One way to model model uncertainty is to have uncertainty about model parameters. Although there can be times where this technique is adequate, it does capture the true nature of model uncertainty. Model uncertainty refers to situations where a baseline model is missing dynamics found in the real world system. Ideally, analysis should be robust to this type of uncertainty.…
Bond EconomicsMore generally [in economics], the possibility of models being incorrect are effectively underweighted. The assumption is that prediction errors are the result of external shocks hitting a known model (albeit one with parameter variability), and not the effect of missing dynamics. More specifically, the possibility of the interaction of the preferred methodology with models that do not conform to assumptions is not taken seriously enough.
Parameter Uncertainty Is Not The Same Thing As Model Uncertainty
Brian Romanchuk
3 comments:
Can you give an example of parameter uncertainty in electronics?
Every electronic device deviates from its theoretical specs. Those deviations are parameter uncertainty if you keep the same model of the circuit.
Sounds like the definition of tolerance.
Components have a tolerance rating. Part of this is may be due to manufacturing limitations, or the effect of aging.
Analog and digital circuits have an operating range. To exceed the range is to render their function unreliable.
High frequency circuits have to be manually calibrated.
Electronics is a good example, is it not?
Its a proven application of mathematical modeling/theorems.
Post a Comment