AAA Insurance Group – Club CSAA
Received 1st Place in SOA Actuarial Research Conference
Provided research on the topic of modeling Insurance Reserves, under the
direction of CSAA. Discovered that time series modeling can also be used
for predicting future losses compared to prior methods like General Linear
Regression. By using Time Series analysis, we were able to conclude time
dependent correlation factors that would better forecast losses for Home
and Auto insurance.
Data Mining Class
This project ended up being a ‘Version 2.0’ of the CSAA Time Series Analysis research in my prior project. Following our research, the reserving group of CSAA asked that we look into finding any possible correlations to the type of ARIMA’s that were fitted to
the individual series by state and/or coverage. Example, Bodily Injury, an
auto insurance coverage that is required in almost all states may experience
a seasonal component if it’s earned exposure is above a certain limit. The
same could be asked by state. If such a classification exists, the application
of this would be useful for understanding what type of reserving trends they should expect for new states they wish to open coverage’s in.