Can Dollar-Cost Averaging improve your performance?

A student raised an interesting question regarding the topic of dollar-cost averaging.

Dollar-cost averaging (DCA) is a technique recommended by financial advisors to their clients to invest their money based on monthly intervals instead of a lump sum all at once. It made sense to do so primarily because most clients earn monthly salaries, so DCA was the most logical thing to do.

In my last early retirement masterclass, a student was interested in knowing whether there is any advantage in splitting a lump sum of capital to be invested over time to exploit the short periods where the stock market has been a bargain. 

The best way to investigate this question is to do it with a simulation involving actual market data.

We assume a capital of, say $1,000,000, and then we consider several scenarios. In all cases, a bit of Python programming can answer this question quite convincingly.

Our program looks at three scenarios:

1. The first scenario is where the lump sum is invested from day one of the simulations. (“Lump-sum”)
2. The second scenario is where we divide the capital based on the number of years of the simulation with a fixed amount invested at each fiscal year. (“Annual DCA”)
3. The third scenario is where we divide the capital into the number of months of the simulation with a fixed amount invested at the beginning of every month. (“Monthly DCA”)

We employ ten years of closing price data on the Vanguard Total World Stock Index Fund (Ticker: VT) ending 1 January 2021 to conduct our simulation and compare the number of stock shares we will own when the simulation is over.

Our results are as follows:

Analysis of 10 years dollar-cost averaging scheme for VT

Scenarios	Resultant Shares
Lum-Sum Units	20733.98
Annual DCA sum units	17174.44
Monthly DCA sum units	16859.19

Clearly, the advantage goes to lump-sum investing at the first day of the simulation for global equities. As we increase the frequency of DCA contributions, the performance becomes progressively worse.

The reason why this is so is that lump-sum investing increases the time spent invested in the markets. Suppose you invest 10% of your capital in year 1, 90% of your wealth is not doing anything for you after day 1 of the simulation. We should also note VT has annualised returns closer to 10% during the period of simulation.

Interestingly, when we apply this same exercise to the STI ETF (Ticker : ES3), the reverse occurs;

Analysis of 10 years dollar-cost averaging scheme for ES3.SI

Scenarios	Resultant Shares
Lum-Sum Units	301204.83
Annual DCA sum units	317864.18
Monthly DCA sum units	318718.71

The STI ETF’s annual returns over the past ten years have not been fantastic at less than 3% per annum, we’ve also traced a V-shaped pattern after the pandemic hit Singapore.

As such, DCA results in superior performance with monthly contributions doing slightly better than annual contributions.

For the same reason, Gold (Ticker : GLD) benefits from DCA as well.

Analysis of 10 years dollar-cost averaging scheme for GLD

Scenarios	Resultant Shares
Lum-Sum Units	7246.38
Annual DCA sum units	7989.08
Monthly DCA sum units	7718.53

How about Value-Cost Averaging (VCA)?

During the DCA discussion, a student proposes an improvement to DCA called Value-Cost Averaging or VCA.

When value cost averaging, our allocation is modified by the ratio of previous price paid for the stock over the current price. Suppose we have a capital of $1,000,000 that we intend to allocation over ten years with $100,000 allocated every year. If the stock price has halved from $1 to $0.50 from the previous year, we will buy $100,000 x ($1 / $0.50) or $200,000 worth of stocks.

When we do VCA we have to consider the possibility of running out of capital before the simulation is over. In some other cases, we may even have money left after the simulation, so we will have to convert remaining capital into company stock at the last closing price.

So we modify our code and repeat the process for VT. Our output is as follows:

 The final output does not differ much from DCA and lump-sum investing continues to be favored.

Let’s do the same for the STI ETF:

VCA seems to work over lump-sum investing. Also, VCA is seen to improve the performance over DCA subtly.

Finally, for GLD:

VCA beats lump-sum investing. VCA improves the performance over DCA and the difference is more pronounced than in the case of STI ETF.

Conclusion

Here are some of the lessons from this exercise:

• It is doubtful that any argument for dollar-cost averaging can be based on any empirical outperformance, a different approach I best for other asset classes. 
• Suppose the instrument produces decent returns and has low downside risk, lump-sum investing wins because there is more time for your capital to generate compound interest.  I would expect a value investing strategy that emphasises dividend payouts like my ERM program would benefit more from lump-sum investing.
• If the underlying instrument is highly volatile and does not produce a dividend, with overall returns being historically quite low, DCA is superior to lump-sum investing. In any case, the VCA strategy should be preferred over DCA whenever possible. I would expect a commodities fund or commodities ETF to benefit significantly from DCA and VCA.

In any case, the frequency of salary payments may limit retail investors to a DCA/VCA so you should not lose too much sleep if you are unable to do lump-sum investing.