Loading...
Please wait, while we are loading the content...
Similar Documents
Testing Periodicity in Time Series Models - A Recommendation of Bootstrap Methods
| Content Provider | Semantic Scholar |
|---|---|
| Author | Herwartz, Helmut |
| Copyright Year | 1996 |
| Abstract | Conditioning the data generating process of a time series on the season periodic time series models have become a promising tool for the analysis of processes exhibiting patterns of seasonal variation. From an econometric viewpoint such models have the appealing property that they are able to embed assumptions like seasonally varying utility functions, seasonal technologies etc. Since seasonal patterns are a dominant component of time series variation one may prefer to perform ones analysis with seasonally unadjusted data. The importance of periodic models for empirical work can be expected to increase with the interest in the analysis of seasonally unadjusted data. Testing for periodic parameter variation is usually done by means of asymptotically valid approximations. The paper examines small sample properties of the likelihood ratio statistic which is in widespread use. Among some other points it is shown that heteroscedastic error distributions may severly increase the empirical size of the test procedure. Bootstrap procedures are characterized by superior properties in small samples. The issue on heteroscedasticity is conveniently addressed by the socalled wild bootstrap. The research of this paper was carried out within the Sonderforschungsbereich 373 at Humboldt University Berlin and was printed using funds made available by the Deutsche Forschungsgemeinschaft. Helpful comments from Helmut L utkepohl and Michael Neumann are acknowledged. |
| File Format | PDF HTM / HTML |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
