With the Federal Reserve seemingly committed to its current rate-hike cycle despite mixed economic data (though the market appears less convinced at the moment), the ongoing panic about the appropriateness of the “4% rule” for retirement asset withdrawals seems to have dulled a bit for the time being.

Of course, even though short-term interest rates are on the rise, longer-term rates have remained largely unchanged10-year and 30-year Treasury yields have both declined by roughly 20 basis points year-to-date. So, even though the withdrawal-rate debate may not currently be raging, it remains highly relevant to the portfolios of retirees everywhere.

Whether the 4% rule is “safe” or not, though, is a debate that largely misses the point. Like many personal finance rules of thumb, the problem with the 4% rule is not just its inflexibility, but its oversimplicity. Like the traditional portfolio “glide path” recommendations (“100-minus-your age” in stocks, for instance), the 4% rule is often thoughtlessly applied as a one-size-fits-all mathematical calculation, ignoring any and all particulars of an individual’s overall financial picture that might make such a recommendation inappropriate or inadequate.

In a past article, I discussed the differences between risk tolerance, risk capacity, and required return, suggesting that a planning-based approach toward determining the necessary investment return was much more important and helpful in determining investment risk than any easily-remembered glide-path rule of thumb ever could be.

The same is true with retirement year withdrawal rates: What makes most formulaic recommendations so incomplete is that they include little or no mention of a retiree’s expenses, either now or in the future. Just as a portfolio risk assessment must include some consideration of future financial needs, so too must an asset withdrawal plan. Retirees should not be thinking about how much they can withdraw from their portfolios, but instead how much they must withdraw.

Starting with the ultimate goal in mind can and will enable retirees to also properly determine how much risk is optimal for their portfolio; I’ll help explain how.

Summarizing the withdrawal rate debate

To understand where the 4% withdrawal rule falls short requires some discussion of the rule’s history, as well as some of its proposed replacements.

Contrary to what many seem to believe, the 4% “rule” is not gospel, nor is it even a particularly long-standing part of the financial planning landscape. In fact, the rule dates back only to 1994, stemming from the research of one California-based financial planner, Bill Bengen. As the New York Times reported, “Mr. Bengen’s original analysis assumed the retirees’ portfolio was evenly split between stocks and bonds, and he tested whether the paycheck could persevere through every 30-year period dating from 1926. It succeeded.”

As even Bengen later conceded, the initial analysis was fairly simplistic. Not only had he considered a basic two-asset portfolio with little dynamic response to market conditions over time (as opposed to the seven to nine asset classes, with rebalancing, that most planners now recommend for a diversified portfolio), but it also required a 100% success rate, thereby making it overly conservative for most “average” market environments. Using the 4% rule, in most situations, would result in significant leftover cash at death.

Bengen, in fact, later raised his recommended withdrawal rate to 4.5% as a result—one could argue, then, that knocking down the 4% withdrawal threshold in response to current market conditions is unnecessary, since the 4% figure itself already accounted for some fairly harsh market conditions in the history that Bengen studied.

Nevertheless, the debate persists. One recently-shared piece suggested that the 4% rate should be replaced by a “your age divided by 20” withdrawal rate. Other methods include a more dynamic (but much more difficult to understand and implement) “floor and ceiling” method, which withdraws different amounts depending on annual stock market returns.

There are also recommendations that center around studying market P/E ratios, “bucket”-based strategies that hold large cash reserves, and rules that respond directly to trends in inflation, as measured by the CPI. Some research, from Michael Kitces and Dr. Wade Pfau, even suggests increasing portfolio risk in the later retirement years, rather than decreasing it in accordance with traditional asset allocation theory.

None of these rules or methods is necessarily right or wrong, and in fact, all of them are useful to consider (indeed, as the statistician George Box once said, and I’m always paraphrasing, “all models are wrong, but some are useful”). Where they fall short, though, is in their insistent focus that we should be focusing on the income side first, rather than on the expense side.

What that sets up, then, is a bit of a moral hazard on the part of the retirees: If they’re withdrawing more than they actually need in the early years of retirement, simply because a financial planning aphorism says that they can, will they be incurring discretionary expenses in those years that they otherwise might not? And could those added expenses earlier in retirement in fact endanger their ability to meet their basic needs in their later years, especially if unexpected expenses arise, or if they live longer than originally expected?

Clearly, there is another aspect that needs to be considered, and an expense-focused withdrawal rule may be just what we need.

Considering the expenses first

One of the first personal finance articles that I wrote for public consumption on the internet surrounded the topic of inflation, and the need for us to rely much less on summary measures of cost pressures over time. After all, no two individuals’ spending patterns are exactly alike, so assuming, as CPI does, that there is any one “basket of goods” that can summarize an individual’s budget is foolhardy and unhelpful.

In reality, we are all subject to our own “personal inflation rate,” and it’s our own expense patterns—and the changes in those expenses over time—that will ultimately need to be covered by our retirement assets. Why, then, would we not start there, rather than starting with a retirement balance and then dividing by a seemingly arbitrary number?

Any number of factors can impact an individual’s expenses, and thus, personal inflation rate. Certainly, regional variations come into play (those in cold-weather states will spend more on heating oil, whereas those in warm-weather states will presumably face lower costs for fresh produce), as will income levels (the poorest 20% of Americans spend 60% of their income on food and energy; the richest 20% of Americans spend less than 10%). Spending patterns will also change as we age, with health expenditures taking up a greater share of the expense pie with each passing year, generally replacing travel and other discretionary expenses.

Finally, the issue of housing is a major one: The CPI includes a 30% weighting to housing cost (based largely on “owner-equivalent rent”), but not all people have the same exposure to changes in housing cost. For homeowners, housing cost is largely a fixed expenditure until the mortgage is retired, and then the out-of-pocket cost drops precipitously to near-zero, depending on the property tax rates of the region in question. For renters, though, housing cost can sometimes fluctuate wildly, and is generally much harder to project or predict.

Should a renter, then, be more or less comfortable with the 4% withdrawal rate than someone who owns their home outright? Furthermore, how should the 4% rule be treated if someone retires with 5 years remaining on their primary mortgage? That 4% annuity might not even cover the mortgage in the early years, but it might be more than enough once the mortgage is retired. What’s a retiree to do?

Finally, family aspects and health-related considerations must come into play. Nationwide, general life expectancy continues to rise (despite a slight blip downward last year), having gained about 3 years on average since Bengen first released his 4% rule research. That said, the gains in life expectancy have not been evenly distributed, For example, white women, in particular, have not shared in the recent life-expectancy gains.

Taking into consideration demographics and family history will go a long way toward determining how many years of retirement expenses need to be taken into consideration (remember, Bengen based his research around a 30-year horizon; some retirement horizons may be much longer or shorter than that).

In addition, the prospect of long-term care is an important one for many retirees. After all, nearly 70% of current 65-year-olds are projected to need some form of long-term care at some point in their lives, and costs of high-quality care are continually on the rise. For some retirees, the issue of long-term care may be a moot point, because they either have in-force insurance policies to cover them, or helpful family members who stand ready to step in. For others, though, some form of “self-insurance” may be necessary, and increased savings (via lower discretionary expenses) in early retirement years may be absolutely vital.

Simply put, an assumption of linear expenses in retirement is a simplistic one, and a fundamentally flawed one, and no withdrawal method should assume that such an assumption is reasonable. Instead of focusing on how much we can “afford” to withdraw from our retirement accounts each year, we absolutely must flip the equation on its head and consider instead how much we in fact need to be withdrawing. If it turns out that we need more than can reasonably be withdrawn from our accounts without bleeding them dry, then we will know that expense adjustments are necessary, and we can make them before it’s too late.

But starting on the income side and then fitting expenses around that income is a failure of basic financial planning and budgeting. While we may tell those in their prime working years to “pay themselves first” in order to save enough for retirement, a similar approach in retirement requires quite different advice. Consider your basic expense needs, assess your various resources, and then determine which adjustments (if any) need to be made.

And if, after taking this new approach to retirement sufficiency analysis, you’ve determined that you have more than enough in your retirement accounts to cover your expense needs, congratulations. You then have two options: You can either increase your discretionary expenses (or your bequests), or you can dial back your portfolio risk to a more comfortable level, since your required return is obviously lower than the traditional advice has suggested is necessary.

Don’t just put your investments in a 50-50 portfolio and mindlessly withdraw 4% (or 3%, or 5%, or any other formulaic number) each year. Determine what you need, and then fit your portfolio approach and your withdrawal strategy around that. That’s in keeping with the true spirit of financial planning, and simplistic withdrawal rules should be left to the textbooks where they belong.