We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Performance of the ADF test in stationary series within structural breaks.
- Authors
Coelho Amaral, Mariane; Silveira, Anderson; Dias de Mattos, Viviane Leite; Konrath, Andrea Cristina; Ricardo Nakamura, Luiz
- Abstract
The study of time series has been developed constantly, given the large volume of observed and measured data over the years. An important characteristic of time series is stationarity, which is mostly analyzed by unit root tests. It is a consensus in the literature that structural breaks, when present in the data series, can bias the result of the Augmented Dickey Fuller Test (ADF), the best known and most widely used method of stationarity investigation. So far, however, there is no consensus regarding the intensity that structural breaks can affect the power of the ADF Test, making the decision about using it difficult and possibly leading researchers to errors under those changes. Thus, this article analyzed the influence of level shift (LS) structural breaks in the stationarity analysis in annual time series using the ADF test through the rejection proportion of the null hypothesis. It was observed that this procedure tends to reject the null hypothesis in the presence of structural breaks in a possible confusion with the presence of a unit root. Furthermore, it was noted that, as the initial perturbation ω increased, the power of the test was rapidly reduced, mainly with level change breaks imputed in positions closer to the origin of the data series.
- Subjects
NULL hypothesis; TIME series analysis; DECISION making; RESEARCH personnel; MULTIPLE imputation (Statistics); TIME management
- Publication
Revista Ciência e Natura, 2023, Vol 45, p1
- ISSN
0100-8307
- Publication type
Article
- DOI
10.5902/2179460X75150