Sequencing risk, the Trinity Study, and the 4% rule in Australia

I’ve written plenty of times previously about sequencing risk and the 4% rule aka the Trinity Study and they’re areas which the FIRE community are very much interested in.  For those who are new to the subject, the Trinity study as often used in the FIRE community basically means that each year you can safely take out 4% of the value of your investments when you retired and never run out of money.  Or to put it another way, once your investments are worth 25 times your living expenses you can declare FIRE and retire. 

The actual Trinity study only looked at this for 30 year periods rather than for the rest of your life so it wasn’t based on the situation for someone retiring in their 30s or 40s, but that often gets glossed over.  It also used US data rather than Australian so isn’t really applicable for those of us Down Under either.  I thought it would be good therefore to have a look at what the actual data says about how a 4% (and 3%) drawdown strategy would have worked historically in Australia for someone investing in Australian stocks and fixed income.

To keep it to nice round numbers I started all scenarios with $1,000,000, and 3% and 4% drawdowns so $30,000 or $40,000 a year.  Or to put it another way I’ve assumed that the amount of money you are drawing down is the same each year in real terms, ie if you start off drawing $40,000 a year and inflation is 10% then next year you draw $44,000 in nominal terms but it’s only going to buy you the same amount of stuff. 

I ran a bunch of scenarios with 3% and 4% drawdown rates with the first period starting in 1968 running the numbers through till 2016 and the last one starting in 2010.  I also did asset splits between 100/0 stocks/bonds to 60/40 in 10% increments.  Obviously not all of these are 30 year periods but are interesting nonetheless.  Also I’ve put in a few tables and graphs for those who are more visual, I’m trying to get away from writing a wall of text although this is a very long post so good luck either way!  So how did various drawdown scenarios play out? 

The Results

The below is for all time periods that have been going for at least 30 years.  The results are over the entire time period through 2016 rather than just the 30 years of the Trinity Study so the miracle of compound interest has had longer to work its magic on the earlier portfolios, although interestingly there are plenty of later starts which have done better than the earlier ones.  Keep in mind all of these are inflation adjusted so it is an apples to apples comparison for the dollar amounts, although as I said some have been running for longer periods of time. 

The good news is that over all those time periods a 3% inflation adjusted drawdown rate was 100% safe, as I said above this is using investment returns ending in 2016.  In fact for any period starting up until 2006 you have more money in real terms (ie inflation adjusted) than when you started.  So hooray for the 3% rule!

I don’t show it here but if you started in 2006, 2007, 2008 or 2010 you’ve got less money in inflation adjusted terms as of the end of 2016.  From a look at where the various asset allocations are and how this has played out in the past there is a pretty good chance of recovering from this.  Then again if I had an actual crystal ball I probably wouldn’t be writing about this sort of stuff on the internet so apply the appropriate amount of scepticism. 

Looking at a 4% withdrawal rate next, this time we have some failures and some portfolios which are likely in trouble.  And as we learned from the first table, even though compound interest has had a lot longer to help some of the earlier portfolios it hasn’t actually resulted in them ending up with more money.

So the bad news is a 4% inflation adjusted drawdown rate did not go as well.  This is exactly what you would expect of course, the more money you take out the greater the chance of it running out entirely which is what happened in some of the scenarios.  There were actually only two starting years where this occurred, 1969 and 1970.  All of these ran out of money entirely if you were in a 100/0, 90/10 and 80/20 portfolio. 

Interestingly the first 2 ran out of money within 20 years, but if you had an 80/20 portfolio you lasted 35 or 34 years (1969 and 1970 respectively) which probably gets you through to the age pension anyway depending on what age you retired at.  So this would be classed as a success for the Trinity Study.  In any case that’s a big difference in how long the portfolio lasts for a 10% change in asset allocation! 

If you were 70/30 you’re still going as of 2016 although you’re down to about $425k, again between this and the age pension you’re probably doing just fine.  The only other scenario where you have less in real terms than you started with is a 1973 start with 100% in equities and you’ve now got about $375k.

The reason for the failures was that the inflation adjusted returns in 1970, 1973, and 1974 for stocks were just horrible.  I’m talking negative 19.4%, 32.4% and 42.7%.  It’s actually possible to survive a couple of these years as indicated by the fact anyone starting in 1971 or 1972 made it through ok, but 3 big bad years before you’ve built up much of a cushion in the asset class where you’ve got most of your money was a killer.  The reason 1968 made it through ok was a 39.1% real return for shares in that year, getting you off to a very good start!

Interesting stuff

• Although the 3% rule did actually work out and you ended up with more money than you started with for all scenarios, looking at the sequences there would have been some extremely hair raising rides!  The 1970 100% growth portfolio dropped below $300k on two separate occasions over a 30 year period, and the 90/10 version did it once.  Plenty of other scenarios had very sizeable falls as well.

• It’s not at all unusual to fall below your starting amount at some point even with a 3% withdrawal rate.  If you use a 90/10 asset allocation and 4% drawdown (pretty much what I was planning on doing) then in 13 out of 20 scenarios you fell below $1,000,000 in inflation adjusted terms at some point over the 30 year period, normally within the first 5 years.  If you use 90/10 and 3% drawdown then you still had 11 starting years where you fell below your starting point. 
• If you make it through the first 5 years with more money than you started with you’re probably golden.  There aren’t any failures, and as I said above there are plenty of scenarios where you had less than you started with 5 years in and still made it through just fine. 
• 1975 was by far the best year to start off.  You ended up with at least $20,000,000 and in some scenarios over $30,000,000.  This was a minimum of $5,000,000 more than any other year in the same asset mix and often closer to $10,000,000.
• There can be huge differences in outcomes just a year or two apart.  Start in 1973 and you end up with at least $15,000,000 less than someone starting in 1975.  Start in 1968 and make it all the way through to 2016 with a minimum of just under $4,000,000, start in 1969 or 1970 and you failed 20 years in for some asset allocations. 

Conclusions

Given there are only 20 starting time periods where the scenarios ran for 30 years it’s pretty hard to come up with anything definitive, plus my data and methodology is not what anyone would describe as robust.  Also people would obviously change their behaviour if they were running out of money, would likely have different asset classes as well, etc etc.  And given we don’t have a crystal ball we don’t know how things would work out in the future either, as they say in the marketing material past returns are not a guarantee of future performance.  With that disclaimer out of the way, here is what I took away from this.

• A 3% withdrawal rate seems to have worked out just fine in all 30 year scenarios, as did 90% of the 4% withdrawal rates.   That being said there were plenty of nervous moments depending on the various factors involved.
• Having fixed income in the portfolio does increase your chances of a successful 30 year period with all 70/30 and 60/40 portfolios making it the whole way through.  The tradeoff with this is that it would generally mean taking longer to get to FIRE as you’d need to be building up more money in fixed income which tends to have lower returns.  Having said that it also doesn’t tend to have big drawdowns so it might help you dodge some bullets on your way to getting to FIRE, and you probably don’t have as many nervous moments once you’ve hit FIRE.
• This will likely be the subject of a follow up post, but basically it’s very unlikely that people would be declaring FIRE in some years.  If your portfolio got hit by a 32.4% inflation adjusted down year in 1973 and 42.7% down year in 1974, chances are pretty good you weren’t getting to having a million bucks saved up for retirement in 1975.  So that bit I said above about 1975 being a great year, well you probably didn’t declare FIRE in that year.
• Same deal for 2009, the GFC in 2008 probably put you a long way back.  Maybe you came into some sort of lump sum like an inheritance or won the lottery or whatever so it didn’t matter, but chances are you were building up your wealth over time and your portfolio got absolutely destroyed so you were delaying FIRE for a while.
• There is a huge range of outcomes depending on when you started.  The below is a 100% stock portfolio with 3% drawdown each year over a 30 year period.   You could end up with anywhere between just over a million and over 20 million.  That’s absolutely massive.

Data and Methodology

It was surprisingly difficult to get data to play with here.  I already had the All Ordinaries accumulation index share data so that wasn’t too bad but fixed income was a bit more difficult, not least because I had to decide what type of fixed income wanted to use.  Should it be 10yr Treasury bonds, should it be term deposits, RBA overnight rates etc. 

In the end I went with 90 bank bills, mostly because there was a pretty good dataset going back to 1968 and it’s potentially reasonably similar to the blended rate people would actually get from a mix of high interest savings accounts, term deposits and maybe some bonds?  If people want to argue whether that’s a decent proxy though go nuts with that, I’m certainly not kidding myself that it’s perfect.  Also because the returns were quarterly I just averaged them out to come up with a yearly return. I’m not sure where I got the inflation data from to be honest, I’m guessing the ABS.

The way I ran the scenarios was that if I had a FI allocation I ran that down entirely first before using any of the stock exposure including dividends at all. It’s pretty likely that in the real world you’d be topping up your FI allocation with your dividends but my google fu was weak and I couldn’t find a data set which separated capital gains and dividends for stocks so I did it this way.  Obviously I could assume a dividend yield instead and top up the cash with that but I didn’t.  I also went with a lump sum withdrawal at the start of each year for the cash needs for that year which isn’t the way everyone would do it either, but it was really the only way to calculate it given the data I have.

I’ve also magically assumed no taxes on the income, but given the relatively low income I’ve assumed of $40,000 in today’s terms (so leanish FIRE?) this is probably near enough to being true anyway for a couple holding the investment in joint names.  Management fees and transaction costs aren’t taken into account either, and you probably couldn’t have easily invested in the All Ordinaries accumulation index over most of these time periods unless you were a big institution so there’s that as well.  Basically all of this wasn’t actually doable at the time although it probably is now with just a couple of ETFs.  The joys of progress!

Lastly the accumulation index probably doesn’t take into account franking credits (which would have only existed for part of the time anyway, as well as the cash refund for these) so maybe this would have balanced it out.  I’m not enough of a combined spreadsheet jockey/tax expert to try to figure all of this out, and to be honest looking at the outcomes and the paths taken it likely wouldn’t have made much difference anyway in my opinion. 

So to summarise all of that there are some issues with the data and methodology, and it’s likely not the way people would do things in the real world, but I don’t think any of it would actually change the outcomes much.  The time periods that failed would likely still have failed, and the ones that worked would probably still have worked, although potentially not as well.  There are a few edge cases in there where that may not be true though.  I’d love to see a fund manager or academic with better data and methodology run the numbers, but this is what I’ve come up with and it should be reasonably accurate.

Disclaimer

None of this is investment advice or advice specific to your situation or meant to be relied upon in any way shape or form.  I’m reasonably sure the data is ok and I’ve written the spreadsheets with the right formulas etc, but please do not invest or do anything else based on what some random guy on the internet has written.  Apart from the obvious financial consequences if it doesn’t work out, it’s going to be pretty embarrassing telling your family and friends that you have run out of money because you listened to some idiot on the interwebz.  So for the sake of all of us, please do not do anything based on what I have written!

Follow up

I figure there will be lots of questions about this post and I’m planning a follow up to address those so if you have something you want covered, please comment below and I’ll try to address it!