What could you learn by comparing changes in poverty among client sub-groups? >

Lindsey Longendyke
• Posted in data analysis

Perhaps one of the most exciting and appealing uses of the Progress out of Poverty Index is what we call a “poverty movement analysis.” When an organization collects PPI data from its clients over the course of multiple years, it can compare poverty rates between two points in time to determine if, on average, clients’ poverty levels are improving. For any organization with a mission to reduce poverty, this analysis is critically important.

This analysis does not serve the purpose of a full-fledged impact evaluation, including methodologies such as random control trials (RCTs), but can provide extremely useful insights, such as those Grameen Foundation explored in a recent publication, Drawing Insights on Poverty Movement with Multi-Year PPI Data: A Case Study of Two Filipino MFIs.

The reason RCTs are so insightful is that they use comparison groups to contextualize what we observe about the sample. For our analysis, we had plenty of PPI poverty data on microfinance clients, but no data on non-clients, so we could not use non-clients as a comparison group.

However, poverty movement tracking allows us to explore a different approach to analysis, which also provided valuable insights.  We created comparison groups by segmenting the MFIs’ clients into sub-samples:

  • Length of client membership (new vs. returning and long-term clients)
  • Frequency of taking out loans
  • Activity level of clients (active borrowers vs. clients that eventually became inactive or resigned)

Why create these subgroups? This analysis illuminates the relationship between poverty and the factors above, and in the process can support or contradict assumptions held by management.

For example, management may assume that the longer a client is a member of the program, the greater their progress out of poverty. Alternatively, management may assume that a program’s benefits eventually reach a point of diminishing return and the client’s poverty status plateaus. The analyses did not support the assumptions. While both long-term and short-term clients typically experienced lower levels of poverty over time, their change in poverty did not correlate with their length of membership.

These trends can be used as a starting point for an organization to better understand the profiles of different client segments, begin to hypothesize why certain client groups may start off poorer or be more or less likely to become less poor over time, and set the direction of further research to understand factors and drivers of poverty movement.

It is also useful to compare poverty data from MFI clients against country, province, or municipality poverty data, if such data is available. The report explains, “If, for example, data exists for changes in poverty rates for the municipalities in which a MFI works, this could be compared to the changes in poverty rates for the MFI clients living in that municipality.” Understanding how client poverty is changing relative to overall poverty in the region can help MFI management conduct a better social performance evaluation.

This report is based on data from microfinance institutions, so clients were grouped according to their behavior as microfinance clients. Today, more than 200 organizations around the world are confirmed users of the PPI to measure poverty.  (We believe a much larger number use it based on the rate of downloading of the tools, but we only count those where we have verified use.)  PPI users include social enterprises, healthcare providers, agricultural entities, and even major corporations. These organizations would group their clients differently depending on their mission, services, and business model.

How would you group your beneficiaries? Why would comparing changes in poverty over time among these sub-groups help you improve your organization’s strategy?