According to Estrella and Mishkin (1998), recession predictions can be made directly from financial indicators. This thesis aims to implement a new method of forecasting output growth for large economies, using information from the entire yield curve. After identifying the two common factors for the U.S. economy using a factor approach and detecting the relative power between the FAVAR model (a factor-augmented vector autoregressive model) and the basic VAR model, the predictive results of the factor model are shown to be more precise than the direct forecast of the GDP growth as expected.