18. Data Flashcards
Define ‘personal data’
Data where an individual could be identified when combined with other data
What allows the collection & storage of vast amounts of data
Improved technology
How can data legislation vary by jurisdication?
Objectives & expected behaviour are similar but legislation varies. The US has much less stringent laws
Why must extra care be taken when transferring data between countries?
Data legislation can vary between jurisdiction
Name the 8 categories of ‘sensitive personal data’ (RREPPCST)
- Race
- Religion
- Ethnicity
- Political opinion
- Physical/mental condition
- Convictions
- Sex life
- Trade union membership
Name the 3 qualities by which big data can be categorised
- Very large datasets
- Brought together from many sources
- Can be analysed quickly
How can big data be altered to provide data protection?
Anonymisation can remove any personal data
What is the theory of data minimisation?
That big data is excessive
Complexity of big data is not an excuse for…
failure to comply
Define ‘data governance’ SIAU
The term used to describe overall management of availability, usability, integrity & security of data
What is a data governance policy?
A documented set of guidelines for data management, detailing how data is captured, analysed & processed
What 6 things does a data governance policy detail? RUSCCM
- Use for data
- Roles/responsibilities of individuals
- How data is captured, analysed & processed
- Security/privacy issues
- Details of controls to meet standards
- How adequacy of controls is monitored
Name 3 risks of poor data governance
- Fines
- Reputational damage
- Inability to rely on data for use
What should the data governance policy detail regarding a merger?
The risk of aggregating data & data systems
Give an advantage of combining data & data systems in a merger
Adv: overhead savings
Disadv: cost of converting systems is high
Name 5 risks around data & its suitability for use
- Errors/omissions
- Insufficient data
- Credibility
- Not reflective of future experience
- Form - not in required form for purpose
Give 5 reasons why data may not be a reflection of future experience
- Random fluctuation
- Abnormal events in data
- Changes to data recording
- Change in balance of homogenous grouping
- Socio-economic change
What is the result of a lack of confidence in data
A lack of confidence in conclusions
What is the issue with very small homogenous groups
They can be too small to draw credible conclusions & if merged to form sufficiently sized groups, it may reduce homogeneity
What is algorithmic decision making?
Automated trading to capitalise on price discrepancies across markets
List the benefits of algorithmic decision making
- Quicker, more consistent decision making
- Lower dealing cost
How can advancements in big data aid algorithmic decision making?
Allows for greater accuracy in setting parameters
Name 6 risks of algorithmic decision making
- Algorithm error
- Data error
- Creating instability in markets (plunge & rebound)
- Turbulent conditions can cause market suspension
- May not operate in turbulent markets
Name 3 reasons why data for all tasks should be controlled through a single system
- Audit trail
- Easier access
- Lower chance of data corruption
Why might competitor data be limited in its usefullness?
- Different benefits offered
- Difference in target market
- Difference in approach to valuation (prudence in CBE)
Why does it take a long time to accumulate good data?
Data takes many years to accumulate so must have good systems in place
What kind of questions are used on the proposal form & why?
- Tick boxes to be easily entered
- Unambiguous for accurate information
- Rating factors used to translate qualitative to quantitative
Who provides the data used in employee benefit schemes?
Sponsor (employer)
Why can data be a particularly prevalent issue in employee benefit schemes?
Provided by sponsor (employer) who may not have sufficiently detailed or reliable data
Name 3 good checks on data
- Checks against data from past valuation date
- Checks against accounting data
- Assertations
Name the 3 things to be attested to
- Appropriate valuation date
- Complete
- Assets/liabilities exist on given date
How can data be checked if it’s not possible to check an entire dataset?
Random spot checks
What is important to note when using summarised data (summarised due to insufficient volume or detail)
Recognise the reliability of results will be impacted
Give an example of an industry-wide data collection scheme
IFoA PPO working party
Give 2 benefits of using industry-wide data collection scheme
- Can compare experience with the industry
- Can compare homogenous groupings
Name 6 potential causes of heterogeneity distortion in industry-wide data collection schemes
- Different policies sold
- Different sales methods
- Different underwriting process
- Different risk factors
- Different data systems
- Different socio-economic conditions
Give 3 other distortions in industry-wide data collection schemes
- out-of-date
- less detailed
- not all firms participate so not fully market representative
