I. Introduction
As part of a corporate initiative to improve data coherence across the economic survey program, the annual industrial consumption of energy survey (ICE) has been incorporated into the new economic survey model, which is called the Integrated Business Statistics Program (IBSP).Note1
This report has two main objectives. The first is to inform ICE data users and other program stakeholders of the sampling changes that have been implemented to date as part of the transition to the IBSP. These changes will take effect when data for the reference year 2014 are released, in November 2015. The second objective is to provide context on how the program changes will impact data outputs.
The document is organized into three parts. Following the introduction, Part II highlights the key methodological changes resulting from the integration of the ICE survey into the IBSP framework and expected impacts on the data. Part III presents the conclusions.
II. Sampling methodological changes to the ICE survey under the IBSP
The IBSP provides a common survey framework for the various business surveys conducted at Statistics Canada. The integration of the ICE survey into the IBSP framework, which started with the reference year 2014 cycle, resulted in the following changes to the sampling methodology.
- Sampling Unit
Prior to reference year 2014, the sampling unit for the ICE survey was based on the business establishment. As the IBSP adopts the business enterprise as the sampling unit, the sampling methodology for the ICE survey was updated towards using the enterprise as the sampling unit.
In this new scheme, when an enterprise is selected as in-scope for the ICE survey sample, all the in-scope establishments under that enterprise will be included in the sample. Given that the majority of ICE sample units represent the so-called simple single businesses consisting of a single business establishment, the impact of this change is not very significant.
- Random sampling method.
The IBSP uses the Bernouilli sampling method, while a simple random sampling method (SRS) was used for the ICE survey, before reference year 2014.
The major difference introduced by this new method is that, within each sampling stratum, the sample size is no longer a predetermined number. Given the fairly large overall sample size and the mini-censuses in many strata of the ICE survey, the general impact is not expected to be significant.
- Change of stratification method.
The previous Lavallée-Hidiroglou stratification algorithm used by the ICE survey has now been replaced by a general geometric stratification. This may lead to minor differences in calculating boundaries between sample strata. Overall, the general impact on the data is not expected to be significant.
In addition to the changes identified above, Statistics Canada’s Business Register, which provides the population frame for the ICE survey, is updated on a regular basis. These continuous improvements may impact the ICE survey from time to time. For instance, the determination of annual business revenue for each company/enterprise is currently being modified. Given the important role played by business revenue in the sampling process, this will result in some changes to the ICE survey sample.
The IBSP methodological and analytical approaches take into account core IBSP objectives, including reducing response burden, maintaining data quality, and maximizing the use of administrative data.
III. Conclusion
The new sampling methodology may result in some data variability from previous years in non census industries. For example, establishments belonging to the same enterprise may have a stronger tendency to consume similar energy products. This is not expected to have a large impact as many of the ICE sample units are simple single businesses where there is only one establishment per enterprise. Revenue updates in the BR and possible boundary changes to the strata may also result in minor data changes for the non census industries.