Evaluation of the Quality of Imputation of Goods and Services Tax (GST) Revenue for Late Transactions

• Activity Number: 483
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Survey Research Methods
• Author(s): Joanne Leung*+
• Companies: Statistics Canada
• Keywords: administrative data ; imputation ; sub-annual data ; data revisions

Abstract

Goods and Services Tax (GST) data are used in many sub-annual business surveys at Statistics Canada. A study was done to evaluate existing imputation methods for late GST transactions by comparing imputed revenues to the reported values. In this paper, we will highlight some results obtained, especially the finding that imputing a new value each month only improves the quality for about half of the units. As a follow-up, we explored the possibility of imputing late values once and keep them at subsequent processing. Results showed that this approach reduces revisions to the data without imposing large differences at the macro level. In light of the findings, this strategy was implemented in the GST Processing System in September 2009. We will show the impact of this change using real data. Finally, we will give some recommendations to further improve the quality of imputation.