Get new posts by email:
  

How I Manage my RRSP in Retirement

Spending from retirement savings, or decumulation, in a way that maximizes what you have left to spend after taxes is surprisingly complex.  I’ve done extensive simulations of various strategies for my situation, including strategies that change over time, to find what works best for me.  Here I describe how I’m managing my RRSP in retirement, but it’s important to remember that it may or may not work well for you depending on your particular circumstances.

Looking for the fully optimal financial strategy is futile.  I ran my simulations and chose a simple enough strategy that worked well across a wide range of investment outcomes.  The only reason for changing my strategy is if something happens that is far outside my expectations.  Those who constantly seek perfection waste their time and hurt their outcomes with constant tinkering.

Our portfolio and goals

My wife and I have RRSPs, TFSAs, and non-registered accounts.  I prefer not to discuss exact amounts, but broadly speaking, our combined RRSPs are larger than our combined non-registered accounts, which are larger than our combined TFSAs.  In addition to the exact sizes of these accounts, two other figures that are significant for simulations are our unrealized capital gains in the non-registered accounts and our deferred capital losses from previous years.

My wife and I have roughly the same net worth.  Although we consider all our assets to be owned by both of us, CRA doesn’t see it that way.  We spent decades carefully choosing whose money to spend each year so that we’d have close to the same net worth now.

Our goal is to maximize the amount we can safely spend each year, rising with inflation, for the rest of our lives.  We have no interest in scrimping now just so we can live rich when we’re much older.  Some might even choose to spend more in their 50s and 60s than they will spend later, but I can’t see any logic in living poor early on just to be rich later.

The main tax challenge we face is high taxes and possibly OAS clawbacks on forced RRIF withdrawals after we turn 72.  These taxes will be even higher after one of us passes away, and higher still after the second passes away.  The remedy here is to make modest RRSP/RRIF withdrawals in the years before we turn 72.  The goal is to make lightly taxed RRSP/RRIF withdrawals early rather than heavily taxed withdrawals later.  This gap in tax rates has to be large enough to overcome the value of continuing to defer taxes.  

This is where the simulations help.  At one extreme, we could be spending entirely from our TFSAs to keep our incomes very low.  My simulations show that this “collect the GST rebate” strategy is not optimal for us (nor do I find it palatable).  At the other extreme, winding down our RRSPs quickly is far from optimal as well.  Something in between is best.

Our decumulation strategy

My simulations tell me that we’re best to target a particular income level each year.  Note that our income is not the same thing as how much we spend.  The amounts we spend from non-registered accounts create only modest declared income for taxes.  By adjusting how much we spend from each type of account, we can target different amounts for how much we spend and how much we declare on our income taxes.

Each year since we retired, we spend exclusively from our non-registered accounts until November.  At this point, I’m able to predict reasonably accurately how much income (interest, dividends, and capital gains) we will declare on our taxes.  For some people, this might include a pension or part-time employment income.  Then we each make RRSP withdrawals to increase our declared income to the target level determined by my simulations.

This target level of income is heavily influenced by marginal tax rates.  We live in Ontario.  For 2026, the tax rate on regular income is 19.15% up to $53,891, 23.15% from there to $58,523, and 29.65% from there to $94,907.  For many people, the optimal income target for taxes will be at one of the break points (roughly $53,900, $58,500, or $94,900), but it can be above, below, or between these levels depending on your particular circumstances.

I definitely do not get excited about hitting my income target exactly.  Being off by $1000 makes little difference. The point is to slowly drain the RRSPs/RRIFs at low tax rates to avoid paying higher tax rates later.  Trying to be exact is a waste of time.

Each year, I run the simulations again to see if our income target for taxes changes.  This income target tends to rise as we get closer to age 72, and it can go up or down if the stock market does something extreme one year.

One thing that my simulations have been very consistent about is that we’re best off to take CPP and OAS when we’re 70.  Even if one of us dies at 70, the other will see an increase in standard of living (from a purely financial point of view).

We continue to add to our TFSAs each year.  Once most of our non-registered money is gone, we will use the TFSAs to augment our spending without increasing our taxes.  This will be particularly important when large demands for cash come up, such as buying a new car.

Another detail is that once we turn 65, we’ll make sure to have at least $2000 each come out of RRIFs rather than RRSPs so that we can use the pension income tax credit.

Conclusion

There are many other details in how we handle our money in early retirement, but this article gives an overview of how my simulations have taught us to think about decumulation.  I have found these ideas to be relevant to a few of my friends and family members for whom I’ve done simulations, but I can’t say much beyond that.  Your mileage may vary.

<< Previous Post

Comments

  1. Thanks Michael for sharing this info. I find this very helpful as I look to transition into retirement over the next couple of years.

    ReplyDelete
  2. I really enjoyed this article. I use Adviice software to help out in my planning.

    ReplyDelete
    Replies
    1. There is getting to be a lot of free software available to help with planning. I find I always want to better understand the baked in assumptions, so I end up writing my own software.

      Delete
  3. Thank you James, my situation is very similar and we follow exactly the strategy you outlined - thank you for sharing.

    ReplyDelete
    Replies
    1. You're welcome. Hopefully this approach works as well for you as it has for me.

      Delete
  4. You correctly point out that when one spouse passes there can be significant tax consequences ... my view is that over time in Canada OAS will be income tested for couples likely in the next 5 to 10 years ... I am melting down my rrsp perhaps a little quicker than you ... thanks for your thoughtful column ... JamesD

    ReplyDelete
    Replies
    1. Yes, income taxes rise after one spouse passes and the other controls all savings. But it's important not to react emotionally to this fact. In my case, the tax difference is far less than the reduced expenses. If that isn't true for another couple, they could use a small amount of term life insurance to solve the potential problem.

      Maybe OAS rules will change, but these things are hard to predict. Few governments are willing to anger old people (who tend to vote in high percentages).

      I didn't reveal how fast I'm melting down my RRSP. Perhaps you're melting yours down very quickly?

      Glad you liked the article.

      Delete
    2. I am perhaps interpreting your words "slowly drain the RRSP's" above when I say that I am melting down my RRSP's a little quicker than you.

      Time will tell on OAS claw back for couples ... your point is taken on the boomer voting block however government expenses are climbing ... BC recently froze indexing of tax brackets through to 2030 ... time to update your simulation software !

      For a 65 year old couple there is a roughly 50% chance one dies before age 80. We are often fixated developing an optimal melt down plan that ends with both spouses passing at 90 or 95, however, if I knew now that my spouse was going to die at 80 would I change my current melt down plan. Lots of layers in the onion for this case study

      Thanks again for all your carefully considered posts and for responding to various comments ... JamesD

      Delete
    3. Hopefully that freezing of tax brackets doesn't make it to Ontario. The mortality tables planners use (CPM2014 Mortality Table, based on data from both public and private sector pension plans for 1999-2008, taken forward to 2025 using the CPM Improvement Scale B) only gives data up to 50% chance of living, but ChatGPT came up with about 87 for 70%, which would mean 50% chance of both 65-year olds living to 87. Of course this doesn't matter for planning anyway if the survivor will be better off financially after the first death. This is true for me.

      Delete
    4. At the risk of getting educated, which I gladly accept, this is my reference from a joint life first to die ... https://lifeannuities.com/annuity-mortality-table-joint-life.html

      Delete
    5. There are many different mortality table out there. It's difficult to take hard data from past deaths and figure out how they should be adjusted for the future where people are generally healthier. The institute for financial planning uses tables that say the last to die for two 65-year olds is 94 at 50%. Your link says the expected age of first death is under 91. Average and median are very different here, and they likely are using different mortality data. And after we've gone through all that, then there's how healthy you and your wife are.

      All that said, whether the first to die is expected to be 80 or some higher number, it doesn't change my decumulation strategy at all. The survivor in our case will have even more available spending compared to when there were two of us consuming. The reduced costs exceed the reduction in after-tax available spending.

      Delete
    6. Michael ... my look at the first table in the attachment shows 15.08 years as "the expected number of years to the first death" for a couple aged 65

      Delete
    7. Yes. The data in the planner's document doesn't include expected or median time to first death. So, I looked at the next tab;e in your link for time to second death. This is 25.63 years (age 90.63). But the planners' data says age 94. This shows that your life annuities link and the planners are using different underlying mortality tables.

      Delete

Post a Comment

Popular posts from this blog

Financial Lessons from Poker

Are Financial Advisors the Solution or the Problem for Older Investors?

My Investment Return for 2024

Archive

Show more