What do you do if you have a large lump sum you need to invest? Do you spread the payments out and drip feed them into the market?
Do you worry about putting all your lump sum to work immediately, only to experience a market crash and lose a chunk of it immediately?
I was the same.
That was until I read a paper by Vanguard that back-tested the comparison of a lump sum strategy with a dollar-cost averaging strategy using historical data. It showed that investing all the lump sum immediately is often more optimal that spreading the payments over a period of time. This completely changed the way I thought about drip-feeding my money into the market and I now invest all in one go.
It’s also a pretty obvious result, but as I find with so many things, it wasn’t obvious to me until I had read some research and then done many hours of my own research. So what better way to welcome 2018 than with some analysis on a topic that I think will be useful to many!
Before going any further you need to read the important notes.
Lump Sum or Dollar Cost Average investing? Vanguard’s Paper
Vanguard’s rocket scientists looked at investing in a 60% Equity / 40% Bond portfolio with either a single lump sum at the beginning of the year, or 12 equal monthly installments over the year. They then calculated the rolling 12 year results from 1926 to 2015 and compared the end of year values and calculated two key measurements:
- How many results produced a “win” for the lump sum strategy?
- What is the average outperformance for the lump sum strategy?
In their paper they call the two strategies “Immediate” and “Systematic”, but I’m less fancy and will call the strategies “Lump Sum (LS)” and “Dollar-Cost Averaging (DCA)”.
Near the beginning of the paper there is a curious throwaway comment about looking at different periods, such as 6 months and 36 months, but very few details about this aspect. They also look at the UK and Australian markets and different equity/bond mixes. For both variations they find very similar results.
So what have I brought to this party in terms of new research?
- Extended the historical period to 1802-2015,
- Changed from monthly to annual payments,
- Varied the projection period from 2 years to 30 years,
- Looked at some specific historical periods,
- Also looked at different equity/bond mixes
- I use real returns that take inflation into account. This is important over the long periods I am investigating, but less so for the Vanguard analysis, and it looks like their returns are nominal (i.e. no inflation).
Are you excited to get into the details? I know, me too!
Who Wins? LS or DCA?
Let’s first look at a 100% equity portfolio and compare a LS with a DCA strategy over a 10 year period. Essentially, we are looking at the difference between investing $100 on day 1 compared with investing $10 each year for 10 years.
We find that looking at all rolling 10 year periods from 1802 to 2005 the LS strategy will win 88% of the time. That is a significant advantage to LS over DCA. Moreover, the average outperformance of the LS over the DCA strategy is 2.6% a year. If you wish, you can think of this as a “return drag” of using a DCA strategy, and it’s significant.
But averages don’t really reveal much, the really interesting insights lie in the cracks between the averages, so let’s dive in a bit deeper and look at the outperformance of LS versus DCA for all the starting years of projection. The following chart shows this.
Want to be served up new posts straight to your dinner table? Of course you do! Well I think of everything, just sign up in the email form through the main menu. You can also follow me on Twitter @actuaryonfire
When Does DCA Win? The Great Depression!
In the above chart I’ve highlighted two periods where DCA wins big time – The Great Depression and Stagflation in the 1970’s. If we look at the equity markets in the 10 year period beginning 1928 we see the problem – see chart below.
You can clearly see that a lump sum invested just before the market crash would have lost money for years (in real terms) and at the end of the period bounced back to roughly par. By drip-feeding your investment you could have mitigated this impact.
However, another lever exists to mitigate this impact and that is asset allocation. Simply diversifying into bonds would have eliminated the impact of the Great Depression on a LS versus DCA strategy. The following chart shows the outperformance of a LS versus DCA strategy on a 50% equity / 50% bond portfolio. You can see that the under-performance of the Great Depression period has vanished, but the 70’s Stagflation under-performance remains.
You can understand this better with the next chart that shows the performance of the bond market in addition to equities through the Great Depression. Check out the bonds – they killed it! Even though equity investors had a really rocky-ride, bond investors enjoyed a steady year on year improvement.
So when you hear about diversification of asset classes and the benefits of using bonds, this is a good case study!
But let’s now go back to the 1970’s and face the ugliness of Stagflation!
Did you notice above that when we diversified the portfolio into 50% equities and 50% bonds the LS strategy still underperformed the DCA strategy during the 70’s? You might have thought that was due to the dubious dress-sense and hairstyles of the 70’s, but no!
Where were the much-trumpeted benefits of diversification? Let’s look at equity and bond performance through that period.
Urgh! Have you seen anything so ugly!
Over ten years (ten years!) the equity and bond markets fell in real terms. You would have lost money over that period whatever your asset allocation. So you can see that a LS strategy loses to a DCA strategy over this period. The Irrelevant Investor has a good blog article on this time period, called “The worst Bear Market That Nobody Ever Talks About”.
Thankfully you can see from the outperformance charts above these ugly historical periods are few and far between and LS wins almost all the time (or at least 88% of the time).
Different Projection Periods
So what happens when we change the projection period from 10 years? Let’s now look at different projection periods from two to 30 years. As before we will look at all rolling periods from 1802. Just to keep things spicy we’ll also look at varying the asset allocation and consider portfolios that have 0%, 50%, 60% and 100% equity.
The following chart shows the proportion of cases where the LS strategy wins. (Honestly, I didn’t need to say that, since the title of the chart says it, and you’re a smart reader, but it’s a groovy chart huh?)
I think this chart is fascinating. It shows that for any investment strategy the LS strategy dominates DCA, but for periods over about ten years a LS strategy is more effective over 90% of the time (for equity-heavy strategies).
What’s nice is that for very short projection periods my results agree with Vanguard. Remember they looked at monthly payments over a twelve month period and my shortest period is two years, but it’s pretty cosmic-Zen that they agree.
You can also see the benefits of time on your investment. If you are investing over long periods then there is much greater certainty that you will benefit from a LS, rather than DCA, strategy. The mean reversion of long term equity returns is unrelenting, you can’t afford to be out of the market for long.
Did you notice the weird looking line of the 100% bond portfolio above? It looks very different to the equity portfolios, and with good reason. Bond returns over long periods tend to be driven by inflation, whereas equity returns over a long period are relatively immune to inflation. But even with a portfolio that is very heavy in bonds the LS strategy wins a large proportion of the time.
What do I recommend? Based on historical evidence from my research and Vanguard’s research it pays to be invested in the market on day one. Don’t drip feed, don’t be timid, you will be wrong more than a third of the time.
This analysis and Vanguard’s before it, is dependent on historical experience and is only useful to the extent that the future reflects the past. Vanguard also performed the analysis in the UK and Australia and drew the same conclusions, however I wonder what would have happened had they looked at the lost decades of the Japanese experience? The conclusion may have been different. There is also survivorship bias, in other words this analysis would look quite different during the hyperinflation of the Weimar Republic!
If you believe in a future with economic growth of real GDP then I think these results hold and you should be fully invested in the market and not dollar-cost average. Whether this continues in the future, who knows? But there is a huge weight of historical evidence for mean reversion of equity returns.
Do you dollar-cost average your investments into the market? Have you changed your mind about this like me, or do you continue to dollar-cost average your investments? How excited are you at Part 2 of this post on a scale of 1 – 10? Comment below and let me know what you like, didn’t like, and what you would like to see more of. Thanks for reading, I really appreciate it.
If you want to read some other number-crunching posts why not look at Reprise! Sequence of Returns Risk or College Investing Strategies. Want something less number heavy? Then try Bitcoin: My Best and Worst Investment. My friend and fellow blogger ERN has also written about this subject here.