The gross domestic product (GDP) is a quarterly released key indicator on the state of the economy and is subject to long publication delays. The Bureau of Economic Analysis (BEA) publishes the first estimate of GDP six weeks after the reference quarter. The uncertainty in between the releases stresses the importance to estimate current quarter GDP growth - nowcasting. In this paper I evaluate the real time performance of various econometric approaches to nowcast US GDP growth. In an extensive empirical study I find that the fairly unused methods in nowcasting LASSO and random projection regression overall perform best and are good alternatives to the well-established models in the nowcasting literature. Keywords: Nowcasting, Dynamic Factor Model, Mixed-Data Sampling, LASSO, Random Projection

Additional Metadata
Thesis Advisor Pick, A.
Persistent URL hdl.handle.net/2105/47307
Series Econometrie
Citation
Mostbeck, J. (2019, April 30). Nowcasting US GDP Growth in `Pseudo\' Real Time Using Various Econometric Models. Econometrie. Retrieved from http://hdl.handle.net/2105/47307