Through an econometrics model (GARCH-MIDAS), we show that the disagreements in oil price forecasters' opinions can significantly improve predictive accuracy on the volatility of oil returns, providing a meaningful step forward toward a better understanding of the connection between asset prices and survey-based subjective expectations.

The positive R&D-return relation means that more R&D-intensive firms have higher exposure to systematic risk, therefore, R&D investment is positively related to expected stock returns in equilibrium. Firm value is a concave function of R&D success rate. Therefore, uncertainty about R&D success rate amplifies the systematic risk exposure due to R&D investment. Firm's stock market illiquidity measures this uncertainty and, in turn, affects the positive R&D-return relation.

We should definitely incorporate CDS market signals in forecasting corporate default if they are available. Among the various components (liquidity, physical default, and risk premium) in CDS spreads, only the physical default component contributes significantly to default prediction. Therefore a CDS model that can perform such decomposition will further improve default prediction. 

The performance of mean-variance v.s. 1/N has long been a debate topic in the literature of portfolio management and asset pricing. We show theoretically and empirically that the superiority of mean-variance over 1/N is increased when the assets have lower idiosyncratic volatility. The results justify the common practice of a two-stage process in portfolio management: mean-variance for asset allocation and 1/N for stock selection.

The extent of capital forbearance can be quantified more rigorously using option-theoretic techniques and stock market information. Given changing trends in regulation and deregulation, it is important to accommodate time variation in modeling forbearance. Our empirical analysis shows that the forbearance capital starts increasing three years before the start of the 2008 financial crisis.

Theoretically, we show most existing Gaussian dynamic term structure models can be nested as special cases under a unified HJM-based framework. Empirically, our analysis reveals that the existing models impose restrictive constraints limiting their flexibility in capturing key features of the correlations and volatilities of the forward rates.

When monolines' credit is in deep trouble, municipal bonds insured by them could become cheaper than those uninsured (yield inversion)! Illiquidity due to the troubled credit of monolines provides an explanation for this puzzling yield inversion phenomenon. Our model quantifies the credit and liquidity interaction in the context of insured and uninsured municipal bonds that can be attributed to the yield inversion.

Mispricing in term structures of CDS spreads (North America single names) is large and persistent. The volatility of credit and liquidity risks partially explains the dynamics of mispricing, signaling potential market manipulation. 

It is puzzling to observe a negative correlation between credit risk and credit spreads. Evidence did present such a puzzle in the Chinese corporate bond market (from 2006 to 2013) when the Chinese government had “zero tolerance” for corporate defaults. 

The literature on enterprise risk management methodologies is worth reviewing. 

Factor loadings in Gaussian term structure models are model-dependent even for the same factors, while the unconditional contemporaneous impacts (UCI) are factor-specific but model-independent. Constraining UCI in estimating P-measure parameters gives rise to more sensible estimates for market expectations and term premiums.