couple of questions regarding inflation and the 4% rule

[btownguy]

I understand the "4% rule" as something along the lines of...you can withdraw 4% of your portfolio value in year 1 of retirement, increase withdrawal each year based on inflation, and you have a high (95%-ish) chance of success over a period of 30 years. I'm aware that 3.5% may be the new 4% and that if you plan to be retired longer than 30 years you need to plan for that.

So my question is...what specific inflation metric do most people use if they strictly adhere to this rule. Are people using the CPI-U?

Also and related. How far is the spread between yearly SS payment adjustments and official inflation numbers?

[renue74]

Up until 2015, I had a FA. When we did our annual visits, she always used 3% as the rate of inflation as a conservative benchmark.

I've developed a huge Google sheet that has our retirement projections at different ages and using different scenarios. I still use 3% for that.

[RadAudit]

I guess if I were using a CPI other than just a SWAG, I'd use one that came closest to the overall rate of inflation in the economy. Reason (?) being that the overall rate of growth in the stock prices over time is the rate of increase in real earnings + inflation (I guess). My personal rate of inflation - increase medical expenses, cruises, etc ... - doesn't impact the overall rate of inflation that much. I don't take that many cruises where I'd up the overall CPI.

Probably wrong.

What did the Trinity Study use? Go with that index.

[Artsdoctor]

There are many, many caveats with Bengen's initial 1994 study but if you're really interested in following his guidelines and suppositions, they're here:

http://www.retailinvestor.org/pdf/Bengen1.pdf

He used a lot of historical data that might not be applicable now and you'll find a few of the details in the appendix. He briefly mentions using 3% as an average inflation guide.

[David Jay]

I understand the "4% rule" as something along the lines of...you can withdraw 4% of your portfolio value in year 1 of retirement, increase withdrawal each year based on inflation, and you have a high (95%-ish) chance of success over a period of 30 years. I'm aware that 3.5% may be the new 4% and that if you plan to be retired longer than 30 years you need to plan for that.

So my question is...what specific inflation metric do most people use if they strictly adhere to this rule. Are people using the CPI-U?

You are WAAAYYY over-thinking this. It is a rule of thumb. Nobody should "strictly adhere" to the 4% rule, flexibility is necessary to avoid either under-performance or over-performance.

Also and related. How far is the spread between yearly SS payment adjustments and official inflation numbers?

CPI-W (what SS uses) runs a bit hot compared to Consumer CPI. There is some disagreement over the amount, but perhaps a half of a percentage point per year.

[hogfanboy]

could you calculate you own personal rate of inflation by comparing your increase in spending over the previous 2 years?

[deikel]

No it actually is a 4% 'rule' even in bad times - no need to reduce to 3.5 or somesuch thing, there are current threads worth reading with some reddit from the original researcher doing the study. Its actually more of a 4.8% rule when you retire in very good years....but thats besides the point...

You should take officially published inflation values and increase by those per year (from the previous year technically)

That is the idea, but in reality you will of course monitor based on economic development and actual need at the time.

[delamer]

could you calculate you own personal rate of inflation by comparing your increase in spending over the previous 2 years?

The CPI does not track changes in spending. It tracks the change in prices of a fixed market basket of goods/services. Say you spent half of your income on 500 gallons of milk and half on rent for your apartment in 2016. (Obviously, a simplified example.) If a gallon of milk increased in price by 5% and your rent increased 7% in 2017, your personal CPI would show a 6% increase. It would not reflect a change in spending, like adding cereal to your budget for example.

[3funder]

I'm nowhere near retirement age yet, but I think I'd just withdraw the minimum amount necessary to meet my annual needs plus a small cushion and assume things will work out. This, of course, is predicated on me continuing to maintain a high savings rate.

[Artsdoctor]

No it actually is a 4% 'rule' even in bad times - no need to reduce to 3.5 or somesuch thing, there are current threads worth reading with some reddit from the original researcher doing the study. Its actually more of a 4.8% rule when you retire in very good years....but thats besides the point...

You should take officially published inflation values and increase by those per year (from the previous year technically)

That is the idea, but in reality you will of course monitor based on economic development and actual need at the time.

I think that the most you can say is that the period studied by Bengen is from 1926-1994. Those are historical data and should be viewed as such.

It's tempting to use officially published inflation values because it makes sense. However, he did not do that.

[MathWizard]

If I actually follow that, I'll use the same as SS uses, because that is easy.

I'll probably get stuck going over 4% because RMDs will force me into it sometime in my
mid to upper 70's. I have a substantial ROTH, but the bulk of retirement money will be tax deferred.

RMDs can't be converted to ROTH, so we'll either spend it all or put the excess in CDs. From my observation
of our parents, once we are in our 80's we won't be doing so much traveling, so it may be too much work to
spend profligately

[onthecusp]

If I actually follow that, I'll use the same as SS uses, because that is easy.

I'll probably get stuck going over 4% because RMDs will force me into it sometime in my
mid to upper 70's. I have a substantial ROTH, but the bulk of retirement money will be tax deferred.

RMDs can't be converted to ROTH, so we'll either spend it all or put the excess in CDs. From my observation
of our parents, once we are in our 80's we won't be doing so much traveling, so it may be too much work to
spend profligately

If the RMD comes out of a retirement account and goes into a taxable account CD (in your example) it should not count as part of (or in excess of) the 4% withdrawn for spending. It is still there as part of the overall portfolio. If bad market conditions cause your RMD to get much smaller those CDs will need to be there to provide the remainder of your spending requirements in that event.

To me the 4% rule is a great first pass planning tool. It helps me to easily compare different options and get a rough idea of how much savings are "needed" for a given spending rate, leading to reasonable decisions on how much to save annually etc. It is a poor tool for making the final decision on the question of 'do I have enough to retire,' or to set a particular spending rate. That decision often includes variables that will change in a somewhat predictable way in the future, for example delayed social security, pensions, paying off a mortgage, and changing tax rates due to RMDs such as MathWizard's. That is where tools like firecalc and some of the excellent tax rate information in the wiki come in.

(Edited for clarity)

[EddieGee]

I believe it's impossible to predict inflation rates, especially in the longer term.

Though I'm not certain, I like to believe that the interest rate on short to medium term bonds will equal or slightly exceed the inflation rate. (Right now we are in an exceptional period since the Fed's "quantitative easing" has artificially depressed interest rates).

This calculation however only works for a tax deferred account. If inflation is 7% and you make 8% on your money you will keep up with it in a tax deferred account but in a taxable account the tax on the "income" will put you into the hole.

[dbr]

Planning programs that try to make estimates of retirement outcomes can take various approaches to this problem. I think there is a lot to say for the historical data approaches such as FireCalc or Otar's program where actual historical inflation is used. Others might argue that inflation as in the late '70's is exceptional and should not be in the model.

A different issue is whether or not CPI is the right index. Generically it is because it really amounts to measuring the value of money. For any given individual it makes sense to make estimates that adjust spending after inflation for whatever personal considerations apply. Models can enter changes in spending by adding or subtracting items. I am not sure if there are models that apply a different rate on top of historical data or that inflate some costs, like healthcare, at a different rate from others. I have a personal expense forecast where I do that, but this is not in connection with a complete model.

A different approach if you can change the rate of inflation used is to enter a series of different values and see how sensitive your results are and when the extremes become intolerable. That can give you a sense of how much trouble you might be in.

[Katietsu]

If I actually follow that, I'll use the same as SS uses, because that is easy.

I'll probably get stuck going over 4% because RMDs will force me into it sometime in my
mid to upper 70's. I have a substantial ROTH, but the bulk of retirement money will be tax deferred.

RMDs can't be converted to ROTH, so we'll either spend it all or put the excess in CDs. From my observation
of our parents, once we are in our 80's we won't be doing so much traveling, so it may be too much work to
spend profligately

Spend your RMD if you want. But I do not understand why people seem to equate the need to take the RMD with any requirement to change your asset allocation.

[dbr]

If I actually follow that, I'll use the same as SS uses, because that is easy.

I'll probably get stuck going over 4% because RMDs will force me into it sometime in my
mid to upper 70's. I have a substantial ROTH, but the bulk of retirement money will be tax deferred.

RMDs can't be converted to ROTH, so we'll either spend it all or put the excess in CDs. From my observation
of our parents, once we are in our 80's we won't be doing so much traveling, so it may be too much work to
spend profligately

Spend your RMD if you want. But I do not understand why people seem to equate the need to take the RMD with any requirement to change your asset allocation.

It might be a behavioral concept that one has been "given" money which is now different from what one had before. It is an example of mental framing that is a part of people's behavior.

[FactualFran]

I think that the most you can say is that the period studied by Bengen is from 1926-1994. Those are historical data and should be viewed as such.

It's tempting to use officially published inflation values because it makes sense. However, he did not do that.

The last sentence of the first section of Bengen's paper, which you linked to in a previously post, is: "In all cases I will rely on actual historical performance of investments and inflation, as presented in Ibbotson Associates' Stocks, Bonds, Bills and Inflation: 1992 Yearbook." What inflation data did Ibbotson Associates' present, if it was not a copy of the official values?

[Artsdoctor]

^ I wish I could pull up the original paper but I can't. My own link will not work on the computers I have available. If you take a look at the Ibbotson numbers used in their SBBI tables, CPI-U has been used. From 1926-2015, I think that the average value is 3% per year; the 2016 number is less at just above 2% although I don't have the exact final number. You're right that you'd think he used SBBI tables which took into consideration the CPI-U. In fact, he references this in subsequent lectures when he calculates numbers going forward. But I can't pull up his reference to 3% inflation rate in the 1994 study.

[FactualFran]

Concerning Bengen's use of the average inflation rate of 3%, see the Appendix box on page 9 of the pdf (page 179 in the journal issue) that you provided a link to. For years after the ending year of the actual data, he used the average return of stocks (10.3%), average return of bonds (5.2%), and the average inflation rate (3.0%). For example, the results he calculated for the starting year of 1976 used 16 years of actual data followed by as many years of average data as needed to run the calculation for the desired number of years.

An alternative approach is that once the last year of actual data has been used, the calculation continues from the first year of the actual data. If the actual data is from 1926 to 1992, then a calculation would use the data of 1926 for 1993, the data of 1927 for 1994, and so on.

After I posted the question, I did a web search and found what appears to be a copy of some of the Ibbotson data. In it the inflation rates correspond to the December to December change in the CPI-U. A better choice may have been the change in the average CPI-U for each year.

[Artsdoctor]

Factual,

Not sure why I was having so much difficulty opening the PDF while at work, but I've finally had a chance to re-read the original article. Bengen looked back at all the data he had available and used inflation rates (CPI-U) up until the end of the data period (the end of 1992). In order to predict the longevity of the portfolios which started being tapped in 1944 and beyond, he needed to expolate average returns of stocks and bonds, as well as the average inflation rate of 3% when generating 50 years of portfolio performance. This is why we stopped calculating the longevity of portfolios being tapped after 1976: he only had 16 years of data and had to guess on returns going forward from 1992.