Monday, November 15, 2010

Changes to Kiva's Delinquency Rate

Kiva is changing how we report Delinquency Rates, in response to some longstanding requests from our lenders and field partners.

Historically, we've calculated Kiva’s Delinquency Rate as the remaining principal amount of all loans that are past due, divided by the total principal amount of loans being repaid. Over the years, a few issues have been identified around this calculation:

1. The penny problem

If a borrower was even a few cents below their expected repayment, or just a few days late in repayment, Kiva's Delinquency Rate calculation would consider the entire loan delinquent. This unfairly over-stated the actual delinquency.

2. Borrowers versus Partner late payments

Kiva's Delinquency Rate calculation takes into account both late payments from borrowers, and late payments from the Field Partner to Kiva. This was true even when the late payments were the result of technical billing problems with the Partner.

Some of our lenders have expressed concern that Kiva's Delinquency Rate doesn't separate late borrower payments versus late Partner payments. We'd like to address this concern in the coming months; for now, we'd like to at least make sure that the delinquency rate statistic only takes into consideration the amount that is still outstanding with the lender, and thus measures what is actually late to the lender.

ADJUSTMENTS

As a result, Kiva is making some adjustments to its Delinquency Rate calculations in order to reflect these two points of feedback.

As of today, Kiva’s Delinquency Rate will = the amount of payments that are past due, divided by the remaining principal amount of loans being repaid.

Now, Kiva's Delinquency Rate calculation will address the penny problem. It will also now take into consideration the amount that is still outstanding with the lender, and only measure what is actually late to the lender.

Kiva will continue to provide information similar to the previous definition of the Delinquency Rate under a new statistic called “Loans at Risk” that will be found on the Partner Page. This will show the remaining principal amount of all Kiva loans that are past due/the remaining principal amount of Kiva loans being repaid.

IMPACT OF ADJUSTMENTS

On a practical level, what does this mean for the Kiva community?

Late Paying Partners

For partners who are having problems repaying Kiva lenders, their delinquency rates will be appear higher than they currently do. That's because historically, we've been calculating delinquency rates as: the remaining principal amount of all loans that are past due, divided by the total principal amount of loans being repaid

Say that a partner owes Kiva lenders $10k total and hasn't made any payments in months. According to our older calculation, if that $10k that they owe was part of say, $20k in total loans... their delinquency rate would be calculated relative to that total $20k amount. But in actuality, it would be more accurate to calculate the delinquency rate relative to the $10k that they owe. So the delinquency rate in this case should be close to 100%: of the $10k that the partner owes Kiva lenders, it's late in paying all $10k of it.

In other words: if a partner is having trouble repaying Kiva lenders, its reported delinquency rate is more likely to be appear substantially higher than it has in the past.

Other cases

In other cases, delinquency rates will no longer be affected by the penny problem. If a $1k loan is repaid all but one penny, the delinquency rate will now reflect that a single penny is delinquent - and not the entire $1k.

In these cases, the delinquency rate should be more accurate - which in this case, means it will be lower.

-----------

Overall, we wanted to make sure to address here in our blog these changes in how we calculate and display statistics. We welcome feedback on all of these stats! We believe the new stat will be substantially more useful and accurate, but are looking to our community for feedback.

Please let us know what you think. Thanks so much!