Abstract
We propose two new parametric tests for an unknown threshold in models with endogenous regressors. They are both based on unconventional 2SLS estimators
that use additional information about the linearity of the first stage. This information leads to more accurate residuals and therefore tests with better size properties than the Wald GMM test in Caner and Hansen (2004), which we show exhibits severe size distortions in small samples pertinent to empirical applications. We prove the bootstrap validity of our tests and evaluate their empirical relevance by revisiting the question whether government spending multipliers are larger in recessions. As Ramey and Zubairy (2018), we cannot rule out that they are the same in recessions or expansions.
that use additional information about the linearity of the first stage. This information leads to more accurate residuals and therefore tests with better size properties than the Wald GMM test in Caner and Hansen (2004), which we show exhibits severe size distortions in small samples pertinent to empirical applications. We prove the bootstrap validity of our tests and evaluate their empirical relevance by revisiting the question whether government spending multipliers are larger in recessions. As Ramey and Zubairy (2018), we cannot rule out that they are the same in recessions or expansions.
Original language | English |
---|---|
Place of Publication | Tilburg |
Publisher | CentER, Center for Economic Research |
Number of pages | 32 |
Volume | 2019-030 |
Publication status | Published - 31 Oct 2019 |
Publication series
Name | Center Discussion Paper |
---|---|
Volume | 2019-030 |
Keywords
- 2SLS
- GMM
- threshold
- bootstrap
- government spending