This dissertation consists of two chapters, all of which attempt to shed some light on what constitutes the time-varying risk premia in financial markets. The first chapter demonstrates that monetary policy shocks identified from New-Keynesian dynamic stochastic general equilibrium (DSGE) models explain the risk premia in stock markets. Indeed, the implied ICAPMs explain the value and the industry premia for the periods of 1980 to 2004. In particular, the permanent monetary policy shocks to inflation target capture the value premium and part of industry risk premium once I account for the capital market imperfection endogenously in New-Keynesian models. The shocks to investment technology, as a main determinant of the external finance premium, are also important for understanding the value premium.

The second chapter examines determinants of stochastic relative risk aversion in conditional asset pricing models by utilizing nonlinear state space model with GARCH specification. After imposing general version of the conditional CAPM or ICAPM, I develop non-ad-hoc empirical models and search for valid specifications of relative risk aversion along with appropriate hedging components. I discover that the surplus consumption ratio implied by the external habit formation model is the most important determinant of time varying relative risk aversion. The CAY of Lettau and Ludvigson (2001a) without a look-ahead bias also captures part of relative risk aversion. The short term interest rate(RREL) has explanatory power for hedging components. I use the implied conditional asset pricing models in explaining the cross-section of average returns on either the Fama-French 25 size and book-to-market sorted portfolios alone or with 30 industry portfolios. I find that the chosen conditional CAPM and ICAPM with time-varying relative risk aversion and a hedging component are at least comparable to or better than the Fama-French three-factor model for the sample periods 1957 to 2005.