Held in the charming Quebec ski resort of Mont-Tremblant, the Third McGill/IFM2 Risk Management Conference took place from March 12 to 14, 2010. In addition to the rather good ski, a set of rather interesting papers were presented; among them, the following piqued my interest:

1. “Components of Bull and Bear Markets: Bull Corrections and Bear Rallies” by John Maheu, Tom McCurdy, Yong Song. The authors employ a Markov switching model to separate out bull and bear markets. In particular, they identify four states: bull market, bull correction, bear market, and bear rally. The interesting point of the paper is their insight to incorporate economic restrictions into the HMM parametrization: the transition matrix is appropriately constrained, and the emission distributions each state must obey:
   • Bull market state: mean return > 0
   • Bull market correction: mean return < 0
   • Bear market state: mean return < 0
   • Bear market rally: mean return > 0

   The estimation proceeds in a full Bayesian manner, with Gibbs sampling applied throughout. A private discussion with the authors indicates that the Gibbs chain mixes very quickly, which is probably due to the strong constraints on the parameters induced by the economic restrictions. All in all, an intuitive and tantalizing model.

2. “A Multifrequency Theory of the Term Structure of Interest Rates” by Laurent Calvet, Adlai Fisher, Liuren Wu. This paper introduces a no-arbitrage term structure model that is capable of fitting instantaneously-observed rates perfectly (which is important in many applications), yet provides realistic evolution dynamics. Its most interesting feature—from my point of view—is to propose a low-dimensional representation of a large number of time series, each operating at its own time scale, and that are superposed to yield the final forecast. The individual time series are linked through common parameters, and the time-scale dependence stems from a power law. This decomposition appears useful for a large number of applications in time-series modeling beyond term-structure of interest rates.
3. “Multi-Period Forecasts of Volatility: Direct, Iterated, and Mixed-Data Approaches” by Eric Ghysels, Antonio Rubia, Rossen Valkanov. This paper follows on the previous MIDAS literature to compare various paradigms for long-horizon forecasting of volatility:
   • The “direct” approach tries to have a model predict at horizon N given N-period returns
   • The “iterated” approach iterates N times a one-period forecast
   • The “mixed-data” approach combines a set of one-period returns to yield directly a N-period forecast.

   The authors’ model falls in the latter category, and appears to beat both the direct and iterated approaches on US stock market portfolios, using a cross-section of size, book-to-market and industry portfolios.

4. “Variance Risk Premia, Asset Predictability Puzzles, and Macroeconomic Uncertainty” by Hao Zhou (speaking for himself and not the Federal Reserve Board). One of the wonders of option markets is that implied volatility is generally higher than the subsequent realized volatility—in other words, implied volatility is usually an upwards-biased predictor of realized volatility. This explains why option writing strategies can be so profitable. This phenomenon, called the volatility premium, has been characterized in depth by Bjørn Eraker. However, it remained some mystery as to the economic drivers of this premium. This paper tries to explain it via economic uncertainty (which is incorporated via a general equilibrium model). It also examines the implications for risk premia across equity, bond and credit asset classes.
5. “What Ties Return Volatilities to Price Valuations and Fundamentals?” by Alex David, Pietro Veronesi. It has long been observed that stock price volatility is much higher than can be explained by volatility in the fundamentals. The paper opens with a citation by Engle and Rangel:
   

   “After more than 25 years of research on volatility, the central unsolved problem is the relation between the state of the economy and aggregate financial volatility. The number of models that have been developed to predict volatility based on time series information is astronomical, but the models that incorporate economic variables are hard to find. Using various methodologies, links are found but they are generally much weaker than seems reasonable. For example, it is widely recognized that volatility is higher during recessions and following announcements but these effects turn out to be a small part of measured volatility.” [Engle and Rangel (2008)]

   This paper attemps to shed more light on this issue: they introduce a model where investors learning about unusual fundamental states leads to a V-shaped relationship between volatilities and prices.

6. “Credit Conditions and Expected Stock and Bond Returns” by Sudheer Chava, Michael Gallmeyer, Heungju Park. This paper introduces a variable that appears predictive of stock returns at quarterly horizons. It is computed from a survey of credit standards administered since 1967 by the Federal Reserve to bank senior loan officers (although the data contains a big whole during the 1980’s, corresponding to changes in the survey methodology). The variable seems to introduce fundamentally new information not already captured by the standard predictors, and is robust to a variety of in-sample and out-of-sample tests.
7. “Option Trading: Information or Differences of Opinion?” by Siu Kai Choy and Jason Zhanshun Wei. This was a last-minute addition to the conference and does not appear in the program. The authors investigate whether option traders have special information or they are merely speculators. They provide evidence that in aggregate, option traders are not particularly well informed, and therefore the “opinion difference” hypothesis (i.e. speculation) probably drives most option trades.