|
Activity Number:
|
245
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Business and Economics Statistics Section
|
| Abstract - #305794 |
|
Title:
|
Quantifying the Quality of Macroeconomic Variables
|
|
Author(s):
|
Alex Teterukovsky*+ and Lars-Erik Öller
|
|
Companies:
|
Statistics Sweden and Statistics Sweden/Stockholm University
|
|
Address:
|
Box 24300, Stockholm 10451, , Sweden
|
|
Keywords:
|
forecast errors ; revisions ; entropy ; statistical quality
|
|
Abstract:
|
Methods to quantify the quality of a macroeconomic statistical time series are presented. The measures are based on a combination of how predictable the series is and how much its statistics needs to be revised. An information window is introduced which is based on signal-to-noise ratios (SNR) and provides a snapshot of the quality. A formulation of information in terms of entropy is considered. A nonparametric approach is taken which allows for testing whether a forecast or a preliminary value is informative. Concavity and monotonous convergence of information accrual are discussed. It is shown how the suggested measures signal that either given forecasts or macroeconomic variables (GDP components) themselves, are of dubious quality.
|
