Skip to main content

Table 1 Advantages and disadvantages of various disaster impact methods

From: Estimating economic loss from cascading infrastructure failure: a perspective on modelling interdependency

I-O methods CGE Methods Econometrics Cost-benefit analysis
Simplicity Can address market based mechanisms and behaviours Very good forecasting capability Simplicity
Clear distinction between indirect and direct effects Can model broad range of impacts (e.g., response to price changes) Rigorous well-developed model validation Attempts to capture all costs and benefits
Well suited to distributional impact analysis Can look at distributional impact No major biases in estimating impacts Applies expert knowledge and experience to generate cost estimates
Excellent organisational framework for data collection and display Better suited to long term recovery analysis Incorporates uncertainty Model transparency
Provides transparent view of the economy Can model impacts across a range of macroeconomic variables. Does not assume market equilibrium Single unit-measurement so costs and benefits can be easily compared
Well-suited to short-term recovery periods    Benefits are easy to double count
Ability to integrate with other models (e.g., engineering, econometric)    
Rigidity due to linearity Intended for long-run equilibrium analysis Significant data demand requirements Does not account for economic multiplier effects
Ignores agent behavioural response to disaster Usually provides over optimistic results because of flexibility of response Not well suited to modelling rare events Subjective costs and complications
Inadequately deals with monetary interventions No explicit distinction between direct and indirect effects Difficult to obtain disaggregated, regional data Single unit-measurement assumes all things can be compared
Relies on market equilibrium, while disasters represent a disequilibrium Assumes all agents optimise No explicit distinction between direct and indirect effects  
Characterised as providing over pessimistic results Assumes agents have perfect information Model is based on historical experience which is unlikely to provide good  
   Inadequately allows for economic multiplier effects  
   Research question must be well specified  
  1. Table has been modified and expanded from Okuyama (2008)