More Reflections on Transparency (not about Kiva)

October 27, 2009

In July, I blogged about MFTransparency, which Chuck Waterfield founded to “combat…confusing and predatory pricing” in microfinance. The core idea is to express the various charges and fees in a single number, what I have been calling the Annual Percentage Rate (APR) and MFTransparency says is the Effective Interest Rate (EIR). (Read the Wikipedia entry to become more confused.) MFTransparency has just released its first data sets, for Cambodia and Bosnia & Herzegovina. One visit helps you understand why the data were so long in coming: it was clearly a big job. MFTransparency got good data and has presented it well.

Two old thoughts, then one new:

• Are there moves to bring these effective interest rates to borrowers’ eyes so they can spot predatory pricing? Or are mostly people like you and me seeing them?
• Has thought been put into whether EIRs are the most effective way to convey costs to borrowers? The study by Marianne Bertrand and Adair Morse makes me doubt they are. They found that quoting an APR of 400% had less impact on payday loan takers in the United States than telling them that borrowing $300 repeatedly, two weeks at a time, for 3 months would cost $270 in interest.
• I compared MFTransparency’s results with the nearest analog on the MIX Market, “gross portfolio yield,” which is basically how much money a microcreditor takes in as a percentage of average loans outstanding. GPY depends directly on interest rates, but can be reduced by defaults and other factors. Indeed, MFTransparency’s figures are generally higher, sometimes substantially so. For ACLEDA, Cambodia’s largest microcreditor, MFTransparency’s calculated rates are 35.2–39.9%/year, compared to just 23.3% on the MIX. This makes me wonder whether analyses of microcredit interest rates based on MIX Market data are too sanguine. For my draft chapter 7, and inspired by the reassuring CGAP analysis of MFI interest rates, I crafted a graph (below; let me know if it is confusing) showing the distribution of microcredit interest rates using inflation-adjusted MIX figures. No region’s median inflation-adjusted GPY exceeds 25%, which is not much more than my credit card charges. Most of the CGAP report’s conclusions probably stand—for example, interest rates probably really are falling, and gross profiteering appears rare—but I do wonder whether I need to adjust the language in my draft chapter to convey that rates may be significantly higher than this graph suggests.

Counts of MFIs by region and average interest rate, 2007, with 5th, 50th, and 95th percentiles vertically marked

Note: Rates are gross portfolio yields, adjusted for consumer price inflation. They do not factor in the hidden costs of forced savings and overpriced credit-life insurance. From left to right, vertical lines show 5th, 50th, and 95th percentile interest rates.

Tweet  

1. Sean Kline Says:
   October 27th, 2009 at 7:49 pm

   As the founder of one of Bosnia’s MFIs I’m keen to see who uses these good new (apples to apples’) pricing data and to what effect. Let’s keep pushing here…

2. Scott Gaul Says:
   November 2nd, 2009 at 8:20 pm

   It is unlikely in this case that the rates would be ‘significantly’ higher due to the differences between stated interest rates and realized income measures (like yields). However, there is a bit of apples-to-oranges comparison which may be the culprit in the example cited here; or rather, apples to apples-and-some-oranges.

   The figures on the MFTransparency site for ACLEDA add up to about 125,000 clients (http://www.mftransparency.org/.....utions/85/). The figures from MIX (for which I work) add up to about 215,000 clients for 2008 (http://www.mixmarket.org/mfi/acleda/data). You can see more detail on the products by expanding the total borrowers or gross loan portfolio values on the MIX site. The gap is probably caused by the fact that some higher-balance / lower-rate products offered by ACLEDA are included in the MIX figures and excluded from the MF Transparency ones. (If you want more detail, you can see p. 26 of ACLEDA’s audits for the year – here: http://www.mixmarket.org/sites.....FS2008.pdf.) Thus, the yield calculated across products is overall lower than the yield for the individual products.

   Defaults would cause differences between APR / EIR and yields. You can get an index of the effect of defaults by looking at the write-off ratio for any MFI (write-offs / average GLP, the same denominator as yield). However, ACLEDA’s write-off ratio has not exceeded 0.5% in the last several years, and thus should not cause a 12 – 16% difference in interest rates / yields. More data is here for Cambodia in general: http://www.mixmarket.org/mfi/t.....0=Cambodia.

   MFTransparency (I believe) also adjusts rates upwards for products that have compulsory deposits. This is a moot point in the case of ACLEDA (which only has voluntary deposits), but could affect comparisons for other institutions.

3. David Roodman Says:
   November 3rd, 2009 at 12:15 am

   Thank you, Scott. From what you write, it sounds like MFTransparency’s numbers can give you a better measure of the interest rate a fairly poor person would pay—which makes perfect sense since that’s what they set out to carefully measure. Come to think of it, I interviewed the head of ACLEDA once and he said that they were proud to be making larger loans to better-off people as part of their business. Sounds like MFIs that move upscale more will appear to have lower rates in analyses like that by Richard Rosenberg, Adrian Gonzalez, and Sushma Narain that use portfolio yields to proxy for interest rates.