Social Security and Safe Spending Rates

A reader writes in, asking

“I’m a big fan of Morningstar and the stuff they’ve released recently on ‘safe withdrawal rates.’ My question is how I should be thinking about future Social Security payments, when I calculate my ‘safe withdrawal rate.’

To me, it makes sense to add the NPV (calculated conservatively) of the future Social Security payments that I expect my wife and I to receive to our current portfolio, before I do the math on what our starting annual withdrawal number looks like.

Have you published anything on how to think about that question?”

That’s a great question, and it gets directly to the limitations of safe spending rate research. That is, such research is very helpful for determining approximately how much a person should have saved before retiring, but when trying to use it as an actual spending plan, it comes up somewhat short.

The biggest issue isn’t that the strategy of spending a fixed (inflation-adjusted) amount from the portfolio every year is necessarily a bad strategy (though it does have some drawbacks). Rather, the issue is that such an idea is simply not applicable for most real-life households.

That is, in most households, it’s rare that (inflation-adjusted) spending from the portfolio will be kept constant from one year to the next, because the amount of non-portfolio income changes meaningfully over time — for example as the person semi-retires, then fully retires, then Social Security begins. And for a couple, there are even more distinct phases, because there are twice as many retirement dates and twice as many Social Security start dates.

As far as considering the expected present value of your lifetime Social Security benefit to be a part of the portfolio, and then calculating an initial spending amount accordingly, the issue I see with that is that it depends significantly on what real interest rates are at the time of the calculation. And the higher that real interest rates are (i.e., the higher the discount rate used in the PV calculation), the lower the PV will be, which would indicate spending a lower dollar amount. And that’s rather backwards (i.e., higher real interest rates should indicate that you can spend at a higher rate).

My preferred way to incorporate Social Security into the analysis is to consider it a reduction in spending, for the years in question.

You can do this manually. Calculate what your non-portfolio income will be, year-by-year (including Social Security, earned income, and anything else). Then you can carve out a piece of the portfolio to “replace” that income in the years in which it won’t exist. And then you can spend at a fixed (inflation-adjusted) rate from the rest of the portfolio. (In the context of Social Security, this is often referred to as creating a “Social Security bridge.”)

Very basic example: Bob is single. He’s 65, just retired. He plans to file for Social Security at age 70, at which point he’ll get $36,000 per year. He could allocate 5 x $36,000 = $180,000 to something safe (e.g., a short-term bond fund or a 5-year CD ladder). And he could spend from that chunk of money at a rate of $36,000 per year. And he would spend from the rest of his portfolio at a fixed (inflation-adjusted) rate, which would lead to a fixed (inflation-adjusted) total spending rate also.

As mentioned above though, a real implementation of this idea is likely to be more complicated than this simple example, because there may be a phased retirement — or perhaps a pension or annuity that starts on some particular future date. And there may be two people involved, which would mean even more dates at which the level of non-portfolio income will shift.

Also, ideally, the above calculation would be done on an after-tax basis.

And, admittedly, all of that can get rather cumbersome when taking a DIY approach.

If you want, you can use financial planning software for this, because it can do all of this math (including the taxes) very quickly. The software I use for retirement spending analysis is RightCapital. It’s great, but it’s priced for advisors. The only reason I mention it is that, because I’m happy with the software that I’m using, I’m not also spending time test-driving a whole bunch of other software packages. So I can’t confidently recommend one software package or another for individual users. I have heard good things about the following, but I have not tested them myself.