Income Harvesting Strategy paper (accounts or buckets?)

[LOL!]

I wanted to draw attention to a new paper by Zachary S. Parker in the Journal of Financial Planning: Income-Harvesting Strategy: Acheiving Inflation-Adjusted Income from a Lump-Sum Asset

This paper describes an algorithm to accomplish what its title states. The algorithm consists of the use of 4 different accounts:

1. Income guarantee account consisting of say laddered CDs and/or bonds, an SPIA, or money market funds. The interest and dividends from this account are spent every year, but you can also spend some of the other money in this account. It should be good for 4.5 to 8 years of all your spending needs. The account has about 25% of your total assets.

2. Equity withdrawal account consisting of equities (no fixed income) in a typical slice-and-dice asset allocation of equities. Withdrawal are made from this account to replenish the Income guarantee account unless there is a 5% or more decline in the value of this account. This account has 30%.

3. Equity harvest account consisting of equities as in the previous equity account. Only gains are harvested from this account. If the account has no gains, then no money is taken from this account. In years with gains, then some of the gains are harvested as needed to other accounts. This account has about 34% of your initial assets.

4. Derivative Protection account consisting principal protected assets with maturities of less than one year such as CDs or very short term bonds. 10% of one's initial assets are in this account. This account is used only in a one-year or consecutive two-year period when the equity harvest account declines by 30% or more. The money gets transfered to the equity harvest account.

OK, this looks like a 65% equities, 35% fixed income asset allocation to me. And it also looks like a Buckets approach. What I found interesting was the algorithm of how to move money among the accounts, the testing and the goal of the testing described in the paper: 90% success rate, initial assets doubled in 25 years and relatively high (> 5%) initial withdrawal rate.

Maybe some of y'all can read the article and make comments on it?
Maybe it can become an algorithm for a withdrawal scheme in FIREcalc?

 

[rogersteciak]

My concern is that two of the four accounts depend on the generosity of Mr. Market for income (e.g., equity withdrawal account and equity harvest account). Another account depends on principal-protected derivatives. What are these derivatives? If I don't understand how something makes money, I will not invest in it.

I like having a portion of my retirement accounts based on "total return" (e.g., the two equity accounts), but I also want a portion of my retirement money to be from an equity income account (i.e., dividend growth). I have no simulation that my approach is any better or worse than the approach analyzed in the article. But intuitively, I like having an income stream that is not dependent on Mr. Market, yet, unlike fixed income, tends to stay ahead of inflation (as long as the inflation is not hyperinflation).

Dividend growth depends on having a good underlying economy and well managed companies that are paying the growing stream of dividends. Ultimately, fixed income depends on the same factors (although for corporate bonds, interest is ahead of dividends in priority when a company is in trouble).

My problem with depending too heavily on Mr. Market (i.e., stocks that are expected to appreciate in price over time) is that the underlying economy can be strong and the companies growing their earnings can be well managed, but if the "idiots next door" decide they want to pay less over time for a dollar's worth of earnings, a solid growing company could very well see its market price either decline or remain flat. That hurts when I need to sell shares to raise the money I need to pay living expenses.

For very long-term investments and for money that is way more than what I need to pay living expenses, I have no problem working with Mr. Market to help me build wealth. But for money I need in the next decade or two to pay living expenses, I don't want my financial fate to be tied to Mr. Market any more than necessary.

 

[LOL!]

My concern is that two of the four accounts depend on the generosity of Mr. Market for income (e.g., equity withdrawal account and equity harvest account). Another account depends on principal-protected derivatives. What are these derivatives? If I don't understand how something makes money, I will not invest in it.

I'm sorry, but it appears you not only did not the paper, but also that you did not read my summary of it. Somehow the word "derivatives" stuck in your head. What are these derivatives? First, they ain't derivatives ... that's just the name for the account. The account holds CDs with a maturity of one-year or less.

I like ...

Dividend growth depends on having a good underlying economy and well managed companies that are paying the growing stream of dividends.

That's fine, but if you own the equities suggested in the article, then you own pretty much own all equities including those "dividend growth" ones you want.

My problem with depending too heavily on Mr. Market (i.e., stocks that are expected to appreciate in price over time)

Not a problem as the article described several reduced-equity-exposure asset allocations at the end.

I guess I'm very frustrated by your response because despite that fact that I put a link in so that folks could easily read it, I don't think you bothered. Rather than commenting on the article, you told us about your investing philosophy which is not very far off from the article anyways.

 

[cute fuzzy bunny]

Sounds like a Buckets strategy to me too. I like buckets!

 

[Notmuchlonger]

This thread reminded me..

YouTube - Bob Dylan - Buckets Of Rain

 

[mickeyd]

Hey LOL,

Nice paper, thanks for the link.

Assets that would be considered suitable for this account would be laddered bonds, laddered certificates of deposit, single-premium immediate annuities, or money market accounts. Based on the current fixed-income environment, a net assumption after fees and expenses of 2 percent was applied to this account.

Since SPIA is listed here, I wonder if pensions should be included also?

 

[rogersteciak]

I'm sorry, but it appears you not only did not the paper, but also that you did not read my summary of it. Somehow the word "derivatives" stuck in your head. What are these derivatives? First, they ain't derivatives ... that's just the name for the account. The account holds CDs with a maturity of one-year or less.



That's fine, but if you own the equities suggested in the article, then you own pretty much own all equities including those "dividend growth" ones you want.


Not a problem as the article described several reduced-equity-exposure asset allocations at the end.

I guess I'm very frustrated by your response because despite that fact that I put a link in so that folks could easily read it, I don't think you bothered. Rather than commenting on the article, you told us about your investing philosophy which is not very far off from the article anyways.

The theme of the article is that besides fixed income from debt-based instruments, you harvest capital gains, along with any incidental dividends, for income. For capital gains to happen, Mr. Market has to be in a good mood.

This is an OK approach to use and much of my current retirement financing is based on it. But I've been thinking about what would happen if we spent many years in a down or sideways market (like the one we're in right now) and have decided to use fresh money to build a dividend-growth portfolio that is in parallel (and acts as a margin of safety) with my Work Less, Live More total return portfolio.

A dividend-growth portfolio does not care what Mr. Market thinks about what the dividend-paying stock is worth (other than when you initially buy it, since you want to get it at a discount to its intrinsic value whenever possible so that its initial yield starts out higher). You hold dividend-growth stocks primarily for the dividends (i.e., their "eggs") they will continue to produce over time rather than for any appreciation (i.e., the "chicken dinner") that might eventually happen.

This is a major philosophical difference between the investment philosophy I use and the approach described in the article. My intent is to share my ideas so that people can be thinking about which approach might be better for their needs.

With regards to the word derivatives, the author should have called it "principal-preservation emergency reserve fund" (or some such thing) rather than give it an exotic name that throws people off.

 

[unclemick]

Hmmm - I've always had trouble with the bucket approach. I can see stuff where a pension plan matches fixed income(CD's, bonds, etc) of known maturity coming due to match known expenses for X yrs and holding the rest in equities to fight inflation provide growth.

But how do I compare my balanced index, let's say 3% yield plus 1-2% float in Vanguard Prime MM. I top up MM once a year (12/31) and write checks to local bank ATM and pay my bills in dribs and drabs. So my 'bucket' is say 5% at the start of the year and goes down - ?so I have a one year bucket or does the Norwegian widow get to count her SEC yield as a back up bucket during the yr even though it's dynamically rebalanced during the yr.

Now I can freeze frame my portfolio % assets on Vanguard's website - so to compare with a 'bucket portfolio' I convert the buckets to a 'typical' mixture of stocks/fixed/cash at a point in time to guestimate relative expected growth of the two portfolio's in retirement?

My brain does not deal with buckets well. SEC yield's as % yes.

heh heh heh -

 

[clifp]

Good link LOL.

First, I think this was one of the more readable articles in the FPA journal. So often I have real trouble following the authors. He did a good job explaining all of his assumptions so it was easy to follow, and made easy for me to poke holes

My investment philosophy is very similar to Rogers, so I am not huge bucket fan, but buckets have uses beside mopping floors.

BTW, an equity income (dividend) portfolio would work quite well as the equity harvest portfolio in most years dividends would flow in the income guarantee account and would grow at above inflation rates.

The major problem is how the author treats inflation. While the overall assumption of 3% inflation isn't unreasonable, that fact that inflation is lumpy pretty much defeats the author goal of providing inflation protection withdrawals.

The other oddity is that inflation withdrawal are only increased every 5 years. The way I interpret this is every 5 years you get a 16% (3% compounded for 5 years) increase in withdrawal, this means even in times of moderate inflation by year 4 you've had to make some cutbacks.

The author points that his strategy fails during the great depression, no surprise the Great Depression is a bitch to retirement plans!

What he fails to realizes is that his withdrawal plan fails during the 1970s due to inflation.

Let say Joe retired with 1 Mil in 1973 and withdraw 4% with inflation protection and adjust for actual [-]government?[/-] inflation (i.e. Firecalc) While Bill is using the bucket system. By 1977 Joe is withdrawing almost 55K, but Bill is still withdrawing 60K. In 1978 Bill gets a 16% pay raise ~10K and Joe gets a 6.5%. Moving ahead past the double digit inflation years of 79-81. In 1982 Joe has moved ahead withdrawing $92.5K vs 70K for Bill. Finally in 1983 Bill gets another 16% "pay" raise, Joe is up to $95K and inflation settles down to a 2-4% level for many years.

However, despite Bill starting with a 6% withdraw rate and Joe @4%, by the end of the inflation era Bill is spending only 80K vs Joe's 95K. I think would be tough to have gone through most of the 70s with only one increase to your withdrawals. So while his strategy doesn't run out of money, I don't want to have to spend like that.

 

[Nords]

My intent is to share my ideas so that people can be thinking about which approach might be better for their needs.
With regards to the word derivatives, the author should have called it "principal-preservation emergency reserve fund" (or some such thing) rather than give it an exotic name that throws people off.

I think LOL's intent was that you should read the article and figure out whether the author has any huge logic gaps or other inconsistencies...

You, uh, you did read the article-- right?

 

[FUEGO]

Good read - seems a bit like data mining to a certain extent. Also, relies heavily on slice and dicing into large/small, value/growth, intl large/small, etc which historically beats the SP500. Pretty amazing that they got "success" with a six percent withdrawal and 2% investment fees where success is meeting the initial spending needs and doubling the portfolio value after 25 years (in nominal terms I believe). I did think the assumption that inflation adjustments are only made every 5 years gave this method a little wiggle room. If inflation were to run at double the historical 3% assumed rate (in other words, 6%) for a period of five years, one would have lost a full third of their purchasing power before an inflation adjustment was made. It might not break your typical retiree, but it would be a painful belt tightening for most.

It reminds me of the paper posted a few months ago (maybe Milevsky's?) that essentially illustrated that the highest portfolio values and survivability can be obtained by spending down your "fixed income" portion of your portfolio first, and then when that account is depleted spending from your equities. This is really a roundabout way of increasing your equities allocation to 100%, much like the "income harvesting" approach does over time.

Although I do like the ideas that they put forth. I may do something sort of similar by having two pots of money (let's call them "buckets"). One is fixed income/bonds/cash, the other is equities. I can spend the dividends from the equities, then use a small portion of the fixed income bucket to make up any income shortfall not provided by the equity dividends. In practice, this might look something like having a 70% stock (bucket 1), 30% bond/cash (bucket 2) portfolio. With my slightly value tilted, dividend tilted stock bucket, I can expect a 3% dividend yield, or 2.1% of my total portfolio value (ie 3% of 70% of the portfolio). To reach a total withdrawal of 4%, I will need 1.9% from the fixed income bucket.

Well, the fixed income produces, say, a 4.5% yield (a mix of bonds, cash, CDs, etc). That 4.5% yield on 30% of the portfolio gives 1.35% towards my 4% total withdrawal. 2.1% + 1.35% gets me to 3.45%, with the other 0.55% being taken from the fixed income capital. Taking out only .55% of the portfolio principal from the 30% fixed income bucket would last about 50 years.

The long term goal is for the equity dividends to grow faster than inflation. Over the last 6 or 7 decades, I believe dividend growth has been around 2% real per year. The hope is that the dividends are enough to fully fund your 4% SWR before the fixed income bucket is depleted.

This is a work in progress, but I think it has some promise for a more sustainable, better sleep-at-night feeling if you are living almost entirely off of dividends and interest. Heck, with the current yield environment, one could almost take a 3.5% withdrawal rate now and implement the strategy I'm promoting without really dipping into the principal now.

Or just buy some pssssst wellesley and live off that yield (or maybe 50/50 wellesley/wellington for a little growth enhancing boost).

 

[FUEGO]

Oh, to get back to a discussion of the OP's linked paper - some of the cutoff thresholds seem arbitrary and data-mined (though the author essentially states as much in the paper). For example - the 30% down year or two that triggers the derivative protection account investment. Why not 25% or 35%? What if the market is down 14% each yr for 3 yrs in a row, yet the derivative protection "bomb" wouldn't be activated. Same with the allocations between the guaranteed income acct and the two equity accts - these allocation percentages were optimized based on past data, so I'd expect them to do really well in-sample, but not necessarily as well out of sample (ie - in the future).

The author does state that the 6% withdrawal rate is at the upper limits of what the model allowed. The paper recommended a lower withdrawal rate should be used to increase portfolio survivability. There were also a few points where the author stated that an adviser (or us, as investors) would have to make a judgment call or a one-time correction to fix things mid-course, but it isn't clear what decision tools one should implement to distinguish between "sticking to one's guns" and "making a mid-course correction". Other than simply using common sense.

 

[LOL!]

Thanks for all the comments. Sure the numbers and cutoffs are data mined, but so what? I think the "derivative protection" or "reserve" account is a great way to force you to rebalance when the market has tanked.

I would also be curious what the results would be if he blew-off the 2% advisor fee and used a 0.25% Vanguard-range fee. That 1.75% a year difference is huge and should be enough to assuage any concerns about some of the other numbers.

 

[FUEGO]

I would also be curious what the results would be if he blew-off the 2% advisor fee and used a 0.25% Vanguard-range fee. That 1.75% a year difference is huge and should be enough to assuage any concerns about some of the other numbers.

Definitely - maybe instead of 6%, you could get 7.5-7.75% withdrawal rates.

Just for kicks, I ran the slice-n-dice portfolio like the one presented in this paper through firecalc. It doesn't translate exactly, but I tried to meld the asset classes into something similar - ie there is no international in firecalc. A 6% withdrawal rate is 90.7% safe, but this assumes 3% CPI bumps each year starting in the first year (unlike the Income Harvesting Strategy). The firecalc run also assumes a 25 year time horizon (like the Income Harvesting Strategy paper) however it uses the 0.18% expense ratio assumed in firecalc instead of the 2% fee used in the Income Harvesting Strategy paper. My guess is the international might earn you a little higher SWR in firecalc, since it is clear that the Income Harvesting Strategy paper used some sort of historical optimization to arrive at their optimal asset allocations.

A little sensitivity analysis - using actual CPI in firecalc (instead of 3% per year) took the survivability from 90% to 80%.

Edited to add: the Income Harvesting Strategy study used DFA historical returns - not sure what firecalc uses, although it is over the same period - 1927 to present.

 

[clifp]

Thanks for all the comments. Sure the numbers and cutoffs are data mined, but so what? I think the "derivative protection" or "reserve" account is a great way to force you to rebalance when the market has tanked.
.

I really like the concept of the derivative protection account. The name is horrible, something like strategic reserve or "in case of financial meltdown break glass fund" is much better.

Given the purpose of the fund I'd like see it invested in something other than cash. i.e some type of bear or hedge fund or commodities. One idea that that I briefly looked is having the fund use the interest received each year to buy out of the money puts on the S&P 500. Most years these expire worthless but during a bear market instead of having only 105K to invest perhaps you have 120K.

 

[Nords]

One idea that that I briefly looked is having the fund use the interest received each year to buy out of the money puts on the S&P 500. Most years these expire worthless but during a bear market instead of having only 105K to invest perhaps you have 120K.

IIRC one of the bigger names in the business was doing this-- a portfolio of mostly Treasuries with a put ladder. Was it "Black Swan" Taleb?

 

[cute fuzzy bunny]

Taleb goes for a portfolio of low risk bonds at about a 70-90% level and then soaking up emerging markets, extreme cigar butts and longshots with the rest. The big ballast reduces the risk and all you have to do is shoot the moon once with the rest.

 

[Marquette]

IIRC one of the bigger names in the business was doing this-- a portfolio of mostly Treasuries with a put ladder. Was it "Black Swan" Taleb?

Swedroe was floating a portfolio as an exercise that was 80% safe stuff and then 20% commodity futures and emerging market stuff. Not necessarily a bigger name, but another expert toying with this.

 

[ats5g]

IIRC one of the bigger names in the business was doing this-- a portfolio of mostly Treasuries with a put ladder. Was it "Black Swan" Taleb?

I think it was Zvi Bodie.

 

[walkinwood]

I find these schemes very arbitrary and confusing. I found the same thing to be true of the bucket scheme from Raymond Lucia.

I feel that in essence they are schemes that change your asset allocation according to portfolio performance. I would be more receptive if they built a scheme based directly on modifying the asset allocation based on portfolio performance. i.e If portfolio changes by x%, change equity allocation y%.

While I'm venting...

How hard is it for these guys to use real CPI numbers? The 3% assumption makes his results very suspect when you take the high inflation / low return environment of the 70's into account.

Also, why not aim to maintain the real value of the portfolio after 25 years instead of the arbitrary 2x? In reality it comes close to the real value assuming a 3% inflation, but why not deal with that implicitly?

Thanks for listening.

 

[LOL!]

First, I have not read Lucia's buckets stuff.

Second, I think this paper by Parker is destined to be one of those "classic" papers that get written up in the NYTimes, WSJ, Money magazine, Kiplinger's, Scott Burns, etc. It's algorithm will eventually be incorporated into FireCalc the same way that Bernicke's and Clyatt's methods are. Parker may even be able to write a book and make some money for himself that way, but his ideas are pretty darn simple.

These reason it's gonna be a classic is that he promises a 6% withdrawal rate most years and not a measly 4%. And he says he can do all this with a 2% expense ratio. His methodology appears to be just as valid as anybody else's. And while he's got an overall roughly 65:35 asset allocation like just about everybody else, he gives you an exact algorithm on what to do each year.

You can take that 2% expense ratio and do-it-yourself for 0.25%. If you like add the 1.75% e.r. reduction then to the 3% inflation rate. Or you can take that 6% and withdraw only 4% a year. Both these changes should give you even more confidence that it's gonna work out pretty well.

 

[cjking]

I'm not impressed by the article. The strategy's complexity is a sign of weakness. They've added in little tweaks to overcome inadequacies in the basic strategy that testing it against historic data highlighted. This means the strategy is tuned to the historic dataset, rather than designed to cope with a generic unknown future.

Here's a simple strategy I've spent a whole half-hour developing, that will cope as well as can be expected with any possible future, and doesn't require you to presume you know anything about the mean, volatility or path of future returns.

Invest your money however you think prudent, and each year withdraw capital/n, where n is your life expectancy in years.

Note that life expectancy does not decrease by 1 each year. At 50 it might be 32, i.e. you expect to live to 82, but at 82 it might 8 (expected to live to 90) and at 90 it might be 4. (These numbers all made up, can't be bothered to look up correct values.)

I've done a crude Monte Carlo simulation of this in Excel, and it works pretty well. It does give unattractively low withdrawal rates at young ages, but that's probably prudent.

 

[walkinwood]

First, I have not read Lucia's buckets stuff.

Second, I think this paper by Parker is destined to be one of those "classic" papers that get written up in the NYTimes, WSJ, Money magazine, Kiplinger's, Scott Burns, etc. It's algorithm will eventually be incorporated into FireCalc the same way that Bernicke's and Clyatt's methods are. Parker may even be able to write a book and make some money for himself that way, but his ideas are pretty darn simple.

These reason it's gonna be a classic is that he promises a 6% withdrawal rate most years and not a measly 4%. And he says he can do all this with a 2% expense ratio. His methodology appears to be just as valid as anybody else's. And while he's got an overall roughly 65:35 asset allocation like just about everybody else, he gives you an exact algorithm on what to do each year.

You can take that 2% expense ratio and do-it-yourself for 0.25%. If you like add the 1.75% e.r. reduction then to the 3% inflation rate. Or you can take that 6% and withdraw only 4% a year. Both these changes should give you even more confidence that it's gonna work out pretty well.

LOL!,
I respectfully disagree. I think his approach is ad-hoc, and feel his validation is sloppy in 3 areas: a) He hasn't used actual CPI, b) he is using a 25 year horizon which is quite low for today's retiree especially an ER, and c) he hasn't tried to use any other technique to simulate more 25 year periods (bootstrap or monte-carlo).

This paper Decision Rules and Maximum Initial Withdrawal Rates
by Jonathan T. Guyton, CFP®, and William J. Klinger[FONT=Verdana,Arial] [/FONT]

, also published by the FPA Journal, shows how to obtain a 6% initial withdrawal rate with a 95% chance of success and a 90% chance of maintaining purchasing power of the portfolio at the end of 40 years. This method uses portfolio and withdrawal rules (like forgoing an increase when the portfolio does badly or increasing withdrawals more than inflation when portfolio does well for a time).

I think that Guyton's method is also sort of arbitrary, but is more in line with what a lay person would do in response to their portfolio's performance. It also does not call for changes in asset allocation in response to market conditions.

The focus of research seems to have moved away from initial withdrawal rates and is more economics based with focus on a lifecycle model, consumption smoothing etc.

Full disclosure: I plan to use the method advocated by Bob Clyatt - the 4%/95% plan. I'm theoretically already using it, but am only in my first year of ER. IIRC, Bob is in his 8th year.

 

[LOL!]

Walkingwood, this is the kind of discussion I was hoping to see in this thread.

I read the Guyton paper(s) a while ago, so I will have to go back and revisit them. I remember being impressed with that paper as well.

As for not using actual CPI numbers, that doesn't bother me because one could pick any rate for inflation you wanted to (such as 3%) and different people have their own personal rate of inflation that depends on their spending pattern.

He clearly states why he used 25 years: because the data allows for that. As for not using a monte-carlo method, he has simply used the method used by FIRECalc as I understand it.

As for all the movement of $ among accounts ... that may be voodoo because once you have a 65% equities and 35% fixed income asset allocation and you spend and rebalance as you go along, then you pretty much have a 65:35 asset allocation. How you divvy up your buckets in your mind or in your actual accounts doesn't really matter. Nevertheless, the paper shows that a 65:35 asset allocation works.

 

[charlie]

At 74, the thing I am beginning to fear most is leaving my wife Lyn, who goes glassy eyed on financial matters, with a complicated portfolio to manage.

Personally I am leaning toward investing my IRA totally in Wellesley with all distributions going to a money market holding 1 year of expenses. I am already
investing my taxable account in Wellington spiced up a little with Tax Managed
Small Cap and FTSE All World Ex-US. All distributions in the taxable go to a MM as well. My withdrawal strategy is spend my RMD out of the IRA MM account,
replenish the 1 year bucket as needed in UP years and never sell fund shares in a
DOWN year. I also hope to never sell fund shares in the taxable account. The
taxable MM is used for large expenses like new cars, vacations and emergencies.

Please excuse the hi-jack, but I just think the KISS method is better as we grow
older.

Cheers,

charlie