On July 24, 2020 the Parliamentary Budget Officer (PBO) reported on costs incurred from the Government’s policy of allowing employees to use pay code 699 paid leave for other reasons. Guidance from the Treasury Board Secretariat (TBS) allowed for employees to request 699 leave for a variety of reasons including being sick with COVID-19, having to quarantine, not being able to access the technology they need to complete their work, and having to care for dependents.
Our original cost estimate included data up to the end of May 2020 for 62 of the 88 federal public service organizations governed by the Financial Administration Act Schedules I, IV and V. The Government subsequently shared monthly data from March to June 2020 for 72 of the 89 federal public service organizations through information request IR0504 (July 2020). [i] Using these data, our office published an updated cost estimate on August 24, 2020.
The Government recently provided us with monthly data from March to September 6 for 84 of the 90 federal public service organizations (see endnote [i] for an explanation of the number of organizations by period). These 84 organizations represent 99% of federal public servants employed as of August 31, 2020. For all weeks the number of hours used was revised from the previous data provided. It is still unclear why some federal organizations are unable to provide data.
Consistent with our prior findings, use of 699 leave continues to decrease (Figure 1-1). As can be seen, after reaching a peak of 1.7 million hours in the week ending April 5, the number of weekly hours of leave has decreased by 90% to 168 thousand hours in the last week of August. The number of weekly hours of leave has slightly increased (by about 20,000) in the last week of data available (first week of September). This represents a 12% increase from the previous week (August 30), but still an 89% decrease from the peak of April 5.
Another interesting observation in Figure 1-1 is an increase in the number of hours in the last week of every month (basically the week containing the last calendar day of each month since some weeks span over two different months). This seems to suggest that some employees or some organizations might enter all their 699 leave at once at the end of the month. Indeed, TBS told the PBO that at the outset of the pandemic, organizations were initially instructed to capture leave in their HR systems as long as it did not negatively impact operational priorities or cause undue strain to network or resources. Thus, the small increase in 699 leave usage in the last week of data available (from August 31 to September 6) could be explained by this phenomenon. We have included in Figure 1-1 a four-week moving average to better represent what the weekly usage of 699 leave would have been if not for this end-of-month anomaly.
This lag in data input of 699 leave also has an impact on the monthly totals presented in Table 1-1 below. Indeed, the monthly total of hours provided by TBS corresponds to the sum of the weekly totals for all weeks ending in that month. For example, the peak of 1.7 million hours was reached during the week of March 30 to April 5 and was thus calculated in the total for the month of April. However, a non-negligible number of hours recorded in that week must have been used in the month of March (either on March 30 and 31, or earlier in the month and only inputted at the end of the month). Similarly, another spike of 1.5 million hours was recorded in the week of April 27 to May 5 and was counted in the total for May. This would suggest the costs presented in Table 1-1 are somewhat underestimated for March while overestimated for May.
As explained in the previous reports, TBS was unable to provide actual personnel expenditures for the survey period. Using 2018-2019 personnel expenditure data provided through information request IR0074, the PBO estimates a cost of nearly $1.1 billion between March and September for leave taken under code 699, inclusive of pension and other benefits.[ii] This includes PBO’s extrapolation of costs for the departments for which no data are available. Note again that this cost does not involve additional spending for the federal government, but is instead a cost in terms of lost hours of output.
Cost Estimate ($ millions) |
Employee Count | Average expenditure per employee ($) | |
March | 160.1 | 60,188 | 2,661 |
April | 295.8 | 74,031 | 3,996 |
May | 307.4 | 68,542 | 4,485 |
June | 142.9 | 46,213 | 3,092 |
July | 97.8 | 31,055 | 3,150 |
August | 65.4 | 22,200 | 2,947 |
September | 12.0 | 8,619 | 1,393 |
Total | 1,081.5 | 113,362 | 9,540 |
The total cost shown in Table 1-1 represents a 31% increase from our previous estimate. With the additional data provided on departments that were missing from the previous update, the cost estimate increased by 10% from our previous estimate over the period of March 15 to June 28 (about one third of the 31% increase). The addition of the months of July and August, as well as the first week of September represent the remaining two thirds of the increase, or about 21%.
As previously reported, the Canada Revenue Agency (CRA) is responsible for most of the 699 leave, representing (as of September 6) 46% of the total cost since the beginning of the pandemic. Table 1-2 shows the five organizations with the highest hours used per month and their respective percent of total hours. The CRA continues to account for a decreasing share of hours used, from 77% in March to 17% in August. Most departments reported their total highest hours in May, but as explained above, this likely includes April hours which were entered at month-end. Only 4 departments reported more hours used in June than in April or May.
The PBO had requested a breakdown of hours by the type of 699 leave being taken (leave for technology reasons, sick leave, family leave, etc.) by departments. The CRA had already provided this information for the period of March 15 to June 6, 2020. In TBS’s initial response to our request, received in July, it was specified that only two of the 88 departments and agencies currently have system capabilities to provide a breakdown of the authorized reasons for extending “Other Paid Leave” and that options were being assessed to explore the feasibility of providing that breakdown for additional organizations.
In the latest data update provided, there is now a breakdown by type of leave for almost all departments for the months of April to August.[iii] Figure 1-2 below presents this breakdown by month. Excluding hours for which the reason is not provided, work limitations were the main cause for using 699 leave in the beginning of the pandemic (April and May), followed by family care. For the following months, family care was the main reason for taking 699 leave (representing between 27% and 34% of all 699 leave).
Considerations
As noted above, the data received for this update contained revisions from our original reporting. TBS has informed the PBO that departments report data at irregular intervals, and they expect leave usage will fluctuate until all organizations have caught up with their data entry of leave transactions. Notably, data for the first week of September (August 31st to September 6th) contained nearly 20 fewer departments than previous weeks and is likely incomplete. As the second wave of COVID-19 cases is clearly under way, the PBO will continue to monitor and report on the cost of 699 leave usage.
Lastly, it was indicated in our previous reports that the usage of 699 leave was likely underreported by many organizations. Indeed, in the latest data update provided by TBS the total hours of 699 leave for the months of March to June was revised upwards for some departments, which confirms the previous numbers were underestimated. Furthermore, the CRA had mentioned that one of the reasons why it was accounting for such a large share of 699 leave (apart from the fact that it is the largest federal department in terms of workforce) was a strong reporting culture within the Agency. If we look at the intensive margin (the number of hours of leave used per employee using 699 leave), then the CRA was definitely above average in the months of March and April, then right on average in May and June, and finally well below average in July and August. This could suggest underreporting by other organizations in the first four to six weeks. However, we know for a fact that most CRA employees were not equipped to work remotely until mid-April, which corroborates these large numbers. It is possible that other departments were better equipped to operate remotely and thus actually used less 699 leave.
Supplementary data
[i] The number of organizations has increased by one in each subsequent data set provided by TBS. This seems to be explained by the fact that some organizations were being combined in previous data releases and are now reported for separately. For example, the numbers for Public Safety previously included the following organizations: The Correctional Investigator Canada, RCMP External Review Committee, Civilian Review and Complaints Commission for the RCMP. These are now reported on separate lines in the most recent data update.
[ii] The PBO’s estimate in the first report (published on July 24, 2020) only accounted for salary at and average cost of $309 per employee per day. The average cost across government inclusive of pension, benefits, and other personnel costs is $481 per employee per day and was used to estimate costs in the update published on August 24, 2020 as well as in this update. Per-department costs were used in the estimate whenever possible. The CRA has lower than average per-employee costs which decreased the overall average cost.
[iii] The breakdown data for the month of March is scarcer, which is not surprising given the fact that even the CRA had not implemented these types of 699 leave codes in its time tracking system until the first week of April. It is not clear however, how federal organizations were able to identify these hours retroactively, given that we were told in July that only two departments had the system capabilities to provide such a breakdown.