quantile nonlinear ardl - bounds testing procedure and critical values
Posted: Mon Mar 30, 2026 9:13 am
Dear EViews community,
Several interesting new features have been added to the 14th version of the software, and hopefully more will be added in future versions.
After watching a video about the quantile ARDL model (QARDL) on the official YouTube channel (see link here https://www.youtube.com/watch?v=VlSRrT0 ... ws&t=4m33s: 4m, 33s), one can realize that the asymptotic critical values of Pesaran, Shin and Smith (2001, PSS) are reported for the bounds testing procedure (BTP) of QARDLs. However, those asymptotic [sample of T = 1000] critical values by PSS are valid only a. for (N)ARDL models estimated with OLS, and b. when the estimation sample is large.
Bertsatos, Sakellaris and Tsionas (2022) show that new critical values are required for the BTP of quantile (nonlinear) ARDL -Q(N)ARDL- models and provide EViews codes that generate new critical values up to 13 regressors. As a result, using the PSS critical values for the BTP in quantile-based error-correction models is misleading. We provide EViews codes in the supplementary material of the paper, and such codes generate critical values for 11 cases [5 standard Cases I-V as in PSS, and 6 new Cases VI-XI with linear and quadratic trend] and for user-selected number of observations. We further provide codes for generating critical values for interdecile and interquartile BTP for 3 cases [Case III, V, XI] to explore whether there is evidence of time-varying estimates or quantile-dependent co-integration.
Indicatively, one suggestion for the EViews Team could be the following: instead of reporting the asymptotic critical values of PSS [T = 1000], you could run our code for T = 1000 and standard tau values [e.g. for the 9 deciles, or τ = 0.25 and τ = 0.75] and use these values in the EViews output of QARDL BTP.
Next, Bertsatos, Sakellaris and Tsionas (2023) propose a panel bounds testing procedure for dynamic fixed effects (DFE) models in large panel datasets, extending PSS (2001) and BST (2022), and provide an EViews code in the Appendix of the paper. This code generates sample-specific and lag-specific critical values for 11 Cases and up to 13 regressors. Specifically, the user can select the number of cross sections [N = 1 indicates time-series and can be seen as an extension of the previous BST codes] and time periods, the number of regressors and number of lags in the estimated (N)ARDL, the distribution of the error term, and the number of simulations [default and minimum number is 2000].
I hope that EViews users and developers will find our papers useful for future projects and papers, as well as for incorporating new features of the ARDL model family into EViews.
Kind regards,
Georgios Bertsatos
Research Fellow at the Centre of Planning and Economic Research (KEPE), Athens, Greece
References
Several interesting new features have been added to the 14th version of the software, and hopefully more will be added in future versions.
After watching a video about the quantile ARDL model (QARDL) on the official YouTube channel (see link here https://www.youtube.com/watch?v=VlSRrT0 ... ws&t=4m33s: 4m, 33s), one can realize that the asymptotic critical values of Pesaran, Shin and Smith (2001, PSS) are reported for the bounds testing procedure (BTP) of QARDLs. However, those asymptotic [sample of T = 1000] critical values by PSS are valid only a. for (N)ARDL models estimated with OLS, and b. when the estimation sample is large.
Bertsatos, Sakellaris and Tsionas (2022) show that new critical values are required for the BTP of quantile (nonlinear) ARDL -Q(N)ARDL- models and provide EViews codes that generate new critical values up to 13 regressors. As a result, using the PSS critical values for the BTP in quantile-based error-correction models is misleading. We provide EViews codes in the supplementary material of the paper, and such codes generate critical values for 11 cases [5 standard Cases I-V as in PSS, and 6 new Cases VI-XI with linear and quadratic trend] and for user-selected number of observations. We further provide codes for generating critical values for interdecile and interquartile BTP for 3 cases [Case III, V, XI] to explore whether there is evidence of time-varying estimates or quantile-dependent co-integration.
Indicatively, one suggestion for the EViews Team could be the following: instead of reporting the asymptotic critical values of PSS [T = 1000], you could run our code for T = 1000 and standard tau values [e.g. for the 9 deciles, or τ = 0.25 and τ = 0.75] and use these values in the EViews output of QARDL BTP.
Next, Bertsatos, Sakellaris and Tsionas (2023) propose a panel bounds testing procedure for dynamic fixed effects (DFE) models in large panel datasets, extending PSS (2001) and BST (2022), and provide an EViews code in the Appendix of the paper. This code generates sample-specific and lag-specific critical values for 11 Cases and up to 13 regressors. Specifically, the user can select the number of cross sections [N = 1 indicates time-series and can be seen as an extension of the previous BST codes] and time periods, the number of regressors and number of lags in the estimated (N)ARDL, the distribution of the error term, and the number of simulations [default and minimum number is 2000].
I hope that EViews users and developers will find our papers useful for future projects and papers, as well as for incorporating new features of the ARDL model family into EViews.
Kind regards,
Georgios Bertsatos
Research Fellow at the Centre of Planning and Economic Research (KEPE), Athens, Greece
References
- Pesaran, M.H., Shin Y., & Smith, R.J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(1), 289-326. https://doi.org/10.1002/jae.616
- Bertsatos, G., Sakellaris, P. & Tsionas, M.G. (2022a). Extensions of the Pesaran, Shin and Smith (2001) bounds testing procedure. Empirical Economics, 62, 605-634. https://doi.org/10.1007/s00181-021-02041-3
- Bertsatos, G., Sakellaris, P. & Tsionas, M.G. (2022b). Correction to: Extensions of the Pesaran, Shin and Smith (2001) bounds testing procedure. Empirical Economics, 62, 635. https://doi.org/10.1007/s00181-021-02085-5
- Bertsatos, G., Sakellaris, P., & Tsionas, M. (2023). A Panel Bounds Testing Procedure. Theoretical Economics Letters, 13(7). https://doi.org/10.4236/tel.2023.137102