Hi, Can anyone please explain to me the ODP bootstrap model. 1. It assumes that the variance exceeds the best estimate by a constant factor. Does the variance refer to variance of incremental claim amounts? 2. Why does the core reading of pg 19 say that bootstrapping the ODP provides only parameter variance? 3. How is the constant factor estimated from past data and is this where the bootstrap comes in? Thanks.
Don't know whether this helps? Found it on the CAS website. http://www.casact.org/education/spring/2006/handouts/clark.pdf
It can refer to incremental claim amounts or claim numbers. Bootstrapping the ODP is similar to the simulation technique described on pp16-18 of Chapter 14. We assume that incremental claims follow an ODP distribution, and we take samples to obtain a set of alternative data sets (ie incremental triangles). The variability across these data sets is determined by our ODP parameters. In other words, the output tells us the variability caused by our parameters; our parameter variance. Since the variance is proportional to the mean, a reasonable approach might be to look at the past data and take the mean of the incremental claims and divide this by the variance of the incremental claims. This then becomes an input into the model. Once you have decided on your ODP parameters, you can follow the bootstrapping process on pp16-18 of Chapter 14.
