What’s Wrong With Kiva’s Portfolio Yield Statistic?

I have criticised Kiva for being non-transparent in a number of ways, but one of my pet-peeves is their insistence upon using the portfolio yield statistic instead of an actual interest rate. Why is this so deceptive a statistic? What could Kiva do better?

First of all, portfolio yield is defined by the MixMarket as the “Interest and Fees on Loan Portfolio/ Loan Portfolio, gross, average”. Kiva get their portfolio yield statistics from the MixMarket, and add further clarity: “The Portfolio Yield is generally based on audited financial information and is a better indication of the cost of borrowing money from a Kiva Field Partner than the simple interest rates reported by our Field Partners because it:  a) Includes any fees associated with loans and,  b) Is expressed in one-year increments (similar to the way an APR works). Please note that Portfolio Yield does not yet include the concept of mandatory savings.”

So, Kiva acknowledge some limitations of this statistic, but let’s look at these limitations more closely.

First, the figures are not always based on audited data. In fact, there is no requirement for audited financials on the MixMarket, and the authenticity of documents and statistics uploaded to the MixMarket is not verified. They are self-reported by the MFIs, and we all know how sensitive the topic of interest rates can be, so there is a strong incentive to massage this figure downwards. Also, the portfolio yield statistics are not always up-to-date. And many MFIs do not report to the MixMarket. In the case of Kiva’s partner Tujijenge, for example, they have not filed data to the MixMarket since 2008. Even when they did bother to file data to the MixMarket, they failed to include the portfolio yield. However, it is fair to say that there may be some flaws with the MixMarket data itself.

Secondly, the figure takes the income over the period (a year), and divides this by the average portfolio. This is questionable, as the implicit assumption is that the portfolio grows linearly over the year. If the portfolio grows exponentially, this figure will be biased downwards (I can provide a mathematical proof if anyone is interested). But this is a relatively minor point.

The next point that Kiva raise is that the yield includes all fees. To correctly report the yield it is correct to include all fees, the question is whether the MFIs do this. It is extremely easy to avoid reporting fees in the audited accounts, and this is not necessarily illegal in the countries in question. One trick is to do a thing called “off-setting”. The MFI takes these fees and nets them off against some of the branch level expenses, before consolidating the accounts. This is not illegal, and the final audited accounts are unchanged, so the auditor doesn’t mind. For example, assume an MFI makes $1.000 in interest and $100 in fees on a constant portfolio of $2.000. Its total operating expenses for the year are $500. This MFI can report this in two ways:

  • Total income is $1.000 + $100 = $1.100, on a portfolio of $2.000, so the portfolio yield is 55%
  • Its operating costs ($500) are 25% of the portfolio, i.e. it costs the MFI $0.25 for every dollar outstanding
  • Total profit is $1.100 – $500 = $600

Or it can off-set the fee income to pay local branch expenses, and report the following:

  • Total income is $1.000, on a portfolio of $2.000, so the portfolio yield is 50%
  • Its operating costs ($500 – $100 = $400) are 20% of the portfolio, i.e. it costs the MFI $0.20 for every dollar outstanding
  • Total profit is $1.000 – $400 = $600

Now, to most people this makes very little difference, the totals are the same. The MFI is not stealing, it is simply moving money from one pocket to another. At the end of the year it still spent $500, got $100 in fees and $1.000 in interest either way, the portfolio was $2.000 in either case, and the profit is the same. By moving from the first method to the second the MFI reduces the portfolio yield and reduces the operating cost ratio – what it gains in one it loses in the other. Essentially a yield of 55% and an operating cost ratio of 25% is the same as a yield of 50% and an operating cost ratio of 20%. So, why would an MFI want to move from the first option to the second?

Simple: the second example appears less expensive to the poor, and more efficient in terms of operating expenses, both of which endear the bank to the investors. The other point to make here is that there is a spectrum of combinations of yields and operating cost ratios which mean the same thing – one of which is reported by the MFI to the MixMarket and reproduced on Kiva. How do we know if this is accurate? Does Kiva go through the account consolidation process? Do any investors? Ask Triple Jump or Blue Orchard if they check for this – if they are reading this blog I expect they are furiously taking notes. Does this happen with Kiva MFIs? Who knows, but it is certainly not illegal in most jurisdictions, and would make complete sense for the MFIs to try to do this. So, let me stress, I am NOT suggesting this is happening with Kiva partners, I am merely suggesting that it is entirely rational for MFIs to do this, and given the broader criticisms I have made elsewhere regarding the due diligence performed by these jokers, I would be extremely surprised if they detect this flaw.

But then we come to the most important part: forced savings. These have a dramatic impact on the cost of capital. The FAQ section of MFTransparency states:

“Security deposits [forced savings usually] are left out of the price calculation in formal regulations of many countries, but is a very serious loophole which can be exploited to dramatically increase the true cost of the loan to the client.” [emphasis added]

This couldn’t be much clearer. The critical fact is that all costs, all cashflows, all fees must be considered. If I have to deposit $20 in order to obtain a $100 loan, I am only actually receiving a net $80. The APR reflects these deposits, and any interest that may be earned on such deposits, to construct the APR to reflect ALL expenses according to microfinance best-practice. Chuck Waterfield, the founder of MFTransparency, is a veteran microfinance practitioner, probably the world expert on microfinance interest rates, a man of great integrity, respected worldwide, honest and hard-working and the data he produces should be taken extremely seriously. He is a staunch advocate for the poor and an example to the sector of genuine transparency. Portfolio yield excludes the effect of forced saving and should therefore NOT be considered a proxy for the APR in any MFI that engages in such practices.

Finally, the portfolio yield, for all its flaws, is simply an average. If an MFI charges rates of 40% to 100%, is it fair to state that the average is 70%? It might be accurate, but do Kiva users want to lend money at 100%? They have no way of knowing. Perhaps the Kiva loans are all lent at 100%, and the MFI’s other loans funded elsewhere are the 40% loans. The average will be the same, but is this transparent?

But, perhaps the weirdest question of all is why the MFI cannot simply state “Kiva loans are lent at xx%”. Why should this be so hard? They have various loan products with various interest rates. The MFIs know precisely how much each client is paying. Why can they not simply say “we will lend Kiva money at xx%?”. I will leave this question open for the time-being, but think about this. If you go to the bank and ask for a loan, would you be a little surprised if the bank replied “sure, the interest rates are usually between 10% and 20%”? Perhaps an approximation as you make an initial enquiry is acceptable, but Kiva state this average after the loan has been disbursed.

So, by way of summary, let’s just compare the stated portfolio yields of a few MFIs that use Kiva and are also reported on the website of MFTransparency – this will give us a feel for how well the portfolio yield stacks up with the actual APRs charged to the poor. The graphs below follow the same format: the black line is the average for the country, and consistently demonstrates that the interest rates are lower on larger loans. The bubbles are the loan products of the specific bank in question, with the APRs recorded on the vertical axis, the loan size on the horizontal axis. I will focus on Africa as this is where the divergences between APRs and portfolio yields tend to be the largest.

Juhudi Kilimo – Kenya, Kiva stated rate: 25.7%

It appears this MFI charges substantially more than the Kiva rate in all cases

Selfina – Tanzania, Kiva stated rate: 25.23%

It’s hard to imagine how an average of only 25% could emerge from this data

Hluvuku Adesma – Mozambique, Kiva stated rate: 42.69%

Although this MFI is cheaper than the average in Mozambique, very few loans appear to be even close to the Kiva stated rate

Christian Rural Aid Network – Ghana, Kiva stated rate: 33.64%

Once again, it appears the actual APRs and the Kiva portfolio yield diverge dramatically

UGAFODE – Uganda, Kiva stated rate: 47.34%

Real APRs are way in excess of the Kiva stated rate

Amasezerano Community Banking – Rwanda, Kiva stated rate: 39.96%

Possibly Kiva loans are the cheap ones, but again, the main loan products cost in excess of 60%

Yehu Microfinance Trust –  Kenya, Kiva stated rate: 37.39%

Is a pattern emerging here?

 Urwego – Rwanda, Kiva stated rate: 50.63%

They have a few cheaper loans, but the main products are above 60% APR

BRAC – Tanzania, Kiva stated rate: 47.44%

Once again, the vast majority of loans are way in excess of the portfolio yield

 Tujijenge – Tanzania, Kiva stated rate: 66.1%

Actually this looks fairly accurate, although some are costing approaching 100%, 66.1% is certainly not a fair average. (This MFI is deeply flawed for other reasons, see my recently blog post)

So, what can we conclude from this? First, it appears that the actual APRs charged to the poor are consistently higher than the portfolio yields stated by Kiva. This lends additional weight to not trusting this statistic as a fair proxy for the interest rate. By extension we can therefore assume that most Kivans are unaware of the actual rates that the poor are paying, as they rely on this flawed statistic. Do we believe that Kiva are unaware of this divergence? This is a double-edged sword, as if they claim they have no idea, one must question their broad knowledge of microfinance. Kiva has channelled hundreds of millions of dollars to the poor via these MFIs, and if they are unaware that the poor may be paying massively more than that stated, we may have a serious problem with Kiva managerial competence. But if they are aware, then why would they continue citing a knowingly flawed statistic?

I do not believe Kiva management are incompetent. I believe they are fully aware of this and chose to use the portfolio yield statistic simply because it presents a rosy impression of interest rates. I am yet to find a case when this does not act in Kiva’s favour, i.e. an MFI that is charging APRs lower than the portfolio yield. Kiva can defend this stance by claiming this is “verified information” from the MixMarket, and that they do not actually state that the portfolio yield is the interest rate, so legally they are protected. But are they protected morally?

Any claim that it is too complicated to calculate the real interest rate is nonsense. MFTransparency seem fairly able to do it. They publish an interest rate calculator on their website for free download. MyC4, another microfinance lending platform who occasionally partner with the same MFIs as Kiva, are able to publish the actual APRs to two decimal places. Calculating APRs is not rocket-science.

Are the MFIs lying to the MixMarket? Perhaps sometimes, but in fact the portfolio yield may in fact be accurate – it just isn’t a proxy for the actual cost of capital to the poor clients. Is Kiva lying to the Kivans? No, it is presenting factual information and the interested reader can find out that Kiva do actually warn (in the small print) about too closely equating portfolio yield with the interest rate. But could Kiva do better? Yes. And are we all turning a convenient blind eye to the rather sensitive topic of extortionate interest rates? Most certainly. Is this transparent? In my opinion, no.

Kivans like to believe they are helping the poor, and in order to achieve this Kiva needs to provide them with minimal, but reassuring information. Some nice photos, a little story, and as favourable an impression of the actual interest rates as possible, as this is an emotive topic that will irritate many Kivans. They can get away with rates of 30%, 40%, even 50%, but they have to avoid rates which will raise too many questions, and by citing a statistic known to be deeply flawed, but reassuring the Kivans, is the best way to do this. Kiva endorsed the SMART Campaign initiative on transparent pricing blah blah… is this transparent? I repeat, again, the quote from Wagane Diouf in his testimony to the House of Representatives:

“I get seriously worried when these institutions [MFIs] start mobilizing funds from institutions that attract capital from individuals in the US and other western countries, such as, um, I won’t mention their name, but institutions that have web-driven mechanisms to attract investments, but the financial reporting of the institutions that are receiving these funds are not up to standard at all, they are very poorly regulated, it’s a very opaque part of the industry.”[emphasis added]

 

 

 

Did you like this? Share it:
This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

5 Responses to What’s Wrong With Kiva’s Portfolio Yield Statistic?

  1. Pingback: Obligatory Viewing » Confessions of a Microfinance Heretic | Blog

  2. Pingback: End of Year Microfinance Sector Wrap-up » Confessions of a Microfinance Heretic | Blog

  3. Pingback: Kivanomics 101, or: DKYC* |

  4. Pingback: How Can I Invest in Microfinance? Part 2: My Zidisha Experience » Confessions of a Microfinance Heretic | Blog

  5. Pingback: Kiva, Microfinance and World Poverty

Leave a Reply

Your email address will not be published. Required fields are marked *

Please type the characters of this captcha image in the input box

Please type the characters of this captcha image in the input box

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>