AIR Currents

July 24, 2013

Editor's Note: In this article, George Davis, Vice President, Business Development, and David Wilson, Senior Product Manager, Product Management discuss how the Property Specific Database in Touchstone™ can be used to improve companies' exposure data and their understanding of their risk.

Introduction

AIR's next generation software platform, Touchstone™, offers companies a variety of tools that allow them to truly own the risk—and the first step is a deep understanding of their exposure data.

In this article, we discuss how Touchstone leverages AIR's Property Specific Database to augment incomplete data and correct incorrect data.

Touchstone's Data Quality Analytics

Accurate data are of paramount importance to reliable modeled losses. In Touchstone's Data Quality Analytics module, AIR provides companies with the tools needed to measure and improve their data quality, and to understand its potential impact on losses. The module includes a comprehensive suite of solutions, including scoring, benchmarking, validation, and augmentation.

Briefly, data scoring can help companies decide where to allocate resources for data quality improvement for maximum return on investment. Companies can benchmark their data against that of the industry by construction, occupancy, coverage, and line of business. They can validate their data using completeness and reasonability checks to identify where data are missing, determine whether a property is accurately geocoded, or whether coded risk characteristics make sense. Validation results provide guidance on steps companies can take to enhance exposure data at the portfolio level through data augmentation.

Finally, Touchstone's data augmentation feature can fill in missing or questionable values in exposure data with replacement values, building characteristics, and geocodes for individual buildings. Successful augmentation requires two critical factors. First, matching addresses within a portfolio to actual buildings on the ground, not average distributions by geography. Second, the source data must be of the highest integrity. This is where AIR's Property Specific Database (PSD) comes in.

George DavisBy: George Davis
Vice President, Business Development

David WilsonBy: David Wilson
Senior Product Manager, Product Management

Edited by Sara Gambrill

Developing the PSD

The PSD is an extensive and detailed database of residential and commercial properties in the U.S.1 It contains the exposure data necessary for catastrophe analyses, including property-specific replacement valuation, building characteristics, and geocoded location.

The PSD is a blended data set drawn primarily from data sources unique to Verisk Analytics, AIR's parent company, as well as census and tax data. The blended data approach enables AIR to assemble the most reliable data for each attribute. Verisk data sources contributing to the PSD include the following.

  • ProMetrix® property information provides detailed high quality data on millions of commercial properties in the United States. The information is compiled exclusively from actual physical surveys of individual commercial buildings conducted by one of ISO's hundreds of trained building inspectors.
  • 360Value® provides individually calculated, component-based replacement cost estimates for residential and commercial property records, supporting a complete range of building types. Powered by Xactware's comprehensive and up-to-date reconstruction cost information, 360Value provides reliable replacement values consistent with recent claims data.
  • ISO PushPin™ is a database of property-specific information for U.S. residential and commercial properties. It comprises aggregated public records data collected from county tax assessors across the United States. The assessor records undergo a proprietary process to clean and format them for use by the insurance industry.

All of the data elements in the PSD (see Data Elements callout) undergo an extensive quality assurance process to ensure their suitability for use in catastrophe modeling. The PSD has been validated through a multi-step process involving overlapping methods to test data accuracy and reliability, including the use of reasonability rules to identify and filter out unrealistic data, validation of questionable data using satellite and aerial images, and identifying and researching unusual data shifts of property specific records. AIR then assigns each element a quality grade, which allows the user to determine if the PSD data element is a more suitable option than the data they already have.

Individual structures in the PSD are identified by an ISO Building Identification Number (BIN). Because millions of buildings in the U.S. have ambiguous addresses and many properties have multiple addresses for the same building, while still others have a single address but represent multiple buildings, companies face challenges when trying to identify and appropriately geocode their risks. The ISO BIN dispels the ambiguity that can be associated with building identification because it uses an alpha-numeric key tied to the property specific information needed for catastrophe modeling.

Data Elements Available in the Property Specific Database

  • ISO BIN (Building Identification Number)
  • Geocode (Lat/Long)
  • Building replacement value
  • Construction
  • Occupancy
  • Year built
  • Number of stories
  • Square footage
  • Roof shape
  • Roof cover
  • External wall type
 

A Significantly Expanded PSD in Touchstone V1.5

Key to successful augmentation of portfolio exposure data is ensuring high match rates with the buildings in Touchstone's PSD. With the latest release in Touchstone Version 1.5, the PSD contains 68 million residential property records and more than 6.5 million commercial property records. Updates to the database are integrated into every Touchstone release.

Also in Version 1.5, the replacement cost estimates for both residential and commercial properties have been updated to account for changes in construction costs during the past year.

Cat Bond Figure 1 Event 1
Figure 1. Percentage of residential (left) and commercial (right) coverage by state in Touchstone's PSD
 

New Consistency Check in Touchstone V1.5

With every release of AIR's software, the information in the PSD is updated and enhanced, leading to its overall improvement. The release of Touchstone v1.5 is no exception. The augmentation feature in the PSD now provides a consistency check that indicates when user data is consistent with PSD data. The consistency check offers additional assurance of the accuracy of a company's data above and beyond actually augmenting data.

Configurable Augmentation Rules

Touchstone allows users to define rules for augmenting data from the PSD or with user-customized values, giving users complete control of the augmentation process. Exposure data can be modified to:2

  • Only augment values for a given column which are "unknown" in the current client data
  • Only augment replacement costs values when the PSD estimate is over 20% higher than the current estimate
  • Only augment geocodes when the PSD geocode is a more detailed resolution (e.g., the current geocode is less than or equal to postal centroid)
  • Identify tenant policies which should not have building-level replacement cost augmentation
  • Set column values to user-specified values. For example:
    • If the building replacement value is updated, adjusts the coverage-specific replacement values (for Other Structures, Contents, and Time Element) by the corresponding ratio of the difference between the augmented building replacement value and the old building replacement value.
    • Adjust secondary characteristic field values for sensitivity testing
    • Clear the geocoding data prior to re-geocoding.
 

A Hypothetical Case Study Illustrates the Benefits of Data Augmentation with PSD

In this business case, we are using a fictitious insurance company called Quality Insurance. They have a potential hotel chain client with close to 1,000 locations who is asking for an insurance quote. Quality Insurance knows that this is a competitive situation and that they need the most accurate view of risk possible to provide the best quote. It comes down to understanding the accuracy of the client's exposure data, and—if the client's data is questionable—finding expedient ways to augment the client's data to improve its quality.

Touchstone provides the right tools to meet the challenge. Right away, Touchstone's Exposure Dashboard Summary reveals possible reasons for concern (see Figure 2). Specifically, a significant percentage of the exposure is missing information on building height; the construction type, too, is missing for a substantial number of risks.

Cat Bond Figure 1 Event 1
Figure 2. Touchstone's Exposure Dashboard Summary shows areas where data are missing or unknown

What the underwriters do not know is the significance of the unknowns. The scoring functionality in Touchstone's Data Quality Analytics can provide this insight. The Data Quality Score is a measure of how missing exposure data impacts uncertainty in modeling results, and this score is lower than what the underwriters would like to see.

They decide to augment the data using Touchstone's PSD. Figure 3 shows the results: 27% of the risks in the portfolio, representing locations with unknown or incorrect construction type, have now been augmented with property specific information from the PSD, and 37% of the locations in the portfolio have been augmented with height information. The underwriters now have a more accurate view of their client's exposure data and when they re-score the data, the improvements are immediately apparent.

Cat Bond Figure 1 Event 1
Figure 3. Data augmentation changes using the Property Specific Database

The detailed loss analysis that the underwriters will run prior to developing the quote for the hotel chain will have less uncertainty and, therefore, the underwriters will have more confidence in their pricing proposal. Furthermore, comparing the percent change in replacement values and modeled loss before and after augmentation will give underwriters a good indication of potential insurance-to-value issues for the portfolio as a whole and its impact on loss estimates.

Conclusion

It is now widely acknowledged that the catastrophe loss estimates generated by detailed catastrophe modeling are only as good as the quality of the exposure data entered. Through the Data Quality Analytics module and leveraging AIR's Property Specific Database, Touchstone offers innovative approaches to assess and even improve that quality of exposure data, enabling companies to increase the reliability of their catastrophe risk analyses.

1 The benchmarking, scoring, and validation Data Analytics features are also available in Europe.

2The user can choose whether augmented data overwrites the existing data. The default option is to create a new exposure view, retaining the original data.


Loading...

Close

You’re almost done.
We need to confirm your email address.
To complete the registration process, please click the link in the email we just sent you.

Unable to subscribe at this moment. Please try again after some time. Contact us if the issue persists.

The email address  is already subscribed.