The mechanism by which they turned garbage mortgages into AAA was 1) tranching, ie having junior and senior tranches, and 2) the assumption of low correlation (not all borrowers will run into trouble simultaneously).
Arguably, here we have something equivalent to the assumption of low correlation: if a few people here or there dump UST, you can indeed maintain the peg by issuing LUNA against it. But if many people simultaneously do it, then the whole edifice collapses.
> they turned garbage mortgages into AAA was 1) tranching, ie having junior and senior tranches, and 2) the assumption of low correlation (not all borrowers will run into trouble simultaneously)
Even the junior tranches eventually paid. They just weren’t liquid like other AAA assets (namely, Treasuries) in a crisis. The ratings described the tranches mostly accurately in terms of payout probabilities. Not in terms of liquidity.
The correlation assumption is common to these two. But…
> we have something equivalent to the assumption of low correlation: if a few people here or there dump UST, you can indeed maintain the peg by issuing LUNA against it
LUNA and UST were simultaneously issued by the same system. They are fundamentally correlated in a way even American mortgages are not. If the bandwidth of arbitrageurs is broken, the peg fundamentally fails. While one mortgage not paying increases the risk of another not paying, it doesn’t directly cause it in the way we see here.
Also, holders of junk mortgages still got a deed of real property against their loan. What will most Terra/Luna holders get?
> They just weren’t liquid like other AAA assets (namely, Treasuries) in a crisis.
Yes. I think that a big contributor to the crisis was that everyone knew that there was some loss on under-water-real-estate, but due to the complicated nature of the traded products nobody knew where those losses would manifest, and thus trust collapsed across the board.
The mechanism by which they turned garbage mortgages into AAA was 1) tranching, ie having junior and senior tranches, and 2) the assumption of low correlation (not all borrowers will run into trouble simultaneously).
Arguably, here we have something equivalent to the assumption of low correlation: if a few people here or there dump UST, you can indeed maintain the peg by issuing LUNA against it. But if many people simultaneously do it, then the whole edifice collapses.