Looking for any interesting papers/methods about forecasting financial meltdowns.

Dulby · Post by **Dulby** » Tue Nov 13, 2012 7:24 pm

This is more for fun, but there was a good [url=/Show%20Post.aspx?PostIDKey=138074]thread[/url] going about a year ago about methods to predict meltdowns. They are fun to backtest and see how accurate they were. Anything using stochastic vol would be interesting but haven't found much there. Last time around Nonius posted this one which was interesting. It's very long but basically models the connectivity of possible bankruptcies and defaults assuming no Fed. The Hindenberg Omen was mentioned which is fun to look at, especially since it flashed before 08' and doesn't have a bad track record. Wiki Omen

This working paper uses the Q ratio to determine the probability of a future draw down. Warning, It's by Taleb's friend Mark Spitznagel. It has very little math but it still is fun to look at his method historically. It predicts a 20% chance of a more than 40% correction in the S&P over the next few years.

It would be fun to see your guys' favorite methods or papers on the subject. Anything a little more quantitative would be interesting and possibly using stochastic vol like the last thread requested. I would have replied there but it had been dead for over a year.

(first post)

athletico · Post by **athletico** » Tue Nov 13, 2012 8:37 pm

Sornette has been pretty quiet this year... anyone know what's going on with his Financial Crisis Observatory? Last post was almost a year ago: http://fcofbe.blogspot.com/2011/12/new-fbe-candidate.html

sfca · Post by **sfca** » Wed Nov 14, 2012 9:43 pm

There was a paper posted this week on the Fed's website http://www.federalreserve.gov/pubs/feds/2012/201277/201277abs.html called "Credit Spreads as Predictors of Real-Time Economic Activity" that could be interesting to you. It studies the forecastability of recessions based upon bond market credit spread data. But instead of using the I-spread of some such thing, it is based on the distance to default from a Merton model. This paper does not do much with Merton's vol input, but someone who wanted to enhance it could certainly go that direction. You seem interested in stochastic vol as a predictor so that is a logical enhancement. That paper specifically points to financial stress (as opposed to other recession generators like oil price shocks) as being more forecastable with this model. Note, I only skimmed it so I may be misstated some of the results.

In contrast to that approach, I kind of think the market is probably underpricing the current risks of recession or crisis. Because of the money being pumped into the system by the Fed and the international capital flows, both riskless and risky interest rates probably have lost some informational content. I also like Minsky's work on financial fragility in which there are scenarios where lower spreads can actually lead to higher risk of crisis.

Cheng · Post by **Cheng** » Thu Nov 15, 2012 9:08 am

KMV based their model on the distance to default approach. I remember vaguely that they did some research on good predictors of upcoming default. DTD had the highest explanatory power so there could be some benefit following this route.

ig · Post by ig » Thu Nov 15, 2012 11:07 pm

whether "financial meltdowns" and default (on bond) here means the same?

Dulby · Post by **Dulby** » Fri Nov 16, 2012 9:40 am

I would have to agree. But thanks this was more or less what I was looking for. Sometimes I like the approach where they just pile on as many factors as they feel necessary. Happens more in econometrics. Finance tends to be more "pretty".

sfca · Post by **sfca** » Fri Nov 16, 2012 9:17 pm

I agree that financial meltdowns and bond defaults are not the same, but the paper is about creating a measure of risk that allows fundamental user inputs such as vol. The paper does not address bond defaults. The difference between distance to default and bond defaults is what KMV addresses.

ig · Post by ig » Fri Nov 16, 2012 11:57 pm

the paper dealing with Merton's approach to a company default. it is somewhat different than financial meltdown.

Cheng · Post by **Cheng** » Sat Nov 17, 2012 12:18 pm

Pls define "financial meltdown".

ig · Post by ig » Sat Nov 17, 2012 1:37 pm

for me subjectively financial meltdown does not a formal notion. it usually used for a country with finanacial problem like resession or crisis.

If we compare countries government bond spreads it looks better if we establish common underlying currency. otherwise difference between spreads does not look transparent. it can be used USD or EUR or something more universal. Given such adjustment we could guess what level of the currency exchange of the tested country admits 0-default government interest rate