Generation 2

Bengen (1994)  ushered in the second generation of safe withdrawal research when he published a groundbreaking Journal of Financial Planning article that still sets the basic framework employed by most research to this day. He used historical simulations of long-term U.S. securities index data to define “Safemax” as the highest withdrawal rate, expressed as a percentage of the account balance on the first day of retirement, and adjusted for inflation annually, that allowed for a minimum of 30 years of withdrawals over all rolling historical periods⁴ in the database (i.e. 1930-1960, 1931-1961, etc.).

Bengen concluded the maximum safe withdrawal rate was about 4.1% for stock allocations between 37%-67% and later upgraded that amount to 4.5% when small cap stocks were included in subsequent research.

Cooley, Hubbard, and Walz (1998) further developed the 2^nd Generation model  by showing a 95–98% chance you won’t run out of money applying a 4% withdrawal rate. This study, dubbed the “Trinity study”, and popularized by Dallas Morning News columnist, Scott Burns, was updated in the April 2011 Journal of Financial Planning with similar results.

Other 2^nd Generation researchers have also come to similar conclusions depending on assumptions applied. Results vary slightly from study to study based on asset allocation⁵, data sources, whether or not fees are deducted, frequency of portfolio rebalancing, and much more.

The key point defining all 2^nd Generation research is that each study applies the same basic premises thus producing extraordinarily consistent results. This consistency caused the 4% Rule to become conventional wisdom and be mistaken as “truth” when it is really just a product of the research premises.

To understand the problems with 2^nd Generation research we need look no further than the amazing breadth of dubious assumptions behind the results.

• Safe withdrawal rate research was based on U.S. securities markets history. No foreign market data was included.

• The research typically assumes a static allocation⁶ to stock and bond indexes as the only viable asset classes. Alternative assets⁷ are not included.

• It assumes 30 years of retirement spending regardless of expected longevity.

• It assumes no investment expenses by using historical index data—obviously not true for any real-world investor.

• It assumes a fixed spending amount that grows with inflation and does not adjust based on changes in portfolio value or age—not true for most retirees.

• It seeks to determine the highest beginning spending amount that can be adjusted for inflation without ever running out of money that works across all time periods in the database—a sort of least common denominator concept. Unfortunately, this forces the lowest valid choice as the answer for everyone when other time periods would have allowed greater spending and lifestyle. It implies one size fits all when the data clearly demonstrates a different conclusion.

• All data periods are created equal with no adjustment for valuations or interest rates at the time you retire.

• You must have 30 years of subsequent data to know if the withdrawal rate was actually safe meaning retirements beginning after 1990 won’t be known until after 2020..

While these assumptions make for expedient research, there is a clear sacrifice of accuracy when compared to the real-world retirement you will face.

Below, I examine each of these assumptions in detail to show you the implications so that you can decide how relevant 2^nd Generation research conclusions—the 4% Rule—are to your retirement planning.