System Dynamics Business Strategy Modeling

A stock is an accumulation that persists over time — inventory, installed customer base, cash, employees — while a flow is a rate that changes a stock, such as hiring or attrition. Forrester built industrial dynamics on this distinction because stocks integrate their net flows and thereby give a system memory and inertia. Confusing the two is a classic modeling error: treating a rate as a level, or vice versa, produces qualitatively wrong dynamics. Getting the stock-and-flow structure right is what lets the simulation reproduce realistic behavior over time.

System dynamics works at an aggregate level, representing populations as continuous stocks governed by feedback loops and differential equations, and is ideal when the dynamics are driven by accumulation, feedback, and delay. Agent-based modeling works bottom-up, representing many discrete heterogeneous firms whose interactions generate emergent structure. They are complementary: system dynamics excels at endogenous feedback in aggregate quantities, while agent-based models excel at heterogeneity and local interaction. The choice depends on whether the strategic question hinges on feedback among stocks or on the diversity and interaction of individual actors.

Not in the sense of precise point forecasts. Sterman emphasizes that the value of these models lies in explaining and improving dynamic behavior — understanding why a strategy oscillates, resists policy, or overshoots — rather than predicting exact figures. A well-built model can reproduce the qualitative pattern of past behavior and reveal how different policies shift that pattern, supporting better decisions under uncertainty. Treating its output as a deterministic forecast, instead of as insight into structure and policy leverage, misuses the method.