In my mind I've been breaking the pension model into two pieces:
pendata, which will have relatively fixed data that we never, or almost never, will want to change, and
penmodel, which is parameter driven and sometimes will require us to change underlying data
An important design question is what should be in pendata, what should be in penmodel, and what should be in both?
Here are a few examples:
Base year workforce data -- actual data from the actuarial valuation -- should be in pendata. We're probably never going to change it.
Base year salary data -- same thing
Plan assumptions -- not our alternative assumptions that we may look at, but those used by the plan in its AV, such as discount rate, actual benefit factors, actual COLA rules -- should be in pendata
Policy assumptions -- such as alternative benefit factors, alternative COLA assumptions, etc., should be in penmodel
Return scenarios -- should be read into penmodel
But there are some gray areas. For example, we'll want to have the AV mortality rates in pendata, but we may want to swap out alternative mortality tables in penmodel so that the model can address questions such as what would alternative mortality or mortality improvement assumptions mean to plan funding.
While most choices about whether to put data into pendata, or develop it in penmodel, or both, are pretty clear, at some point we should have a discussion to work out some gray areas.
In my mind I've been breaking the pension model into two pieces:
An important design question is what should be in pendata, what should be in penmodel, and what should be in both?
Here are a few examples:
But there are some gray areas. For example, we'll want to have the AV mortality rates in pendata, but we may want to swap out alternative mortality tables in penmodel so that the model can address questions such as what would alternative mortality or mortality improvement assumptions mean to plan funding.
While most choices about whether to put data into pendata, or develop it in penmodel, or both, are pretty clear, at some point we should have a discussion to work out some gray areas.