Experimental indexes of economic activity in the provinces and territories, August 2020
Archived Content
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
Released: 2020-12-07
Experimental indexes of economic activity show that the economies of all provinces and territories experienced growth over the summer as restrictions due to the COVID-19 pandemic were relaxed. Economic activity in most provinces and territories in August 2020, however, remained below that registered in February 2020. Alberta remains the province with the lowest activity compared with its pre-pandemic level.
First released in July 2020 for the January 2002 to March 2020 period, the experimental indexes were developed to address the need for a comprehensive measure of economic activity at the sub-national level in advance of the annual estimates of gross domestic product for provinces and territories. The indexes were created using four different statistical methods to combine a range of economic indicators from a number of areas (e.g., labour market, merchandise trade, manufacturing production, and wholesale and retail trade, etc.) into composites.
Since the first release, methodological improvements have reduced the variability of the estimates within and between the four approaches. Going forward, analysis will be focused on the approaches based on principal component analysis (PCA) and least absolute shrinkage and selection operator (LASSO).
Largest declines in activity registered by April/May 2020
Restrictions on the economy as a result of the response to the COVID-19 pandemic manifested themselves in Canada starting in March 2020. At that point, differing degrees of restrictions on economic activity began, with stronger restrictions being implemented in British Columbia, Ontario and Quebec, and less stringent measures in other provinces and territories. Concomitant with the onset of the pandemic, a sharp reduction in oil prices produced a second negative impact for the oil producing provinces of Alberta, Saskatchewan and Newfoundland and Labrador. The result was a steep decline in economic activity in most provinces and territories that culminated with a trough in April and May compared with February 2020. The largest declines occurred in Alberta and Saskatchewan, while Manitoba was largely unaffected. The PCA-based indexes generally show larger changes than the LASSO-based indexes because they include a larger set of input series in their construction and hence capture more variability.
Economic activity levels increased over the summer
As summer unfolded, the prevalence of the virus decreased and restrictions were lessened. Some types of economic activity, such as retail sales or employment, then began to recover. As a result, both activity indexes (PCA and LASSO) rose for most provinces and territories over June and July.
In August, both activity indexes show that the levels of economic activity in Alberta, Saskatchewan, Ontario, and British Columbia remain the most impacted when compared with February 2020. This is consistent with the dual effects of the pandemic and the decline in oil prices on the economies of Alberta and Saskatchewan, and the fact that Ontario was the last province to lift the most restrictive physical distancing measures. The indexes also show that Nova Scotia, Prince Edward Island, Quebec and Manitoba have been impacted the least. Up to August, Manitoba was a jurisdiction where the effects of COVID-19 were less prevalent, and Quebec was among the first provinces to ease restrictions.
For some of the smaller economies, where fewer inputs series are available, the results need to be interpreted with more caution. Only one index is available for each of the Northwest Territories, Newfoundland and Labrador and Nunavut. The PCA-based economic activity index for the Northwest Territories suggests a larger decline (the 4th largest decline overall for the PCA index) than might be expected given that the territories were among the jurisdictions that were less affected by COVID-19 closures. Newfoundland and Labrador's LASSO-based economic activity index shows a decline which is comparable to that for Saskatchewan and may reflect the effect of declining oil prices on its economy. The index for Nunavut shows a higher level of activity in August compared with February, but is based on the smallest number of input series of any of the indexes.
The results are mixed for New Brunswick and the Yukon. In August, the PCA index suggests that New Brunswick was among the most impacted regions, while the LASSO index shows a level of economic activity above that in February. Similarly, the PCA index for Yukon shows a higher level of activity compared with February, while the LASSO-based index is near the February level.
Chart 1
Percent change in experimental indexes from February 2020 to August 2020 by index, province and territory
Note to readers
In July 2020, a set of four experimental economic activity indexes for the provinces and territories was released by Statistics Canada to provide improved information on the impact of the COVID-19 pandemic on regional economic performance. To construct the indexes, a regional dataset containing approximately 1,300 data series was compiled. Then, different statistical techniques were used to combine the data to produce measures of aggregate economic activity. One index uses principal component analysis (PCA) to transform the input series into a measure of aggregate regional economic activity. A second, simple economic activity index uses employment, retail sales and manufacturing sales to form an activity index. A third index is based on the least absolute shrinkage and selection operator (LASSO) which selects a sub-set of the inputs for each province and territory which are then used to create an activity index. The fourth index is calculated as a weighted average of the PCA index and the simple index.
With this release, the experimental activity indexes are being updated with data up to August 2020 as well as methodological updates and the incorporation of new data points. The methodological improvements produce more reliable information from the PCA and LASSO based indexes, and they are the focus of this release. The simple index continues to be reported on table 36-10-0633-01 as a check on the more complex methodologies while the weighted index is being discontinued. The weighted index was constructed to address the large variance found in PCA indexes, and this has now been addressed through the methodology improvements.
Updates have been made to improve the quality of the indexes by better controlling for aberrant data points, by incorporating the 2019 provincial and territorial gross domestic product (GDP) estimates, and by incorporating revisions to monthly series. Together, these changes lead to revisions for the entire history of the indexes.
Improvement to the methods used for addressing aberrant data points in 2020 due to significant changes from the effect of the COVID-19 pandemic have been incorporated. When the statistical models are used to predict the monthly changes in the indexes, an adjustment is made to account for differences between the variances of annual and monthly time series. When data for 2020 are included in this adjustment, the resulting monthly series exhibit a high degree of variability due to the presence of large pandemic related monthly changes. To prevent these 2020 data points from influencing monthly estimates, the variance adjustment is now made using data from 2002 to 2019 only. On a year-on-year basis, the growth rates of revised indexes and the growth rates of the originally published indexes are highly correlated (rho > 0.9), but the scale of the change is now lower. This produces indexes that are more stable, and more similar to the scale of changes found in the underlying time series. The change in how the variances are treated produces improved activity index quality, and is the largest source of changes for the indexes.
The 2019 estimate for provincial and territorial GDP has been incorporated into the statistical models. The inclusion of the additional data point leads to moderate changes in parameters, and this generates a historical change. The effect of including the additional data point is larger here than in most settings as the number of annual data points for estimation is relatively small (2002 to 2019) and this makes some model parameters more sensitive to expanding the time series when new data points become available.
Revisions from monthly statistical programs have also been included. These revisions predominantly affect the monthly input series from January 2020 onwards. To limit the effect of in-year revisions on historical values, the base year for the economic activity indexes is being re-set to January 2002 from January 2020.
Contact information
For more information contact us (toll-free 1-800-263-1136; 514-283-8300; STATCAN.infostats-infostats.STATCAN@canada.ca).
To enquire about the concepts, methods or data quality of this release, contact Ryan Macdonald, 343-998-1890; (ryan.macdonald@canada.ca), Economic Analysis Division.
- Date modified: