The 4% rule has become something like a religious topic to retirement planners. In some way, it’s become a holy number like pi—a guideline, a rule of thumb, or a concept to be snubbed. Even if you use safe 4% withdrawal rates as a framework for clients, you probably have to come up with ways to innovate around it.

Bill Bengen himself, the author of the rule, has said on occasion that there are a variety of numbers that could work depending on when you retire. The 4% rule assumes you can take a safe annual withdrawal of 4% and then can adjust for inflation. He has adjusted that rule and called it SAFEMAX, 4.5%, when the allocation includes small-caps, and the rule was designed to cover the worst 30-year retirement periods, especially the era of high inflation that started in 1968.

Advisors at the Atlanta firm Brightworth say they have found their own new wrinkle for the number. According to research run by one of the firm’s financial planners, Wesley Wood, the simple tweak of making more frequent withdrawals, not just modeling a lump sum withdrawal for the year, can allow advisors to squeeze extra basis points out of a portfolio and increase its longevity by years.

“We really wanted to re-examine a lot of research that we had read but hadn’t done ourselves,” Wood says. “So we took monthly historical data from different indices and built some sample portfolios and back-tested them with rolling monthly periods starting in 1926 through the present.”

What the Brightworth team found intriguing was what happened when the retirees took withdrawals monthly or quarterly, as most do, instead of taking the annual lump. Doing it that way often allowed the investors to take out 20 extra basis points a year. “So instead of 4% it might be more like 4.2%,” Wood says. “It might not sound like much, but that’s 5% more in spending,” he says. “So it could mean an extra vacation or a nicer car. Giving more to charity or something like that.

“It wasn’t necessarily what we went out to test. We just wanted to validate the research that was already out there. But it was something we stumbled upon.”

The math generally works, he says, because the market historically has been up more than it’s been down. So if you’re taking a lump sum distribution for all of your spending needs the first day of the year, you have less in funds to go up over the course of the year. “If you take it monthly or quarterly, you’re leaving it a month in the account to grow.” He calls it the opposite of dollar cost averaging.

There are also some behavioral advantages, he says: Clients are used to receiving a paycheck every two weeks, and so it’s easier for them to budget that than take a single lump-sum withdrawal on the first day of each year. “Most research that I’ve read [assumes the withdrawals are] annual,” Wood says, and most of his clients are taking the quarterly and monthly withdrawals. Again, if you keep it invested, you participate in more growth, he says.

The firm looked at rolling 30-year periods starting every month from January 1926 to the present, for approximately 770 payment periods, he says. The 4% rule is generally based on the worst 30-year period, he says, so many periods can get you extra withdrawals beyond 20 extra basis points.

Coming after a 17-year bull market and just before a decade of stagflation, a December 1968 retirement was the very worst period, no matter what kind of withdrawals you were taking, Wood says. Bengen’s own research has confirmed as much.

Still, if you took out monthly withdrawals or quarterly withdrawals, “it increased the rate you could take out by about 20 basis points,” Wood says. More often than not you were 20 basis points better in the rolling periods, though there were a few instances, he says, when the annual withdrawals would have been better (generally during years of market pullbacks). More often than not, though, you were 20 basis points better off with more frequent withdrawals.

The firm also used different allocations, using a portfolio melding the Russell 2000, the S&P 500 and the Bloomberg Barclays U.S. Aggregate Bond Index, mixing a 50% allocation to the S&P with a 10% allocation to the Russell index and a 40% allocation to the bond index. These allocations were then tweaked (the bond portfolio went anywhere from 30% to 50%).

The math was fairly linear and done in Excel with rebalancing, Wood says.

These findings also have ramifications for the longevity of portfolios, Wood says. In other words, it means something tangible and important for human beings who are living longer. “My great-grandmother recently passed away; she was 97,” he says. “When she was born, her life expectancy was around 60. So it would have been really easy not to protect for longevity for her and I’m glad her advisor did.”

So with the 50% large-cap/10% small-cap/40% bond portfolio, chosen for the worst 30-year period starting December 1968, Wood says 4% annual withdrawals run out somewhere around year 33, plus 10 or 11 months (he adds that other models often don’t include those small-cap allocations). With monthly withdrawals and small-caps included in his model, it’s running out around year 40. “So it’s seven extra years.”

Then you can take the 30-year period starting in September 1, 1929. In that case, the retiree never ran out of money, even after 50 years (the limit for the model). The reasons for that are complicated—one is the performance of small-cap stocks over that period, and the other is the period of deflation it covered during the Great Depression. Annual and monthly withdrawals both survived the turmoil.

Wood says that many of the models for 30-year-return results, more than half of the ones he’s seen, are using just the S&P 500 and bonds. “But our clients are much more diversified than that.” Typically adding other asset classes improves the safe withdrawal rate, he says.

These kinds of tweaks can offer dramatic changes, especially when retiring in one off year can throw off your sequence of returns and determine whether your portfolio is a smooth train ride or a mine car going off a cliff.

Wood takes three years close together as an example. Starting December 1, 1966, with annual withdrawals, the safe withdrawal rate was 4.88%, That number was 4.41% if you retired in December 1, 1967—a 47 basis point difference just one year later—and in December 1968, that safe rate had fallen to 4.09%.

Now go to monthly withdrawal rates. According to Wood’s calculations, the safe rate was 5.09% for December 1966—a 21 basis point upgrade from the annual withdrawal rate. The safe rate for December 1967 was 4.62%, another 21-point boost; and it was 4.29% for the locust year of December 1968, the beginning of the most challenging retirement period.

There’s much better news for more recent retirees: He takes the period starting with January 1990, which encompasses the tech bubble and the 2008 financial crisis. With annual withdrawals, a retiree could safely take out 7.4%, he says. With monthly withdrawals, it’s 7.5%, so it’s one of those situations where there is not much difference. “I think it’s because the first years matter the most, and if we’re starting in 1990, you’re getting so much growth at the beginning of your time, you’re getting 10 years of growth before you’re getting the tech bubble. By then you get another eight years of growth before you hit the Great Recession.”

He says that his model differs from Bill Bengen’s because Bengen has more in small-caps in his 4.5% withdrawal rate model—20% not 10%—which makes a big difference. “Another reason is that he’s doing annual withdrawals instead of monthly.” The data set might have also made a small difference (Brightworth used Morningstar data).

Brightworth co-founder and advisor Dave Polstra and Wood plan to present these findings at the Inside Retirement conference on May 2.

Bill Bengen himself weighs in on what might happen with more frequent withdrawals: “Although I have not tested withdrawals more frequently than annually with my models,” Bengen says, “the conclusions you cited make sense. I don’t know any other researcher who has used quarterly withdrawals. I used annual withdrawals to generate my 4.5% maximum safe withdrawal rate. My withdrawals were scheduled at the end of each year, not the beginning (with a little tweak to partially offset the timing). Other researchers I have spoken to computed safe withdrawals using annual withdrawals made at the beginning of each year. They generally arrive at significantly lower safe withdrawal rates, about 4.1% to 4.2%.

“Thus, it seems likely to me that using quarterly withdrawals, an intermediate approach, would yield intermediate results.”