The four rules
This page explains the four Guyton-Klinger decision rules. The Guyton-Klinger calculator is the page that runs them on a portfolio you type.
Jonathan Guyton and William Klinger, “Decision Rules and Maximum Initial Withdrawal Rates,” Journal of Financial Planning (2006), published four decision rules, not a formula.
- Initial withdrawal. Year-one spending is the initial withdrawal rate times the portfolio. 4% of $1 million is $40,000.
- Inflation rule. Raise spending with inflation each year. After a year the portfolio lost money, skip the raise if it would put the withdrawal rate above the initial rate. A skipped raise is not made up later. In the last 15 years of the plan the raise is capped at 6%.
- Capital preservation. If this year's withdrawal over the current portfolio sits above the initial rate × (1 + band), cut spending by the adjustment. The 2006 defaults are a 20% band and a 10% cut, so 4% trips at 4.8%. The rule stops 15 years before the end of the plan: planning to age 100, there are no cuts from 85 on.
- Prosperity. If the current rate sits below the initial rate × (1 − band), raise spending by the adjustment. 4% trips a raise at 3.2%.
Why a constant return understates the rails
On a 5% every-year path the withdrawal rate barely moves, so the rails almost never trip. That is an honest teaching case and a bad stress test. The calculator therefore has a historical mode (every rolling S&P window since 1928) and a 1,000-path mode. White Coat Investor's explainer is the best prose on the rules; this page plus the calculator is the thing that actually runs them.
Versus risk-based rails
Kitces and Tharp argued the tripwire should be remaining probability of success, not the current withdrawal rate. A 5.5% withdrawal with ten years left may still be fine; a 4.2% withdrawal with forty years left may not. The risk-based calculator is that critique made runnable. The 4% vs guardrails page is the comparison most people mean.
To apply the four rules to a portfolio, use the Guyton-Klinger calculator.