ALM vs Capital Model

[Nicholas.Campbell]

Stupid question I'm sure, but exactly is the difference between the two? My interpretation is they are the same thing except possibly a capital model focuses more the extreme outcomes and can allocate risk? But the notes go on to say that an ALM can be used to model capital also!

To me the notes uses the term 'ALM' when talking about determining investment strategy and 'Capital Model' when talking about determining capital requirements - but essentially they are the same. This confused me in CA1 too.

Thanks,

Nick

 

[Hemant Rupani]

Normally statistical models, whether stochastic or deterministic, differ because of their different purpose and xyz constraints.

Now to ALM- it can also generate extreme outcomes by stress test, and it also outcomes some risks(i.e. Market, credit, liquidity etc). It can also provide, say, 10000 simulation of output, so VaR, expected value etc can be measured.

Now, what risk-based capital requirements can we generate from ALM?
From ALM we can get simulated Balance sheet as a outcome. So we can apply any risk measure to determine risk-based capital.
But there are some risks(like premium risk, reserving risk etc) that cannot be measured by ALM, so we need to set other model for this. For eg, we can do by setting density function for attritional, large separately(for both frequency-severity, if detailed analysis needed).

 

[Nicholas.Campbell]

Normally statistical models, whether stochastic or deterministic, differ because of their different purpose and xyz constraints.

Now to ALM- it can also generate extreme outcomes by stress test, and it also outcomes some risks(i.e. Market, credit, liquidity etc). It can also provide, say, 10000 simulation of output, so VaR, expected value etc can be measured.

Now, what risk-based capital requirements can we generate from ALM?
From ALM we can get simulated Balance sheet as a outcome. So we can apply any risk measure to determine risk-based capital.
But there are some risks(like premium risk, reserving risk etc) that cannot be measured by ALM, so we need to set other model for this. For eg, we can do by setting density function for attritional, large separately(for both frequency-severity, if detailed analysis needed).

Thanks, but I'm still a bit confused. We need both premiums and claims for an alm. If we model them stochastically, premium risk and reserve risk could be easily obtained as well as overall risk.

Or are you saying an Alm wouldn't model premiums and claims stochastically, only market variables?

Thanks, Nick

 

[Hemant Rupani]

Thanks, but I'm still a bit confused. We need both premiums and claims for an alm. If we model them stochastically, premium risk and reserve risk could be easily obtained as well as overall risk.

Or are you saying an Alm wouldn't model premiums and claims stochastically, only market variables?

Thanks, Nick

Hi Nick, I know ALM model premium and cashflow.

But, we can't use it for measuring insufficiency of premium charge or insufficiency of reserve held. Because, in ALM, we consider only cashflows of Assets and Liabilities.
(you can refer section 10: Risks of Chapter 18: Investment principles and AL matching, to see the risks relating to the investment strategy)

we need to remove the investment income(can't be treated in premium risk) and remove baseline profit(to make it in realistic basis) to measure premium risk. Now you can see the model is no more AL model. same apply for reserving risk capital measure.