Sequence of Return Risk

Yes I was having a bit of fun joking. Most won’t be fortunate enough to have significantly more wealth than they need so inflation will be a dominant issue.

That nails it of course. Each of us is unique both financially and behaviourally. Whatever our beliefs we should consider ourselves fortunate that we Have a plan and are at least making an effort to save and invest unlike the vast majority of the population. If the likes of you and I are unable to cope with an adverse sequence of risk event then god help most others.

 

Sequencing risk on the road to FIRE – returns aren’t smooth and why it matters

Another good read about sequence risk with some analysis of the Australian market.
Unfortunately the author comes to the same conclusion we have... there isn’t much that can be done other than having a healthy cash bufffer at the retirement date.

 

The way I understand Sequence of Return risk is that having a bond tent/cash buffer to prevent forced selling of equities in the wrong moments will at worse result in a bloody nose as opposed to being spread all over the bitumen having no buffer at all during wild rides in the market. It’s hard to come out completely unscathed in any significant crash event unless you have a large asset based to begin with.

 

If you could live comfortably off dividends with a safety margin to mitigate risks then it should be OK. However, if certain risks came to fruition then dividend income would be impacted e.g. big market correction like the GFC with extended recovery period and a change in rules for franking credits; both plausible events.

Say Allan Koholic has $1million in shares paying 4% fully franked dividends. Al will earn around $57k. Assume 35% drop in dividends and loss of franking credits and Al now gets $26k. That's less than half and may require sell-off to sustain living expenses. I think the important point that we've discussed before is having a few years living expenses in cash.

What would it take for the big banks to cut dividends?

 

Thinking out loud: assuming you're dollar cost averaging into the market. Then one approach may be to decrease investments into the market as you approach retirement thereby building up the cash buffer / bonds. If the market crashes then start to buy in using the accumulated money where the dividend yield should be higher given the market has crashed. If no crash, maintain the cash buffer and slowly increase investments back into the market AFTER retirement maintaining the cash buffer for a worst case scenario. It will reach a point where either the market crashes and you reinvest or the calculations say that even in the worst case scenario you will be fine without any sort of a cash buffer.

 

Sounds familiar:

Sequence of Return Risk

 

Playing around with some Australian data to see what effect historical sequences have had on different asset allocations.

In this chart, both options started with capital of 17X annual expenses and a 50% bond 50% equity allocation.

Option a did not rebalance the bond allocation – rather all expenses were funded from bonds until they were exhausted – creating a rising equity glide path.

Option b rebalanced keeping the 50/50 static allocation.



For the first 20 years the static allocation looks superior, generally higher capital value and less volatility, but beyond 20 years the more pleasant (less volatile) ride afforded by not gliding back to equity is starting to have an impact.

 

I've done some basic ASX historical returns analysis. Data is annual returns (including dividends) from 1900 to 2017.

I wanted to look at the relationship between return in a given year (x-axis) vs returns for years to follow (y-axis)

Below are probably relevant to the sequence of return risk discussion.







Clearly, the 2 and 3 yr charts show the sequencing risk is significant with shorter time horizons.

5 years is where things get interesting. There were 4 instances of negative forward 5year CAGR returns. 3 of them were only by a small amount, so as not to be utterly devastating

1968 (forward 5 year CAGR of -0.57%)
2006 (forward 5 year CAGR of -2.12%)
2007 (forward 5 year CAGR of -1.98%)

The biggy was 1969, which recorded a forward 5 year CAGR of -9.13%...
1970: -16.2%
1971: 4.3%
1972: 26.4%
1973: -23.3%
1974: -26.9%

An interesting observation is the returns for the 4 years that had negative forward 5 year CAGR were elevated. Not substantially, but enough to be noteworthy

1968: 42.5% (rank 9 of 118)
2006: 25% (rank 19 of 118)
2007: 18% (rank 42 of 118)
1969: 14.7% (rank 58 of 118)

Based on these stats, I'd be comfortable with around ~5yrs of living expenses as a cash buffer. Roughly >99% chance of success ain't too bad...

Data source: https://www.marketindex.com.au/site...stics/historical-returns-infographic-2017.pdf

 

Hi @Befuddled

Nice to see you looking at the data.

Nominal rates is not the full picture for sequence risk. Inflation plays a huge part.

for example, with your 1969 example CPI for the relevant years was

1970: 5.3%
1971: 7%
1972: 4.7%
1973: 12.5%
1974: 16.7%

That gives a CAGR of -19.55% in real terms or reduces a 1969 $ to 33.7cents in real purchasing power terms by 1974.

Actual wage inflation was substantially higher than reported CPI. High inflation remained a problem for more than another decade, making recovering in purchasing power terms very difficult.

 

@dunno where can I get inflation data back to 1900?

That's a missing piece of the puzzle that I want to integrate into this analysis.

 

http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/6401.0Sep 2018?OpenDocument
Will take you back as far as 1948 with what should be pretty acurate data.

Secret squirrel site for lots of interesting historical data (including long dated series on house prices for many countries) will take you back to 1870 for AUS CPI. How accurate the data is that far back is anybody’s guess.
http://www.macrohistory.net/data/

 

What If You Retire At a Stock Market Peak?

 

Cheers, Enjoyed that if it's possible

Excellent to compare real returns and how the best period of $ returns was the worst outcome.

 

Plugged in the CPI data from macrohistory. Real returns over 2y, 3y, 5y, 10y periods






On a real basis, 5year time frames become unreliable, with 13 instances of negative 5 year forward CAGR, including a series of 6 consecutive years where the forward 5 year CAGR was negative (between 1968 - 1973). Interestingly the CPI between 1974 and 1977 were in the 12-15% range.

Even on a 10 year forward-looking basis real returns can be negative. 5 instances:
1964: -2.6%
1967: -1.69%
1968: -3.66%
1969: -1.66%
1972: -3.23%

Ouch. For me, the takeaway from this is that the sequence of return risk is elevated in times of high inflation.

Another observation is that on a real basis, the 10 yr forward CAGR after a negative year was at least 2.47% p.a (1965), with a maximum of 15.21% p.a (1916).

Why is that interesting today? The all ords finished 2017 on ~6167. If we subtract 4% for dividend yield, it needs to finish above 5920 by the end of the year to register a non-negative year. It currently sits at 5749, so there's a good chance this year will be one of those negative return years.

 

What if Sam is Japanese and is still waiting for their market to recover 20+ years on from the last peak?

 

I’ve heard frugality is the in thing nowadays.

 

Great Post @Befuddled thanks for sharing your analysis.

I couldn't agree more - Just looking at nominal returns you miss at least half the picture. Investing for me is all about maintaining/increasing purchasing power over the long term. Nominal doesn't tell you anything about purchasing power.

Initial valuation matters big time to sequence of return risk. Your analysis here is an extremely robust measure of valuation in that if the return for current year is negative than the market must be cheaper than the previous year and from this simple measure of cheapness you can already see the data confirming that valuation matters to sequence risk. i.e. no historical negative real 10-year CAGR returns starting from a negative year.

Starting valuation in relation to sequence risk is my favoured area these days for mental exercise. I think there may be a case for adjusting your asset allocation at peak sequence of return risk years if equity markets are expensive. (determining a robust expensive measure is the trick)

Though it doesn't really stand out because you adjust the axis scales, the other very important point you analysis bears out is that return volatility reduces as time increases. The axis for one-year volatility is -40% to 60+%. whilst the axis for 10 year is -5% to 20%.

So in you excellent post I see three significant concepts considered via first hand examination of the data.

Inflation matters;
Starting Valuation Matters;
Time reduces volatility.

Once again, thanks for your post.

 

@dunno is back!

Cheers,
Oracle.

 

Couldn’t resist acknowledging @Befuddled post.

 

The old Japan nugget.


This is a historical price chart for Japan. I can’t find a total income chart, I suspect it would have exceeded the 90’s peak by now.

Japan has had low or negative inflation since the peak, so nominal returns have not been undermined to the same degree as other countries.

Realistically if you were about to retire in Japan just prior to the 90’s bubble bursting you should have benefited from the run-up and been a squillionaire requiring such a low withdrawal rate to maintain a high income on an international comparative basis that even the lost few decades wouldn’t worry you.

Valuation matters: whilst excessive valuation is tricky to see in the present, I suspect a near retire exposed to only Japanese equity may have been able to at least glimpse some of the overvaluation and risk then present and take asset allocation decisions appropriate to their near retirement circumstance. If they didn't take steps and they only had just enough capital to maintain a 4% withdrawal rate - yes those few would be toast in what was to come.

The next chart is GDP per capita on purchasing parity terms. (how rich on average is each citizen of a country on equalized purchasing power terms)



Despite their terrible share market (their property market suffered similarly) Japanese wealth per person is hanging in there. Actually their wealth Gini coefficient is lower than Australia’s which means their wealth is distributed more evenly. I suspect their average retiree actually is in a superior position to average Australians. Things didn’t seem so bad when I was last there.

We had a similar advance in GDP per capita wealth during the mining boom. Is it going to take us as long to bed that in as Japan did post their boom? How long before the technology/resource cycle swings back to favouring resource?

I suspect on a close impartial analysis of Japan, things on a purchasing power basis are not as bad as those that want to scare you make out.