By effectively utilizing comprehensive traffic crash data, our Location-based crash risk scores are proven to improve the profitability of auto insurance carriers by 3 to 20% through better pricing and risk selection
Many Auto Insurance carriers have already incorporated location risk through territories into their current pricing models. These territory definitions are usually derived from garaging addresses based on their own loss experience or though a consortium exchange of competitive loss data. Some sophisticated carriers may even use population density; proximity to high traffic locations etc. to further refines rating territories into smaller geographical units like 5-digit zip, 9-digit zip, and census tracts, etc. However, these sources and methods only paint a part of the picture because of limited access to industry wide loss history. In addition to this, our roads are constantly changing every year with some becoming safer while others becoming more prone to automobile crashes.
With more drivers hitting the road and combined loss ratios on the rise, to improve their bottom-line auto Insurers need to monitor where crashes occur and dynamically incorporate crash location data into their rating territories beyond just the loss history on garaging locations.
Crash Locations: A Crucial Piece of Missing Information
TNEDICCA has made available a more comprehensive view of loss data. Since traffic accident related claims are likely to be a subset of traffic crashes, TNEDICCA’s location-based crash risk scores represent a more comprehensive view of industry claim losses across geography. More importantly, the probabilistic nature of crash frequency, hence claim frequency, are broader than the current defined territory
“The first step in reduction of the frequency of accidents is to determine whether the cause of an accident belongs to the system or to some specific person or set of conditions. The split is possibly 99 per cent from the system,1 per cent from carelessness”
-W. Edwards Deming.
A fundamental insight from Deming’s teachings is that process design drives outcomes more than behavior of the individuals
An implication of this fundamental insight into the context of traffic crash analysis is that only a small set of locations consistently accounts for a very large portion of total traffic crashes. Hence, it is crucial to know where these crashes occur
Approximately, the top 10% of crash locations
account for more than 66% of all crashes.
A New Frontier: Location Based Crash Risk Scores
TNEDICCA has taken up the tedious task of collecting and curating detailed comprehensive crash data and has been focused on building crash risk scores based on geocoded crash locations that has not been available in the past.
Our location-based crash risk score in a simplest term is the weighted summation of crash frequency around a given address; usually a residential/garaging address that represents a complete picture for more accurate auto insurance pricing.
TNEDICCA Location Based Risk Score aims to answer: Two important Questions
Question 1
Can auto insurance carriers use the location-based crash risk intelligence to predict the frequency of crash-related insurance claims?
Hypothesis
If the answer to this question is “yes”, we should observe a positive trend in claim frequency per policy when plotting location-based crash risk scores as shown below
Result
The actual retro-test results indicate very strong signals that location-based crash risk scores can be used to predict claims frequency for collision, property damage liability, and bodily injury claims
Illustration: location-based crash risk scores vs. an insurance carrier’s claims frequency. The higher the location risk, the higher the claims frequency.
Anonymized actual results from a retro-test with a regional carrier: the higher the location risk, the higher the claims frequency
Question 2
Have these location-related risks been incorporated into carriers’ current pricing through other sources of information?
Hypothesis
For insurance carriers, it’s all about the incremental value of new information. Had the location-based crash risk fully incorporated into the current pricing through the use of other factors, the prices and losses would look much like the charts below (i.e., you would see them as parallel)
Result
Empirical results indicate that the claims risks represented through location-based crash risk scores have not yet been fully incorporated into the current pricing processes. Hence, there are opportunities for rate modifications to further improve overall profit
Overall Results by Coverage
