Diamonds no longer forever

Part of the mission of MIX is to promote transparency for microfinance. At a high level, transparency has increased over the past several years - we've gone from 28 institutions providing confidential information to the first MicroBanking Bulletin in 1997, to 162 with public information on MIX Market in 2003, to over 2000 with profiles and data now on the MIX Market site.

Figure 1: Global transparency over time

But do we have more and better data on the issues that matter? Is the quality of microfinance data improving? This week we are making one change to the MIX Market site to shed some light on these questions.

MIX Market uses a 'diamonds' system to score the data and third-party documentation available for each MFI. Until now, the diamonds reflected any data that the MFI had ever provided. Consequently, the diamonds did not give a good indication of the current level of transparency for the MFI or the evolution over time.

Starting this week, diamonds are no longer forever. Each profile on the site now features annual diamonds and the diamonds for each MFI profile reflect just the most recent year of data. The updated rules for diamonds are in our FAQs. Within each profile, you can see the annual diamonds for each individual year. We can also learn about how transparency for the sector as a whole if we aggregate the annual diamonds data. [Fn1]

The coverage and quality of data have increased over time. Figure 2 shows that in 2010, we have data for roughly 1200 institutions. Of these, over 750, or about two-thirds, provided some form of third-party validation, either audited financial statements or an external ratings report. In 2002, the year MIX was founded, fewer than 150 institutions provided audits or ratings, representing just over 40 percent of the total coverage. By 2005, this had climbed to 54 percent of the data, covering 650 institutions. (The increasing number of 1 diamond institutions reflects those that reported at some point in the past, but no longer do.)

Figure 2: Distribution of diamonds over time

Figure 3 shows average diamonds by region. There is general improvement, with LAC typically receiving the highest scores, but with recent declines in Africa and MENA. The graph shows only scores for MFIs that have reported since 2000, so the decline is not purely due to a tradeoff between breadth and depth as the dataset has expanded. (To see the same trends for all institutions, see here.)

Figure 3: Average diamonds by region (10-year panel data set)

While the level of third-party validation has increased overall, some industry efforts supported more disclosure in the past. The Financial Transparency Awards encouraged institutions to publish audits on MIX Market, while the Rating Fund, and its successors, have subsidized ratings and, in some cases, required public disclosure of ratings. Figure 4 shows that external ratings peaked around 2005 in most regions, corresponding with the height of the Rating Fund.[Fn2] The percent of institutions disclosing rating results has dropped by half, from 16 to 8 percent globally, and close to zero in some regions.

Figure 4: Percent of 5-diamond institutions by region and year

What else do we expect to see correlated with transparency? Figure 5 lets you explore the relationships between the annual diamonds data and other variables from MIX Market and investment data from our recent 'Tipping Point' report. Has increased transparency gone hand-in-hand with increased foreign investment? Does increased transparency lower the cost of funds? Are institutions with higher risk less likely to be transparent?

Figure 5: Diamond scores and potential drivers of transparency

This change to MIX Market marks the first step in improving our 'data on data.' Annual diamonds give us a window into some aspects of transparency, but they don't include important disclosures about social performance, funding investments, products and the range of other data that has become available in recent years. We are now beginning to look at what else can be done to reflect the volume and quality of information available on different providers and welcome your suggestions and ideas in the comments below.

Fn1: The data set covers 35,671 individual MFI years. All of the data is available for download here. Since an institutions receives a 1-diamond score by default once the profile is visible, we excluded 1-diamond scores for years before the institution provided other data, which leaves 13,983 records.

Fn2: This likely underreports the volume of ratings, as not all MFIs that are rated disclose their ratings.