When it comes to retirement planning, one of the ideas that I’ve been keen to explore is how do we measure risk in a drawdown portfolio.
The common industry standard on this is to use volatility. Based on previous work we have done, we found there’s very little correlation between the volatility of a portfolio over any given time period and sustainable withdrawal rates.
So if volatility isn’t a good measure of risk in a retirement portfolio, what is the alternative? We decided to stress test portfolios, factoring in withdrawal amounts and asset allocation, against a wide range of market conditions. We came up with four criteria to determine what a successful portfolio in retirement looks like and to visualise risk:
- Success rates/ probability of success. That is, if a portfolio is tested over a 30-year period, in how many scenarios out of 100 does the client not run out of money? This is crucial because it gives a sense of how sustainable the retirement income is.
- How long does it take for the pension pot to run out in the most severe scenarios?
- What is the cumulative income over, say, 30 years? Although a strategy of adjusting income based on portfolio performance would generate a 100 per cent success rate, in order to achieve this the client’s income has to be cut. This may not look like a ‘successful’ strategy from a client’s point of view.
- What is the legacy? That is, the median average balance left at the end of the period.
Our measure of success was to see whether a particular strategy did better than another on three out of four of the above criteria. So what did we find when we ran the numbers? The results might surprise you.
Forget what you know about rebalancing
Let’s assume you start with a simple portfolio made up of 50 per cent equities, 40 per cent bonds and 10 per cent cash. Alongside taking withdrawals, we looked at the impact of rebalancing annually and every two years. We also examined the impact of rebalancing when any of the asset classes have moved by 5 per cent and 10 per cent from what was set originally, as well as the impact of a ‘do nothing’ strategy, with no rebalancing at all.
We based our analysis on historical data between 1926 and 2017, over rolling periods.
What we found is when it comes to the first four rebalancing methods, there’s very little difference – there’s no statistical significance whether you rebalance annually, every two years or by 5 or 10 per cent bands. Less frequent rebalancing seems to produce slightly better outcomes as you’re giving equities more time to grow.
But perhaps what’s more interesting is the strategy that seems to do the best is the ‘do nothing’ strategy, where you just leave the portfolio alone. Obviously, in a sense you are cheating by doing this, as it means the portfolio becomes riskier over time. The equity element will end up overwhelming the bond element because they grow faster – it’s not rocket science.
Yet even in the worst-case scenarios, the portfolio lasted longer where there was no rebalancing carried out. Based on these metrics, we feel comfortable saying the traditional approach to rebalancing in retirement doesn’t seem to add much value. And this research has been corroborated by work carried out elsewhere, such as in the US.
As well as allowing equities to grow, a ‘less is more’ approach to rebalancing works because strangely enough, the withdrawals themselves have a smoothing effect. For example, where you withdraw across all asset classes, in a bull market most of that withdrawal will come from equities because that is the asset class that has done better. It’s not exactly the same effect as physically rebalancing, but it does help to ensure that in most cases the portfolio doesn’t go completely out of control.
I have to admit when I first saw these findings and read about them elsewhere, I was quite surprised. I had been a big fan of traditional rebalancing, and this idea that we should keep equities within certain risk boundaries.
These results will be a shock to some advisers, as it’s always been done that way in terms of regular rebalancing. It just goes to show the ‘activity bias’ that is present, and that rebalancing can be a way of justifying adviser fees to clients. But the numbers don’t lie, and they point us to the perhaps uncomfortable conclusion that sometimes less is more on rebalancing.
What about asset allocation?
Our numbers had some more surprises in store. Suppose you take your bond allocation, and substitute bonds for something else. Our substitutes were Treasury bills, to see if this changed the risk parameters in any significant way. The answer was no.
Then we asked, what if we replace global bonds with UK bonds? Again, this doesn’t change the outcome in any significant way. Then we thought, let’s go crazy. What happens if we substitute bonds for gold? In fact, the outcome actually improved.
Now, to be clear, the numbers for gold are terrible. Since 1926, gold has delivered an average return of 7.6 per cent a year, but the volatility is 20 per cent. On its own, it’s a terrible asset class. But when we mix it with other things, in some aggressive inflationary environments gold tends to come into its own.
There are issues with how you capture that return, so I’m not saying pile in on gold, but I use it to make the point that nothing beats equities. Bonds are the traditional diversifiers, but I would say be careful how you use them.
Overall, from a sustainable income point of view, the idea that when clients are in retirement we should be derisking a portfolio and selling down equities has zero empirical basis. There’s no support for it in the extensive historical data we have on the behaviour of asset classes.
At our recent Science of Retirement conference in London, we heard from the legendary Nick Murray. He had a great analogy which compared bonds with a rattlesnake which I can’t recount in detail here, but the upshot was bonds should come with a warning not to go near them. Why pursue fixed income in an environment of rising living costs?
The point of all this is we can’t pander to clients. Previously, drawdown was used for discretionary spending. Clients weren’t relying on drawdown for a substantial part of their income.
But in an environment where we’re increasingly moving away from guaranteed income, and clients are facing longevity and inflation risks, I just don’t see how the argument to derisk in retirement holds up any more.
This article is based on a presentation given by Abraham Okusanya at the Finalytiq Science of Retirement conference 2018.
Start the discussion