Financial Access Studies Clash over Whether Glass is Half Full or Half Empty

November 6, 2009

The biggest controversy to hit microfinance since the Compartamos IPO erupted this week as easygoing, bookish economist types at two East Coast microfinance research institutions dueled over the longstanding empirical question of whether the glass is half full or half empty. CGAP researchers, citing industry-consensus best-practice guidelines, accentuated the positive in observing that about 2.5 billion adults worldwide have accounts with formal banking institutions (about 2.5 billion do not). Academic researchers at the NYU-based Financial Access Initiative adopted a more skeptical stance, pointing out that Half the World is Unbanked.

OK, so I won’t quit my day job for nightclub comedy.

Two new studies are indeed out tallying how many people have credit or savings accounts with commercial banks, savings banks, post office savings banks, cooperatives, government development banks, and microfinance institutions. The FAI report is short and sweet and draws on data from Patrick Honahan, who is now governor of Ireland’s central bank. In contrast, the CGAP report harvests new information collected by surveying financial regulators in 139 countries.

Measuring access to formal financial services is hard. Usually what is measured is the number of savings and loan accounts outstanding among certain groups of institutions. But consider these distinctions: between having access and actually using it (which is what shows up in the data); between accounts and account holders (100 million accounts does not mean 100 million people since many people have more than one account); and between households and individuals (if I have a savings account, does that count as access for my wife?). And data are incomplete and infrequently updated.

In 2004, CGAP researchers Robert Peck Christen, Richard Rosenberg, and Veena Jayadeva did path-breaking work in this area, counting some 750 million savings and loan accounts at “alternative financial institutions,” ones meant to reach poorer people, in developing countries circa 2000. (Read page 1 of their report for lots of caveats.) Earlier this year, I contacted Rich in search of guidance in updating this data. His advice was: Don’t do it unless you are prepared for a LOT of work collecting and culling information from hundreds of institutions around the world. I didn’t do it.

But to return to the time line: In 2006, the World Savings Bank Institute had Stephen Peachey add savings bank accounts to this data set (CGAP had counted post office savings accounts only), bringing the total to 1.4 billion. Then Patrick Honohan worked through all this data. He corrected some apparent errors. And where he could, he calibrated with results from a different kind of data source, household surveys that more directly measure how many families are connected to the formal financial system. Honohan estimated, for a large number of countries, the share of adults so connected. (Gated journal version.)

The new FAI report crunches Honohan’s country-by-country percentages into a grand, global total. That means it is actually based in substantial part on a) CGAP data that is b) about 10 years old.

In contrast, CGAP’s new report is based on brand new data. On the one hand, I think it may be less complete, poorly covering, e.g., postal savings banks, which put up big numbers in the old CGAP data. On the other hand, questions have been raised about those postal numbers: many of the accounts appear to be tiny and/or dormant. In the event, the new CGAP study arrives at a similar grand total: 2.7 billion instead of 2.5 billion people are unbanked. That difference is well within the large margins for error in these data.

If you think about this for about ten seconds, you realize that it’s not so surprising that a lot of people have bank accounts and a lot don’t. The real interest comes from subdividing the data. The old CGAP study, for instance, cuts the data by region, institution type, and credit vs. savings. In first reading it, I was struck by the modesty of microfinance’s role in the bigger picture. The new studies look across countries, showing that, yes, richer countries have more access but there is diversity at any given national income, pointing to scope to improve access without just waiting for economic growth.

Perhaps the authors can comment on the pros, cons, and meanings of their tabulations.

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

1. Jake Kendall Says:
   November 6th, 2009 at 11:45 am

   Hi David,

   Thanks for posting on our report (and on this topic). You have broken down the issues nicely.

   Your final point is a very important one to me and one that I reflect on (indirectly) in a recent blog post on cgap:
   http://microfinance.cgap.org/2009/11/04/the-global-access-gap/

   Microfinance is a drop in the bucket, both in terms of how many are served (100-200m by most estimates) vs. how many are not (2.5bn+ by my estimate and by FAI’s) and in terms of how many are served by the regular banking sector plus other types of institutions like credit unions (approx 2.5bn world wide… e.g. there are over 800m bank accounts just in Japan!).

   To me this speaks to the importance of branchless banking and other innovations that have the power to bring the poor into the mainstream banking sector. At least for deposit services, this seems like the only way we will reach them in large enough numbers to fill the gap.

   Any thoughts?

2. Jonathan Morduch Says:
   November 6th, 2009 at 1:08 pm

   David,

   Glad to see the post. We’ve been pushing for better numbers on financial inclusion for the past few years, and it’s exciting to see that serious data collection efforts are likely going to take off soon, with World Bank and IMF leadership.

   In the mean time, we wanted to squeeze more out of the available evidence, with a strong bias toward wanting household-level data. It was far more time-intensive (and concept-intensive) than we had anticipated. After all that work, we’re glad that our household-based tabulations turn out to be in the same ballpark as CGAP’s account-counting exercise.

   Honohan’s estimates cover 160 countries based on population data around 2002–2003. So they’re more recent than you wrote (6–7 years old, not 10 years old), but, you’re right, there’s an inevitable lag. Fortunately, the lag doesn’t seem to make a big difference empirically. We scoured news reports and more recent studies (like the Finscope studies) and updated the data for 12 large countries that account for about 2 billion people. The aggregate numbers don’t change much: with the most recent (2007–9) financial measures, we estimate 2.4 billion adults who do not use formal financial services (relative to our original estimate of 2.5 billion). Given the similarity, we stayed with Honohan’s numbers for consistency.

   The really interesting finding for us was the diversity of experience—with some countries having much more financial inclusion (and some much less) than you’d predict based on income or urbanization. The glass is more than half full once you start looking closely at positive outliers.

3. David Roodman Says:
   November 7th, 2009 at 6:58 pm

   Jake, I think scatter plots like the one I linked to raise several questions. For the countries that have achieved high access for their GDP/capita, what were the key government steps behind these successes? What is the quality of these seemingly high quantities of financial access? How are new technologies changing what is possible? In India, I think a lot of the recorded access came through branch, not branchless, banking. Certainly branchless banking has been key in some places, such as Brazil. Probably the full global story is richer, though. A recent CGD task force, on which Jonathan and I both served, looked at principles for policy to support access.

4. Richard Webb Says:
   November 7th, 2009 at 11:35 pm

   David. Your chapters and blogs are enormously illuminating. On access, it seems to me that its measurement and discussion should include terms. When loans cost 50% p.a., the entire population might actually obtain loans, but nobody can afford to borrow very much for very long. There is an unmet demand, or “lack of access,” for the extra debt that would be incurred at lower rates. Likewise for the longer term debt required for many investments or personal needs. Yes/no statistics on borrowing are of course a relevant first approximation, but miss a very large part of the problem, especially as the number of borrowers grows.

We value frank and constructive exchanges and encourage you to use your real name in your comments.