Information Overload: can technology address MFIs' reporting burden?

Microfinance reporting has something to learn from how university applications and tax returns are filed.  Every year in the United States, millions of high school students complete lengthy university admissions applications hoping to secure one or more spots at their top choices, all of which have their own admissions applications.  Thanks to the Common Application, students can provide most of the necessary information through a single application, only answering questions from individual universities if they fall outside that application.  Similarly, every year, millions of taxpayers must file tax declarations to multiple levels of government to determine their tax obligations for each one.  Online tax preparation software (such as TurboTax or H&R Block) determines what authorities the filer must report to, and then gathers the information to meet those requirements.  The filer only enters each piece of information once.  In each case, technology offers solutions that reduce the burden on individuals; microfinance should take a cue from these solutions.

Like university admissions officers reviewing the latest pool of high school candidates, microfinance analysts, investors and regulators all clamor for more information to help them make sense of increasingly complex microfinance sectors.  Recent evolutions in the industry have reaffirmed a continued need for more timely, detailed information: whether to understand how MFIs adapt their product offerings in response to market conditions, or to analyze market concentration, particularly in the fallout from the Indian AP crisis.  Unlike university students – or the typical American tax payer – MFIs must respond to each of these report requests individually, and the number of reporting stakeholders has only grown over time.  MFIs maintain reporting to increase public exposure through public information platforms, to comply with the needs of lenders – numbered at nine for the typical MFI in 2010 – or for regulators, networks and others.  A recent MIX survey on African MFI reporting helps quantify this ‘reporting burden’: surveyed MFIs prepared nearly 40 external reports each year on average and dedicated 60 or more days of staff time to just filling in forms for reporting.  Yet, as the Common Application and TurboTax examples demonstrate, technology can help bring efficiency to this reporting marketplace.

Given the importance of data for market growth and stability, MIX embarked on a ‘reporting genome’ project last year with the goal of understanding the dynamics of MFI reporting:

• What information is asked of MFIs?
• To what extent do those requests overlap?
• Can technology provide a solution to the ‘reporting burden’?

MIX started its analysis with a group of information stakeholders with broad footprints and with the greatest incentives to adopt international industry reporting standards: cross-border funds, international DFIs and MFI networks.  Each reporting format was mapped against MIX’s XBRL taxonomy, a common data framework based on IFRS and widely-accepted reporting standards in the microfinance sector.  To date, eighteen reporting formats had been reviewed and mapped.  The analysis that follows points to a few important conclusions:

1. Reporting formats for MFIs are still quite varied.  This proliferation of data labels and formats increases the time that MFIs must dedicate to reporting, and increases the risk that the same information will be reported differently in different formats.
2. There is significant overlap in the actual data requested of MFIs.  The ‘core’ data set covers more than one hundred variables and provides a focus for creating greater efficiency in MFI reporting.
3. Technology can help reduce the reporting burden.  Technology can intermediate the multiple demands of various stakeholders into a single, common request for MFIs.

MFIs still face a myriad of reporting formats…

As far back as 1995, the global microfinance community launched various initiatives to increase transparency through a common language for talking about MFI performance and tracking industry development.  While this first concentrated on financial and operational results, standards efforts have more recently included social performance metrics.  If successful, these initiatives should also reduce the reporting burden on MFIs as more actors use standard terms to ask for the same information.  

How successful have these reporting standards efforts been at creating that common language?  If the microfinance community were to judge success by the sheer number of MFIs with standardized data available on MIX Market, the efforts would appear quite successful: since 1995 there has been a ten-fold increase in standardized data available on MIX Market for basic data points. 

Unfortunately, these results mask important details.  Behind the scenes, MIX analysts – and their peers in similar positions elsewhere – have actually performed most of that standardization after the fact, taking disparate source documents reported according to varying standards to create a coherent fact set.   As a result, the cost of converting reports into meaningful data -- essentially a cost of doing business --remains unnecessarily high for the industry.  Although difficult to quantify, this 'reporting tax' serves as an important disincentive for smaller institutions to seek external capital for their growth.  From their vantage point, there are still too many reporting formats out there.

How would an MFI judge success on standards from its vantage point?  Most MFIs would probably count the number of times that they have to report the same data point to a different actor or report the same information in slightly different ways.  MIX’s ‘reporting genome’ provides a useful proxy for quantifying this duplication of effort by looking at the actual language of those reporting formats that are most likely to be sensitive to international industry standardization efforts.  If standardization is successful, everyone who wants to know the same information should ask the same question.

Across the 18 reporting formats reviewed, there remained a surprisingly large variation in the number and ways of asking the same questions.  First, as Figure 1 shows, the number of data points for any information category, such as portfolio or balance sheet, varies widely.  While numbers alone do not indicate different standards, they do mean that the MFI must situate its information differently in response to each format. Imagine a balance sheet that asks only one question 'Assets', versus one that asks for 20 pieces of information, including details of 'Trade and Other Receivables.' Then imagine another that asks the same 20 questions, but with a different breakout for 'Trade and Other Receivables.' The differences in granularity alone represent different standards for reporting for MFIs.

Figure 1: Reporting formats ask for a range of details by type of data

Even when reporting formats ask for the same information, they do it in different ways.  To take one example, the reporting formats reviewed variously referred to an MFI’s own borrowed funds as ‘Borrowings,’ ‘Loans,’ ‘Funding Liabilities,’ ‘Debt,’ ‘Loan Funds,’ or ‘Borrowed Funds.’  Breakouts compound this variation, as different formats asked to see borrowings classified as short- and long-term (or current and non-current), by local or foreign currency (and sometimes hedged or not), by type of counterparty (government, bank, non-bank financial institution) or by relationship to the MFI (related-party).  An MFI required to use all these formats may have as many as 70 different ways of reporting information about its borrowings alone.

…but these formats ask for a common data set…

In the current system , in which MFIs enter data into individual reports , the labels and presentation of each reporting format matter because they mean that MFIs must maintain their own methods for classifying information for each of the reporting formats used by their partners.  As the ‘borrowings’ example shows, however, different labels may not mean different data.  If all those labels were harmonized and mapped against one overarching format, how big would that common, core data set be and how much variation is there around that core?  Where is there convergence and where are data standards – and requests of MFIs – still evolving?

Analysis of the 18 reporting formats clearly outlines a common, core data set across various information types.  The bars in Figure 2 represent the total number of distinct data points requested across all formats for a given data set.  A distinct data point means, for example, that ‘Number of Borrowers’ is different than ‘Number of Female Borrowers’ and ‘Number of Male Borrowers.’ The light blue portion totals up all unique data points asked across the entire sample of reporting formats.   The dark blue portion of the bar shows the number of data points that matched across multiple formats after removing differences in labeling only.  This common data set is then compared against each individual reporting format. The red (median) dot represents the number of unique data points in addition to the core set of data points asked by the typical reporting format.

Figure 2: Reporting formats show strong overlap in common data

 The picture that emerges is one of surprising harmony.  Despite the large number of questions catalogued (over 700), a core of fewer than 150 data points meets 90 percent of the needs of the typical report format.  Take the balance sheet as an example.  For an MFI that reports to all the funders and networks surveyed for this project, just 46 data points cover all common information needs for the balance sheet requests of all the actors, and only four additional data points are required to complete the data requests unique to each report. Remove the variation in labels, and below the surface, standards begin to emerge.

Even where reporting requirements varied the most – portfolio information – a common core and a growing periphery exist.  As the heat map in Figure 3 demonstrates, gender and delinquency breakouts are common requests across all formats, and a substantial number also request data by product or location.  Further down the list, breakouts start to reflect the growing sophistication of MFI financing (hedges, on- and off-balance sheet financing), as well as the need to understand competition (client exclusivity) and concentration (geography). 

Figure 3: Common breakouts asked on portfolio information

...meaning that a solution that focuses on common data holds promise.

Given the rapidly evolving nature of the sector, the microfinance industry’s desire for more information will not cease, but the current system for gathering information represents a high cost to all participants.  MFIs must devote an increasing amount of time reporting on non-standards forms – a disproportionately high tax for smaller, start-up providers – while those requesting the information face greater delays and a higher likelihood that data may be misreported.  If the multitude of forms sent to MFIs presents an important obstacle, the common data requirements that emerge from them point towards potential solutions.

Once again, the Common Application and TurboTax examples offer insights into how the microfinance industry might successfully address the ‘reporting burden’ through a combination of technology and coordinated action.  The use of technology and data standards means that software like TurboTax can interpret data requests from various agencies to determine how to minimize the information request of tax filers.  Coordinated action among universities that participate in the Common Application ensures that no student is required to answer the same question more than once, while all admissions offices get the information that they need. 

As a first step, technology also holds promise for reducing the MFI reporting burden and enabling a more efficient information chain through common data standards, such as XBRL.    Data standards shift the burden away from the form and onto the data model.  Looking at the MFI reporting burden issue from a data-based approach, rather than a forms-based approach, has already shown the scope of common reporting standards in practice, even if they are masked by different forms and labels.  In the broader financial and regulatory world, many regulatory bodies have already adopted this standard for their reporting, including the RBI in India, regulator to NBFC-registered MFIs.   For several years now, MIX has also implemented XBRL for modeling and capturing microfinance data, allowing MIX to harmonize data reported according to hundreds of different reporting standards from more than 110 markets where MFIs report to MIX.

How might use of a standard language like XBRL help alleviate an MFI’s reporting burden?  Figure 4 shows that the MIX’s XBRL taxonomy already covers, or can be easily extended to cover, 90% of the data points requested across all reporting formats analyzed.  As a data model, rather than a series of forms, new data points are added to the model based on a common concept, not a common label.  In other words, whereas MFIs currently input their ‘borrowings’ under ‘borrowed funds’, ‘debt financing’ or several other labels in their current reporting, they need only connect that value once to ‘borrowings’ in the XBRL data model.

Figure 4: How does MIX's taxonomy cover reporting requirements from funders and networks?

Beyond technology and the promise that a common data standard brings, coordinated action by funders, networks and other market actors is required to turn the theoretical promise into reality.  Here, too, there are positive signs in the industry.  MIX’s ‘reporting genome’ project is one such example of cooperation that can lead to common action for the benefit of individual institutions as well as the sector as a whole.  We are grateful to the 18 funders and networks that made their reporting formats available to MIX as part of this analysis, expressly motivated by the desire to find a way to meet their information needs without placing additional demands on MFIs.  Through their participation, those actors have already taken a step forward towards supporting common action for the common good.  As MIX develops its information platform to facilitate how data travels between market actors, whether through public data standards such as XBRL, or reporting services like its new MIX Gold service, it will continue to engage information stakeholders with concrete ways to build the collective action necessary to tackle this crucial reporting challenge.