Should Industry Concentration Cause Consternation?

On my post about Beth Rhyne, Scott Gaul of the MIX commented:

What are the causal impacts that we could expect institution-building to have? Healthier or deeper financial sectors, higher savings rates, more competition, etc.?

I replied:

I think that’s a good question. Beth discusses metrics at the level of the institution---is it self-financing? is it growing?---but not at the level of the country. One metric might be a measure of industry concentration, such as a Herfindahl index, where lower would be better within certain bounds, since it would indicate a more diverse, competitive sector.

Scott again:

A nice attribute of looking at country-level impact is that there is already literature associating country-level financial sector development and economic growth. Does microfinance have any causal impact on financial sector development, through institution-building?

I liked his suggestion that I should look beyond whether individual microfinance institutions (MFIs) are succeeding to whether the financial system is improving. Among his suggestions, I honed in on measuring the concentration of a country's microfinance industry---whether market share is monopolized by a single giant or divided among many players. That measure's link to the growth of individual MFIs seems loose enough to avoid near-tautologies like "growth of MFIs increases use of microfinance" but tight enough to avoid grand questions about what causal links you can prove with statistics. Industry concentration interests me because declines in it may indicate rising competition, which seems to lead to lower interest rates and more diverse and flexible products.(On the causality issue: a statistical literature seems to show that countries with bigger financial systems grow faster. Following Robert King and Ross Levine's paper, Finance and Growth: Schumpeter Might be Right, financial sector development is most often measured as total outstanding loans to private non-financial businesses as a share of gross domestic product. Microcredit increases that---if you count microenterprises as business, which the statisticians probably don't and growth economists might prefer not to since the link microenterprise is arguably more about subsistence than growth. At any rate, the leading study showing that private credit causes economic development turns out to have problems in ruling out reverse causality. I think the financial system is aptly compared to the human heart. A financial crisis is an economic heart attack. A healthy child's heart grows as she grows. It does not follow that surgically enlarging her heart would help her.)After the public dialogue with Scott, I worked with him behind the blog scenes to assemble data from the MIX in order to measure and graph microfinance industry concentration (raw data here).The standard measure of industry concentration is the Herfindahl-Hirschman Index (HHI). You compute it by squaring each company's market share, then adding those numbers up. Values range between 0 and 1. Consider some extreme examples. If there is one company, a monopolist, its market share is 1.0. The sum of squared market shares is then also 1.0: maximum concentration. If another company muscles into a 1/3 market share, the HHI drops to (²/₃)² + (¹/₃)² = ⁵/₉. If there are 100 companies with 0.01 share each, the HHI works out to 0.01---darn near the lowest possible value of 0, indicating minimal concentration. The Herfindahl-Hirschman Index has a long name, but it is mathematically elemental. In fact, technically, the HHI should be called the Herfindahl-Hirschman-Roodman Index, thanks to one of those great coincidences in the history of science. As a Harvard student in the late 1980s I became piqued that my proud college boasted an average class size of 14 while herding students into giant courses to meet distributional requirements. Sure there were lots of small classes, but hardly anyone took them! I figured out that average experienced class size was the sum of squared class sizes divided by the sum of class sizes, which is a small step from the HHI. Beat studying. Oddly, no one paid any attention to me. At any rate, I think it is appropriate that economists got naming rights to the HHI, since the largest course on campus was the equivalent of Economics 101. It typically had a thousand students. To quote Ringo Starr, "Got to pay your dues if you want to sing the blues."I digress.This graph shows microcredit industry concentration over time in select countries with relatively robust industries. "Market share" here is percentage of all microcredit borrowers in a country. Caveats follow:The biggest problem we faced in making this graph was missing data. Over time, more MFIs have reported data to the MIX. If an MFI's data is missing one year, and that missing entry is treated like 0, and then the MFI shows up with 100,000 clients the next, that would make the index of industry concentration plummet. That is why different countries have different date ranges in the graph. We also had to fill in some key missing data points as best we could. Scott, for instance, found and added numbers for Jamii Bora in Kenya.Another problem is defining the universe, the "microfinance industry." There are no sharp lines in the data between microfinance institutions and credit unions, between those who serve the less-poor and the more-poor. All reporting institutions are counted. The effective scope of "microfinance" may vary. Still another problem is that microfinance markets are local, not national. A country might have three big MFIs, each with a monopoly in a different region. That would show up in the data as three MFIs splitting the national market three ways. You could imagine that being a particular problem in India, though in fact the big MFIs are clustering in the same states.What do the data show? In most of the countries, the industry is becoming more diverse. Exceptions are Bangladesh, where the industry is maturest at the time the data start and the HHI is flat; Kenya, where Equity Bank has vaulted upward in recent years, perhaps because of a loan product catering to salaried workers; Bosnia and Herzegovina, where some big players grew fast and are now in trouble; and India, where there is much talk and some fear of bubbles. In fact, the MIX documented (page 5) increasing concentration throughout Eastern Europe, where microfinance is now seen as most overheated.Unexpectedly, there is an intriguing pattern: increasing concentration tends to be a sign of trouble. The flipside is that diversification may signal healthy development in the financial system. Not that I'd bank on that relationship.