Risk-Based Guardrails vs Guyton-Klinger: An Advisor’s Decision Guide

Back to Resources

Two households retire in late 2007 with the same portfolio, the same monthly spending, and the same commitment to adjust if markets turn. One follows Guyton-Klinger guardrails. The other follows risk-based guardrails. By the depths of the Global Financial Crisis, the Guyton-Klinger plan has called for a 28% cut to income. The risk-based plan has called for pulling back spending by 3%. On $10,000 a month of spending, that is the difference between living on about $7,200 and about $9,700 through the worst markets of a generation.

Both frameworks exist to answer the same client question, “How much can I spend?”, but they answer it with very different machinery. Guyton-Klinger guardrails, introduced by Jonathan Guyton in 2004 and formalized with William Klinger in 2006, watch a single number: the portfolio withdrawal rate. When that rate drifts too far from where it started, spending is cut or raised by a fixed percentage. Risk-based guardrails, the methodology developed at Income Lab and explored in a series of papers by Justin Fitzpatrick and Derek Tharp, watch the risk of the entire plan: investments, cash flows, longevity, mortality, and inflation. These complex interactions are analyzed and recalculated monthly, but expressed as portfolio balance guardrails a client can actually see and understand.

Guyton and Klinger deserve genuine credit. They put guardrails on the map for financial advisors, and the instinct behind their work, that retirement spending should adjust as conditions change, is exactly right. The rest of this guide is about acting on that instinct well: how each framework works, where the original mechanics struggle in real life, and why withdrawal rate guardrails like the Guyton-Klinger method, though well-meaning, should not be used in real life.

Where guardrails came from

In 1994, William Bengen showed that a retiree drawing an inflation-adjusted 4% from a 60/40 stock-bond portfolio would have survived every historical period he tested. It was the industry’s first defensible answer to the spending question (“How much can I spend?”), but a static one: it never told anyone when they could spend more, and in most historical scenarios it left retirees with far more money than they had hoped, having underspent and worried for thirty years.

Jonathan Guyton challenged that stasis directly. His October 2004 paper in the Journal of Financial Planning, “Decision Rules and Portfolio Management for Retirees: Is the ‘Safe’ Initial Withdrawal Rate Too Safe?”, argued that retirees could start at a higher withdrawal rate if they agreed in advance to adjust spending by a set of rules. In March 2006, Guyton and William Klinger formalized the framework in “Decision Rules and Maximum Initial Withdrawal Rates”. Before their work, an advisor’s realistic options were a fixed rate or improvisation. After it, advisors had a disciplined, explainable way to say “we will adjust, and here is when.”

In principle, this was a huge advance in retirement planning and management. After about a decade, however, other researchers reported that the actual math and process behind Guyton-Klinger withdrawal rate guardrails didn’t work all that well. Wade Pfau (2015) and Karsten Jeske (2017) independently published studies showing that, in historical and hypothetical (Monte Carlo) scenarios, the Guyton-Klinger approach often led to substantial unneeded reductions in spending.

The next step came in 2021, when Derek Tharp and I published “The Retirement Distribution ‘Hatchet’: Using Risk-Based Guardrails To Project Sustainable Cash Flows” on Kitces.com, laying out a guardrails methodology triggered by total plan risk rather than withdrawal rates. In March 2024 we followed with “Why Guyton-Klinger Guardrails Are Too Risky For Most Retirees (And How Risk-Based Guardrails Can Help)”, which ran both frameworks through real historical stress periods side by side and reinforced Pfau and Jeske’s conclusions.

What risk-based guardrails measure, and how they adjust

Withdrawal-rate guardrails like Guyton-Klinger ask if the portfolio withdrawal rate has strayed too far from the original target withdrawal rate. Risk-based guardrails start from a different question. Instead of asking “how far has the withdrawal rate drifted?”, they ask “how is the whole plan doing?” Each month, the analysis re-evaluates everything that bears on the answer: the portfolio balance and its allocation, every cash flow in and out (a Social Security benefit starting in two years, a mortgage that ends in 2031, a new roof next spring), the household’s longevity and, for couples, the chance that one spouse outlives the other, return and inflation expectations, including variability of inflation and the plan’s sensitivity to these factors, and the inflation that has accumulated since the plan last changed. Then the plan either calls for a change or it doesn’t. Most months it doesn’t.

The analytics behind risk-based guardrails are substantial. But the output is a simple set of numbers, expressed in dollar terms, that a client can hold. The first is the retirement paycheck (or, if you prefer, “spending capacity”): the total monthly spending the plan supports right now, funded from all income sources together. The other two are the guardrails themselves, expressed as portfolio balances which, if we hit them today, would trigger the need for an adjustment.

Consider Mark, 67, and Linda, 65, with $1.6 million in investable assets. Their retirement paycheck is $9,200 a month, sourced from Mark’s Social Security, Linda’s pension from the school district, and withdrawals from their IRAs. Their upper guardrail is $1.95 million: if the portfolio reaches that balance, the plan calls for a raise. Their lower guardrail is $1.15 million: at that balance, the plan calls for slowing down. Between the two sits the no-change range, a range of portfolio balances where nothing needs to change. These guardrails are point-in-time estimates, and they should change. If Mark and Linda had the same guardrails ten years from now that they have today, something would be very wrong; markets move, inflation accumulates, and the two of them get older. Guardrails need to change to reflect that. (The inability of Guyton-Klinger withdrawal rate guardrails to change over time to reflect changes in the plan and the world is one of the key weaknesses of that approach.)

The upper guardrail is permission to spend more. Mark and Linda can reach it three ways: the portfolio grows faster than expected, they spend less than expected, or inflation runs lower than expected. When they get there, the increase goes all the way back to the target spending level for their chosen risk posture, here from $9,200 to roughly $10,200 a month. The raise is full-speed on purpose: underspending and regret is the most common negative retirement outcome we see, and the upper guardrail is the structural defense against it.

The lower guardrail is a yield sign, not a stop sign. Suppose the portfolio drops to $1.15 million. At that balance, reverting to a spending level that completely resets risk could require a large change. But more often than not, that would be unnecessary and lead to “whipsawing”. In order to avoid unnecessary cuts, the plan calls for closing about 10% of the gap between current spending and that original risk target, roughly $600 a month, and then it keeps monitoring. If conditions keep deteriorating, the plan may call for another measured step. The gradualism is anti-whipsaw engineering: deep cuts made on a temporary drawdown have to be reversed when markets recover, and a plan that swings a household from $9,200 down to $6,600 and back again has not protected anyone. And the conversation also doesn’t have to end in a spending cut at all: shifting the size or cadence of an expense (a new car every 7 years instead of every 5) or revisiting Social Security claiming timing can do the same work.

Two more mechanics matter. A minimum-change threshold, 5% by default, filters out adjustments too small to matter, whether they come from a guardrail or from inflation that has cut purchasing power by 5% or more, so clients are not whipsawed by noise between their reviews at whatever pace the practice runs (quarterly, semi-annual, annual). And the framework’s risk posture is explicit and chosen, with a default of 20% overspending risk and 80% underspending risk.

There is no probability of success score anywhere in this framework, by design; the full argument for eliminating the score lives in why probability of success can mislead advisors. For the complete methodology, see the complete guide to retirement income guardrails.

How Guyton-Klinger guardrails work, and where they break down

Guyton-Klinger guardrails deserve description in their own terms. The retiree starts at an initial withdrawal rate, with the 2006 paper supporting starting rates around 5.2% to 5.6% for equity-heavy portfolios, meaningfully above the 4% rule. A set of decision rules then governs the plan. The capital preservation rule: if the current withdrawal rate rises more than 20% above the initial rate (a 5% starting rate drifting above 6%), cut spending by 10%. The prosperity rule: if the rate falls more than 20% below the initial rate, raise spending by 10%. The inflation rules govern when annual cost-of-living raises are granted or skipped after bad market years. These are fixed percentage triggers, and inside the Guyton-Klinger framework that is exactly the point: rules simple enough to run in a spreadsheet.

The mechanics break down in four specific places.

Real spending is not a steady withdrawal rate

Real retirement cash flows are lumpy. A client retires at 64, bridges on portfolio withdrawals until Social Security begins at 70, and pays off a mortgage at 72. Withdrawals start high and step down, the pattern Derek Tharp and I called the retirement distribution hatchet. Under Guyton-Klinger, that planned, healthy spike in early withdrawal rates is indistinguishable from a failing plan: the framework cannot tell the difference between “the portfolio is in trouble” and “Social Security simply hasn’t started yet,” so it cuts spending exactly when the client is supposed to be spending the most.

Withdrawal rate is one risk signal among many

A retirement plan carries risk from longevity, from mortality in joint plans, from inflation, from taxes, and from the sequencing of market returns. The withdrawal rate compresses all of that into a single ratio, and the compression discards most of the information. One household at the same withdrawal rate might be 85 with a pension covering essentials; another might be 62 with nothing but the portfolio for thirty more years.

The triggers never age with the client

Withdrawal-rate guardrails are static across a client’s lifetime. A 6% trigger set at age 60 applies the same way at age 85. But 6% may be perfectly fine (or even too low!) at 85, with a much shorter remaining horizon, and genuinely dangerous at 60. Real guardrails must update as the client ages, and a fixed band around a fixed starting rate never does.

The cuts run far deeper than they need to

These three structural problems compound into the practical one: under real historical stress, withdrawal-rate guardrails often produce large, unnecessary reductions in spending. In our 2024 Kitces.com analysis, the Guyton-Klinger retiree took a 28% income cut through the Global Financial Crisis while the risk-based retiree took 3%. In the 1970s stagflation, the comparison was 54% versus 32%. In the Great Depression, 45% versus 8%. In the dot-com bust, 36% versus 0%. The pattern is whipsaw: the capital preservation rule triggers repeatedly on the way down, stacking 10% cuts, and then the prosperity rule pushes income back up after recovery (often long after). The client lived every one of those cuts, and most of them turned out to be unnecessary.

This is not one team’s view. Research independently produced by Fitzpatrick & Tharp (Kitces.com, 2021 and 2024), Wade Pfau (“Making Sense Out of Variable Spending Strategies for Retirees,” Journal of Financial Planning, October 2015), and Karsten Jeske (Early Retirement Now, February 2017) converges on the same finding: withdrawal-rate guardrails break down under real-world conditions.

For the full walkthrough of these stress periods, see why Guyton-Klinger guardrails are too risky for most retirees. From here, this guide turns from critique to the decision an advisor actually faces.

The two frameworks, dimension by dimension

One row deserves a second look, because both frameworks use the number 10% and mean entirely different things by it. Guyton-Klinger cuts 10% of spending each time a trigger is breached, and in a prolonged decline those cuts stack. Risk-based guardrails close about 10% of the distance between current spending and the recalculated target, a far smaller first step, followed by more monitoring rather than another automatic cut.

Which framework?

Both risk-based and withdrawal-rate guardrails answer the same client question: “How much can I spend?” It’s understandable that some advisors may have adopted Guyton-Klinger guardrails methodology early on. After all, its promise seemed impressive and it was relatively easy to do with a spreadsheet. Pfau and Jeske’s warnings about the extreme weakness of the approach in practice might have been missed, or might have been heard but ignored. After all, running Guyton-Klinger guardrails after the Global Financial Crisis (when they first became well known) certainly didn’t lead anyone to huge unneeded pay cuts! Instead, rising markets probably led to increases in spending. What’s to worry about?

However, now that markets have seen some testing (2022 inflation, etc.), it’s time to reevaluate. Furthermore, dependable technology (Income Lab) is now available to deploy and manage risk-based guardrails. For most modern advisory practices, that means risk-based guardrails are the default for retirement income planning and management. Social Security claiming decisions, required minimum distributions (RMDs), multiple account types with different tax treatment, and bridge periods before benefits begin all break the steady-withdrawal-rate assumption Guyton-Klinger depends on, and joint plans and variable spending plans sit even further outside it. If the practice runs software that can recalculate the whole plan monthly, the case for keeping a withdrawal-rate rule of thumb largely disappears.

To see how the decision plays out on a real client plan, Book a walkthrough.

How Income Lab makes risk-based guardrails practical

There are no simple rules of thumb here. Risk-based guardrails require sophisticated software, which is why Income Lab built its retirement income planning software around the framework from the start.

The spending capacity engine calculates the retirement paycheck as a dollar amount, not a probability score, and bases this (and the guardrails) on the full plan: accounts, cash flows, longevity, mortality, inflation. Track & Monitor then recalculates the guardrails monthly, and monitored plans have inertia by design: the plan calls for a change only when a guardrail is breached or accumulated inflation crosses the minimum practical threshold.

When a client asks what guardrails would feel like to live with, the Retirement Stress Test answers with stories instead of statistics. It runs the client’s own plan through real historical periods, the Great Depression, 1970s stagflation, the dot-com bust, the Global Financial Crisis, and shows when each guardrail would have been hit and what the adjustments would have been. Walking a client through two or three of these periods does more for their understanding than any explanation of the math behind the methodology.

FAQ

What is the difference between Guyton-Klinger guardrails and risk-based guardrails?

Guyton-Klinger guardrails trigger on the portfolio withdrawal rate: when it drifts a fixed percentage above or below the starting rate, spending is cut or raised by 10%. Risk-based guardrails trigger on the risk of the whole plan, accounting for investments, cash flows, longevity, mortality, and inflation, recalculated monthly and expressed as portfolio balances that update as the client’s circumstances change.

Are risk-based guardrails better than Guyton-Klinger guardrails?

Under real historical stress periods and hypothetical (Monte Carlo) scenarios, yes, by a wide margin. In the 2024 Kitces analysis by Fitzpatrick and Tharp, Guyton-Klinger called for a 28% income cut through the Global Financial Crisis versus 3% for risk-based guardrails, with similar gaps in stagflation (54% versus 32%), the Great Depression (45% versus 8%), and the dot-com bust (36% versus 0%).

Who developed risk-based guardrails?

Risk-based guardrails as a published methodology originated at Income Lab. Justin Fitzpatrick, PhD, CFA, CFP© (President, Income Lab) and Derek Tharp, PhD, CFP© (Associate Professor of Finance, University of Southern Maine) published the framework on Kitces.com in 2021 in “The Retirement Distribution ‘Hatchet’: Using Risk-Based Guardrails To Project Sustainable Cash Flows,” as well as in several subsequent works. Guardrails as a withdrawal-rate concept were introduced earlier by several researchers, but were made especially well-known by Jonathan Guyton in 2004 and formalized with William Klinger in 2006.

Can risk-based guardrails be implemented without specialized software?

No. The framework requires recalculating the risk of the entire plan every month, accounting for longevity, mortality, inflation, taxes, and every non-portfolio income stream. That math is not a rule of thumb, and there is no spreadsheet version of it. The older framework is hand-operable precisely because it ignores most of what the newer one measures.

Do risk-based guardrails use Monte Carlo?

They can. The framework is method-agnostic: the trigger is total plan risk, however that risk is assessed. In Income Lab, the default analysis runs plans through actual historical sequences of returns and inflation, but traditional and regime-based Monte Carlo analysis methods are also available. (Monte Carlo is a statistical method; probability of success is a metric derived from it. Risk-based guardrails can use Monte Carlo as an engine while eliminating probability of success as the thing clients are shown.)

Recommended reading

• The complete guide to retirement income guardrails
• Why Guyton-Klinger guardrails are too risky for most retirees
• Derek Tharp and Justin Fitzpatrick’s risk-based guardrails research
• Why probability of success can mislead advisors
• How much can a retiree spend?

The framework an advisor chooses is, in the end, a structure for answering one client question: “How much can I spend?” Guyton-Klinger answers it with a withdrawal-rate rule of thumb. Risk-based guardrails answer it with total plan risk, recalculated monthly and delivered in dollars. The decision is which answer matches the practice you actually run and the clients in front of you. Book a walkthrough to see both halves of that answer on a plan like the one in your next review meeting.