Optimizing the age to start CPP/QPP and OAS.

Let's start with CPP/QPP.

For simplicity, we deal mostly with CPP here, but I have explored QPP more thoroughly in English here and en Fran├žais ici.

You can take CPP as early as age 60 (with a 0.6% penalty for every month before you turn 65) and as late as 70 (with a bonus of 0.7% for each month that you delay the pension past age 65). There's also an effect of the indexation of the YMPE that the CPP calculation is based on, but we'll ignore that for Figure 1. It shows the total amount of CPP money received, before tax in today's dollars, over your lifetime. The math is quite clear: unless you die young, it seems best to delay as long as you can.

Fig.1 - Total CPP.

This math doesn't consider anything else however. If you have a certain amount of income that you need to live on, as most people do, then delaying CPP might not work if you need to get the money from your savings instead, which could run out. If you withdraw from registered retirement accounts instead of taking CPP, then your current taxes will be higher (but lower for your heirs) and it could lead to OAS clawback (which we'll get to later). If you don't need the money to live on, you could potentially stay invested longer and potentially invest more. All of this can have tax consequences (good and bad) for you now and for your heirs later.

To study the problem I added a CPP optimizer to the WITHDRAWAL OPTIMIZER of the MoneyReadyApp that I have described previously. In that post, I discussed how we could use a smart algorithm to determine the best strategy for withdrawal from your different types of savings accounts rather than brute force. But here, we'll just use brute force. I mean that we'll just set the age to start CPP to different values, run the WITHDRAWAL OPTIMIZER for each, and pick the optimum solution.

There are at most 11 years to try (60 to 70), although you can start CPP in any month, so there are actually up to 132 months to try. To reduce the problem we'll just start with the month after your birthday and jump by years. We have to do this because most people also include a spouse. Since spouses can help each other out with expenses, what one does affects the other, so the problem is multiplicative, and we have to try the 11 years of the spouse against each of the 11 years of the main user. That's 121 possibilities to try (17,424 if we used months!). Each of these possibilities requires 1 TIME MACHINE run, 1 Optimizer algorithm run, and finally one more TIME MACHINE. Running through 121 possibilities takes about 9 minutes on my local machine. The result is the best of those 242 TIME MACHINE runs.

I ran it on my database of anonymized users, in the same way that I tested the WITHDRAWAL OPTIMIZER, but removing users who had already started CPP. It took a while to run, but the result was worth it—the success rate was phenomenal: 97.56% of users had an increased legacy. Remember the WITHDRAWAL OPTIMIZER had a 75% success rate on its own. With CPP optimization, the legacy had a median of 10.16% (up to 161%), corresponding to a median of $179,369 after tax in today's dollars. This also means the user could spend more if they did not want to leave a larger legacy.

Most users (70.6%) had chosen age 65 to start CPP. The CPP OPTIMIZER chose 70 most often as the preferred start date, but in only 38.2% of cases (Figure 2). It certainly prefers starting CPP at 65 or later (63.2%) than earlier than 65 (26.47%). Only 17.6% of users stayed at their preferred start age. For most, the age was increased (52.9%). Despite the math we discussed earlier, for a significant portion, the optimized age was decreased (29.4%) (Figure 3).

Fig.2 - Frequency of age to start CPP preferred by users, and after optimization.

Fig.3 - Result of optimization of user's set age to start CPP.

An example where delaying CPP can be bad is shown in Figure 4. This couple, with a current savings of $600,000, eight years from a very early planned future retirement at age 40, the optimized CPP age is 61 for both. The graph has several peaks and valleys, and missing values outright, at points where they run into cash-flow problems and the TIME MACHINE run is not considered. This is a case where they need the CPP, even if it's quite reduced due to their short careers. This graph also demonstrates the interdependency of the two spouses.

Note that you can click and drag the graph to rotate it.

Fig.4 - An example of CPP optimization for a couple.

However, even if you don't need the money, it might be worth taking the pension early to not draw upon investments or even to invest more earlier if you are a good investor. For example, another young couple plans to retire in their early 50s, but are good savers and investors with a portfolio growth of 7% per year, and with generous defined benefit pension plans (Figure 5). They are optimized to take CPP early because it allows them to invest more. It's a riskier plan because it assumes both the markets and the investor will perform well for many years.

Fig.5 - Another example of CPP optimization for a couple.

But in most cases the recommendation is to take CPP later, as in Figure 6, where the optimized age to start CPP was increased from 65 to 70 for both spouses. Because there are so many interconnections between the two spouses, their accounts, withdrawal strategies, and their market expectations, it is difficult for them to predict their optimum strategy. This makes the CPP OPTIMIZER a useful tool.

Fig.6 - Another example of CPP optimization for a couple.

So what about OAS?

You can start collecting OAS at age 65. The TIME MACHINE also estimates the Guaranteed Income Supplement (GIS) and Allowance pensions; we include those in OAS when applicable. If you delay receiving your OAS pension past age 65, your monthly pension payment will be increased by 0.6% for every month (up to a maximum of 36%, at age 70). There is the Old Age Security pension recovery tax, that can claw back your OAS pension. For 2021 the clawback starts at a net income of $79,054 and is 15% of every dollar of income you have over that threshold. The OAS is completely eliminated at an income of $128,149.

So that's another 6 years of possibilities to consider (65 to 70), or 36 for spouses. Of course, this could be multiplicative with the CPP optimization. That results in a manageable 11 x 6 = 66 possibilities for a single person. It's fewer than the CPP possibilities for a couple, so I added the OAS optimization for single individuals. This allows them to also get a cool 3-D graph, this one for CPP Age versus OAS Age.

I ran this over all the singles in my database. I was hoping the results would show the OAS age tracking the CPP age, since the effect of adding or reducing income in any year should be the same whether the income is CPP or OAS. I was hoping that CPP and OAS results were mostly independent, not combinatorial. But the results were all over the place.

As an example, here is an interesting case where the user had originally set 70 for both CPP and OAS (Figure 7). The CPP optimizer alone brought the optimum CPP age down to 67. But when optimized for OAS together with CPP, the optimum age for OAS became 65, and brought the optimum age for CPP up to 70. This is proof that we can't just optimize CPP, pick that age for it and then optimize OAS (or the other way around).

Fig.7 - Example of CPP and OAS optimization for a single person.

The major problem is that for couples, the surface is now in 5 dimensions, and it's a not-so-manageable 4,356 possibilities. At least not for a website operating in real-time. So unfortunately at this time, I can't bring you a complete CPP and OAS Optimizer for couples as it would take too long and/or the computing costs would be too high. But despite the proof of Figure 7, I tried doing the 36 combinations of OAS for a couple, after the 121 CPP combinations. With this approach, we may not find the global maximum, but may be able to find a higher local maximum.

But am I making a mountain out of 5-D molehills? How much difference does OAS make? For low-income seniors, it could be a lot but they would probably prefer the money earlier than later as it is needed. For high-income seniors, it could be mostly clawed back anyway. For the "middle class", it is usually not an important source of income, but a nice one to have, and I suppose that Canadians try to optimize their taxable income so that their OAS is not clawed back.

Figures 8 and 9 show the distributions for the preferred and optimized OAS starting ages. Over 80% had chosen age 65 for starting OAS, and for 91.7% the start date did not change or was increased. Overall, the results (Figure 10) show that once CPP is optimized the result of the OAS optimization has minimal benefit for most people.

Fig.8 - Frequency of age to start OAS preferred by users, and after optimization.
Fig.9 - Result of optimization of user's set age to start OAS.
Fig.10 - Average % increase of legacy.


Optimizing when to take CPP is important. The main reason is that you can't change your mind. Once you start taking it, how much you receive is set for life (but it's indexed for inflation). For most people, delaying CPP is best. The exceptions are people who need the income, who think they will have a shorter than average life span, and for people who don't need the income and have (or can hire) the ability to have high returns on their investments over their lifespan. The MoneyReady App now brings you the ability to optimize your CPP start age, for both spouses when applicable, in conjunction with the withdrawal optimization strategy.

Optimizing when to take OAS is not so important. Again, once you make your decision, it can't be changed, although it will be indexed also. Low-income seniors are the ones to pay particular attention, but in most cases, they need the money, and delaying it is of little benefit. For all others, it's not OAS that will make or break a successful financial plan. Most people prefer to take it at 65 and that's fine as it won't make much difference to your heirs and I encourage you to spend it on making your golden years the happiest they can be.

However, the MoneyReady App now brings single individuals the ability to optimize your OAS start age, along with optimizing the CPP and in conjunction with the withdrawal optimization strategy. But we can do that only for single individuals, or couples where the optimization is only applicable to one spouse, because the number of combinations for the full OAS and CPP optimization becomes too large for a couple. In that case we only optimize for CPP and leave your selected OAS start ages as is. You can always change those ages manually and re-optimize.

But I don't want to leave a legacy! The optimizers maximize your after-tax legacy (from your accounts only, it ignores real estate and loans). If it is too high, you can spend more in your lifetime in order to bring it back down. You can easily add to your planned EXPENSES entries in the MoneyReady App and run the TIME MACHINE and the optimizers, to determine if your plan will work and to leave the desired legacy. This is a bit tedious to do, so we'll be working on an EXPENSE OPTIMIZER to see if we can calculate a level of before-tax expense required to bring your legacy to a level that you chose. Now available, see the MoneyReady App CHOOSE YOUR LEGACY buttton.