Drawdown rules help

jamesd

For anyone else who can read javascript here's the G&K spending rule from cfiresim source code, i is inflation, j current year, form the input form and sim the simulation info.

"guytonKlinger": {
calcSpending: function(form, sim, i, j) {
if(j == 0)
{
   return form.spending.initial;
}
var initialWithdrawalRate = form.spending.initial / sim[i][0].portfolio.start;
var exceeds = (form.spending.guytonKlingerExceeds / 100);
var cut = form.spending.guytonKlingerCut / 100;
var fall = (form.spending.guytonKlingerFall / 100);
var raise = form.spending.guytonKlingerRaise / 100;
var currentWithdrawalRate = sim[i][j-1].spending / sim[i][j].portfolio.start;
var exceedsRate = initialWithdrawalRate * (1 + exceeds);
var fallRate = initialWithdrawalRate * (1 - fall);
var floor = SpendingModule.calcBasicSpendingFloor(form, sim, i, j);
var ceiling = SpendingModule.calcBasicSpendingCeiling(form, sim, i, j);
var currentYearInflation = ((sim[i][j].cumulativeInflation - sim[i][j-1].cumulativeInflation) / sim[i][j-1].cumulativeInflation + 1);
var simulationDuration = form.retirementEndYear - form.retirementStartYear + 1;
var yearsLeftInSimulation = simulationDuration - j;
if(currentWithdrawalRate > exceedsRate && yearsLeftInSimulation > 15)
{
   return Math.max(sim[i][j-1].spending * (1 - cut) * currentYearInflation, floor);
}
if(currentWithdrawalRate < fallRate)
{
   return Math.min(sim[i][j-1].spending * (1 + raise) * currentYearInflation, ceiling);
}
return sim[i][j-1].spending * currentYearInflation;
}
},

In another place, cFIREsimOpen.js you can see that the rebalancing happens without any special G&K rules, just based on the annual rebalance flag being set or not.

westv

The way I understood it was that cash was kept outside the portfolio.
So each year you sell sufficent of the best performing portfolo asset to rebalance and any surplus after income is put outside of the portfolio in a cash account.
The cash account can consist purely of excess profit or profit and a cash buffer.
After a year when all assests have fallen then income is taken from the cash fund first.
Does that sound right?
Interesting subject.

coyrls

mrstout wrote: »

If we accept this to be the meaning, how do we explain this statement on 'rebalancing' from the Guyton Klinger paper? It says:

"Each year's entire withdrawal was made on the first day of the year from the portfolio's assets according to the decision rule in effect. At that time, overweight asset classes were also rebalanced to the target allocation unless the portfolio decision rule was applied. At the end of each year, asset class returns were applied before any rebalancing or the source of next year's withdrawal was determined."

This is contradictory. You rebalance unless a decision rule is applied. Following this logic, whenever a decision rule is applied (which is always) you do not rebalance. It is in effect saying you would never rebalance.

Alternatively, if the act of selling overweight assets and adding to cash (which is a decision rule being applied) is the rebalancing then how does that fit with this statement?

My thought is (and please understand that I haven't done the research or gone into the detail to confirm it, but I'll say it anyway) that, as you have pointed out, you are also potentially building up cash in your portfolio, such that for a given year, your income could be 100% from cash. If income is from cash only, then no portfolio decision rule would have been applied and so you would need to rebalance.

mrstout

jamesd wrote: »

It is fiddly to understand what is meant by the portfolio rebalancing section but logic comes to the rescue so some extent. The portfolio used is split between only equities and fixed interest. If equities are up and overweight, they are sold, bringing the weights into line, making rebalancing unnecessary. That's what I think the portfolio decision rule means. So if that sale wasn't triggered, do the income taking then rebalance.

I've been thinking of a worksheet for this stuff and maybe a Google spreadsheet that people can use each year. But not just for G&K, also covering where I can the rest that I recommend, particularly Guyton's sequence of return risk taming method.

That sounds like it could be what they mean. Thanks. The ambiguity in some of the published papers does not help implementation.

I am also looking to include Guyton's sequence of return risk taming into my spreadsheet. But again I find it less than clear. I cannot post links but your other thread links to the research (without the www bit the link is onefpa.org/journal/Pages/Sequence-of-Return%20Risk%20Gorilla%20or%20Boogeyman.aspx) Sequence-of-Return Risk: Gorilla or Boogeyman? (Guyton)

Under the heading "Dynamic Approach" explains dynamic allocation policies (in this case used in tandem with dynamic withdrawal rules) to trigger changes in equity allocation based on Shiller CAPE.

The relevant explanation states "...in modeling the dynamic allocation policy I use a 65 percent neutral equity allocation with just under a 25 percent shift to 50 percent (80 percent) when an over- (under-) valuation occurs."

Surely if the initial allocation is 65% equities, an over-valued equity allocation would be reduced 15% (if the new allocation is 50%) or increased 15% (if the new allocation is 80%)? When does a 25% shift occur?

mrstout

westv wrote: »

The way I understood it was that cash was kept outside the portfolio.
So each year you sell sufficent of the best performing portfolo asset to rebalance and any surplus after income is put outside of the portfolio in a cash account.
The cash account can consist purely of excess profit or profit and a cash buffer.
After a year when all assests have fallen then income is taken from the cash fund first.
Does that sound right?
Interesting subject.

coyrls wrote: »

My thought is (and please understand that I haven't done the research or gone into the detail to confirm it, but I'll say it anyway) that, as you have pointed out, you are also potentially building up cash in your portfolio, such that for a given year, your income could be 100% from cash. If income is from cash only, then no portfolio decision rule would have been applied and so you would need to rebalance.

jamesd wrote: »

It is fiddly to understand what is meant by the portfolio rebalancing section but logic comes to the rescue so some extent. The portfolio used is split between only equities and fixed interest. If equities are up and overweight, they are sold, bringing the weights into line, making rebalancing unnecessary. That's what I think the portfolio decision rule means. So if that sale wasn't triggered, do the income taking then rebalance.

I've been thinking of a worksheet for this stuff and maybe a Google spreadsheet that people can use each year. But not just for G&K, also covering where I can the rest that I recommend, particularly Guyton's sequence of return risk taming method.

Is cash presumed to be outside the portfolio?
Taking income as cash is one of the decision rules.
Is the portfolio only equities and fixed income (bonds)?

Looking at the paper (cornerstonewealthadvisors.com/wp-content/uploads/2014/09/08-06_WebsiteArticle.pdf) table 1 shows 3 portfolios being tested (50%, 65%, 80% equities) and all 3 have a 10% allocation to cash.

westv

The 10%: allocation is "91 day Treasury Bills"
Is that cash as we mean here?

mrstout

westv wrote: »

The 10%: allocation is "91 day Treasury Bills"
Is that cash as we mean here?

The 3 model portfolios list 8 asset classes with corresponding indexes. The equity percentage allocations are 50%, 65%, 80% with the remainder being either 'Fixed Income' (Lehman Agg. Bond) or 'Cash' (91 dat T Bill).

I think in the portfolio the T Bill may be a proxy for cash returns on their monte carlo simulations. They make a distinction between Fixed Income and Cash both in the model portfolios, and in the 'decision rules'. I would interpret this as meaning that all their portfolios start with 3 asset classes: equities, bonds, cash. The cash allocation was always 10%.

If as has been suggested the 'rebalancing' they refer to is the process of applying the decision rules to sell overweight assets and increase cash, has anyone worked out what they meant by the problematic bit I asked about:

"Each year's entire withdrawal was made on the first day of the year from the portfolio's assets according to the decision rule in effect. At that time, overweight asset classes were also rebalanced to the target allocation unless the portfolio decision rule was applied. At the end of each year, asset class returns were applied before any rebalancing or the source of next year's withdrawal was determined."

We are accepting that applying the decision rules is the action of rebalacing, but in their above statement the 'unless' would mean they were only rebalancing when no decision rule has been applied.

mrstout

mrstout wrote: »

That sounds like it could be what they mean. Thanks. The ambiguity in some of the published papers does not help implementation.

I am also looking to include Guyton's sequence of return risk taming into my spreadsheet. But again I find it less than clear. I cannot post links but your other thread links to the research (without the www bit the link is onefpa.org/journal/Pages/Sequence-of-Return%20Risk%20Gorilla%20or%20Boogeyman.aspx) Sequence-of-Return Risk: Gorilla or Boogeyman? (Guyton)

Under the heading "Dynamic Approach" explains dynamic allocation policies (in this case used in tandem with dynamic withdrawal rules) to trigger changes in equity allocation based on Shiller CAPE.

The relevant explanation states "...in modeling the dynamic allocation policy I use a 65 percent neutral equity allocation with just under a 25 percent shift to 50 percent (80 percent) when an over- (under-) valuation occurs."

Surely if the initial allocation is 65% equities, an over-valued equity allocation would be reduced 15% (if the new allocation is 50%) or increased 15% (if the new allocation is 80%)? When does a 25% shift occur?

Got it. The "25% shift" means as a proportion of 65%.

mrstout wrote: »

The 3 model portfolios list 8 asset classes with corresponding indexes. The equity percentage allocations are 50%, 65%, 80% with the remainder being either 'Fixed Income' (Lehman Agg. Bond) or 'Cash' (91 dat T Bill).

I think in the portfolio the T Bill may be a proxy for cash returns on their monte carlo simulations. They make a distinction between Fixed Income and Cash both in the model portfolios, and in the 'decision rules'. I would interpret this as meaning that all their portfolios start with 3 asset classes: equities, bonds, cash. The cash allocation was always 10%.

If as has been suggested the 'rebalancing' they refer to is the process of applying the decision rules to sell overweight assets and increase cash, has anyone worked out what they meant by the problematic bit I asked about:

"Each year's entire withdrawal was made on the first day of the year from the portfolio's assets according to the decision rule in effect. At that time, overweight asset classes were also rebalanced to the target allocation unless the portfolio decision rule was applied. At the end of each year, asset class returns were applied before any rebalancing or the source of next year's withdrawal was determined."

We are accepting that applying the decision rules is the action of rebalacing, but in their above statement the 'unless' would mean they were only rebalancing when no decision rule has been applied.

Still have not figured this though!