Risk assessment Study Guide

📖 Core Concepts Risk assessment – systematic process to identify hazards, estimate their likelihood and consequences, then judge whether the risk is tolerable. Hazard analysis – first stage; catalogs what could go wrong. Risk evaluation – second stage; decides if risk is acceptable, needs mitigation, or must be transferred. Acceptable risk – a risk we understand and tolerate because mitigation costs exceed expected loss. Mild vs. wild risk – Mild: follows normal‑like distributions, predictable (law of large numbers). Wild: fat‑tailed (Pareto/power‑law), may have infinite mean/variance, hard to predict. Linear vs. non‑linear systems – Linear: predictable response to input changes; Non‑linear (complex): unpredictable, often requires dynamic risk assessment (DRA). Expected risk (R) – sum of each possible loss multiplied by its probability: $$R = \sum{i} L{i}\, p{i}$$ Annualized Loss Expectancy (ALE) – quantitative metric for financial risk: $$\text{ALE} = \text{SLE} \times \text{ARO}$$ --- 📌 Must Remember Risk = Hazard × Likelihood × Consequence (implicit in the definition). Acceptable risk exists when mitigation cost > expected loss. Mild risk → normal distribution, Wild risk → fat‑tailed, infinite variance. ALE = Single Loss Expectancy (SLE) × Annualized Rate of Occurrence (ARO). Dose‑Response safety factor – typically 10× per unknown step (e.g., inter‑species, low‑dose extrapolation). Audit risk = Inherent risk × Control risk × Detection risk. PEC/PNEC ratio > 1 indicates potential environmental threshold exceedance. --- 🔄 Key Processes Standard Risk Assessment Identify hazards → Analyze (probability & loss) → Evaluate tolerability → Document & decide on mitigation. Dynamic Risk Assessment (DRA) (emergencies) Continuous monitoring → rapid re‑evaluation of hazards → immediate control actions → post‑event debrief. Quantitative Risk Calculation (ALE) Estimate SLE (monetary loss per event). Estimate ARO (events per year). Multiply → ALE. Dose‑Response to Safe Dose Determine NOAEL/LOAEL → apply safety factors → derive Reference Dose (RfD). Risk Matrix Classification (diving example) Plot Likelihood vs. Consequence → assign level: Unacceptable, Marginal, Acceptable. --- 🔍 Key Comparisons Mild risk vs. Wild risk Mild → normal/near‑normal distribution, predictable. Wild → Pareto/power‑law, infinite mean/variance, unpredictable. Linear system vs. Non‑linear system Linear: predictable input‑output; simple static assessment works. Non‑linear: unpredictable, requires DRA & scenario stress testing. Quantitative (ALE) vs. Qualitative assessment Quantitative: numeric ALE, good for financial decisions, may miss non‑quantifiable factors. Qualitative: uses words/ratings, captures perception, ideology, and “unknown unknowns”. --- ⚠️ Common Misunderstandings “All risks can be quantified.” – Quantitative methods ignore non‑measurable factors and can mislead for wild risks. Treating wild risk as mild. → Leads to severe under‑estimation and insufficient controls. Assuming low probability = low concern. – Large potential loss (high $Li$) can make a tiny $pi$ highly significant. Safety factor of 10 is universal. – It is a rule‑of‑thumb; actual factor may differ based on data quality and uncertainty. --- 🧠 Mental Models / Intuition “Risk = Exposure × Severity.” Visualize a sliding scale: high exposure low severity vs. low exposure high severity → similar risk levels. Fat‑tail intuition: Imagine a handful of huge dice (wild risk) vs. many small dice (mild risk); the occasional huge roll dominates overall outcome. Cost‑Benefit threshold: If Cost of mitigation > Expected loss, accept the risk. --- 🚩 Exceptions & Edge Cases Wild risk with infinite variance – traditional variance‑based decision tools fail; use scenario analysis or stress testing. Voluntary risks (e.g., smoking) – often under‑estimated because individuals feel in control; risk perception diverges from objective risk. Small sub‑populations (≤0.1%) – risk may be hidden in aggregate data; must assess exposure vs. susceptibility separately. Regulatory thresholds (PEC/PNEC) – a ratio >1 flags concern but does not quantify actual ecological damage. --- 📍 When to Use Which Use a quantitative ALE when you have reliable loss and occurrence data (financial, insurance). Use a qualitative matrix for safety‑critical, low‑data contexts (diving, wilderness activities). Apply DRA in non‑linear, rapidly changing emergencies (fire, disaster response). Choose mild‑risk models for large‑sample, repeatable processes; choose wild‑risk models (Pareto, stress scenarios) for rare, high‑impact events. --- 👀 Patterns to Recognize “High loss × low probability” appearing repeatedly → suspect wild‑risk characteristics. Repeated “voluntary” language → likely under‑perception bias. Presence of safety/uncertainty factors → dose‑response assessment is underway. Risk matrix cell labeled “unacceptable” → expect a mandated mitigation step before proceeding. --- 🗂️ Exam Traps Choosing ALE for wild risk – exam may present a scenario with fat‑tailed loss; ALE assumes finite mean → wrong. Confusing “acceptable risk” with “no risk.” – acceptable means tolerable, not eliminated. Selecting linear‑system methods for a complex system – will be penalized for ignoring non‑linearity. Assuming a PEC/PNEC ratio quantifies ecological damage – the ratio only signals threshold exceedance, not magnitude of impact. ---