An Introduction to System Dynamics

In the 1950s, General Electric found itself struggling with three-year employment cycle at one of its Kentucky plants. The problem was proving persistent and, because the regular business cycle could not explain the severity of the problem, it was decided that outside assistance was needed. Plant management sought help at the offices of one Jay Forrester with the then newly-formed MIT School of Management.

Forrester, who had a background in electrical engineering, was interested to see how his experience could be brought to bare on management theory. He therefore proceeded to assess GE’s problem as an artifact of a complex system. And where plant management had seen only the oscillation of staff, he saw an intricate network of feedback loops that exerted influence in ways that wasn’t readily observable.

The insights that Forrester gathered would prove seminal to the field of System Dynamics—an approach to understanding the behavior of complex systems over time—that he would later establish. Today, it’s perhaps the better known as one of many aspects within Systems Theory. And it’s concerned with the structure of systems—including the many circular, interlocking, and sometimes time-delayed relationships they exhibit.

System Dynamics makes use of stocks, flows, and feedback loops to understand and visualize complex systems. These “building blocks” help describe how seemingly simple things can produce baffling nonlinearity. In the GE case, Forrester was able to show that it was the stock-flow-feedback structure of the GE plant—including the existing corporate decision-making structure for hiring and layoffs—that created the problematic employment cycle.

Free interpretation of Forrester’s GE assessment drawn as a Systems Dynamics model. New hires (inflow) add to the number of staff (stock), which, in turn, is depleted by staff layoffs (outflow). Two balancing feedback loops (B1+B2) help regulate the system while three delays work to create oscillation.

Management at the GE plant looked to incoming orders and inventory build-up when determining future staffing needs. And if they found a discrepancy between what they had versus what they thought they needed, they would react accordingly. The problem with this system was that it had significant delays built into it. It could, for example, be many months before a perceived hiring need would ever be satisfied. This created a mismatch that, over time, grew into a three-year employment cycle.

As Forrester would later explain: “it became evident [to me] that there was potential for an oscillatory or unstable system that was entirely internally determined. Even with constant incoming orders, one could get employment instability as a consequence of commonly used decision-making policies.” Plant management had, in other words, itself created the employment cycle by putting in place policies that were unable to satisfy staffing needs in a timely manner.

System Dynamics is often used to create models for advanced computer simulations. But it can also, as this GE example illustrates, be used to create simple back-of-the-napkin visuals. Regardless of application, it finds its strength in its ability to simplify complex systems—allowing for the holistic analysis of dynamic, and often nonlinear, systems and their behavior over time.