Common menu bar
Breadcrumb
Estimating and Forecasting Aggregate Retirement Flows in the Canadian Labour Market - December 2005
4. Existing Policy Research and Coordination Directorate (PRCD) Time Series Estimates of Retirement Flows
Policy Research and Coordination Directorate (PRCD) currently uses two time series of retirement flows: 1) the self-reporting based series (SRS); and 2) the age distribution based series Age Distribution Series (ADS).
The Self-Reporting Series (SRS)
The Self-Reporting Series (SRS) estimate of historical retirements is derived simply as the total number of workers in the Labour Force Survey (LFS) that separated from their job in a given year and self-reported retirement as the reason. Although intuitively this approach appears ideal, given that retirements can be easily estimated from the Labour Force Survey (LFS), it suffers from several drawbacks. First, as was established previously, the retirement process is complicated and can take on many forms and therefore, attempting to identify retirement through a simple survey question is oversimplified. In fact, according to Rowe and Nguyen (2002), "retirement as a self-reported event appears to be relatively infrequent...therefore in many cases, the job separation that ultimately ended a career must have been a layoff, an illness or disability, or a family related event." This point is supported by Figure 2, which demonstrates that on average over the 1976 - 2005 period less than half of all the job separations among workers aged 55 and over were due to retirement. Therefore, by including only those who self-reported retirement, the SRS potentially underestimates greatly the annual number of retirement events. Secondly, this estimate of retirements is unable to distinguish between a permanent retirement and a temporary one2, which is needed to satisfy PRCDs definition of retirement established in the previous section. As was already discussed, retirement is a complex process and self reporting retirement as the reason for a job separation does not imply a permanent separation from the labour force.
The Age Distribution Series (ADS)
In light of the shortcomings with the SRS, PRCD developed an alternative time series of retirement flows. Annual retirements in this case are simply estimated as the average annual number of employed workers within 5 years of the LFS median retirement age. In fact, this estimate does not produce an actual yearly retirement estimate but instead a 5 year moving average estimate. For example, in 2002 there were approximately 1.5 million workers that were within 5 years of the median retirement age of 61 (or aged 56 and over), implying an annual retirement number of 300,000. Although this methodology is a useful tool for identifying retirement pressures, it does not provide an estimate of actual job openings arising from retirements since it measures at best, the number of potential retirees. In addition, median retirement age estimates are critical inputs in the estimation, which in itself, creates two potential difficulties. First, the retirement age estimate is derived by Statistics Canada using information from the LFS. As was mentioned above, the actual number of people claiming to have retired in the Labour Force Survey (LFS) represents only a portion of all retirement events in the labour market and therefore, the retirement ages derived from this information may not be representative of actual retirement behavior in the Canadian economy. Secondly, this retirement estimate is sensitive to small changes in the median age of retirement.
The SRS and the Age Distribution Series (ADS) estimates have produced significantly different evolutions of retirement flows since the mid-1990s. While retirements derived from the ADS have been trending upwards (given the aging of the workforce and a falling retirement age), the SRS has produced a rather low and stable retirement estimate. In 2003, the ADS retirement estimate at 315,000 (or a 2.2% retirement rate) was more than double the SRS estimate.
- 2The LFS is not a longitudinal survey; therefore it is not possible to track workers following job separation in order to establish the permanency of the separation.