Chapter 18 Flashcards
(6 cards)
List 4 limitations of models
- Results are as reliable as model used
- Results are as reliable as data used
- Results depend on suitability of assumptions
- Could be prone to random error
How to minimise limitations of models
- Consider lots of suitable models
- Use expert opinion when choosing a model
- Ensure that the data is reliable
- Do data checks
- Goodness of fit of model v actual experience
- Sensitivity analysis on assumptions
What are the pros and cons of building a new model
Pros:
* Fully customisable
* Transparency in assumptions, calc and methods
Cons:
* Resource intensive
* Time consuming
* Updates are an internal issue
What are the key requirements of a model and the operational issues that need to be considered
- Needs to fit purpose at hand
- Need to decide on time horizon: time to maturity
- Need to decide on stochastic or deterministic
- Results and workings must be communicable
- Independent verification
- Development and refinement
- Not too complex
- Documented
- Must meet any statutory requirements
Pros and cons on using a commercial modelling product
Pros:
* Reliable
* User-friendly
* Has been validated and tested
* Meets regulatory requirements
* Saves time
Cons:
* Can be expensive
* Hard to deeply customise
* Limited transparency
* Depend on provider to fix bugs
Pros and cons of using an existing model in an organisation
Pros:
* Familiar
* Cost effective
* Saves time
* Consistency between products
Cons:
* May not fit purpose
* Outdated
* Hidden bugs
* Bad performance under large volumes of data
