Catherine Hood Consulting

The "Case Studies" Project

Back in 2004, my friend and colleague Kathy McDonald-Johnson and I started work on a book titled Diagnostics for Seasonal Adjustment and Time Series Modeling. Unfortunately, due to problems at the Census Bureau, Kathy and I weren't allowed to finish the book. I've recently begun to sort through old notes, and as I finish pieces of the book, I will post them here.

Diagnostics for Seasonal Adjustment

and Time Series (Case Studies)

Note: The papers below are in Portable Document Format (PDF). To view the files you will need the Adobe® Acrobat® Reader which is available for free from the Adobe web site.

Papers on this page are still in draft form. Comments are always welcome.



Contents

Introduction

Spectral Graphs

Seasonal Adjustment Diagnostics

  • Descriptions and Examples
    • Ms and Qs (X-11/X-12 only)
    • Sliding spans
    • Revision histories
    • SI Ratio graphs (X-11/X-12 only)
    • Other miscellaneous diagnostics
  • Questions Answered
    • Is the series seasonal?
    • Can the series be adjusted reliably?
    • Can I improve the adjustment?
    • What’s the best span of data to use for the adjustment?
    • What filter length is best?
    • How can I tell if there is different variability in different months?
    • When do I need to reduce the number of extreme values chosen in X 11/X 12?
  • What do I do when the diagnostics seem to conflict?

RegARIMA Modeling Diagnostics

  • Descriptions and Examples
    • Model Adequacy Diagnostics and ACF/PACF graphs
    • Model Comparison Diagnostics and AIC and OSFE graphs
  • Questions Answered
    • Are there calendar-related effects (besides seasonality) in my series?
    • What is the correct set of outliers for my series?
    • How do I find the best possible ARIMA model?
    • What's the best span of data to use for modeling?
    • How do I compare regARIMA models? What do I do when the diagnostics seem to conflict?

Diagnostics for Composite Series

  • Descriptions and Examples
    • Smoothness measures (introduced in X 11-ARIMA/88)
    • Ms and Qs
    • Sliding Spans
    • Revisions
    • Spectral diagnostics
  • Questions Answered
    • How do I know whether the direct or indirect adjustment is better? What do I do when the diagnostics seem to conflict?
    • If I have a group of series that is published at the aggregate level, when do I need to look at the diagnostics for the aggregate series?

Several example series, going all the way through from start to finish


Back to the Top

Page design by David Joyce at Exit 42 Design
Web content by Catherine Hood Consulting, Copyright 2006-2010
Last modified: 5 Jan 2012