ETF Safe withdrawal rate

Guys,

Is 3.6% withdrawal rate safe for a 50 year retirement period, for a couple without kids (50% VAS, 50% VGS)? We wouldn't own a PPOR and would rent a small apartment.

I'm a bit confused, there are a lot of articles saying that 4% withdrawal rate is not really safe, hence I'm planning our early retirement with 3.6% withdrawal rate. Any reference to a useful article (relevant for Australia) would be awesome.

Thanks

 

This updated study includes 50 year periods and mentions 3.5%
https://thepoorswiss.com/updated-trinity-study/

This one mentions an 8% failure rate for 4% over a 50 year period
Here's How the "25 Times Expenses" Portfolio Has Performed Over the Past 89 Years - Four Pillar Freedom

The trouble with fixed spending strategies is that they can still fail and target worst cases so most commonly ends up with lots of excess capital.

Look at variable dynamic strategies like https://www.vanguard.ca/documents/literature/dynamic-ret-spending-paper.pdf

 

Thanks so much! Very useful info.

I really like the Vanguard approach of dynamic spending. I haven't seen though that they include income tax and CGT in their calculations? I assume that would further decrease the safe withdrawal rate?

 

There's a lot more to consider when factoring in your retirement "safety" than just a %. Things like:
1. Are you / your partner eligible for the pension? Assuming (everything about retirement planning is assuming and guessing) that you both are and it exists when you're 70ish that's a massive safety net. It doesn't kick in when you have no money either, you can still have a decent amount of investments and get it.
2. Are you planning 3.6% living a barebones lifestyle where you cannot temporarily reduce it? Or does that include a lot of holidays etc. you could postpone if there is a tough year? Big difference.
3. How are your families genes? What are the odds of needing long term care? I plan on that in my calculations for one of us because unfortunately we know it's likely coming, and it makes a big difference.
4. Are you saying 3.6% from right now? We are 10ish % from the peak, so this should be factored in.
5. A PPOR is treated very kindly by the government when it comes to means testing for most things, tends to have modest associations with share prices, and is a decent inflation hedge. Have you considered paying the minimum deposit on one and getting a loan BEFORE you quit work? It will likely be much harder if not impossible after you retire to get a big loan. If you can get one with good bones that needs a reno it's something you can do with all of your spare time - then you have the option of selling it tax free later on.
6. Can you switch to part time for a few years to ride out any potential horrible sequence risk the first few years? Do you have skills you can sell? Are you willing to do jobs like house washing/mowing etc. if things go bad? A few years of that can prevent you drawing down during bear markets which is the ONLY reason these simulations fail. If you MUST retire from all forms of paid retirement that is different to a skilled person choosing to. The former has few options for topping up their spending, the latter has much more potential.

That's a few of the MANY things I've thought about while planning ahead. My guess is that at least one of you will be employable for many years ahead. So the odds of failure are probably low. The psychology of it will be the toughest thing....

If you are steadfast on 3.6% then I recommend downloading the data for USA from Schiller and the AUS data from the RBA sites. You can then build your own models. I like David Swensens model for Yale which is a simple adaptive strategy that accounts for market booms and busts. Play with these data and you'll get a better idea of how things might play out.

Good luck!

 

@Realist35

Acknowledging that you aren't using Super which has mandated minimum withdrawal amounts which increase every few years however, each fund has a calculator which allows changes to the assumptions and aims for zero balance at the end of X years.

Other assumptions include a tax free environment (which managed funds aren't), a rate of growth/return, single/married, optimistic or pessimistic outlook etc.

Although you are confident of your own abilities to invest, have you considered putting $x into a SMSF based on the 3 years non-concessional & concessional contributions for both you and your partner & investing that in the same manner ie 50/50 in VAS & VGS in order to create a tax-free income stream after 60? That way, you would still have the same starting point, same draw down (taken from ½ of the investment pool), smaller overall tax exposure/expense due to the concessions in super, greater growth in the super (due to lower taxes).

I suggest a mixed strategy for the reasons that you don't need to access all of the funds invested in the short term ie before you turn 60 (no idea of how far off that is for you), capital withdrawn is tax free (after meeting release criteria) but also note the required withdrawals commencing at 4% (which you can reinvest outside of the fund).

*Not licensed to give financial, marital, health, lifestyle, legal or any other advice which may have consequences on your life.

 

1. There is a very interesting thread on this topic under the FIRE forum
2. SWR has nothing to do with whether you own a home
3. 50/50 VAS/VGS is as good as any SAA IMO, however, at the moment you retire, your risk is highest, and that can be mitigated with low risk assets to cover a couple of years of spending (so you don't withdraw equities at their low point)
4. Safe means absolutely NO chance of going to zero. 4% will be safe in most cases. There are scenarios where you could do 8% and still be fine. It is stochastic but at this point unknown
5. Remember you can always fall back to aged pension if disaster strikes later on when your spending will certainly decrease
6. The SWR should be variable - one does not have to withdraw 3.6% every year. In times of trouble one would withdraw less, and vice versa. If mandated to withdraw, you can reinvest if you want.

 

Thanks so much for such a detailed reply!

I wanted to share here an awesome online calculator I have come accross. It backtests your portfolio and the success rate.

FI Calc

 

Thanks for the share.

A very good calculator offering a range of withdrawal approaches and explanations.

 

Good one.

The distributions show what I was trying to say in point 4 - it is very unlikely to go to 0 and one will generally end up ahead. The red starting years demonstrate that starting with a very bad few year(s) is the only way to fail - this can be mitigated with a bond tent and a variable withdrawal rate.

Good luck.

 

The calculator has an option to export the data in Excel after you run a scenario. I analysed the data for a bit, specifically the last couple of years of retirement, and to me there is an issue with the calculator.

Therefore I looked at other ones and came across these two amazing tools. They both do back testing.

cFIREsim
Monte Carlo Simulation

The cfiresim calculator gives lower chance of success when I input the exact same parameters as in FI Calc. I am still happy with the outcome though. However the portfolio visualiser calculator is giving much lower success rates. Hm, I wish one of you gurus can figure this out

 

They are two completely different methods. Put simply, Cfiresim uses historical returns directly, portfoliovisualizer.com uses summary statistics from historical returns to create a model of these data. They both have strengths and weaknesses.

Personally I've done my own "simulations" as none of them did quite what I wanted. But even then it is a hollow feeling of knowing that none of them can be right because none of them can predict the future.

Give yourself a buffer, remain employable in case you have to be, understand that this is all random chance and outside your control or prediction. It's about the best we can do.

 

New paper, just starting to look at it now ;

The Safe Withdrawal Rate: Evidence from a Broad Sample of Developed Markets by Aizhan Anarkulova, Scott Cederburg, Michael S. O'Doherty, Richard W. Sias :: SSRN

Synopsis ;

We use a comprehensive new dataset of asset-class returns in 38 developed countries to examine a popular class of retirement spending rules that prescribe annual withdrawals as a constant percentage of the retirement account balance. A 65-year-old couple willing to bear a 5% chance of financial ruin can withdraw just 2.26% per year, a rate materially lower than conventional advice (e.g., the 4% rule). Our estimates of failure rates under conventional withdrawal policies have important implications for individuals (e.g., savings rates, retirement timing, and retirement consumption), public policy (e.g., participation rates in means-tested programs), and society (e.g., elderly poverty rates).

 

Two things this paper shows.

First the fact it comes up with such low withdrawal rates shows the madness of modelling a fixed withdrawal rate from volatile assets.

Second. How important international diversification is.

Because they are using fixed withdrawal, domestic only equity and incorporating domestic markets that have had catastrophic failures. The withdrawal rates are actually the straight line that gets below the mathematical probability of picking up those catastrophic examples in the blocked bootstrap.

Looks like a fairly solid database but the premise of the research is meaningless. Except maybe to educate/scare those that think international diversification isn't necessary and/or flexibility in spending from volatile assets is not necessary.

The headline safe withdrawal rates though are unnecessarily alarmist to any half decently constructed internationally diversified portfolio in the hands of anybody who would react even slightly rationally to troubled markets

 

Dynamic withdrawal is a total no brainer, and agreed I think anyone following the markets with half a clue would reflexively tighten the belt when needed.

The similarity of SWR at 40/60/80 equities is interesting.

There is a thread on RR now where one of the authors is engaging in Q&A btw.

 

Not surprising to me given the design, but in reality, a wrong conclusion based on flawed design. Think about it --- They have widened the database to include failed states. They have then only looked at it from the perspective of domestic investment. So, in their total bucket of scenarios to draw from in the bootstrap process they have a couple of catastrophic failure scenarios. The results they create is the chance of if, and how soon they draw out the catastrophic scenario and how many iterations they run. Because they are using a fixed withdrawal to get under the lowest observation everything is smashed low, the SWR and the effectiveness of additional growth assets.

In reality you handle this potential for total loss by diversification - you don't sensibly hold just 1 company! In their study it would have been much more meaningful if they didn't just hold one country in each bootstrap pick. You could have then drawn some conclusions about investing instead of only being able to draw conclusions about 'math's' in a bootstrap process that has a fixed variable and a catastrophic incidence in the potential sample.

There are exceptions, but my summary of RR is that it has become largely infected by group think and appeal to (academic and visible author) authority. Not a place I choose anymore to spend much of my limited time other than reading mainly the links.

The market is the only thing I ultimately respect - I wouldn't use my time effectively if hanging out there.

 

Indeed, scenarios where 50% home country + 50% all other countries was run would make sense. I don’t have issue with inclusion of failed states but your point on diversification is well made.

Same page on RR, I skim it every now and then. They really went down a rabbit hole over there.

 

Thanks for sharing - I couldn't help myself and read it knowing that most likely it wouldn't be very good. It wasn't.

There's a lot of problems with this "research", of which @dunno has mentioned a couple. For those interested here are some other issues I have with it.

Scientifically:
1) The simulation process relies on normal distribution estimates for the modelling, but these data are not and cannot be normally distributed. It is very well established (see Nassim Taleb's work - I'm confident he would hate this paper but he would never read it) that share market returns are not normally distributed. This is one of the major reasons banks/hedge funds etc. go bust over and over again, they use models reliant on normal distribution to predict the outcomes of something that is not normally distributed. You can bootstrap all you want to use the central limit theorem to get normally distributed looking data, but the reality is its not.
2) You can see evidence of this lack of normal distribution in their Table 2 data, where they show a lot of countries having MONTHLY standard deviations of returns in stocks and bonds above 10% per month! This occurs around monthly mean returns of between -1 and 1%. This implies that they expect ON AVERAGE these countries to have a monthly return of more than 20% or worse than -20% for 5% of months - the equivalent of happening more than once every two years. The worst example is german bonds, with a monthly standard deviation of 46%. This implies that your money is likely to go down/up in value by 46% or more once in every 3 months on average.
3) This is likely happening because when a market collapses it happens quickly. This causes a massive tail effect, in essence a 100% drop in value on a single day (i.e. whatever the current value is the day before closing turns into 0, hence it is a 100% drop). This impacts the normal distribution in the same way that Bill Gates walking into a room of people with no money makes the average wealth look much higher, but also throws the standard deviation of the wealth out of whack. This makes it looks like the people with no money have a wide range of wealth, including some having hundreds of millions of dollars, but they all actually have nothing except Bill Gates.

I can't see anything in the methods which describes how they deal with this. My guess is they don't and choose to either ignore it or don't know this issue exists (hence why they haven't tried to fix it).

4) The have international stock data in their dataset, they talk about it a lot in the paper and use it for the target date example, so why don't they use it for the other comparisons? Eyeballing it you can tell it will increase the SWR markedly, because all the countries with very low domestic return values have relatively high international return values. Hence it should washout the poor returners, which is the only thing this paper assesses because it is focused on the negative tail end (i.e. only the worst outcomes).

Apologies for the rant, but this stuff really annoys me. It's just another poor science article that attempts to scare people, and the truth seems to be hidden. This sort of thing makes me ask the question - What do you think the odds are that one of the co-authors will soon be running a fund promoting their skills as risk experts? Then if they do, what are the odds they blow up because they are repeating the cycle of people not knowing what they are doing pretending to know numbers? I don't have a lot of sympathy for people who lose their money bring greedy (i.e. wanting to get rich quick), but I do have a lot of sympathy for people who are just trying to live comfortably on their nest egg they have earned and then get swindled or fed misinformation.

 

Couldn't agree more. The assumptions made by so many of the "academics" and people talking about SWR and expected returns boggles my mind. Why would we think that we have seen the worst or best share returns, and hence can confidently model prior data to predict future returns? Every generation so far has had some "never before seen" moment in the share markets, why do we think that will stop now? For how long have we had a majority/large proportion of the population of developed countries pumping money into stocks on a regular basis (eg. super in Australia, 401K in the USA)? What effect will this have on future returns?

I am optimistic that a well-diversified portfolio of a wide range of sectors of global stocks indexed somewhat to market cap, but with some smaller cap stuff thrown in there, will provide the returns I require over the next 40+ years. To write a paper giving a specific SWR with a decimal place in the value and state it with any confidence to me is ridiculous.

 

Just re-read those posts....I must have been grumpy yesterday !

 

I didn't notice any difference.

Anyways, welcome to the club.