You may want to check out a recent C. D. Howe Institute report titled Saver’s Choice: Comparing the Marginal Effective Tax Burdens on RRSPs. The report concluded that TFSAs are a more tax-efficient retirement savings vehicle than RRSPs for many Canadians because the effective rate of tax payable on retirement income is often higher than that the tax imposed on regular income during working life. To make a comprehensive comparison of effective tax rates, the authors of the report use a metric they call marginal effective tax rate (METR):
The marginal effective tax rate (METR) is the tax rate bearing on an incremental dollar of income, or the next dollar earned. For individuals, comprehensive METR measures take into account the income thresholds and statutory rates of the personal income tax system, as well as the impacts of tax deductions and credits and income-tested federal and provincial benefits.
The authors then compute the difference in METRs between working-life income versus retirement income and produce graphs such as the one shown below (the graph shows the METR delta for a Single, Ontario resident replacing 60% of working life income) for three provinces (Alberta, Ontario and Quebec) and three income replacement rates (80%, 70% and 60%).
The graph confirms our suspicions that lower income Canadians are better off to save for retirement in a TFSA. Recall that withdrawals from a RRSP result in a claw back of the Guaranteed Income Supplement (GIS) boosting the effective tax rate to as much as 70%. On the other hand, withdrawals from the TFSA are not counted in calculations for income-tested benefits handing the TFSA a massive advantage for low-income Canadians.
However, there is a surprise in the findings. The report’s METR comparisons seems to suggest that Canadians in middle tax brackets (such as those Ontario residents earning between $54,000 and $81,000 and aiming for a 60% replacement rate) might be better off saving for retirement in a TFSA.
Unfortunately, I think the analysis, which seems complicated enough, is still too simplistic. What matters in comparing the TFSA and RRSP is not the METR, it is the AETR (average effective tax rate) on contributions to a retirement account and the AETR at the time of withdrawal. Let’s take a simple example. A one-income household with an annual income of $100,000 will likely have a AETR of 43% in their working years. For a 80% replacement rate, the AETR on withdrawals is likely to be 30% or less because (a) due to income from the CPP, most of the GIS is clawed back anyway, whether or not the household receives another dollar of income from a RRIF and (b) income-splitting opportunities that allow the household avoid the OAS claw back. Add children to the mixture and the analysis becomes even more complicated because RRSP contributions will not only result in tax deductions but also increase CCTB payments.
Ultimately, the TFSA versus RRSP question is largely academic because who knows how tax rules, brackets and rates will evolve and change in the future. For now, it seems clear that low-income households can skip the RRSP and save in a TFSA. Everyone else can probably safely cover all bases by contributing as much as they can to both the TFSA and RRSP. If they save predominantly in one vehicle rather the other, they are still likely to come out ahead because what matters most is that they saved for their retirement, not which savings vehicle turned out to be marginally better in retrospect.