Integrating System Dynamics into Learning, Monitoring and Evaluation

The conventional way of designing Learning, Monitoring, and Evaluation (L, M&E) indicators, typically through log frames, assumes that change happens in a linear and predictable manner. Inputs lead to outputs, outputs lead to outcomes, and outcomes eventually lead to impact. While this structure offers clarity and ease of reporting, it often fails to reflect how change actually unfolds in complex real-world systems.

In reality, social and ecological systems are dynamic, interconnected, and constantly evolving. Change does not move in a straight line. Instead, it emerges through feedback loops, time delays, and interactions between multiple actors and forces. People’s beliefs, aspirations, and values also play a significant role in shaping system behavior. As a result, conventional indicators that focus only on isolated variables or end results tend to miss deeper system shifts. They need to capture endogenous feedback — the reinforcing or balancing loops within the system — and relative change, which tells us how one part of the system shifts in relation to another.

If we want to better understand how change happens, our approach to L, M&E needs to evolve. Indicators must move beyond measuring outputs and begin to capture relationships, feedback, and patterns over time. They should help us learn about how systems behave, not just whether predefined targets are achieved. Systems are often shaped by external triggers as well. These could be markets, policy or climate. How they respond to these external triggers also changes their inter-relationships.

System Dynamics Modelling (SDM) and Systems Thinking (ST) offer a useful lens for this shift. By focusing on system structure, feedback loops, and accumulations such as stocks and flows, SDM allows us to identify leverage points and patterns that may otherwise remain invisible. It also enables us to think about how interventions interact with existing system dynamics, often producing unintended or disproportionate effects.

This note presents two conceptual models to illustrate how indicators can be reimagined using a systems perspective. The first example focuses on endogenous feedback within a system, while the second highlights the importance of tracking relative change rather than absolute values. Together, they offer a starting point for rethinking how L, M&E systems can support deeper learning about systems change.

Concept Model 1: Understanding Endogenous Feedback

In this example (Figure 1) , we model how the population of girls in school is changing over time. It is driven by two main flows: the inflow of ‘New admissions’ and the stock feeds back into admissions through a reinforcing loop of social acceptance and decreases by the outflow of ‘Dropouts’. As more girls remain in school, their presence can influence community perceptions. Increased visibility of girls’ education can strengthen social acceptance and raise aspirations among families. This, in turn, can lead to more girls being enrolled. In other words, the stock of girls in school feeds back into the flow of new admissions, creating a reinforcing loop. This loop is endogenous, meaning it is generated from within the system itself rather than being imposed externally. Over time, such reinforcing dynamics can significantly amplify change.

Figure 1: Understanding the dynamics of the number of girls in a School

A standard intervention by an agency would aim to reduce the dropout rate in order to increase the number of girls in school. Such an intervention, if successful, would measure the impact on the extent of reduction in dropout rate, i.e. measuring the change in the outflow. This is the conventional way of monitoring impact, and that would typically get reported.

Graph 1: Simulation run showing impact of interventions.

However, in the above case, we are ignoring the role of the endogenous reinforcing loop, which might also impact the admissions as the number of girls in the school increases. Thus, it is imperative to look at the endogenous feedback loop, which may be incidentally strengthened through the intervention. The new indicator here would be to assess the change in the strength of the feedback loop of social acceptance. The impact measurement must therefore account for the system-level change that our intervention might be indirectly creating. Similarly, another intervention through social mobilization and motivating the parents to educate their girl child could lead to more girls joining the school (the flow of project-driven new admissions). The impact would be seen through the increase in the new flow of admissions of girls through the project intervention. But this again could strengthen the endogenous feedback loop of organic new admissions. Thus, it should also account for the change in the strength of the feedback loop of social acceptance.

The two interventions, reducing dropouts and adding new admissions through interventions, will create a disproportionate impact on the system. L, M&E systems need to be cognizant of these and, as much as possible, try to capture them.

As a process:

1. The first step in such cases is to identify endogenous feedback loops in the system that can create positive or negative change.

2. The second step is then to think about which ones to strengthen or weaken to get the desired results.

3. This is then followed by consideration of intervention design, activities, and indicators for L and M&E.

Beyond tracking dropout rates and enrolment numbers, we may need indicators that capture the strength of social acceptance or aspiration within the community. While such indicators are more difficult to quantify, they provide insight into how the strength of the reinforcing loop is changing over time. It is not only the question of whether our interventions shall be impacting other parts of the system; the question also is to ask how it is impacting and what we can do differently to create the change in the desired direction.

Concept Model 2: Tracking Relative Change in Climate Finance

A second way to rethink indicators is by focusing on relative change rather than absolute values. This can be illustrated through the example of climate finance. In many policy discussions, progress is measured by the total amount of funds allocated to climate action. While this provides useful information, it does not fully capture how priorities are evolving within the system.

The model has three stocks: funds allocated for adaptation, mitigation and the total funds for climate action. Traditional indicators would track how much funding is allocated to each category.

Figure 2: Understanding policy impact on fund flows for climate action

For example, the stock of funds allocated for adaptation is an accumulated value over time (1). Monitoring this will show us how much money has been cumulatively spent over time. But another alternative would be to think about the proportional change in the ratio between adaptation and mitigation fund (2), i.e. has the proportion been changing? If historically mitigation has been getting more funding, then it creates a bias towards more climate action for mitigation. If both adaptation and mitigation actions have to be in balance, then adaptation funding needs to start getting prioritized. We need to start allocating more funds, not just in absolute values but in proportion to what is being allocated to mitigation. The tilt in the ratio (refer to graphs 2 & 3) between the two would indicate if we are seeing a shift in the system.

Graph 2 & 3: Indicators showing fractional change in adaptation funds and the ratio between adaptation and mitigation funds.

Another important dimension is adequacy. Similarly, the fund allocation should also be compared against the requirements or needs. This way, we can sense the adequacy of the funds being allocated. These relative indicators, such as ratios and adequacy measures, provide a richer understanding of system behavior. They move beyond scale and focus on direction, balance, and sufficiency. What is commonly referred to as the adaptation deficit needs to be addressed by enhancing climate action, supported by adequate funding. An overall growing fund allocation value could show us success in policy change, but until it becomes adequate to meet the needs/deficit, it will remain short of creating the desired systems shift. Hence, what we monitor, in addition to the absolute values, should include the relative change in the system, e.g. availability against requirement, i.e. the adequacy indicator (3).  When we monitor such a relative change in the system, we begin to understand the direction of systems change, not just its scale. That’s a key shift in perspective.

Reflections and Guiding Questions

L, M&E for systems change is not about achieving flawless measurement; rather, it is about deeper learning and understanding the system and the change it is going through. The process should begin by asking fundamental questions. Some are listed below:

• What system shifts do we really care about?

• How much invested we are to learn about these shifts?

• What indicators could tell us about them?

• How can we identify and curate the indicators?

• How can we create the desired information systems to be able to populate the indicators? and

• What will it take to operationalise the above?

Systems dynamics modelling and systems thinking as methods could help us structure our thought process to be able to ask the right questions that can guide us on what type of indicators we need, why and what information they will give us.

Typically, indicators should allow us to:

• Follow a line of enquiry into the system to understand why things are happening the way they are happening,

• What potential changes could be thought about to change the outcomes we are seeing

• What inner systems shifts, i.e. our own beliefs and values, are needed to bring about outer systems change.

Ultimately, we need a perspective change in operating L, M&E, from proving success to learning about the systems’ behaviour and the systems’ change, both outer and inner. By integrating systems thinking and system dynamics into L, M, and E, we can move towards more meaningful and responsive ways of understanding impact.