INCREASED DATA QUALITY
Why bad data destroys any risk management
Bad Data quality It does not simply lead to minor inaccuracies – it leads to erroneous decisions, false certainty, false priorities, and expensive measures that end up costing taxpayers more money. Risk to pass by
Typical consequences of poor data quality:
- Misjudgments in management
- low acceptance within the company
- inefficient and expensive Control measures
- distorted Exposure calculations
- wrong Risk classes
- Lack of transparency
Briefly: Bad Data quality is its own risk in risk management.
Problem No. 1: Risk names
- too long
- too vague and creative
- not one-sided
- double (in different languages)
- whole sentences instead of terms
A risk novel is not a novel. It is the First line of defense against chaos.
Problem No. 2: Risk description
- too technical
- too abstract
- incomplete
- unintelligible
If one Risk Responsible he cannot explain his own text, that is the Description almost worthless.
Problem No. 3: Risk assessment (methodology)
- Presetting incorrect distribution functions
- Models depend on the MC simulator used
- Assignment of the evaluation to the risk manager
Often they are wrong distribution functions used for risks (e.g. Triangle– instead of Three-point distribution). This led to massive error results (larger than system or model errors).
Problem No. 4: Risk assessment (Know How)
- Selecting distribution functions
- Defining and estimating parameters and moments
- Understanding mathematical models
That is unrealistic. „The choice of Distribution function is not feasible for 90% unskilled (!) RVs.“
The persons responsible for risk management are No financial mathematicians. They are supposed to appraise, not model.
CORRECT distribution functions
The classic - Triangular distribution vs. Three-point distribution
Choosing the right one Distribution function has a massive influence on the Simulation results and calculated risk exposures.
In practice, a frequently recognizable Errors in the selection of the Distribution function, that risks are falsely attributed to Triangular distributions are characterized/evaluated, although they:
- distributed discreetly,
- asymmetrical,
- heavy-tailed
are. Examples of risk include cyber attacks, plant fires, compliance violations, natural disasters, etc.
Briefly: These risks need to be taken into account Three-point estimates (Best, Normal, Worst Case), none Triangular distributions.
WHY?
One Triangular distribution generated:
- invented probabilities
- Overweighting the most likely damage
- massive underestimation of tail risks
- Artificial limitation of distribution
Briefly: One Triangular distribution It is too simplistic and methodologically flawed for these risk types.
In the risk assessment with distribution functions, the following conditions exist: 3 serious risks:
80%
the Risk managers are included in the selection of Distribution functions and their evaluation is overwhelming.
70%
the central risk managers give the VT functions They also carry out the evaluations beforehand.
60-80%
the results are distorted because they are unsuitable Distribution assumptions and parameters that drive the errors more strongly than the model itself.
Risk Responsible Checklist
Checklist for risk managers
Below is a short checklist for Risk managers, to increase the data quality.
-
Clear and concise risk names (max. 3-5 words; nouns at the beginning)
-
Clear descriptions (trigger, cause, effect)
-
Error reduction (empirical data, plausibility, avoidance of extreme assessments)
-
Realistic and understandable estimate
-
Accurate information (derivation and explanation of the content)
Data quality determines the acceptance of risk management
Data quality determines whether risk management is an effective Control instrument or one Documentation tool It is. Faulty or inconsistent data distort results more than any methodology, while high-quality data stabilize evaluation, minimize estimation errors, and only then make risk management reliable. Data quality is therefore the Most important success factor in the entire risk process.
