This dissertation consist of two main parts. The first part empirically studies the standard consumption-based asset pricing model of Lucas (1978), Breeden (1979) and the consumption-based asset pricing model with recursive utility of Weil (1989) using quarterly data from Thailand from 2000 to 2016. The equity premium is calculated using FEDR returns, which includes dividend and right benefits and considers the effect of stock split. The key advantage of FEDR returns over the SET total returns is that the former returns are available since April 1975 while the latter is available only after January 2002. The result confirms that the models, which are the stan-dard consumption-based asset pricing model of Lucas (1978), Breeden (1979) and the consumption-based asset pricing model with the recursive utility of Weil (1989), assuming consumption growth are iid and Markov processes, cannot explain observed expected equity premium and observed expected risk-free return in Thailand data. Thailand's quarterly financial data from 2000 to 2016 exhibit the equity premium and risk-free rate puzzles. The second part shows that the long-run risk model (Bansal and Yaron, 2004) can solve the equity premium puzzle and risk-free rate puzzle in Thailand based on the data between 2000-2016. The calibration result gives a plau-sible value of risk-aversion which is less than 3 and a plausible value of elasticity of intertemporal substitution which is in a range between 1 and 2. In addition, the generalized method of moments (GMM) estimation results confirm that consump-tion growth and dividend growth are long-run processes, driven by a persistent latent variable and time-varying economic uncertainty. These together suggest that news about consumption growth and volatility have significant impact on asset prices. In addition, robustness checks for all models are performed using a diagnostic test, called Hansen-Jagannathan bound.