Abstract
In this paper, we consider both the partial unit root and the near partial unit root processes in nonlinear transition autoregression models. Our simulations show that when these time series data are used in ordinary least squares regression, spurious regression occurs. However, if we re-estimate the regression by adding an AR(1) term, spurious regression can almost be eliminated.
Original language | English |
---|---|
Pages (from-to) | 189-191 |
Number of pages | 3 |
Journal | Economics Letters |
Volume | 118 |
Issue number | 1 |
DOIs | |
Publication status | Published - Jan 2013 |
Keywords
- Monte Carlo simulation
- Partial unit root
- Spurious regression