Introduction to retirement spending models

When most people think of retirement planning, one of their first questions is, “What will it cost me to live as a retiree?” Answering this question can seem overwhelming for many people. The guidelines presented here to help answer this question are referred to as models of retirement spending.

Retirement spending models are primarily used to help calculate an estimate for the Total Personal Savings needed at retirement. Price inflation must also be included in this calculation. It is common practice to separate nominal retirement spending into two independent components: real spending, and an inflation adjustment. This article deals only with the real spending component. See Inflation and retirement spending for a discussion of the inflation adjustment component.

Estimating your retirement spending is easier if it is broken down into 2 steps: 1) Estimate an Initial Spending at the start of retirement, followed by 2) Extrapolating that initial spending as retirement progresses until eventual death.

Generally 2 types of models are used to describe a retiree's Initial Spending:
 *  Replacement Rate (or ratio) models estimate the total income needed in retirement by multipying total income before retirement by a constant, the Replacement Rate. Generic estimates for a retiree's replacement rate range from 0.7 to 0.85. But more personalized replacement rates can be obtained from the tables published by the GSU/Aon RETIRE Project.


 * Budget or Expense models come in a variety of forms. They all have in common the estimation of total retirement spending by working with estimates of spending in numerous, smaller budget categories.  The most common type of budgeting approach involves the use of a  Single budget to describe all retirement expenses.  The use of a  Dual budget (having a lower Essential spending and a higher Preferred spending) provides a more flexible framework for extrapolating spending as retirement progresses.

Models used to extrapolate retiree spending until death usually build upon one of the Initial Spending model estimates. Such models can be grouped into 5 broad categories:
 *  The Constant (Real) Spending model assumes that the real spending at the start of retirement continues unchanged until death. Although this approach is widely used, it doesn’t correspond to the typical spending patterns exhibited by retirees.  It usually (but not always) leads to an overestimate of the savings needed at retirement.


 * The Stages of Retirement model breaks up retirement into 3 or 4 age ranges, each of which has a different real spending pattern. Spending within each range can either be constant or varying.  This model does a much better job of representing the diverse spending patterns exhibited by retirees.


 * Investment Returns Dependent models allow spending in retirement to vary based on how total savings or investment returns change over time. These models are best combined with Dual Budget models of initial retirement spending.  These models mimic the tendency of retirees to vary their spending based on their net worth.


 * Flexible Spending models allow numerous individual categories of spending to be separately incorporated into the total spending plan. Each spending category can be assigned its own start and stop age, as well as spending amount.  This model is excellent at representing large step changes in spending during retirement.  Many retirement calculators will combine elements of this model with others already described.  This combination supplies a very flexible framework for accurately describing a diverse range of retiree spending patterns.


 *  Life Cycle models calculate how a household should smooth their spending over their entire lifetime, not just in retirement. They provide a detailed spending and savings plan that should be followed in the pre-retirement years.  Doing so will provide a similar lifestyle for both the pre- and post-retirement years.

All of these model types have been implemented in actual retirement calculators. Specific examples of each can be found in the article Retirement calculators and spending.

Introduction
When most people think of retirement planning, one of their first questions is, “What will it cost me to live as a retiree?” Answering that question can seem overwhelming for many people. Frankly, coming up with an exact answer is probably impossible: there are simply too many future unknowns. But that does not mean that useful estimates of retirement spending are impossible to obtain. What is needed are some guidelines to assist in the process of estimating retirement spending. These guidelines will be referred to as models of retirement spending.

When thinking about future retirement spending, there is always a question of how much detail to put into the models / guidelines. This question is at the heart of any economic model: how much can the reality of the situation be simplified and yet still maintain usefulness. As will be seen, many models of retirement spending are possible, ranging from the extremely simple to the detailed and complex. This article will review the main modeling approaches currently being used for retirement spending, explain their key simplifying assumptions, and give a general feel for their strengths and weaknesses.

Safe Withdrawal Rates
Many studies on retirement spending start with an estimate of future retirement savings and then estimate the spending that might be achievable in retirement. Such studies are often especially concerned with estimating the maximum spending that won’t lead to premature depletion of personal savings, known as the  Safe Withdrawal Rate. These models treat the maximum amount of savings as an independent variable. The amount of spending is adjusted so the savings last as long as planned.

This article series works from the other side of the equation and treats retirement spending as the independent variable in the retirement planning process. The future retiree first develops an estimate of their desired spending in retirement. From there, the planning process makes assumptions about longevity, inflation and investment returns, and then estimates the future savings required to potentially sustain that retirement spending.

Inflation
The effects of inflation on retirement spending are also excluded from this discussion. This is not meant to imply that inflation is an unimportant consideration. Just the opposite, inflation is probably the retiree’s worst enemy!  But the majority of retirement planning approaches treat inflation as an independent, adjustable variable. In doing so, they separate nominal retirement spending into two components: real spending, and inflation adjustments. This article follows the same approach. Inflation adjustments are covered in the article  Inflation and retirement spending.

Retirement calculator models
There is a strong connection between retirement spending models and retirement planning calculators. Most software used for retirement planning, whether explicitly or implicitly, is written assuming some model of retirement spending. The adjustable options included in the software reflect the choice of spending model. It is an unfortunate fact that some retirement calculators have great strengths in many aspects of their design, yet are needlessly weak in how they incorporate retirement expenses.

The usefulness of retirement planning software will be limited by its least realistic assumption. It makes little sense to develop retirement software having numerous investment type choices along with a detailed Monte Carlo treatment of potential future returns, if the real retirement spending is estimated using only a simple replacement rate model.

The article Retirement calculators and spending contains an extensive list of calculators which have been categorized in terms of the retirement spending model(s) they use. Refer to this article to find real-world examples of the spending models discussed here.

Organization
Having a classification system for retirement spending models is useful because it emphasizes similarities and differences. In this article the following classification will be used to organize retirement spending models:

Initial Retirement Spending
 * Replacement Rate models
 * Single Budget models
 * Dual Budget models

Spending as Retirement Progresses
 * Constant (Real) Spending models
 * Stages of Retirement models
 * Investment Returns Dependent models
 * Flexible Spending models
 * Life Cycle models

Initial retirement spending
These models guide the pre-retiree to develop an estimate of their anticipated spending early in retirement, often targeting spending for the very first year of retirement. There are three main approaches for developing this estimate:
 * 1. Replacement rate (or ratio) models
 * 2. Single budget (or expense) models
 * 3. Dual budget models

Replacement rate models
Replacement rate (or ratio) models represent the simplest, most generic approach to estimating retirement spending. Rather than directly yielding an estimate of spending, replacement rate models give an estimate of pre-tax, gross income after retirement. This model is based on the following equation:


 * Gross Income (retired)  =  Gross Income (pre-retirement)  &times;  Replacement Rate

Assuming the pre-retiree has an idea of their gross income just before retirement, all that is required is a replacement rate multiplier to derive a gross income just after retirement. An actual retirement spending amount, if needed, can be calculated from the gross income by subtracting federal, state and local taxes.

In its simplest form, this model assigns the same replacement rate to everyone: 0.75 being a commonly suggested value (75% of your pre-retirement income is needed for retirement), but sometimes a range of 0.70 to 0.85 is suggested. In this form the replacement rate model is extremely generic and for many retirees the least accurate. Replacement rates are know to change with key household characteristics such as gross income and marital status. Tables containing these less generic replacement rates, as well as a more detailed discussion, are located in the article  Replacement rate models of retirement spending.

Although many free retirement planning calculators on the Internet incorporate the replacement rate model, these calculators are typically very simple. Even when using simple Internet calculators, replacement rates from the GSU/Aon tables are preferred to a generic estimate (e.g. using 0.75 for everyone). If a more detailed retirement calculator is being used, the GSU/Aon replacement rates should be further adjusted (individualized) to increase their accuracy.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%"|
 * Provides a quick and easy estimate of gross income in retirement.
 * Well suited for use by persons who are far from retirement. The uncertainties in the GSU/Aon replacement rates will likely be on the same order of magnitude as uncertainties in other aspects of the retirement plan.
 * width="50%" |
 * GSU/Aon replacement rates need further adjustments for best accuracy.
 * Require a supplemental calculation to obtain after-tax spending.
 * Persons far from retirement will need to estimate their gross income just before retirement in order to use replacement rates.
 * Generally less accurate than a Budget model approach for persons who are close to retirement.
 * }

Single budget models
Budget (or expense) models come in a variety of forms. They all have in common the estimation of total retirement spending by working with estimates of spending in numerous, smaller budget categories. The most common type of budgeting approach involves the use of a single budget to describe all retirement expenses.

Two approaches are typically used to obtain a single budget spending estimate: a Current Spending approach and a Bottom-up approach. A Current Spending approach starts with the total (current) spending just before retirement. Each spending category is then examined to see what adjustments, either up or down, are anticipated upon retirement. These spending adjustments are then incorporated to arrive at a total spending just after retirement.

The Bottom-up approach develops a personalized spending budget “from scratch” by estimating the retirement spending for every conceivable budget category. A detailed budget worksheet should be used to help insure that no category of retirement spending is overlooked. These category estimates are then summed to yield the total spending estimate. This approach is more time consuming than the Current Spending approach, but has the potential to yield a better estimate. The Bottom-up approach also supplies a better starting point for projecting how the budget might change as retirement progresses.

The in-depth article Budget models of retirement spending provides links to detailed worksheets useful for developing a single budget spending model. Many retirement planning calculators also incorporate worksheets for developing budget spending estimates. However it is better to develop the budget using an independent, detailed worksheet. Doing so minimizes the possibility that important spending categories get overlooked. For example, spending on big, infrequent purchases are very important, but tend to be omitted by most calculators’ built-in worksheets.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Capable of supplying more accurate spending estimates than a replacement rate model.
 * Better suited to handle spending estimates for big, infrequent expenses than a replacement rate model.
 * A Bottom-up budget model provides a particularly good starting point for incorporating real spending changes over time.
 * width="50%" |
 * Requires more time and effort than a replacement rate model.
 * Realistic spending estimates are difficult to determine unless personal spending is already being tracked.
 * }

Dual budget models
Dual Budget models are a direct extension of the Bottom-up budget model that incorporate two total spending estimates. The first or Essential budget represents the lowest level of retirement spending that can be accepted. The second or Preferred budget represents a higher level of retirement spending that is actually desired. The retiree’s spending in any year is assumed to fall within the range bounded by these two budgets.

The easiest way to estimate the Dual budget models is by using a worksheet designed for this purpose. Each budget category on such a worksheet accepts two entries: an essential spending amount and a discretionary spending amount. The sum of both gives the preferred budget spending estimate.

Dual budget spending models are particularly suited for combination with Withdrawal Methods that allow real spending to vary. Examples of such withdrawal methods are the Constant Percentage method and the Floor and Ceiling method. The Dual budget model supplies an additional restriction on the calculation of total savings at retirement: real spending can’t rise above the preferred budget nor fall below the essential budget.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Provides a more nuanced model of retiree spending that a single budget model.
 * Better suited for combining with Withdrawal Methods that allow variable real spending.
 * width="50%" |
 * More work to develop dual budget models than a single budget model.
 * }

Spending as retirement progresses
The models previously described supply a starting point for the more important problem of modeling how real spending changes as retirement progresses. A large number of such models have been proposed and incorporated into  retirement calculators. These models are summarized below. More detailed descriptions of all the following models can be found in the article Models of spending as retirement progresses.

An important criterion for evaluating retirement spending models is how well they can describe the actual spending patterns of retirees. These actual spending patterns are obtained primarily from the various Surveys of retirement spending. From these surveys, as well as from the practical experiences reported by professional financial planners, certain key facts have become clear:
 * The average retiree exhibits a slight drop in real spending at retirement, followed by a steady decline in real spending as they age into their late 70’s or early 80's. This decline in real spending is voluntary and not a result of limited financial resources.
 * A substantial fraction of retirees (perhaps as many as 25%) enter retirement involuntarily. They exhibit a sharp drop in real spending at retirement.  If the involuntary retirement was health related, such retirees may subsequently exhibit a medical expense induced, real increase in total spending near their death.
 * Some smaller percentage of retirees exhibit an increase in real spending at retirement.  This is often driven by a jump in travel or other leisure activities.  After a certain time period these special activities end, and real spending patterns once again match those of “average retirees.”

Constant (real) spending models
This is the simplest of all approaches used to model spending as retirement progresses. The real spending at the time of retirement is assumed to continue unchanged until death. Inflation is not ignored, but as explained in the Introduction it is treated as an separate, adjustable variable in the retirement planning process. From a nominal spending perspective, this model states that spending for any year in retirement is equal to spending in the first year times an independent inflation adjustment.

Comparison with Retiree Spending. Constant real spending models do not reflect the reality that the average retiree’s spending steadily drops as retirement progresses. Many retirement planners have pointed this out as a major shortcoming. They have also pointed out that such models lead to an overestimation of the total savings needed at retirement.

A constant real spending model could conceivably be appropriate in cases where retirement is involuntary and financial resources are limited. This situation would force the retiree to immediately drop down to an essentials only spending budget. To the extent that they really are living at their minimum acceptable level, their nominal spending would be expected to grow at somewhat the rate of inflation (constant real spending) as they were forced to pay ever increasing market prices for those essential goods.

Although  Life Cycle spending models sometimes yield a constant real spending in retirement, they do so from an entirely different Economics Perspective. This approach is discussed in its own section below.

Calculators. Constant spending is the most commonly used model in free Internet retirement calculators. This is especially true for calculators that are deterministic, less so for those that utilize Monte Carlo or Historical Returns calculational approaches. There are however some free Internet retirement calculators that allow you to choose models other than constant real spending. Refer to the article Retirement calculators and spending for specific examples.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Easy to understand, so they're useful for illustrative (teaching) purposes.
 * Easy to implement in a retirement calculator or spreadsheet.
 * width="50%" |
 * Does not match the reality of how average retirees spend money as they age.
 * Often leads to an overestimation of the total savings needed at retirement.
 * }

Stages of retirement models
Many studies have recognized that a typical retirement can be divided into 3 to 4 Stages or Phases. Within each Stage retirees tend to exhibit similar patterns of physical activity and spending. Stages of Retirement spending models are built upon these observations. In the very simplest Stages of Retirement model the real spending within each Stage is assumed to be constant. All the changes in real spending occur between the stages. If this spending is modeled as dropping by ages 75-80, it approximates the average retiree spending pattern. If a budgeting approach is used to estimate the real spending within each stage, these can be referred to as  Step Change Budgets. But a changing (often dropping) replacement rate could just as easily be used to model each successive stage.

Numerous studies have pointed out that retirees steadily reduce their spending as they age rather than exhibiting in a few sharp drops in spending. Based on this observation several retirement planners have suggested that Stages of Retirement models be made even more realistic by allowing real spending to vary gradually within one or more of the stages. Well known examples of this approach are Ty Bernicke’s Reality Retirement Planning model and William Bengen’s Prosperous Retirement model.

Comparison with Retiree Spending. The use of 3 to 4 Stages of Retirement, each of which can have a different real spending, allows for much better modeling of retiree spending patterns. All three of the basic patterns discussed earlier can be mimicked. And if the stages allow for a gradual annual spending drop rather than just a constant real spending, then an even better match to actual retiree spending can be achieved.

Calculators. Calculators having a Stages of Retirement spending model are much less common than those using a constant real spending model. Referring to the  Calculators and Spending article, only a few of the calculators explicitly incorporate this model. Some of these have included Ty Bernicke’s Reality Retirement Planning approach.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Allows the diversity in retiree spending patterns with aging to be incorporated into the retirement calculation.
 * Significantly improves model accuracy without excessively increasing complexity.
 * width="50%" |
 * Additional work needed to estimate budgets or replacement rates for each Stage.
 * There is some uncertainty in the most realistic annual percentage spending drop to age 75.
 * Somewhat more difficult to implement in a retirement calculator than constant real spending.
 * }

Investment returns dependent models
Investment Returns Dependent models allow spending in retirement to vary based on how total savings or investment returns change over time. It isn’t surprising that retirees would cut back on their spending during times when they don’t feel as financially secure. For retirees having savings invested in stocks or stock mutual funds, stock bear markets would be such a time of reduced spending. Conversely, when a retiree feels financially secure (e.g. during a stock bull market), they would naturally increase their real spending. These periods of increased and decreased spending can be modeled by using Withdrawal Methods that contain a dependency on either the previous year’s total savings or its investment return. An extensive summary of these methods is given on the Variable Withdrawals in Retirement page at Bob’s Financial Website.

A  Dual Budget model can be added as a natural complement to such variable withdrawal methods. The Essential budget spending level would set a lower limit on spending as retirement progressed. The Preferred budget spending level would set a corresponding upper limit on spending. These constraints would be included with those supplied by the withdrawal model when calculating the approximate savings needed at retirement.

Comparison with Retiree Spending. By itself an Investment Returns Dependent spending model is not able to adequately describe the key observations on retiree spending discussed earlier. It has no ability to match the steady decline in real spending as retirees age. But it does supply an element of reality that other models lack: the ability to reflect how retirees alter their spending in response to changes in their net worth. A combination of these two spending tendencies would be very beneficial.

Calculators. Calculators having an Investment Returns Dependent spending model are much less common than those using a constant real spending model. However, referring to the  Calculators and Spending article shows that more calculators incorporate some variation of this model than those explicitly incorporating a Retirement Stages model.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Reflects the tendency of retirees to adjust spending up or down as their net worth changes.
 * Nicely complements variable withdrawal methods used to insure savings survival in retirement.
 * More realistic than a Constant Spending model.
 * width="50%" |
 * Doesn’t match the reality that average retirees spend progressively less money as they age.
 * Less able to mimic the broad range of retiree spending patterns than a Stages of Retirement or a Flexible Spending model.
 * }

Flexible spending models
A Flexible Spending model allows numerous individual categories of spending to be separately incorporated into the total spending plan. Usually there is no need to enter every conceivable budget category as a separate spending item. Rather, these models usually allow you to first enter a base spending amount via another model (Constant Spending, Stages of Retirement, etc.). Then any spending that still isn’t properly represented can be entered as a separate, individual spending item. Typically the starting age, ending age and spending amount for each category are the only required information.

This type of modeling flexibility is particularly important for retirees anticipating one or more large expenses that would abruptly start or stop during retirement. Example of such expenses include the ending of a home mortgage, the purchase of a vacation home, or the payment of long-term medical care expenses. Thus it offers the potential for detailed personalization.

Often a Flexible Spending model can be used to mimic the spending pattern of a  Stages of Retirement spending model. This is accomplished by using one flexible spending category to contain all the spending within one of the retirement stages.

Comparison with Retiree Spending. Provided there are enough flexible spending categories/entries available, this model can do an excellent job of representing the diversity of retiree spending patterns discussed earlier. If there are only a limited number of categories/entries available (e.g. 3-5), then representing a steady decline in retiree real spending through their late 70’s or early 80’s would not be possible.

Calculators. Calculators having a Flexible Spending model are much less common than those using a constant real spending model. However, referring to the  Calculators and Spending article shows that a handful of free calculators do incorporate some variation of this model. This spending model is found in most purchased retirement calculators.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Capable of providing a personalized description of retirement spending for a broad range of retirees.
 * Best model for handling large changes in spending that might occur at irregular times during retirement.
 * width="50%" |
 * Entering the personalized spending data can be very time consuming for some calculators.
 * Not well suited for mimicking investment returns dependent retiree spending.
 * Additional effort required to incorporate this model into a retirement calculator.
 * }

Life Cycle models
A Life Cycle model predicts how a household should smooth their consumption over their entire lifetime, not just in retirement. Economic Utility theory predicts that people would be happiest if they neither over-spend nor under-spend each year as they age, whether retired or not. This approach to modeling retirement spending attempts to incorporate this fundamental economic reality. In doing so, Life Cycle models take into account all types of pre-retirement expenses as they generate a year by year savings plan that is consistent with a lifetime of smoothed consumption.

It is not a simple problem to estimate how much to save each year before retirement so that a comparable lifestyle could be maintained each year after retirement. This is especially the case if uncertainties in future investment returns, life spans, and health care expenses are to be taken into account. Calculators using the Life Cycle approach make use of a computational technique called Dynamic Programming to overcome these difficulties.

Comparison with Retiree Spending. As commonly implemented, Life Cycle models generate a constant (real) discretionary spending target throughout retirement. This is clearly not in agreement with the average retiree spending patterns discussed earlier. But the problem appears to be more one of model implementation rather than of a weakness in Utility theory. To the extent that a Life Cycle model is designed to smooth lifetime consumption (the actual goal) rather than lifetime spending (a simple proxy), it would exhibit better agreement with observed spending patterns.

Calculators. Calculators incorporate Life Cycle models less frequently than other spending models. This is not surprising, given the complex programming required to implement the model. The  Calculators and Spending article shows just a few calculators that employ the Life Cycle model.


 * {|style="border-collapse: collapse;" border="1" cellspacing="0" cellpadding="2" width="90%"

STRENGTHS WEAKNESSES
 * - valign="top"
 * width="50%" |
 * Gives specific, annual spending and saving goals both before and after retirement.
 * Able to recommend optimum age to start Social Security.
 * Able to fully integrate federal and state taxes into the calculation of optimal spending.
 * width="50%" |
 * Simple implementations that yield a constant real spending will not match average retiree spending patterns.
 * Complicated to fully implement into a retirement calculator.
 * }