Economic Capital

[secondlife]

Its look like the book is ($63) is as dry as the study note.

What is EC? The VAR (this was a brand new term as statisticians did not know what confidence intervals were like the Sharpe Ratio/coefficient of variation) was a failure. Instead of coming up with a measure for one portfolio/division, come up with a measure for the whole company considering correlations among the divisions/portfolios. I am not sure they mentioned anything about fat tails/rare events in it. So basically EC is a wholesome approach.

Do I have it right E?

[E]

Originally Posted by secondlife

Its look like the book is ($63) is as dry as the study note.

What is EC? The VAR (this was a brand new term as statisticians did not know what confidence intervals were like the Sharpe Ratio/coefficient of variation) was a failure. Instead of coming up with a measure for one portfolio/division, come up with a measure for the whole company considering correlations among the divisions/portfolios. I am not sure they mentioned anything about fat tails/rare events in it. So basically EC is a wholesome approach.

Do I have it right E?

My sense, and this is echoed repeatedly in that Basel Committee paper I linked to, is that VaR was widely used because it's easy to calculate and explain. However, it's now generally regarded as too weak compared to alternatives. It didn't hold up well during the crisis, or more generally any time tails get very long and non-normal.

[secondlife]

Originally Posted by E

My sense, and this is echoed repeatedly in that Basel Committee paper I linked to, is that VaR was widely used because it's easy to calculate and explain. However, it's now generally regarded as too weak compared to alternatives. It didn't hold up well during the crisis, or more generally any time tails get very long and non-normal.

I am not criticizing/arguing with you, but easily explain part was the problem.That committee approved "VAR", would you expect them to say "we intentionally misrepresented facts and Taleb was right?"

[E]

Going on a slight tangent, I think it's unfortunate that Taleb's name is now synonymous with the black swan concept. The concept certainly existed before his book. But I think now people view these things as being black and white or in this camp or that camp.

It causes people avoid adopting certain approaches because of political side-taking. Both sides have merits.

There is no right or wrong approach in this kind of modeling. All approaches are fundamentally wrong. We're paid to constantly look for the least wrong.

[secondlife]

Originally Posted by E

There is no right or wrong approach in this kind of modeling. All approaches are fundamentally wrong. We're paid to constantly look for the least wrong.

All models are wrong, but if your model is less wrong than your competitor's/buyer's model you will make money. I am sure somebody must have said this.

[Paintball Wizard]

Originally Posted by E

I think it's unfortunate that Taleb's name is now synonymous with the black swan concept.

Taleb =

[Paintball Wizard]

Originally Posted by secondlife

All models are wrong, but if your model is less wrong than your competitor's/buyer's model you will make money. I am sure somebody must have said this.

No, if your model is simple, yet conclusive (i.e., not a black box and not overly sensitive to inputs or changes), then people (i.e., clients or senior mgmt) will keep asking for your model. There honestly is no right/wrong with high severity but low frequency (i.e., the 1/200) events. When it hits the fan, what you said will never be enough; but if you really held enough for that time, you wouldn't have a going concern.

[secondlife]

Originally Posted by Paintball Wizard

No, if your model is simple, yet conclusive (i.e., not a black box and not overly sensitive to inputs or changes), then people (i.e., clients or senior mgmt) will keep asking for your model.

Conclusive models in Finance/Insurance/Social Sciences? May I ask you to read E. Derman's writing? Taleb never said that he is an expert of "black swan" prediction and he is certainly not the first one to talk about rare events. But the people (mainly news paper columnist) who peddle Taleb's writing imply that.

[diegol]

Originally Posted by E

This large paper requires a lot of coffee to get through, but it's chock full of the stuff:

http://www.bis.org/publ/joint25.pdf

(I know because I outlined it for the APM course I instruct.)

Thanks for this!

I was looking for a more academic take on EC (ie, a paper describing concepts from scratch rather than practical applications of EC). So after a few days now after my first post, I think I'll start with the second paper I mention there.

[secondlife]

Solvency valuation might be an alternative.