Measuring Cross-indebtedness in Latin America

While over-indebtedness can be difficult to observe in practice, we can use simple surveys to estimate one proxy for over-indebtedness: cross-indebtedness, or the number of institutions from which the typical borrower has a loan. In markets with strong credit bureaus, MFIs have good data to support such surveys. Data on multiple borrowing lets us test the relationship with social and financial performance metrics, and provides a check on aggregate outreach figures.

Definition of over-indebtedness and multiple borrowing                                                                                                                                                  
Typically, we think of over-indebtedness when a borrower’s debts exceed their repayment capacity. However, this can be difficult to observe in practice. One common proxy for over-indebtedness is whether a borrower has multiple active loans outstanding. Since the borrower may have loans from several different institutions, we can try to measure the number of institutions from which they have borrowed instead. We call this measure – the number of MFIs from which a client borrows – ‘cross-indebtedness.’ As credit bureaus have opened in some markets in Latin America, MFIs have more information about multiple borrowing by clients.

Methodology                                                                                                                                                                                                                    
Last year, MIX surveyed MFIs in Latin America to ask how many of their borrowers have loans from more than one institution. A sample questionnaire is included below:

FIGURE 1 HERE

For 2009, 139 MFIs in 15 countries reported on these questions, although we will focus on just Ecuador and Peru here. Using some arithmetic,¹ we extrapolate from the survey responses to come up with an estimate for the average number of institutions from which each client has borrowed.

The result is a ‘cross-indebtedness’ metric for each MFI. We interpret this as the average number of MFIs from which each borrower has one loan or more. We can then aggregate these results for all MFIs in the sector to produce estimates of total sector cross-indebtedness.

Updating aggregate outreach figures

Table 1 shows these results for Peru and Ecuador for 2009²:

From the table, we can see that, prior to adjusting for multiple borrowing, the average borrower appears to have 1.1 loans in both markets, and outreach is to 595 thousand people in Ecuador, and 2.8 million in Peru.

FIGURE 2 HERE

As a first result, we need to adjust total outreach downwards to reflect multiple borrowing by clients. The graphic above shows this dynamic: since most aggregate figures count borrowers by MFI, if a borrower has loans from more than one MFI, a simple total across MFIs will double-count those borrowers.

The cross-indebtedness figures have a fairly dramatic impact on outreach figures. The number of borrowers is reduced by 33% in Ecuador and 27% in Peru, due to the effects of cross-indebtedness. If these numbers are typical globally, we would expect to see high-level outreach figures reduced by tens of millions of borrowers.

Adjusting aggregate outreach figures also affects market penetration rates and supply/demand estimates. Some recent research has pointed to market penetration (or saturation) as a risk factor for microfinance crises and these types of adjustments can be important to improving those measurements.

Relationship to social performance                                                                                                                                                                                                                               
We can also use the institution-level data to see whether MFIs that adopt the client protection principles through the Smart Campaign appear to have clients that take on fewer loans. The first client protection principle addresses over-indebtedness and asks MFIs if they “ensure that credit…will not put the borrowers at significant risk of over-indebtedness.” When we look at responses to the Smart Campaign principles for Peru and Ecuador, we can see though that cross-indebtedness levels for MFIs that observe the first principle do not differ substantially from those that do not.

FIGURE 3 HERE

Does this mean that cross-indebtedness is not an important risk factor or does this indicate some other process by which MFIs evaluate repayment capacity?

Relationship to MFI performance                                                                                                                                                                                             
If we look at other performance metrics, we can see a potential relationship between cross-indebtedness and write-offs or portfolio-at-risk levels, the prominent measures of credit risk.

FIGURE 4 HERE

FIGURE 5 HERE

Source: MIX

While the relationship is not strong in either case, we do see that higher risk levels are associated with higher levels of cross-indebtedness. More detail and tests of how cross-indebtedness relates to product offerings and growth are covered in our most recent Latin America regional report.

Conclusions 

As long as MFIs continue to provide information on borrowers to credit bureaus, we will be able to validate the relationship between over-indebtedness, multiple borrowing and social and performance metrics such as write-offs, product offerings and growth, and also to explore new relationships to learn the importance of cross-indebtedness for the health of microfinance sectors.

¹The methodology applies the percentages from the questionnaire to the stated outreach figures of the MFI. If all borrowers have debts only with the MFI, there is no adjustment. If, say, 80% of the borrowers have one loan, and 20% have loans from two MFIs, then the cross-indebtedness level will be 1.2 = 80% * 1 + 20% * 2, and so on for higher levels of multiple borrowing. When survey results are more limited we also make some inferences based on average levels for peer MFIs in the country.

²While data from the credit bureaus is confidential, we have private confirmation that the results are ‘close’ to the true figures, although somewhat surprisingly they slightly underestimate the proportion of exclusive borrowers (with loans from only one MFI).