265. Causal Loop Diagrams

A qualitative diagram showing causal relationships between variables in a system. Used for understanding feedback structure without building a quantitative model.

265.1. Anatomy

Variables (nodes): named quantities, e.g., “Population”, “Birth rate”.

Links (arrows): cause → effect relationships, labeled with polarity:

• Positive link (+): cause increases → effect increases (and decreases → decreases). Same direction.
• Negative link (-): cause increases → effect decreases (opposite direction).

265.2. Loop polarity

Trace a loop and multiply link polarities:

• Reinforcing loop (R): odd number of negative links ‘multiplied out’ to positive net effect. Self-amplifying.
• Balancing loop (B): even number of negatives → net negative. Goal-seeking, stabilizing.

265.3. Example: population growth

              ┌──→  Birth rate
              │       │
              │       │ +
              │       ↓
   Population ──+──→ Population
              ↑      │
              │      │
              │ -    │
              └── Death rate

Loops:

• Population → Births → Population: reinforcing (R) — more people → more births → even more people
• Population → Deaths → Population: balancing (B) — more people → more deaths → less population

Behavior depends on which loop dominates. With more births than deaths: exponential growth. With more deaths than births: exponential decay. Balance: steady state.

265.4. Common patterns / archetypes

• Reinforcing loop: exponential growth or collapse (population, viral spread, runaway inflation)
• Balancing loop: goal-seeking, homeostasis (thermostat, predator-prey, supply-demand equilibrium)
• Limits to growth: R + B = S-curve (logistic growth, market penetration)
• Shifting the burden: short-term fix, long-term fundamental solution underused
• Tragedy of the commons: individual benefit ≠ collective benefit
• Eroding goals: gap → lower goal → narrower gap → success theater

265.5. From CLD to quantitative model

A CLD shows structure, not magnitudes. To predict behavior quantitatively, convert each link into an equation (a stock-flow system):

• Variables become stocks or auxiliaries
• Positive links with cause = $𝑥$, effect = $𝑓\left(𝑥\right)$ where $𝑓$ is increasing
• Negative links: $𝑓$ is decreasing
• Define table functions for non-linear effects
• Quantify all parameters

265.6. Common pitfalls

• Drawing too many links: keep CLDs focused — 5–15 variables, 10–25 links max
• Confusing correlation with causation: only draw a link if A causes B (not just correlates)
• Implicit time scales: real links have delays; CLDs don’t show them — note carefully
• Bypassing CLDs and jumping to stock-flow: tempting but loses high-level structure

265.7. See also

• Stocks and Flows — quantitative version
• Feedback Loops
• System Dynamics — overview