Dollar Cost Averaging (Don’t Buy It)

October: This is one of the particularly dangerous months to invest in stocks. Other dangerous months are July, January, September, April, November, May, March, June, December, August and February. -Mark Twain

The Theory

Dollar Cost Averaging (DCA) is a technique where, instead of investing a lump sum all at once, you spread it out over time. For example, maxing out a Roth IRA every year. Vanguard allows a good deal of freedom in setting up automatic deposits, so if I wanted to I could set it up such that the $5000 max is spread out over 52 weekly deposits. (I also have the choice of monthly, yearly, every 2 weeks, or twice a month.) In my case, $5000 divided into 26 equal payments works out to $192.31 every two weeks. By spreading out my investment over a large period of time, I’m Dollar Cost Averaging. There’s no specific time period that you have to divide your investment into, but the most common I’ve seen is anywhere from 1 week to 1 month. Regardless of the time period, the idea behind DCA is the same. By spreading out your investment over time you lower the risk of buying a lot of shares high or trying to time the market. It’s used as an automated strategy to buy more shares when the price is low and fewer when the price is high.

Sounds Good, But…

Does it work? Rather than creating a scenario like “Investor A invests a lump sum in fund XYZ in January while Investor B uses DCA to invest the same amount over a five month period” and then skewing the results so that DCA wins (see the Prudential example), I chose to uses real performance data for a few funds. I’m quite skeptical about contrived illustrations. To do this I set up a Google spreadsheet that pulls in historical fund pricing from a few mutual funds that I’ve held and ran the numbers. I used weekly data (that’s the most frequently Vanguard would let me invest) and set up a simulated account as if I had been investing $5,000/year from January 2000 until the end of 2011 (12 years). I ran three different scenarios. The first assumed weekly investments (in other words DCA), the other two invested $5,000 per year as a lump sum either in January or December of the same year. Total simulated investment was $60k for each scenario, the only difference being when the money was invested. [EDIT] It actually worked out to about $58.5k invested since data wasn’t available for a few of the dates. In order to simply things, I assumed that the money is in a tax advantaged account (in this case a roth although 401k, 403b, etc would also work). I also assumed no withdrawals or anything else that would pull money out.

What I found was surprising. For all three funds I chose to simulate, the returns were actually higher using lump sum investments. For the Vanguard S&P 500 Index Fund (VFINX) using DCA the increase in value over 12 years was $4735. For lump sum investments the gains were $4934 and $6362 when investing the sums in January and December respectively.

There was no specific month I found that was always better. Although there probably is statistically an ideal month to invest, I wouldn’t recommend making investment choices based on it. Of all the funds I looked at, none showed an advantage using Dollar Cost Averaging. I even tried using some large cap stocks and found even greater advantages to investing lump sums vs DCA. Apple (AAPL) was especially depressing (not the difference between methods but the fact that I sold my ~300 shares ~8 years ago).

The problem I see with DCA is that on average you are investing later. Putting $5k in an account in January will give the money more time to gain value than spreading it out over the whole year. Obviously this won’t work if the fund is on the way down, but as a general rule of thumb I don’t see how anyone could argue against investing earlier rather than later.

Conclusion

Dollar Cost Averaging might net you better returns, or it might not. I still do it because it’s easier to budget for periodic investing instead of a yearly lump sum. Any adviser who shows you a pretty graph proving how great dollar cost averaging is should be treated with caution. (Historical price data should always have jagged edges.) Or, better yet, print out a copy of my data and let them scratch their head and complain about how most of their clients don’t ask them questions like this.

References:
http://www3.prudential.com/signature/Dollar-Cost-Averaging.html (They’re looking to sell you something)
http://en.wikipedia.org/wiki/Dollar_cost_averaging
What’s the better way to invest—little by little, or all at once? I actually found this post a few months after writing this and figured it would be good to add. They came to the same basic conclusion, just with more graphs.

6 thoughts on “Dollar Cost Averaging (Don’t Buy It)

  1. Interesting post. I’ve long been a supporter of DCA, but haven’t really done the math as you have here. For me, I do it simply because its easier & If I was investing a larger sum I would be more apt to try and “time” the market looking for the best possible entry point.

    With your December #’s did you run them from Dec ’99 and Jan ’00 or Jan ’00 and Dec ’00?

    • It was Jan ’00 and Dec ’00. What I did (which you can partially see from the linked spreadsheet) was to figure out the share price once per week. For the January/December numbers I then divided my total dollar amount invested by the share price and add it to a running total. The goal was to simulate how the shares would have actually been purchased.

      I was a bit surprised at how the lump sum at the end of the year worked out so much better, I would have thought that investing it all in January would be better since I would have almost another full year worth of gains, but most of the stocks or funds I ran this on actually didn’t work out that way. Just goes to show you timing the market is tough to do, if at all possible. Maybe I’m conditioned to think that way by all the smooth graphs the financial “pros” have shown me.

      • This is perplexing. How is it possible for the Dec contributions to be higher than the Jan? The only thing I can think of (and I didn’t check this against the real numbers) is the particulars of 2000 and 2012 i.e. would the result hold by picking other 12-year periods? For instance, if the market lost value (relatively) in 2000 or 2011 it would have been better to buy at the end of those particular years, but not every year will be like that.

      • I think it’s just how it happened to work out in the 12 year period from 2000 to 2011. I suspect that if you were to do it over a longer period, say 30+ years it would shift so that the January lump sums were advantageous. The only reason I didn’t do it over a longer period was that the google finance functions I used to get the data only let me go back to 2000 even though the funds in question have been around for longer. For some reason it worked for longer periods using stocks, but not mutual funds.

  2. I do DCA but I have read analyses like yours that lump sum is better (only at the start of the year, though!). We still DCA because the money comes from our cash flow monthly – we don’t have the lump sum to invest at the beginning of the year.

    • We do DCA for the same reason. My main problem with DCA is that it’s used as a selling tool, I have yet to find a financial services company that uses actual data to justify it. The prudential link I have at the bottom of this article is a good example, obviously DCA will work great if the market drops 40% after January. Call me crazy, but I think they’re more interested in getting lots of money under management and DCA is a good way to have people do that.

Leave a Reply

Your email address will not be published. Required fields are marked *


CommentLuv badge