This dissertation investigates computational methods for macroeconomic asset pricing
models. It demonstrates that advances in economic modeling often require advances
in computation and highlights a particular case where more demanding computational
methods are required to solve an economic model. It also discusses advances
in computational technology that allow researchers to utilize solution methods
that would have been previously infeasible. In particular, it demonstrates the wide
applicability and potential gains of GPU computing, a parallel computing framework,
and applies those tools to a computationally challenging model which investigates
trading volume in a general equilibrium, complete-markets economy where agents
have heterogeneous beliefs.