You know the game of musical chairs: players sit on chairs arranged in a circle. The music starts and the players start circling – dancing, running – while chairs are progressively removed. Then the music stops and chaos erupts as the players seek to find a place to sit.
In Mexico, the number of chairs remaining is few indeed, even as the MFIs continue to dance. The recently published study by the Microfinance CEO Working Group shows just how few chairs are left. The published numbers seem bad enough: some 21% of the microfinance loan applicants to FINCA Mexico already hold four or more loans. Adding those holding three loans, we get 33% of applicants.
Ok, so a third of borrowers are in deep debt, but it’s probably not every one of them who is in over their heads, though clearly many are. Seems like a serious situation, but still – can an industry deal with a fifth or even a quarter of its clients who are unable to repay? Seems challenging, but in principle, yes. It’s been done before.
Understanding multiple borrowing numbers
I’d love to tell that story about Mexico, but therein lies delusion. I’ve been working on questions of market capacity and credit bubbles for long enough to understand that the numbers aren’t what they seem. First, the survey involved not borrowers, but applicants. By definition, there are no borrowers holding zero loans. However, from the 76% of applicants who already have at least one loan, we can extrapolate the actual levels of multiple borrowing. Naturally, that requires an assumption that these 76% of applicants are representative of the broader population of borrowers. Is that a reasonable assumption?
To a degree, yes. First there's the question of whether applicants to FINCA are not too narrow a group. I don't believe so, and the survey covers a wide geographic spectrum. The bigger problem is that loan applicants are a self-selecting group. A client who has one loan but feels that two loans would be too many, would presumably not be applying for a second. The same argument would then be applied across the rest of the distribution, to clients holding two, three, four loans, and so on. The result is that the number of multiple borrowers would be over-represented in the sample, though we don’t know to what degree. However, one thing is clear: a large number of multiple loans seems perfectly acceptable to both microfinance borrowers and lenders. So, if the distribution extrapolated from the survey is oversampling existing multiple borrowers, one can reasonably guess that it’s a distribution we’re likely to see in a not-too-distant future. For that matter, that future may well be already here, considering that the survey used credit bureau data from nearly 1.5 years ago (Nov 2012 – Feb 2013). Using the 76% of FINCA loan applicants as a view on the broader market of MFI clients is thus quite a reasonable assumption.
And that has very serious implications. First, it’s a simple extrapolation from applicants to borrowers: the distribution now is considerably more concerning – no longer do we have 21% of applicants holding 4+ loans, but now it’s 28% of clients. Still, we’re not yet done.
Consider an obvious fact: while a borrower holding one loan has a relationship with one MFI, someone holding five loans has five MFI relationships. From a sector-wide perspective, MFIs in Mexico have five times more exposure to the second borrower than to the first. In the event of a mass default – an event we’ve seen in Andhra Pradesh, Nicaragua, and elsewhere – the impact from the second borrower would thus be felt by all five MFIs.
Recalculating the figures through this lens of industry exposure, we get numbers that are absolutely stunning. More than half (54%) of the Mexican MFI exposure is to borrowers holding four loans or more. Adding those holding three loans, we get an exposure of 72%. In my prior blog, I showed how the level of multiple borrowing in Mexico far exceeds even that of Bosnia and Andhra Pradesh. But when it comes to exposure levels, Mexico is in a different world altogether.
From multiple borrowing to overindebtedness
So, how many of those clients holding 4+ loans are overindebted? We don’t know. The study did conduct interviews with some clients, and many were clearly struggling with debt, but we don’t know the actual figures. However, one thing seems inescapable – should a repayment strike take place, these borrowers would have the most to gain by joining in.
So why hasn’t there been a repayment strike yet? The answer is simple: the music is still playing. So long as borrowers have access to more credit, the game goes on. And so we get ridiculous metrics, such as 11% of clients holding six or more loans. That figure is so high that it wasn’t even measured in any of the other surveys, whether in Bosnia or anywhere else, where published figures max out at 5+ loans. And yet in Mexico, these clients already account for 27% of the sector’s exposure.
The survey found that even clients holding five loans may be doing well and finding the loans useful. But it also found that among those holding the largest number of loans, some were dealing with downward debt spirals, seemingly unable to cope, other than through additional borrowing. We don’t know the actual numbers. But frankly, I’m skeptical that large numbers of clients can hold so many multiple loans and do so without falling into a debt spiral. I am eager to be proven wrong on this point.
Debt spirals cannot go on indefinitely, and when something cannot continue, it will stop. If the number of clients in this situation is small enough, it is manageable. But if a large share of clients are in fact in a debt spiral, the problems will be more difficult. The music in Mexico will eventually stop. What will be its impact?
From bubble to crisis
Consider for a moment the subprime loans in the US in the lead-up to the financial crisis. It seems everyone knows that loans during that time were pretty crazy – “exploding” loans, “liar” loans, NINJA loans (no income, no job, no assets). But a few facts are in order: not ALL subprime loans were crazy or were held by borrowers who were overindebted. Like the borrower “Carlotta” from the Working Group study, who happily acknowledged having five concurrent loans and described how these were helping her manage her cashflows, so too were there plenty of positive stories of subprime borrowers in the US, many of them immigrants or the poor spurned by banks. For them subprime loans were the only opportunity to buy a home. At the time, many saw these loans as being akin to the inclusive finance we speak of today.
Moreover, despite the widespread attention, subprime loan originations peaked at 23.5% of the mortgage market – not so different from the number of borrowers holding 4+ loans in Mexico. And yet, when the music stopped, defaults skyrocketed across all loan segments, not just in subprime. In the case of the US, the initial wave of defaults was driven by the dual problem of falling home prices and the inability to refinance. Without these two critical outlets, struggling borrowers were forced to default. And as home prices continued to fall, defaults kept rising, eventually infecting the broader economy. The “soft landing” that Alan Greenspan sought to engineer proved elusive.
In Mexico, there are no home prices tied to microfinance to serve as a channel to infect other borrowers. But there are two other channels: the MFIs and the borrower groups. Once defaults begin to spike, the MFIs will inevitably begin tightening their lending standards – the first (and appropriate) response of institutions facing rising credit losses. Meanwhile, cut off from further loans, struggling borrowers will have no choice but to default. For some time, they may triage their repayments, but once they recognize that further loans won’t be forthcoming, they are likely to default to most, if not all, of their MFIs. After all, how many MFIs would be willing to renew a loan to a client known to be already in default to several others?
The impact of the defaults from these borrowers would then become magnified by the widespread presence of groups, which can handle some level of default, but which fall apart once their tolerance threshold is breached. Instead of helping support repayment solidarity, the groups then become channels for spreading default – group members are themselves less likely to repay, if their own loan renewals are conditioned on supporting an excessive number of defaulters. This effect has been demonstrated in several places, including the 2008 crisis in Pakistan and the 2009 crisis in Kolar, India.
So what can be the expected impact of a crisis? Given the levels of multiple borrowing, I believe sector-wide write-offs at the level of 25% are quite likely, and for many institutions, the figures may be higher still. According to calculations from the MIX Market, MFIs in Bosnia and Nicaragua wrote off about 12% of their 2008 portfolios during the subsequent three years (2009-11). Those are industry-wide figures. A good number of institutions wrote off north of 25% of their 2008 portfolios, while a few (like Banex in Nicaragua) collapsed outright, without leaving a data trail on MIX.
Can microfinance survive a Mexican crisis?
Write-offs at this magnitude are disastrous, but yet still survivable. Some MFIs would necessarily collapse, but the sector itself is more resilient. The bigger problems would come not from direct financial impact, but from the indirect effects on the sector's reputation. The trouble with Mexico is that it has a particularly combustible mix of factors that would greatly amplify the crisis, causing exceptional damage to the microfinance sector on the global stage. Hard as it may be to accept, the impact from a Mexican microfinance crisis would be the worst on record, easily exceeding Andhra Pradesh.
As I wrote a year ago, there are three factors whose combined impact could devastate the sector’s reputation:
First, there are the extraordinarily high interest rates, exceeding 80% for many prominent MFIs. Such figures are never easy to explain to a lay public anywhere. They would necessarily color the public perception of an unfolding repayment crisis in Mexico.
Second, there is the extraordinary profitability of the sector. Its most well-known MFI, Compartamos, has been among the three most profitable major MFIs in the world for each of the past ten years. In 2013 alone, it paid its shareholders €154 million in dividends. There’s no point comparing that to other MFIs. For a more suitable reference, consider Credit Agricole, Europe’s largest bank, which paid €302 million in dividends that same year. That’s right, Compartamos’ dividends are over half of Credit Agricole’s.
And third, Mexico has very strong cultural ties to the US. Mexican-Americans make up over half of the US Hispanic population – a group that counts 39 legislators in the US Congress, in both Democratic and Republican parties. And unlike the Indian immigrant community, Mexican-Americans tend to be from lower income classes, whose relatives in Mexico belong to the very social groups that are served by the MFIs.
Here’s how these three factors are likely to interact when the bubble in Mexico pops. First, there would be heart-wrenching stories of struggling borrowers. These are inevitable – one can find them in any country that has experienced a consumer debt crisis. With the strong cultural ties to the US, it won’t take long for such stories to appear in major US media. As was the case in India, those stories will be juxtaposed with the very high interest rates and immense profits. The combination has all the makings of a scandal. Consider the imagery of poor Mexican families, driven to destitution and desperation (even suicide?) by MFIs charging 80% interest, while paying their shareholders bigger dividends than most US banks.
Most importantly, these won’t be random families, but relatives of Americans living all across the United States. How long before this scandal leads to Congressional hearings, hauling in officials from OPIC (the US public investment arm)? How long before politicians in US and Mexico begin denouncing the microfinance activities of public institutions, including the World Bank and the Inter-American Development Bank (both lenders to Compartamos)?
In that context, defending Mexican MFIs will be a largely impossible task. After all, unlike in 2009-10, the sector can no longer claim innocence through ignorance – it’s been nearly four years since the Andhra Pradesh crisis, and nearly six since Bosnia and Nicaragua. The widely endorsed Client Protection Principles, which include prevention of overindebtedness, were promulgated back in 2009. Indeed, it’s easy to imagine scenarios in which the scandal ends up with drastic funding cuts to microfinance by the US and international DFIs – cuts that may not even discriminate between Mexico and other countries. Will that be the time that European funders or private social investors step in? Very unlikely.
Alarming or alarmist?
The period of microfinance crises during 2009-10 tested the microfinance sector, and it proved resilient. Mexico is the one country – the only country – that credibly threatens to bring the entire edifice down.
Does this seem alarming? Of course. But it’s alarming for a reason. I am not crying wolf for the sake of attention. Over the past five years, there have been only two countries that I have warned as potential areas of severe crisis: India (in 2009 and early 2010) and Mexico (since 2013). It’s far too easy to discount these warnings as the work of a shrill, over-the-top blogger. But let me ask this: what is the scenario by which all this is avoided? Is there a “soft-landing” in sight? And what exactly is being done to engineer that landing?
There’s nothing I would welcome more than being proven wrong on Mexico.