![IconGems-Print](images/IconGems-Print.png)
463 – SPEED: Methodological Advances in Time Series: BandE Speed Session, Part 1
To Adjust or Not to Adjust? An Empirical Evaluation of Time Series With Unstable Seasonal Patterns
Demetra Lytras
U.S. Census Bureau
Often a time series exhibits a fairly stable seasonal pattern, making the decision to seasonally adjust it straightforward. However some series may show indications of seasonality but have a swiftly changing seasonal pattern. One school of thought is that all such series should be seasonally adjusted to ensure there is no residual seasonality in the published composite adjustments, while another says that only series that have a stable adjustment should be seasonally adjusted. This paper examines series with unstable seasonal patterns to determine whether it is better to adjust or not to adjust.