Sequence of Returns Risk or Something Else?

I've read many columns and papers about how devastating large losses can be to a portfolio when those losses occur early in retirement. Usually, this risk is referred to as Sequence of Returns (SOR) risk and it is illustrated by showing a few market returns and how they provide the same ultimate portfolio value no matter how they are ordered. Then, the column shows how the order becomes significant when we buy or sell stocks periodically from the portfolio. 

Hopefully, you followed my explanation in my two previous posts, Clarifying Sequence of Returns Risk Parts One and Two.

I agree that large losses early in retirement can be potentially devastating, and that SOR risk is introduced when an investor withdraws constant-dollar amounts from a retirement portfolio. I just don't see them as the same thing.

Let’s imagine two workers who retired on October 1, 2007, each with a 100% stock portfolio worth a million dollars. One plans to spend $45,000 every year from his portfolio. The other plans to buy a fixed annuity with the entirety of his portfolio sometime in the next month. As I explained in previous posts, the constant-withdrawal retiree is exposed to sequence of returns (SOR) risk. The fixed annuity buyer obviously is not.

From October 8, 2007 to March 2 2009, the S&P 500 fell nearly 55% in just 16 months. Our two retiree-investors are both now looking at portfolios worth about $450,000 and I would argue that, despite having different spending strategies, they are about equally screwed.

The risk of the sequence of future returns is of no consequence to the annuity buyer and for the constant-dollar spender, SOR risk will seem like a minor annoyance compared to his recent loss of more than half his portfolio.

Their problem is clearly not Sequence of Returns risk as I have discussed it.

I see their problem as having made an extremely large bet on stocks at the riskiest time in retirement and I refer to it as "early loss" risk.

The bet is extremely large because the typical pattern for a retiree-investor results having his or her their largest portfolio value just before and just after retirement. The following graph shows portfolio value for a worker who invests $6,000 a year while working, and spends $50,000 a year in retirement with a constant 8% rate of return. Even with significant changes to portfolio allocation, the worker will make his largest bets on the stock market (i.e., have the largest portfolio to invest) in the decade before and the decade after retirement.

Why is early retirement the riskiest time? Ignoring the size of the annual bet, which is in and of itself an increased risk, portfolio values in the decades before and after retirement have the greatest impact on terminal portfolio value.

Here’s a simple example. Assume that a retiree can earn 5% every year of a 30-year retirement with no risk and will spend $43,000¹ each year. I replaced the 5% gain with a 30% loss in the first year of retirement, then moved the 30% loss to the second year, etc. To be clear, each scenario contains just one 30% loss and 29 five percent gains, with the loss marching through the years.

A 30% loss in the first year of retirement left only $24,000 in the terminal portfolio. The same 30% loss in year 30 leaves over $962,000.

With no losses, the portfolio would have ended with a value of nearly $1.5M. As you can see from this simple demonstration, the earlier in retirement a major loss occurs, the greater the impact on future wealth.

A 30% loss in year one had 25 times the impact of a loss in year thirty when comparing TPV’s.

(Wade Pfau posted a similar analysis today. He uses maximum sustainable withdrawal rates, instead, but comes to the same conclusion.)

So, put these two charts together and you see that the typical retiree will place her biggest stock market bet at the riskiest possible time.

That’s why I refer to this as “early loss risk” instead of SOR risk and why I see it differently than the SOR risk I have explored in my past few posts.

The obvious solution to this problem is to lower stock allocations early in retirement and perhaps increase your allocation as early loss risk eases². That is exactly what Wade Pfau and Michael Kitces recently proposed.

To some extent, the differences between “my” SOR risk and SWR advocates’ SOR risk is semantic. They’re saying that if you implement a constant-dollar withdrawal strategy and have losses early in retirement you’re in deep doodoo. I’m saying that a large early loss is bad no matter what your spending strategy.

In a more important way, they're different. Large early losses with a constant-dollar spending strategy frequently result in the retiree going broke, while losses with a strategy based on a proportion of remaining portfolio balance threaten the retiree’s annual income, while allowing her portfolio to survive.

Either way, if you hold most of your wealth in stocks early in retirement, you have enormous risk, as many recent retirees and near-retirees learned in 2007. I was one of those, and it gave me an entirely new perspective on stock market risk.

I held 40% stocks at the market peak in 2007. I recently advised a friend to hold 20% or less.

It certainly changed my idea of “conservative allocation”.

In my next post, "Sequence of Returns Risk: What's That Mean", I'll tie up a few loose ends on this topic.

---

¹ I reduced this from $45,000 to avoid portfolio failure in less than 30 years for the worst case.

² It never goes away. Whatever your current age in retirement is the riskiest investment year of the rest of your life.