Working Paper: CEPR ID: DP9654
Authors: Eric Ghysels; J Isaac Miller
Abstract: We examine the effects of mixed sampling frequencies and temporal aggregation on standard tests for cointegration. We find that the effects of aggregation on the size of the tests may be severe. Matching sampling schemes of all series generally reduces size, and the nominal size is obtained when all series are skip sampled in the same way. When matching all schemes is not feasible, but when some high-frequency data are available, we show how to use mixed-frequency models to improve the size distortion of the tests. We test stock prices and dividends for cointegration as an empirical demonstration.
Keywords: cointegration; mixed sampling frequencies; residual-based cointegration test; temporal aggregation; trace test
JEL Codes: C12; C32
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| mismatched sampling frequencies (C83) | size distortion in cointegration tests (C22) |
| consistent sampling frequency across series (C22) | no size distortion (D39) |
| different aggregation methods (C43) | size distortion (F12) |
| mixed-frequency models (C32) | alleviate size distortions (F12) |
| aggregation method (C43) | accuracy of cointegration tests (C22) |