10.6 Implementation

In a full Bayesian analysis, we should have a prior distribution for \(\varphi\) as well

• But for ease of implementation, we’ll use a plug in estimate

• Value used is that obtained from the application of the ODP, estimating the row and column parameters using MLE

The main thing is just picking how strong our prior is and it’ll be more like Chainladder or BF depending on our prior

This section of the paper just goes through the methods we’ve discussed and comparing it to the no intervention Chainladder