Abstract
This paper contains a new approach to analyzing loss statistics which uses stochastic processes. The author views loss statistics as samples from a specific type of stochastic process. The author believes that type of process is the most consistent with the realities of insurance statistics, and he explains why. Using that mathematical framework the author develops a formula for credibility when the complement of credibility is applied to trend. The paper also contains a formula for trending data that is more consistent with the stochastic approach (and hence the realities of insurance statistics) than the trend line.
Credibility, Trend
Volume
Special Edition
Page
341-400
Year
1993
Keywords
predictive analytics
Categories
Actuarial Applications and Methodologies
Ratemaking
Trend and Loss Development
Financial and Statistical Methods
Credibility
Financial and Statistical Methods
Statistical Models and Methods
Publications
Casualty Actuarial Society E-Forum