Actuarial analysis of subprime mortgages and insurance companies

[Indy Coug]

If you've got a couple of minutes, here's some light reading.

http://www.soa.org/files/pdf/research-2009-fin-crisis.pdf

[UtahDan]

If only we knew an actuary who could digest it for us and give a summary of the highlights.

[Indy Coug]

If only we knew an actuary who could digest it for us and give a summary of the highlights.

Read the seven page executive summary.

[Indy Coug]

Here are a couple of key points

Eight, models can create a false sense of comfort. Managers must be alert to the assumptions that go into models and the limitations of model results due to these assumptions. It is critical to challenge the assumptions and subject them to stress tests. For example, the recent crisis highlights the value of independent credit risk assessment.

Nine, stress testing needs to be more dynamic and robust by incorporating a
rich variety of economic scenarios, as well as explicitly considering a company’s own rating downgrades, counterparty rating downgrades, the failure of liquidity suppliers, and increased correlations in asset returns, between products, and across different business lines or business units during times of distress.

[beelzebabette]

Here are a couple of key points

This is the part I found most interesting:

We argue that the primary cause of the crisis lay in the widely held belief that housing prices could not decline significantly on a national basis. This optimistic belief was shared by policymakers, economists, and market participants in general, permeated the models used by rating agencies to assign inflated ratings to securities built from subprime mortgages, and was reinforced, for a time, in market prices through a self-fulfilling prophecy.

[I.J. Reilly]

This is the part I found most interesting:

Wait, are you indicating to me that markets can be irrational???

Whoa.

[Indy Coug]

This is the part I found most interesting:

Another very important point. To this point, I'm unaware of any of the ratings agencies being held to the fire their gross oversight or complicity in this scandal. Anyone else know?

Despite that, that doesn't absolve the insurance industry or anyone else that were major purchasers of these securities for not questioning the quality of the assets they held and for not understanding and sufficiently hedging against the associated risk, particularly when they were assumed to be almost "risk free".

[beelzebabette]

Wait, are you indicating to me that markets can be irrational???

Whoa.

It's the chain reaction from that single assumption, like Indy highlights with his excerpts. I know the fallout from the faulty ratings has had a significant impact on my employer, e.g.

[beelzebabette]

Another very important point. To this point, I'm unaware of any of the ratings agencies being held to the fire their gross oversight or complicity in this scandal. Anyone else know?

I haven't heard of anything. I'm interested to see if anything happens. I'd like to know eventually as well if there wasn't a knee-jerk overreaction dropping most corporate bond ratings afterward as compensation for bad analysis, or if the drop was justified.

[statman]

Here are a couple of key points

Models absolutely can be misleading. The problem is, they're usually right, and ignoring them is usually a mistake.

And that brings up an interesting note that I've never seen mentioned about the senior management at the banks at the heart of the crisis of the last 18 months: They are all very familiar with models. But based on when models became a driving force in the financial world (especially risk models) almost all of the senior management are 5-10 years too old to actually have model-building experience. They know what deciles and rank-ordering mean, and they know that there are some vague assumptions about models when they're being built. And they know that htey can 'stress-test' models to see what they predict in different scenarios. BUt they've never done it themselves. They've never made the decision - "okay, I realize that there is some finite risk that there might be a catestrophic collapse of our real estate market, but since I can't really estimate that risk mathematically, I'm just going to ignore it as being inconsequential." they've always had someone do all that work for them - someone who's repeatedly told to "simplify their discussion of models with their superiors." Or to "just tell them what they need to do - don't go into the details of the models." And they've seen that the guys who get ahead in the banks, are the ones who can best "sell" a model, not the guy who develops the best models, and best informs management about the potential pitfalls associated with them.

In other words, the guys making the decisions, know about some potential dangers about using models, but they've surrounded themselves with people who 'get things done' rather than those who actually know what they're doing. They never REALLY understood what was at stake if the model's assumptions proved to be totally wrong.

[statman]

Another very important point. To this point, I'm unaware of any of the ratings agencies being held to the fire their gross oversight or complicity in this scandal. Anyone else know?

Despite that, that doesn't absolve the insurance industry or anyone else that were major purchasers of these securities for not questioning the quality of the assets they held and for not understanding and sufficiently hedging against the associated risk, particularly when they were assumed to be almost "risk free".

Come on Indy - they had a formula to calculate the value of those assets. A FORMULA! And it contained a correlation coefficient! It had to be right...

(Here's a cool article on the formula most responsible for the financial crisis:

http://www.wired.com/techbiz/it/maga...urrentPage=all )

[statman]

they were assumed to be almost "risk free".

They had reasons, beyond the numbers, to assume they were near "risk free." The instruments were backed by bundles of 'compliant' mortgages, presumably backed by the full faith and credit of Fannie Mae and Freddy Mac and the US Government.

Then Lehman filed for bankruptcy, and in a firesale, sold bundles of Mortgage Backed Securities - again, made up of mortgages presumably backed by Fannie & Freddie - for ~25 cents on the dollar.

Not exactly risk free...

[Moliere]

Another very important point. To this point, I'm unaware of any of the ratings agencies being held to the fire their gross oversight or complicity in this scandal. Anyone else know?

They lost some credibility. This was stated on the afternoon news edition of NPR where they discussed the failed ratings and how people are now more vigilant in performing their own analysis of a company's or financial products security (duh!!!).

The problem with the rating agencies is that they are not independent of the entity. If my company wants a rating for a product they have to pay for it. The agencies obviously have some incentive to get it right or no one will believe them, but this lack of independence often times causes them to lag in downgrading a rating.

An independent rating agency would be a better solution, but I have no idea how one could go about setting up such an agency.

[I.J. Reilly]

They lost some credibility. This was stated on the afternoon news edition of NPR where they discussed the failed ratings and how people are now more vigilant in performing their own analysis of a company's or financial products security (duh!!!).

The problem with the rating agencies is that they are not independent of the entity. If my company wants a rating for a product they have to pay for it. The agencies obviously have some incentive to get it right or no one will believe them, but this lack of independence often times causes them to lag in downgrading a rating.

An independent rating agency would be a better solution, but I have no idea how one could go about setting up such an agency.

I was thinking about this yesterday (not particularly too hard, so I may be way off base here) and I thought that if instead of paying money to the rating agencies themselves, those that were required to be rated or those that desired a rating would instead pay into a pool of money and would be randomly assigned one of the 3 ratings agencies. The agencies would be assigned the companies more or less in proportion to the percentage of the ratings that they cover currently, so something like (and this is just a guess) 20% for Fitch, 40% for Moody's and 40% for S&P. This would resolve the issue of forum shopping for a better rating and reduce the pressure on an agency to change a company's rating just to maintain their business. Since between the three of them they already have a monopoly because of their NRSRO status, this would just formally lock them into a specific percentage of the business.

Another option is to have the ratings that are mandated for banks and such be conducted by the SEC and deregulate ratings as much as possible to open up competition and let the agencies compete on their track records, i.e. "when so-and-so company rates a bond, 98.5% of the time that bond performs to expected levels," or something like that.

Anyways, just a couple of thoughts that popped into my head.

[beelzebabette]

I was thinking about this yesterday (not particularly too hard, so I may be way off base here) and I thought that if instead of paying money to the rating agencies themselves, those that were required to be rated or those that desired a rating would instead pay into a pool of money and would be randomly assigned one of the 3 ratings agencies. The agencies would be assigned the companies more or less in proportion to the percentage of the ratings that they cover currently, so something like (and this is just a guess) 20% for Fitch, 40% for Moody's and 40% for S&P. This would resolve the issue of forum shopping for a better rating and reduce the pressure on an agency to change a company's rating just to maintain their business. Since between the three of them they already have a monopoly because of their NRSRO status, this would just formally lock them into a specific percentage of the business.

A counter argument to this is that it reduces the competitive incentive to be most accurate. I know my employer is required to get a Moody's rating as part of its bonding--meaning it's not always the business/entity itself selecting its rating agency.