Syndicate content

#6 from 2017: What is a systems approach, anyway?

David Evans's picture

Our Top Ten blog posts by readership in 2017. This post was originally posted on February 27, 2017.

“It makes me a little crazy when you keep saying systems.” – Jowhor Ile, in And After Many Days

At home, we have a porchlight at the entrance to our house. If I flip the switch for that light, there is about a 50-50 chance it will turn on. The reason? There is another switch in the basement that controls the electricity flow to the porch, and the porchlight will only come on if both switches are on.

This – slightly adapted – analogy came from Justin Sandefur at the Center for Global Development, in an effort to explain what a systems approach is and how it can improve development programming.

If you’re like us, there is so much talk about systems that it can be easy to get lost. At a recent event, we asked a mixed group of operational teams and researchers, “How confident are you that you know what a systems approach is?” Nearly 40 percent had little to no idea.

How confident are you that you know what a systems approach is?

To take education as an example, a systems approach to education recognizes the following:

1. An education system is made up of different actors (students, teachers, administrators, political leaders), accountability relationships (management, politics), and design elements (financing, information) (see Pritchett or Scur).

2. Changes to one part of the system are moderated by other parts of the system. For example, the effectiveness of investments to get children to school will be limited (or enhanced) by the quality of the schooling.

3. A change to one part of the system leads to changes in other parts of the system: increased public provision of school supplies won’t increase learning if parents subsequently reduce their pre-existing investments in school supplies, as indicated by what happened in India and Zambia (Das et al.).

A systems approach seeks to explicitly take these separate components and their interlinking movements into account.

Three models demonstrate how a systems approach can apply at each point in the reform process: One identifies the current performance of each element of the system, one answers questions of what happens as elements of that system change, and one seeks to leverage this information to improve reforms.

First, the Systems Approach for Better Education Results (SABER) – represented by Luis Benveniste at our recent event – seeks to measure policy intent and policy implementation in a range of countries. If an education system is made up of a bunch of different components, then understanding where a system is on each component is key to taking effective action. SABER helps countries gather systematic data on their policies around 13 areas, grouped into inputs (finance, teachers, school health, and school feeding), governance (school autonomy and accountability, engagement of the private sector), information (student assessment, management and information systems), and complementary inputs and cross-cutting themes (information and communication technologies, equity and inclusion, and resilience), and levels of education (early childhood development, tertiary education, and workforce development).

It then gathers data on actual practice via the Service Delivery Indicators (overview and reports; data), as in the figure below.

SABER: Implementation

Source: Benveniste presentation

Second, the Research on Improving Systems of Education (RISE) program – represented by Justin Sandefur– takes a systems approach to education research. What does that look like? Policy evaluation with “big questions / big programs, evaluated at scale,” plus a “focus on politics and implementation,” which is “embedded in a unified conceptual framework that can explain failures” (a.k.a. Policy Analysis Plus). That “conceptual framework” is the recognition of how actors influence each other and how the structure of the system influences them.

For example, a recent evaluation in Indonesia found no impact of doubling teacher salaries on teacher effort or student learning. The RISE approach examines this failure within the context of all the other benefits to and requirements from teachers to understand not only the failure, but also how to effectively map the path to success.
The result                                           Embedding that result in an education systems framework

Third and finally, the Global Partnership for Education – represented by Karen Mundy – discussed the development of education sector plans with strong country ownership and financing. Ideally, such a plan draws on benchmarking from SABER (and similar efforts) and Policy Analysis Plus from RISE (and similar efforts) to detail a set of actions that explicitly recognizes the interactions and expected reactions from different actors in the system to any policy change. A systems approach to such a plan would also build in learning and opportunities to shift course, as unexpected results lead education leaders to update their understanding.

Albert Einstein is said to have coined the phrase, “Everything should be as simple as it can be, but not simpler.” Surely, to the frustration of some readers, a systems approach is not a simple thing. It is an approach that explicitly recognizes and seeks to grapple with the interacting players and structures of the education system. But without it, we run the risk of never figuring out how to turn the porchlight on.

Follow PublicSphereWB on Twitter!


Submitted by Duncan on

Two other ideas that can help:
1. Avoid confusing The System (the multiple interacting entities that jointly provide education, healthcare etc) with a systems approach
2. Some systems are complicated; others are complex - there's a big difference both in how they work, and how you try and influence them. Complicated systems can be disaggregated, understood and influencing strategies arrived at. Complex systems cannot be disaggregated (whole not just greater but fundamentally different to sum of the parts), so more open ended, experimental, iterative and humble interventions are required. Not clear to me which categories the systems you describe here seem to fall into.

Add new comment