3.3 Modern Safe Withdrawal Research & Failure-Rate Modeling

Modern Safe Withdrawal Research & Failure-Rate Modeling

Executive Summary: Contemporary decumulation science has progressed beyond rigid historical averages. Research by David Blanchett demonstrates that real retiree expenditures follow a Retirement Spending Smile, declining through early retirement before rising due to late-life healthcare. Concurrently, Dr. Wade Pfau established that sustainable withdrawal rates depend on starting market valuations (Shiller CAPE). Modern modeling replaces simplistic binary failure rates with analyses of the magnitude of failure and shortfall severity.

---

David Blanchett & The Retirement Spending Smile

William Bengen and the Trinity Study assumed that retiree spending remains constant in real terms, increasing each year by headline CPI.

In his 2014 study "Exploring the Retirement Consumption Puzzle," Morningstar's David Blanchett analyzed empirical expenditure data from the U.S. Bureau of Labor Statistics' Consumer Expenditure Survey (CEX). Blanchett discovered that actual retiree expenditures follow a U-shaped Retirement Spending Smile:

1. The Go-Go Years (Ages 65 to 74): Early retirement involves travel, hobbies, and leisure. However, work-related expenditures (commuting, clothing, payroll taxes, retirement savings) cease. After adjusting for inflation, real household spending declines by about 1% a year on average over retirement. The rate of decline varies with the household's spending level and stage of retirement.
2. The Slow-Go Years (Ages 75 to 84): Physical mobility slows and travel diminishes. Real discretionary spending reaches its lifetime nadir (lowest point).
3. The No-Go Years (Ages 85+): Discretionary spending remains minimal, but total spending inflects sharply upward due to out-of-pocket medical bills, in-home care, and assisted living or nursing home costs.

Planning Impact

Because real spending tends to dip through mid-retirement, assuming flat real spending overstates the capital a retiree needs. Using realistic spending paths instead of constant inflation-adjusted spending lowers the wealth needed at retirement. Wade Pfau's application of Blanchett's pattern found almost 15% less in a historical worst case, and some estimates run up to about 20%. Equivalently, the same savings can support a somewhat higher starting withdrawal at the same level of safety. The benefit varies by household. Later replication work also finds the late-life upturn weaker when the same households are tracked over time, so planners should still budget for rising health and care costs.

Wade Pfau: Valuation-Based Withdrawal Rates

Research by Wade Pfau and by Michael Kitces showed that the historical safe withdrawal rate was not a single fixed number. It depended heavily on market conditions at the start of retirement, especially:

1. The Cyclically Adjusted Price-to-Earnings (CAPE) ratio (Shiller P/E10), which compares the S&P 500 price with 10-year average real earnings.
2. Prevailing bond yields and the dividend yield, which shape the income a balanced portfolio can generate early on.

Researchers also caution that U.S. history is only one sample. With today's valuations and yields, forward-looking Monte Carlo models often put sustainable 30-year rates below Bengen's historical 4%, unless the retiree is willing to adjust spending.

---

The Actuarial Approach: The RMD Method

The actuarial approach dynamically recalibrates distributions annually based on remaining life expectancy divisors, commonly implemented via the IRS Uniform Lifetime Table (RMD method):

$\text{Annual Distribution} = \frac{\text{Prior Year-End Portfolio Balance}}{\text{Remaining Life Expectancy Divisor}}$

• Immunity to Depletion: Because the annual withdrawal is a mathematical fraction (dividing by a factor > 1.0), the portfolio can mathematically never deplete to zero.
• Trade-Off: Annual income fluctuates directly with portfolio volatility, posing challenges for retirees with high non-discretionary expenses.

---

Probability of Failure vs. Magnitude of Failure

Standard Monte Carlo simulations report plan viability as a binary probability of success (e.g., 85% success = 15% failure). However, binary metrics ignore crucial qualitative risk dimensions:

1. Timing of Failure: Exhausting funds at age 94 differs fundamentally from running out of money at age 74.
2. Magnitude of Failure (Shortfall Severity): A plan where a client experiences a minor $5,000 shortfall in year 29 is classified as a "failure" just like a plan where a client runs completely out of money 12 years into retirement.

Advisors must evaluate terminal wealth distributions and shortfall severity rather than relying blindly on binary pass/fail probabilities.

---

Case Example & Practical Calculations

The Taylor Retirement Analysis: Probability vs. Severity

Mark and Susan Taylor (age 66) have $2,000,000 in assets. Baseline living expenses require $45,000 from the portfolio (2.25%), but they seek $50,000 extra for early retirement travel ($95,000 total, or 4.75%):

• Static Model: Monte Carlo shows a 76% success rate (24% failure), with failing iterations causing complete depletion by age 81.
• Blanchett Smile + Floor: Modeling a 1.5% annual real decline in discretionary travel through age 78 increases the success rate to 89%. Adding a contingency rule (trimming travel by $15,000 if portfolio drops 25%) eliminates catastrophic early failure, leaving a worst-case minor shortfall of $8,000 at age 93.

---

Exam Tips & Common Traps

[!IMPORTANT] RICP Exam Traps for Section 3.3:

• Blanchett Spending Smile: Real spending declines during go-go/slow-go years (ages 65 to 75–84) and turns upward in late retirement (age 85+) due to medical costs.
• Valuation Metric: Pfau uses the Shiller CAPE ratio and 10-year Treasury yields to predict sustainable withdrawal rates, not trailing 12-month P/E.
• Exhaustion Immunity: The actuarial / RMD method mathematically eliminates the probability of portfolio exhaustion.
• Binary Metric Trap: Probability of failure does not reflect the timing or dollar magnitude of a shortfall.