In a recent post, I argued that probability-of-success-driven guardrails are generally a superior form of guardrail strategy than the more commonly used withdrawal-rate-driven guardrails.

In writing that post and working on some follow-up work on that topics, I’ve come to realize how cumbersome it is to express a given strategy. So, in this post, I aim to present a shorthand that should make it easier to discuss such strategies.

There are five key characteristics of a probability-of-success-driven guardrails strategy that generally need to be conveyed:

1. The upper guardrail (i.e., the probability of success at which spending would be increased)
2. The initial target (i.e., the probability of success initially targeted)
3. The lower guardrail (i.e., the probability of success at which spending would be decreased)
4. Speed of spending decrease (i.e., the movement in the probability of success up toward the initial target)
5. Speed of spending increase (i.e., the movement in the probability of success down toward the initial target)

Granted, it’s easy to envision strategies more complex than the framework above. For instance, the target probability of success could be age varying. However, for the purposes of developing a shorthand that I think will describe the majority of strategies actually adopted by financial planners, I am going to stick with the parameters above.

Using the items identified above, I am proposing the following shorthand:

Upper-Target-Lower (Speed up/Speed down)

In the prior post on probability-of-success-driven guardrails, I described a hypothetical strategy with a 99% upper guardrail, 95% initial target, 80% lower guardrail, and a 10% adjustment back toward the target in the event that an upper or lower guardrail was hit. In other words, 95% is targeted as the initial probability of success, if the plan probability of success increases to 99%, then spending is adjusted 10% of the way back to the target (i.e., spending is increased until the new probability of success is at 98.6%. Similarly, if the probability of success fell to 80%, then spending would be adjusted 10% of the way back to the target (i.e., spending is decreased until the new probability of success is at 81.5%.

So, the shorthand for reporting this strategy would be:

99-90-80 (10/10)

While I think the speed aspect is important to consider because I think it will be a relevant factor in future research on such guardrail strategies, I do think it’s the most difficult to grasp (and much collective thinking is needed on how the speed of adjustments should actually be implemented).

For instance, suppose a retiree’s probability of success falls to 80% and triggers a spending adjustment up to an 81.5% probability of success. I interpret the strategy above as suggesting that if the probability of success the subsequent year is less than 83%, then the same cadence of a 10% difference between the target and lower guardrail would be used to adjust spending. Therefore, if the probability of success the following year is only 82%, then spending would again be cut to bring the probability of success up to 83%. This process would then be followed until the target is reached in 10 years.

But what if markets recover and the probability of success in the following year (after hitting 80% the year prior and being adjusted up to 81.5%), is now 90%? Are the guardrails “reset”? Would hitting 80% again be required to start a new adjustment sequence? What if the probability of success falls to 80% (triggering and adjustment to 81.5%), then rises to 90%, and then falls to 75%? Is the adjustment then to adjust back to 81.5% (i.e., a complete resetting of the guardrails), or should the prior path be treated as a minimum that should be caught up to if the markets haven’t recovered on their own (i.e., adjust up to 84.5%)?

I think these questions are largely unsettled and I think norms will arise with time to help guide the use of these strategies, That said, the “speed” placeholders should still be able to accommodate different conceptions of what speed entails.

However, in the event that speed adjustments are both 100%, I would also propose that the up/down text in parentheses could be omitted.

For instance, if it’s assumed that a strategy calls for a reset entirely back to a target right away (e.g., a shift from 80% back to the target of 95%, given the numbers above), then this 100% speed adjustment can simply be omitted and reported as:

99-95-80

Note how much easier it is to express the guardrails strategy above using the shorthand rather than writing out all of the components.

If someone wants to test four different strategies, reporting using the shorthand can be very helpful. For example:

1. 99-95-80
2. 99-90-70
3. 99-70-50
4. 99-50-25

The numbers above aren’t meaningful. The key point is just how much more efficient it is to communicate guardrail strategies using a shorthand.

One final difficulty is what to do with the probability of success “greater than” 100%. Of course, a true probability of success could never be 100%, but the idea here is that there ought to be a way to convey the degree to which a strategy may be even more conservative than the highest level of spending at which 100% probability of success was obtained.

One idea for addressing this is to use a number greater than 100 (e.g., 120) as indicating that a spending level is a certain percent (e.g., 120-100=20 which is interpreted as 20%) lower than the highest spending level with a 100% probability of success.

So, for instance, 120 would mean spending 20% below the highest spending 100% probability of success scenario. Similar to the issue with speed above, I’m not sure if this is a perfect solution, but it does at least allow us to express strategies that may target particularly conservative spending levels. For example:

1. 120-100-80
2. 150-100-80
3. 175-150-100

A strategy such as (3) above would refer to an initial target spending level of 50% below the highest spending level that achieved a 100% probability of success, with an upper guardrail at 75% below 100% probability of success and a lower guardrail at 100% probability of success. Because speed is omitted above, it is assumed that adjustments are always made fully back to the target level of 50% below the highest spending level that achieved a 100% probability of success.