I understand that the Gaussian Copula model treats asset spreads and correlation as separate inputs–that the levels of asset spreads are independent of the correlation between asset spreads. I reject this assumption on common sense grounds. There is feedback both ways. For some entities that are particularly interconnected to the financial system, their credit risks are high precisely because they are especially susceptible to correlation. From a macro view, if a large number of interconnected entities were to suddenly default simultaneously, the conditional expected recovery would be tiny.

My statement “Assume it’s worthless and bid $0?” was hyperbole. You lost me at “the level of credit spreads can be ignored…”

So, if the problem is insolvency, the Geithner plan can only work if the bidders overpay for the assets and the loss is taken on by the government. On the flip side, the banks who were insolvent, who got overpaid through the auction, get all the upside. Thus, Krugman’s thinking is that, no matter what, the government has to take the losses (either through nationalization or through overpayment). If that is the case, why not keep the upside?

Hopefully, you won’t mind unpacking why you think this is spurious on Krugman’s part (or at least where y’all disagree)?

“If systemic risk is reduced, credit spreads must tighten. Said differently, systemic risk is not defined as correlation.”
in credit markets, high correlation is an indicator of systemic risk, and low correlation of idiosyncratic risk, this is not correlation as it is generally understood, it refers to the relative level of tranche credit spreads in a closed system such as a CDO, for the underlying assets credit spreads, high correlation means the dispersion of the spreads is narrower than when correlation is low. It doesn’t follow that because correlation drops, asset credit spreads tighten, they don’t have to, instead what happens is that the dispersion of asset credit spreads widens. So to go back to your question in your first comment “Did I read that right?” no, sorry, it is implicit in the maths of correlation that the dispersion of asset credit spreads tightens/widens irrespective of the change in the average level of asset credit spreads. and diminished systemic risk leads to lower correlation not higher (your last comment which i assume is a misprint).

“As a bidder, assume the worst for your position…”
Assume it’s worthless and bid $0?
we deal in probabilities, in competitive markets or auctions, if you are certain that 15% of the assets will default then you bid the discount value of the coupon flow until the expected time to default, 0 is the price only for instantaneous default. the assets have some book value now and to get a margin of safety, I would discount it based on some scenario of what could happen in the future and on what i think in my source of risk, right now there is no market for toxic assets so you have to make one…my view is that I would assume future low correlation + lower volatility reducing the value of the put option on the non-recourse loan and price accordingly. the level of credit spreads can be ignored, reason being that the assets are bought, not for sale and funded to term, and whatever spreads over funding and guarantees is locked to term, what matters for the risk of the equity position is correlation i.e the dispersion of credit spreads in the asset pool, not the level.

OK (maybe). But how is that germane to the question I posed? I asked about the all-in state of the world conditional upon diminished systemic risk. Higher correlations may be a predictable outcome thereof, but I posit that it is not the only outcome. If systemic risk is reduced, credit spreads must tighten. Said differently, systemic risk is not defined as correlation.

“As a bidder, assume the worst for your position…”

Assume it’s worthless and bid $0?

The logic here seems quite densely constructed to my slow-working brain. If I try to deconstruct it, I think what you’re saying is that current prices of equity tranches reflect a high correlations relative to a “normal” state due to current greater-than-normal systemic risks. However, if the Geithner plan is a success, then systemic risks will be mitigated and correlations will come down. Equity tranches being long correlation, the prices thereof would be expected to drop in the state of the world where the Geithner plan succeeds, all else equal. Thus, a rational bidder would conservatively assume the success of the plan (lower correlations) and thus place a low bid consistent with this success. Did I read that right?

Assuming I got that right (big assumption, I admit), I’m not sure I agree with the “all else equal” part. I have to ask, would a rational investor not also consider the other effects of a state of the world where the Geithner plan succeeds? I would think that if Treasury were to succeed at curbing systemic risk, not only would correlation normalize, but so too would credit spreads. I’m not smart enough to predict the direction or magnitude such a change would have on a prototypical equity tranche. But I would argue that by the time the private investor got around to submitting a bid, other market participants would have already figured out how to incorporate the likelihood of success into both the correlation surfaces and credit curves.

AIG’s financial products group could get a good bit of profitable business selling default insurance. The mind boggles!