Canadian international trade in services, December 2020

Revisions

In light of travel restrictions and border closures related to COVID-19, atypical values for travel services not related to education and for the passenger fares component of transportation services were reported for most of 2020.

To publish consistent estimates of seasonally adjusted values, seasonal factors from 2019 were applied to the unadjusted data for 2020. While using 2019 factors allowed for a certain amount of coherence and consistency in the seasonally adjusted estimates, it resulted in a larger than usual discrepancy between the annual seasonally adjusted series and the unadjusted series that must be reconciled at the end of the year, as is common practice.

As a result, significant revisions to the seasonally adjusted data were introduced with this month's release, in particular to exports for the first months of the year. These revisions have created a substantial decrease in exports from December 2019 to January 2020. Data for 2019 will be open for revisions in the fall of 2021, and seasonal adjustment models may be revised at that time.

Under more typical seasonal patterns, this decline would not have been as large. Users are advised to use caution when analyzing month-over-month movements for seasonally adjusted exports for the two series, especially during this period, and may wish to include unadjusted data their analysis.

Adjustments to the methodological approach

Circumstances surrounding COVID-19 pose issues for the production of monthly international trade in services statistics, particularly for travel and transportation services.

Travel services and the passenger fares component of transportation services are estimated using tourist and traveller counts as indicators of monthly movement. These data are not available on a timely basis and are projected for the reference month before being replaced the following month with the actual values. This approach works in a typical month but does not work in the current situation, in which travel restrictions to reduce tourist and traveller numbers were implemented very quickly.

To better capture the effects of the COVID-19 pandemic, Statistics Canada is incorporating data from the Canada Border Services Agency's primary inspection kiosks into its monthly international trade in services program. These are electronic customs declaration kiosks that have been installed in most major Canadian airports. While these data represent only a subset of total travellers, they provide relevant and timely insights.

In addition, the methodology used to estimate the monthly international trade in services statistics involves first projecting quarterly benchmark values, then dividing these quarterly values into monthly values using economic indicators. While the projected quarterly value was previously kept constant throughout the production process for all three months of the quarter, with the current volatility, this particular element of the model was revised, and the benchmark values are being adjusted based on the most current data, as needed.

Methodology

Because several data sources used to compile the data from the international trade in services program are available only on a quarterly or annual basis, the monthly statistics on Canada's international trade in services are generated using models when up-to-date information is not available. This follows the methodology used in many countries that produce monthly trade in services data.

In general, for most commercial and travel services, as well as some transportation services, modelling of monthly estimates follows a three-step approach. First, values for the upcoming quarter are estimated using statistical models (auto-regressive integrated moving average). Second, indicator series that act as a proxy for the monthly movement of trade in services are identified through relationships with other economic indicators, such as merchandise trade or gross domestic product by industry. Third, a temporal disaggregation method (the Denton–Cholette method) is applied to distribute modelled quarterly services on a monthly basis, using the predicted values of monthly services generated in the second step. Adjustments are made each month as new information becomes available.