|
Data Quality Analytics
RMS Data Quality Analytics empower insurers and reinsurers to assess
exposure data quality, measure its impact, and target data improvement
where it matters most for catastrophe risk models. Data Quality
Analytics deliver objective and independent insight into the main
elements of exposure data—where it is (location), what it is
(vulnerability attributes), and how much it is (valuation)—allowing
insurers and reinsurers to assess the quality of exposure data input
into catastrophe models.
Key Components
Data Quality Analytics account for severity (hazard and hazard
gradient), the relative importance of attributes, and the loss
contributions, to reveal insights that go beyond simple measures such as
percentages of street-level geocoding or known/unknown
attributes—conventional, but incomplete metrics that can lead to a false
sense of security about data quality. Data Quality Analytics consist of
completeness scores and accuracy assessments. These transparent,
actionable, and easy-to-communicate metrics are used in conjunction with
the RMS ExposureSource database to evaluate data quality and inform
portfolio management and underwriting decisions.
Completeness Scores
Weighted by peril, hazard severity, hazard gradient, line of business,
and other factors, completeness scores measure:
-
Geocoding Resolution Score (GRS): Assesses the
variability in modeled losses resulting from the level of geocoding
resolution
-
Vulnerability Completeness Score (VCS):
Measures the amount of variability in modeled losses resulting from
missing primary vulnerability attributes, such as occupancy,
construction class, year built, and number of stories
-
Completeness Score (CS): Uses a unique
algorithm that combines the GRS and VCS scores based on the relative
importance of each.
-
VCS Improvement Potential: Quantifies the
relative importance of the missing building characteristics on the
VCS—improvement potentials are based on extensive sensitivity analyses
-
Variation in Loss Cost: Measures the potential
variation in model loss estimates resulting from the percentage of low
resolution geocoding and unknown vulnerability attribute information in
a data set. For example, in a case where there are 10 buildings—8 with
known construction type, 2 with unknown construction type—the variation
in loss cost indicates the maximum variation (+/-) due to the 2 unknown
construction types if they were coded to the best case and worst case
attributes.
Accuracy Assessment
Accuracy is assessed using rules and algorithms that assess the
credibility and objectivity of the data, and by comparing data against
trusted in-house and third-party sources.
-
Validation Heuristics: Rules that identify
inconsistent or illogical combinations of geocoding, vulnerability,
valuation, and financial attributes, and investigate suspicious patterns in the data
suggesting bulk coding
-
Comparison against RMS ExposureSource Database:
Data is validated against the ExposureSource database of
location-specific commercial and residential property exposure data,
developed specifically to provide exposure data and optimized for use
with catastrophe models for the insurance industry
-
Comparison against Unknown Portfolio: Industry
comparison metrics indicate how aggressively or conservatively the
portfolio has been coded compared to the underlying industry
inventory—i.e., how loss results for the portfolio as coded would
compare to results if occupancy is known but everything else is unknown
Data Quality Analytics provide objective and
independent insight into the quality of exposure data input into
catastrophe models, by providing transparent (intuitive and easily
understandable), actionable and portable (easy to communicate and
transmit across insurance value chain) metrics and insights to inform
portfolio management and underwriting decisions.
The
ExposureSource Database
The
ExposureSource database is a comprehensive database of
location-specific, U.S. commercial and residential property exposure
data. Developed specifically to provide exposure data optimized for use
with catastrophe modeling for the insurance industry, the ExposureSource
database provides a robust source of exposure data for improved risk
analysis and management.
|
Follow RMS on Twitter