March 18th, 2009 
Yong-Long Lai Was looking at my Kiva Account today and wanted to see if there were people I could lend some money to (yes I was bored)… unfortunately there was none. While I was browsing, I noticed there some cool statistics available on the site.
I’ve obviously been lucky to not have suffered any losses… (maybe it’s my savvy Credit Risk Management skills?)
Anyways, since I work in Credit Risk Management so I figured I’ll do some simple analysis considering the Actual Loss rate that’s been observed is quite ‘high’, definitely a lot higher than what I noticed from when I first opened the account (from memory). I guess today is your lucky day because you get a quick lecture on Loss Calculation using Kiva as an example.
Anyways, here’s the numbers we have to work with
- loss rate of 2.09%
- probability of delinquency of 4.28% (this is actually 1 payment down – i.e. 1 month down. Bank use 90 days or 3 payments down for capital purposes and also other risk related measurements)
The loss rate of 2.09% is equivalent to the Expected Loss and the Probability of Delinquency of 4.28% is equivalent to the Probability of Default. Now we can reverse engineer the Loss Rate or the Loss Given Default.
Expected Loss = Probability of Default x Loss Given Default
Loss Calculation:
LGD = EL/PD = 4.28%/2.09% = 48.8%
Remember I mentioned that Kiva is using a 30 day bad/default definition. If we 1/2 the PD from 4.28% to 2.14% we will effectively double the loss rate (LGD) closer to 97.6% which is effectively full loss. Keep in mind I suspect they haven’t even added economic loss (ie. time value of money – discounting of cashflow) End of the day, these sort of numbers makes sense to me since these are effectively unsecured loans. If you think about it this makes sense intuitively, if you lend somebody money with no collateral and they are unable to pay you then you are most likely to lose all the money you lent them.
I’m just surprised at the actual loss rate from the Kiva portfolio is quite high. Since lenders (i.e. us) don’t receive any interest income from these lending we are effectively taking on quite a bit of risk. That said, the Kiva field partners usually charge borrowers interest… the question is should we be charging the field partners at least 2.09% so that we can cover for the long term loss or make the Kiva field partners absorb the losses. If we don’t, it’s sort of like playing on a roulette table, the more numbers you bet on over time the more chance you have of losing money.
Summary, once your borrower is late on any of their payments (4.28%), you are likely to use 48.8% of your money.
Obviously I don’t actually have the underlying data to check all their number and I suspect there is sampling bias of some sort most likely due to early full losses that Kiva suffered on some of their loans or they didn’t include loans that have not completed their term which would cause both the expected loss and delinquency rates to be higher than what Kiva has reported. I figured this is the case from the following note:
- Many Field Partners do not yet have many Ended Loans due to their short history on Kiva (see “Time on Kiva”). A more meaningful indicator of principal risk is “Delinquency Rate”.
- For loans that are delinquent at the end of a loan term, Kiva allows the Field Partner 6 additional months to attempt collections before deeming the loan as Defaulted.
End of the day, regardless of the results from the loss calculations, I’m still going to continue to lend money to these guys and donate because I think it’s a worthy cause! so make your heart a little warmer today by opening an account and lending some money!
Posted in 
Tags: 
