Most discussions about retirement with consumers examine accumulation separately from decumulation. The research mostly addresses the years before retirement and how to achieve a wealth accumulation target.

Wade Pfau, an associate professor at the National Graduate Institute for Policy Studies in Tokyo, has noticed that the preponderance of research seeks a safe withdrawal rate. The typical retirement planning then uses this rate to compute a target for how much wealth needs to be accumulated so that the desired retirement spending can be funded from this wealth.

Pfau, however, who earned his Ph.D. in economics at Princeton in 2003, wonders in a research paper what would happen if he linked the accumulation and distribution phases together in an integrated whole.

"My findings suggest that a fundamental rethink about retirement planning is needed," he says. "When linking the accumulation and decumulation phases together, the concepts of 'safe withdrawal rates' and 'wealth accumulation targets' end up serving as almost an afterthought. Focusing on them is the wrong way to think about retirement planning."

Pfau's paper, "Safe Savings Rates: A New Approach to Retirement Planning Over the Lifecycle," which appeared in the May 2011 issue of the Journal of Financial Planning, hopes to frame the question of how much one needs to save, recognizing both one's saving for retirement and one's spending through retirement. If there is a "safe withdrawal rate" there should be a "safe savings rate." At what rate of savings does it always work out?

To get to this rate, he set up the problem like this: "The baseline individual wishes to withdraw an inflation-adjusted 50% of her final salary from her investment portfolio at the beginning of each year for a 30-year retirement period. Prior to retiring, she earns a constant real salary over 30 working years, and her objective is to determine the minimum necessary savings rate to be able to finance her desired retirement expenditures. Her asset allocation during the entire 60-year period is 60/40 for stocks and bills. Data is from Robert Shiller's Web page for the S&P 500 and Treasury bills."

Put another way, consider a person making $100,000 today who expects that salary to increase with inflation and who wants to withdraw $50,000 in today's dollars from her portfolio to supplement any other income, adjusting for inflation over a 30-year retirement. Based on market behavior since 1871, a 60/40 mix and ignoring taxes, there was never a time where saving at least X% of her salary for 30 years prior to that retirement didn't achieve those goals. Pfau solved for X.

With the data going back to 1871, Pfau examined 30 years of savings followed by 30 years of withdrawals. Retirements would therefore have beginning dates ranging from 1901 to 1980. He found extreme volatility in the maximum withdrawal rates that turned out to be sustainable, ranging from just over 4% up to 10% in some periods. As one might expect, the higher withdrawal rates correlated to better markets and the lower rates to weaker markets during retirement.

Of course, a retiree does not know what the markets will do ahead of time. So assuming that the goal was to accumulate enough to use a 4% withdrawal rate, Pfau calculated what savings rate was needed to accumulate that amount. This is very similar to how most people approach retirement planning-isolating an accumulation goal to support an isolated set of parameters during distribution.

The savings rates required to accumulate enough to employ the 4% rule were every bit as volatile as the differing withdrawal rates. They ranged from 10.89% to 37.7%. If markets did well during retirement, less was needed to start the final 30 years, and if markets were poor, more was needed. When Pfau noticed that the periods of high required savings corresponded to periods where far less was actually needed during the subsequent retirement, he abandoned the 4% number and used actual retirement results to determine the needed savings. The volatility of the needed savings amount dropped significantly.