Print
Share
Interest Rates by Region
March 25, 2011
Working on chapter 7 now, I just updated and refined this graph showing the distribution of microcredit interest rates by region. It is my attempt to convey the kind of information in Figure 1 of this report by Richard Rosenberg, Adrian Gonzalez, and Sushma Narain. The basis for the figure is the gross portfolio yield data series from the Mix Market, which is just the amount of interest an MFI received during the year divided by its average outstanding loan stock—a rough but ready proxy for the interest rate. I’ve adjusted the figures upward by adjusting in an approximate way for loan losses, which reduce gross portfolio yield below the true interest rate. But I’ve adjusted them downward by subtracting inflation. I have not weighted MFIs by number of borrowers (the graphs look funny when I do), so big and little ones count the same.
I have not adjusted for compounding, compulsory savings, overpriced credit-life insurance, or value-added tax. See my post on Compartamos’s interest rate for more on these factors. Here, Compartamos comes in at 76%/year above inflation.
I tried my Tuftesqe best to make the presentation clean and informative. Comments welcome on the substance and presentation.
Counts of MFIs by region and interest rate, adjusted for inflation but not compounding, 2009
Possibly Related Posts
- OK, This Time I’ve Really Got It: Grameen’s Interest Rate Revisited
- Definitive Measurement of Grameen Interest Rates
8 Comments on “Interest Rates by Region”
Post a Comment
We value frank and constructive exchanges and encourage you to use your real name in your comments.
March 25th, 2011 at 7:26 pm
David, gross loan portfolio already is exclusive of loan losses so you may be double counting in adding them back again. Net loan portfolio is the concept that subtracts loan losses.
March 25th, 2011 at 10:32 pm
This looks right to me; the difference between gross and net loan portfolios are the reserves held against future losses. Actual realized losses (i.e. write-offs) would hit both the gross and net portfolio. So it makes sense to add the realized losses back to the gross portfolio so that the basis for the yield better reflects all the loans the institution made during the period.
On the other hand, it sounds like you are subtracting the inflation rate directly from the yield. Generally, we divide through by it – so the real yield is nominal yield / (1 + inflation rate), not nominal yield – inflation rate. Probably not a huge difference in practice for either adjustment though.
I couldn’t resist trying another variation on the graph, using some data we had for a post earlier in the month: http://public.tableausoftware......ldbyregion?:embed=yes&:toolbar=yes&:tabs=no.
March 25th, 2011 at 11:02 pm
Thanks, Scott. That is reassuring. Actually, I do divide by the inflation rate the way you describe—I just avoided getting into the technicality.
The Tableau graph is very cool! I’m thinking about trying Tableau for something else.
March 27th, 2011 at 5:37 pm
David,
Glad to learn that the denominator is gross loan portfolio. However, two points I’d like to add from my experience analyzing financials of African MFIs:
- Many MFIs in Africa recognize interest on a cash basis. This is important for organizations with non-zero PAR figures as non-collected interest doesn’t show up in the numerator. This “yield gap”, the average actual interest rate on loans vs. portfolio yield, is common and is a sanity check against reported PAR figures. With respect to your most recent blog post, the gap between yield and APR (ignoring savings, etc.) should be smallest where PAR is low.
- At least in Africa, writeoffs tend to be made only after loans are overdue more than 365 days. Because of this, adding back in writeoffs in the current year is not necessarily appropriate as they largely reflect bad outstanding loans from previous years. Furthermore and to the point above, organizations would not be recognizing interest on these loans on a cash basis (or, very likely, on an accrual basis) so it’s unclear what the value of adding back in written off loans is.
Best,
Ben
March 28th, 2011 at 10:18 am
Excellent, thank you Ben. Of course, my purpose is to give a sense of the overall distributions rather than estimate the rates of specific MFIs. To the extent that loan loss rates are stable on average in given countries from year to year (important caveat), then I should be OK. But do you think it would be better to adjust for PAR in a rough way in addition or instead?
March 28th, 2011 at 10:24 am
David,
Not a problem.
On loan loss rates being the same from year to year, I don’t know what MIX’s data says on writeoff ratios on the aggregate. In Africa (anecdotally), writeoffs tend to be zero for a number of years and then spike. However, in other regions I’ve seen them be more stable.
On adjusting yield for PAR, I wish there was a way to do this! Unfortunately, I haven’t seen a formula for this but let me know if you see anything.
Finally, I think yield is actually a useful proxy for a concept you spoke about in your book: the “actual” interest rate paid. I know you addressed this in the concept of the money lender and the actual rate when you look at when payments are actually received. APR tells you what the interest rate is upfront but yield tells you the blended, weighted interest actually paid by clients for an MFI.
Ben
March 31st, 2011 at 12:48 pm
Median #s for sub-Saharan Africa look too low (based on my experience with SSA FSAPs). Commercial banks’ prime rates in East Africa, for example, routinely exceed 15% and MFIs charge a premium of up to 40% on top of these rates.
What countries are included in the MIX sample? I would love to see a similar graphic for a sample of countries in each region.
Thank you!
March 31st, 2011 at 2:11 pm
Ed, would the average inflation rate of 8% in Sub-Saharan Africa help explain the difference? These are inflation-adjusted rates. That 8% figure is weighted by the number of MFIs per country.