Headspace

Conferences are simply exhausting activities - too many ideas from the program activities and from talking with other participants and not enough processing power. I'm quite glad the conference is short.

Behavioral Heterogeneity in the Option Market
The methodology for this paper is very interesting. The author tries to check for the presence of fundamentalists, who believe in a mean reverting conditional volatility, and chartists, who trade on noise. The interesting part of this is that the author considers that the fundamentalists and chartists will switch depending on the perception of profitability of either, even if there is no real profitability improvement (fundamentalists may not be rational agents). The percentage of switching of fundamentalist to chartists is thought of as similar to a time varying behaviour and the GARCH model is applied to estimate the presence.

Market Conditions, Default Risk and Credit Spreads
Credit spreads sizes were regressed on macroeconomic variables such as growth rate, reccession, investor confidence, systemic risk volatility (which I think systemic jump risk is meant) and the cash flow of the company. ("Specifically, we show that, during economic expansions, firms with high cash flow betas have lower credit spreads, ceteris paribus, than firms with low cash flow betas. This relation reverses during economic recessions.") The discussant was openly critical that the paper is an indirect test of the author's earlier model and was almost dismissive of the paper, saying things like he was impressed with how the author got the difficult to obtain data and that the paper was thoughtful and nice. (?!)

Estimating Asset Correlations From Stock Prices or Default Rates: Which Method is Superior?
All the hard work in simulation is lost on me but I recalled the answer is (A).

The Economic Value of Volatility Timing Using a Range-based Volatility Model
The author pitted Engle's return's based DCC (where volatility is measured with day end observation) with his own range based DCC (where volatility is observed as a range over the entire trading day) by estimating the switching cost to achieve a target expected return with a portfolio of bonds, cash and equity. The range DCC was more efficient (costs less) but perhaps that Engle was in the audience (in front of me!), the author said that the difference in cost between the DCC was not great so it could mean that the contribution of range DCC is not very large.

Anticipating Correlation
Engle introduced three models Factor Arch, Factor Double Arch and Dynamic Conditional Correlations, which is a short cut method he devised to compute vast portfolio correlations. A guy behind me at the coffee break exclaimed to another that he was inspired and has an idea that he would go back to work on and I wondered why he would say it because DCC is not revealed to the world in this conference. I was not excited.

When To Sell A Stock, If You Must
I tried but failed to find a copy of the paper for this talk. The author walks the audience through his mathematical model what a profitable strategy is as long as there is transaction cost to trading. He was very thorough with his assumptions: investor will hold as long as it is profitable and will sell as long as it is close to the maximum price or when maturity is due. The math bit is hard even though he kept saying it is simple. He arrived at the idea that as long as it is a good stock (measured by a particular index he derived) one should buy and hold until one needs the money because the good stock will provide the investor with income and there is only a 10% error that the price will be more than the maximum and not sold.

Information Shocks, Jumps, and Price Discovery -- Evidence from the U.S. Treasury Market
Her paper was highly praised by the discussant for incorporating new technology in measuring price jumps and assessing the jumps and tying the jumps back to the news release. However, the stronger finding is that liquidity shocks is better at predicting jumps. Intuitively, the finding is unsurprising as the data tested is the Treasury Bond prices and I would expect news and liquidity shocks to create jumps. I am quite glad for the summary provided by the discussant as I had trouble hearing the author.

Tranching and Rating
Don't get it but very interesting. It seems that slicing up a credit product into tranches, the seller can make more money than not slicing it up. The key to thinking about this is the deceptiveness of ratings and, I suspect, the inability to correctly value these tranches. Bears reading in full.

Nested Simulation in Portfolio Risk Measurement
I should have gone to the other one.