result, the net present value per US dollar lent increases from 11 to 14 cents while the internal rate of return on lending increases from 17.6 to 21.8%.
We document how mandatory information sharing allows loan officers to lend more conservatively at both the extensive and intensive margins[RB1] . This increased conservatism reflects in particular the availability of positive credit-registry information, which provides loan officers with a complete picture of the indebtedness of loan applicants. The resulting improved credit allocation increases loan quality, in particular for first-time borrowers.
At first sight, the increase in rejection rates and the reduction in lending appear at odds with positive cross-country correlations between information sharing and banking sector depth. Our view is that both observations are not inconsistent. We expect that in the longer term, the improved functioning of the Bosnian credit market (and the associated higher profitability of lenders) that we document will stimulate lending. Indeed, our data show how the increased transparency in the credit market due to the new registry already allows well-behaved repeat borrowers to increase their borrowing limits and enjoy better loan conditions.
Bos, J, R De Haas and M Millone (2016) “Show me yours and I’ll show you mine: Sharing borrower information in a competitive credit market”, Baffi Carefin Centre Research Paper No. 2015-8.
Djankov, S, C McLiesh and A Shleifer (2007) “Private credit in 129 countries”, Journal of Financial Economics, 84(2): 299-329.
Giannetti, M, J M Liberti and J Sturgess (2015) “Information sharing and rating manipulation”, Swedish House of Finance, Research Paper No 15-11.
Ioannidou, V and S Ongena (2010) "Time for a change: Loan conditions and bank behavior when firms switch banks”, Journal of Finance, 65(5): 1847-1877.
Jappelli, T and M Pagano (2002) “Information sharing, lending and defaults: Cross-country evidence”, Journal of Banking & Finance, 26(10): 2017-2045.
Pagano, M and T Jappelli (1993) “Information sharing in credit markets”, Journal of Finance, 48(5): 1693-1718.