Abstract
This paper investigates if the recursive detrending method that works well for linear unit root tests also provides good outcomes for nonlinear unit root tests. It is found that the method improves the power of the nonlinear test when only a non-trending mean needs to be removed. The test, however, is no longer performing well if removal of a deterministic trend is required.
| Original language | English |
|---|---|
| Pages (from-to) | 2028-2036 |
| Number of pages | 9 |
| Journal | Economics Bulletin |
| Volume | 30 |
| Issue number | 3 |
| Publication status | Published - 2010 |