Data Flashcards
What is data governance?
a documented set of guidelines for ensuring the proper management of an organization’s data
Data governance’s influence on guidelines
Roles and responsibilities of individuals on data
Way of capturing, processing and analyzing data
Data security and privacy
Controls that will be put in place
How the adequacy will be monitored on an ongoing basis on usability, accessibility, integrity and security
Define what is big data
Increasing use of technology have now made it possible to
collect and analyze
very large data sets
often brought together from different sources
can be analyzed very quickly
What can data be used on?
Analyze history experience
Investigate claims
Underwriting cover
Better personalized experience
Target market for marketing campaign
Throughtout need to comply with all data policies
Give six key risks relating to data
Data inaccurate or incomplete
Data not credible due to insufficient volume
Data not sufficiently relevant to the intended purpose
Data not reflecting what will happen in the future
Data groupings may not be optimal
Data not available in appropriate form for intended purpose
How to can data be used to build a predictive model? (process)
Decide whether all the data is relevant
Need to organize data to show exposure
Data will be grouped into relatively homogenous risk groups using rating factors
Need to balance credibility and having enough distinct cells to build a detailed model
Some adjustments needed to the data
A model can be constructed
The predicted outcomes will be compared to actual outcomes
Costs associated with bad data governance
Getting fined for non-compliance
Having business suspended
Lack of sales or increased lapses
Poor reputation
Not having robust internal information
Sued by clients
Increased risk of loss due to cyber risks
Increased risk of loss due to undetected fraud
Sources for big data
Application forms of current and new clients
underwriting data
social media accounts
browsing history
medical records
credit scores
banking information
travel information
How to can data be used to build a stochastic model? (process)
calculate the basic risk premium
use historic data to derive distributions and associated parameters
may require some dynamic links between distributions
after distributions have been fit, run goodness of fit test
by comparing actual experience with predicted
this assumes that future experience will be similar to
before
explicit adjustments made for trends and prudence
Consider anti-selection if it is a choice
run on the exposure many times to simulate events
distribution of the expected result can be derived
will be summarised and presented for discussion
required and desired risk criteria’s
Required:
The policyholder should have an interest in the risk
The risk is of financial and reasonably quantifiable nature
The claim amount must bear some relationship to the financial loss incurred
Desired:
Individual risks should be independent
Probability of the event occurring should be small
Large numbers of similar risks should be pooled to reduce variance
Should limit on ultimate liability undertaken
Moral hazard should be eliminated as far as possible
Should be sufficient data in order to quantify risk
features of a good data system
should be capable of storing information so that historical data is available for future pricing exercises
should be robust and flexible
should be secure
only certain individuals are allowed to amend the data on the system
checks of movement analyses should be carries out regularly
checks of changes in policy details should be carried out regularly
