FSA: 50% house price falls were a reasonable scenario

Exocet

chucknorris wrote: »

50% falls were the extreme parameter that they tested their financial model to. When you stress test something you test it to extremes, not to what you think it will reasonably go to. So to say that 50% was a reasonable scenario is not correct.

If you were stress testing a model of a dam or sea wall would an engineer merely test it take the forces of what he reasonably expected it to take in the natural environment, the answer is no, he would test it much further.

Hmmm. I almost agreed with this but am not entirely sure on reflection. You can't ask the banks to stress test their business beyond what is reasonable otherwise they would almost certainly fail - or never loan money out again. So perhaps (and I am not 100% on this), perhaps the stress tests have to be 'reasonable' for that er reason.

Whereas the wall idea is a different kettle of fish maybe.

chucknorris

Exocet wrote: »

Hmmm. I almost agreed with this but am not entirely sure on reflection. You can't ask the banks to stress test their business beyond what is reasonable otherwise they would almost certainly fail - or never loan money out again. So perhaps (and I am not 100% on this), perhaps the stress tests have to be 'reasonable' for that er reason.

Whereas the wall idea is a different kettle of fish maybe.

I don't see the difference between stress testing a financial or sea wall model, both are merely hypothetical tests, the real valuable information would arise from the resultant forces required to bring it to the brink of destruction. But your scope of force needs to be large enough to ensure that it is well beyond reasonably anticipated forces.

Lets face it some banks did actually fail

chucky

Exocet wrote: »

Hmmm. I almost agreed with this but am not entirely sure on reflection. You can't ask the banks to stress test their business beyond what is reasonable otherwise they would almost certainly fail - or never loan money out again. So perhaps (and I am not 100% on this), perhaps the stress tests have to be 'reasonable' for that er reason.

Whereas the wall idea is a different kettle of fish maybe.

he's actually right - the point of stress testing is to two-fold.

1. to see the resistance behaviour of a system/process at a pre-determined level (in this case they belioeve 50%).

2. to see the consequences of a system/process at the pre-determined level if the system has broken. (in this case 50%)

apparently the stress tests on the US banks last year were not as tough as they could have been so the results didn't tell us much.
the 50% level looks like that it will provide much more useful data. if it doesn't they may take it to 60% next time or lower it to 40%.

i would have preferred some sort of functional testing as it would provided more specific info.

Exocet

chucknorris wrote: »

I don't see the difference between stress testing a financial or sea wall model, both are merely hypothetical tests, the real valuable information would arise from the resultant forces required to bring it to the brink of destruction.

Ok then, there is a limit to how far I want to be right, and it is not enough to make me want to read any associated articles.

Really2

Exocet wrote: »

You can't ask the banks to stress test their business beyond what is reasonable otherwise they would almost certainly fail

That is the idea, they cover their selves so they don't fail. You always stress test something to an extreme limit so it it stressed.
If they tested it to a reasonable amount it would not be a stress would it.

chucknorris

Exocet wrote: »

Ok then, there is a limit to how far I want to be right, and it is not enough to make me want to read any associated articles.

I'm glad to hear that, I haven't the time to make this a dissertation subject

Really2

chucknorris wrote: »

5
If you were stress testing a model of a dam or sea wall would an engineer merely test it take the forces of what he reasonably expected it to take in the natural environment, the answer is no, he would test it much further.

The twin towers is a good example they only stress tested for the largest plane when they made. (they believed that was the greatest stress for them then)

They had no idea of modern jets and fuel loads they now carry, if they did they would have stress tested for them.

Exocet

Really2 wrote: »

That is the idea, they cover their selves so they don't fail. You always stress test something to an extreme limit so it it stressed.
If they tested it to a reasonable amount it would not be a stress would it.

Oh go on then, I'll argue a little bit more. If you design a stress test for a bank, and they fail the stress test would that have an impact in their share price, perhaps even the wider financial sector? Therefore would you be inclined to make the stress test 'reasonable' rather than 'extreme' in order to avoid upsetting things?

Please don't do a really/graham on me, I haven't had tea yet.

Really2

Exocet wrote: »

Oh go on then, I'll argue a little bit more. If you design a stress test for a bank, and they fail the stress test would that have an impact in their share price, perhaps even the wider financial sector? Therefore would you be inclined to make the stress test 'reasonable' rather than 'extreme' in order to avoid upsetting things?

Please don't do a really/graham on me, I haven't had tea yet.

AFIK they would have to bring their balance sheet in line to cope with the test.
It would most probably low their share price, possibly a rights issue to bolster it's cash reserves.

If you made it reasonable, it would worry investors should something happen the bank would fail would it not?

chucky

Exocet wrote: »

Oh go on then, I'll argue a little bit more. If you design a stress test for a bank, and they fail the stress test would that have an impact in their share price, perhaps even the wider financial sector? Therefore would you be inclined to make the stress test 'reasonable' rather than 'extreme' in order to avoid upsetting things?

it depends what interest rates and market prices etc your test model used and what results you were expecting - ie bond markets, equity markets etc... i doubt if they would be able to re-create a viable test model - otherwise we would be able to predict markets exactly.

i think it's all very subjective