|
Activity Number:
|
123
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2003 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Business & Economics Statistics Section
|
| Abstract - #300241 |
|
Title:
|
Estimating Calendar Month Values from Data with Various Reporting Frequencies
|
|
Author(s):
|
Benoit Quenneville*+ and Michael A. Hidiroglou
|
|
Companies:
|
Statistics Canada and Statistics Canada
|
|
Address:
|
3-G RH Coats Bldg., Ottawa, ON, K1A 0T6, Canada
|
|
Keywords:
|
benchmarking ; calendarization ; interpolation ; forecast ; GLS ; imputation
|
|
Abstract:
|
Statistics Canada now accesses the Canadian Goods and Services Tax (GST) data. Each unit on the GST file has a set of transactions that report the Tax Paid and the corresponding Sales for a set of consecutive days called reporting periods. These periods vary from unit to unit and may also vary for a given unit. This causes two problems: the data points have different timings and durations and are therefore neither comparable nor summable over units; and the data lacks timeliness because a large fraction of the quarterly and annual reports are not available for the current months. A third problem is that the Sales variable can be missing for both past and current reference periods. The objective is to obtain calendar month estimates for all months and for the current months in particular for both the Sales and Tax Paid variables. The method proposed entails benchmarking a monthly indicator to the fiscal data. The resulting monthly benchmarked values are the desired interpolations. The paper also discusses the selection of the indicator and presents an application. (Coauthored with Pierre Cholette).
|
