Why Dollar Cost Averaging is Popular

Sometimes topics crop up in the PF blogosphere, seemingly out of nowhere, and rattle around from blog to blog for a while. Dollar cost averaging is a recent example. The Digerati Life brought it up on September 23, Lazy Man and Money responded the next day, and The Sun’s Financial Diary shared its Stock Chart thoughts on the 28th. There are probably several other mentions out there I missed.

Before I add my voice to the echo chamber, I’ll define the term. Dollar cost averaging refers to buying an investment, usually a stock or stock fund, over time in installments of equal dollar value.

It is often confused with the laudable and similar idea of regularly saving. Setting aside a certain amount of your pay every week or month may look like dollar cost averaging, but it’s not exactly the same thing.  Implicit in the question "is dollar cost averaging a good idea" is the premise that there is an alternative, that you could have invested it all at once rather than slowly as you earned it.

In its purest form, the question of the wisdom of dollar cost averaging boils down to the following hypothetical. Imagine you have $12,000 you wish to invest in the stock market. Do you put it all in now or do you put in $1,000 a month for 12 months?

Assuming that your goal is to maximize your expected return (a fairly safe assumption) and that having that $12,000 in the market is the right thing for you to do (more of a leap, but it’s a premise of the question) then you should put all the money in the market now. Dribbling it out over the next year will only reduce your returns if the market has a tendency to go up, and if it didn’t have a tendency to go up, why would you want to invest in it at all?

Put another way, the dollar cost averaging choice is between buying at today’s price and buying at the average price over the next year. If you think the market is more likely to go up than down, you must also believe that today’s price is more likely to be lower than the average price over the next year. This isn’t really all that subtle or complicated.

And yet dollar cost averaging enjoys widespread popularity. Partly, this may be confusion with the wholesome idea of sticking with a regular savings plan, something which deserves to enjoy widespread popularity.

Further, as SVB at Digerati Life tells us, we’ve just finished a year in which the average price of the market was a lot lower than the year-ago price, which was, as it happens, approximately today’s price as well. So maybe the reason dollar cost averaging has arisen as a hot topic just now is not such a mystery. SVB cites an example recently provided by Wells Fargo, an outfit obviously interested in inspiring regular investments.

But there is more to the appeal of dollar cost averaging than short-term hindsight and marketing hype. There is some interesting psychology. As Ken French (he’s a really important finance professor, trust me) explains in this video, people like dollar cost averaging because it makes them feel better, or at least maximizes their likelihood of feeling better.

Folks feel worse about buying something that goes down than they do about not buying something that goes up. Irrational, but true. So the missed opportunity of maximizing gains by not investing all $12,000 at the start is not so important when compared to the fear the market might go down after the $12,000 is invested. The appeal of dollar cost averaging is that each investment that could possibly be regretted later is reduced in size and made pseudo-automatic so the investor feels less specific responsibility. And, after a year, it is likely that the investor will have at least a few investments to feel positive about, even if the overall average is a loss.

But, as with all investment questions, the bottom line is the bottom line. The goal for your investment portfolio should not be to make you feel happy about how smart you are, it should be to provide for your retirement. Professor French discusses this as an interesting behavioral finance finding, rather than yet another example of emotions conquering logic and costing people money. But that is exactly what it is.

No Comments

  • By Drew, October 5, 2009 @ 1:41 pm

    I dollar cost average because it takes the “emotion” out of long term investing. Not all of us have $12,000 to throw in the market. I’m just a typical 30-something who has managed to pay off most of his non-mortgage debt. Instead of dumping money into just a savings or money market, I also invest in a small but steady stock that also happens to pay a nice dividend. By automatically investing the same amount every week, it removes the temptation to put in more or skip a week depending on the price.

  • By Neil, October 5, 2009 @ 2:36 pm

    Good post. I will sometimes use dollar cost averaging when saving for a vacation (buy my destination currency in installments), but I do this because I have no particular insight into which direction the currency market will go. Averaging gives me a nice middle-of-the-road price, so I’m not as affected by the volatility.

    In investments…well, I guess my regular savings plans qualify as DCA by default, but that’s a side effect rather than a goal unto itself.

  • By Rob Bennett, October 5, 2009 @ 3:10 pm

    It makes people feel good because it causes them to feel that they are paying an average price for stocks. Passive Investors worry (properly) that they are overpaying for stocks. Dollar-Cost Averaging eases these fears. They say: “Hey! I bought at the high and at the medium and and at the low — I’m covered!”

    The reality is that stocks were selling at insanely inflated prices for the entire time-period from 1996 through 2008. Those following Dollar-Cost Averaging were buying some shares at insanely high prices and others at super-insanely high prices. They were not “covered” at all.

    Telling people who invest passively to dollar-cost average is like telling people in a burning building to get out of the room that is on fire and then relax because they have solved the problem. The problem isn’t relative highs. The problem is absolute highs. Dollar-Cost Averaging does nothing to address the real overpriced stocks problem. It makes us feel good without addressing the concern that made us feel bad.

    Rob

  • By Rick Francis, October 5, 2009 @ 3:31 pm

    Frank,

    I don’t think you have done enough math to show DCA isn’t a sound strategy over some period of time- maybe it’s only days or months instead of an entire year.

    It seems to me that DCA might have the highest expected return in the following case:

    The market tends upward slowly over the long term but fluctuates significantly in an unpredictable way in the short term such that there is a time period that:

    #1 Is long enough to average out most of the short term fluctuations.

    #2 Isn’t so long as to lose a significant portion of the long term increases.

    Looking at some price data from the S&P500 it doesn’t seem implausible.

    -Rick Francis

  • By bex, October 5, 2009 @ 4:05 pm

    I think most people do “dollar cost averaging” because that’s how their 401k is set up. Yes… they invest as soon as they have money — on the assumption that in 5 years their value will be higher — but it’s simultaneously dollar cost averaging.

    Also… the average investor is far too easily swayed by fear and greed, so an “autopilot” investment scheme will help them avoid making dumb-ass mistakes.

    Yes… if they save up some cash/bonds and try to time the market or look for bargains they could probably do better… but odds are low that they could do significantly better than regular monthly contributions to index funds… unless they don’t have a day job!

  • By Dan, October 5, 2009 @ 5:02 pm

    Rick,

    That is incorrect. In your example since the short term fluctuations average out and there is a long term up trend, the price at first will be lower than the average price. Therefore, you would be better off buying it all at once.

    The only time when DCA would win is if the long term trend was down.

  • By Jim, October 5, 2009 @ 6:18 pm

    Over the long term you should assume that stocks will go up overall. But you should also assume that in shorter term periods that there will be much volatility. Its the shorter term volatility that DCA is going to help you average out.

    Overall though I don’t know how much the debate really matters. Most people put their money into the market with periodic contributions usually as its deducted from their regular paychecks, so thats basically DCA by coincidence. If you put $5k a year into a Roth IRA and do it on Jan 1 then thats DCA on an annual basis.

  • By Kelly, October 5, 2009 @ 11:00 pm

    As Rick and Jim pointed out, the stock market is incredibly volatile over the short term, compared to the long term return trends. The S&P 500 rose by 1.5% today; last Thursday, it fell by 2.6%. Compared to a long-term average swing of +10%, the day to day variability is huge.

    Dollar cost averaging is a way to lessen this variability, ensuring that you get something closer to an average price. It decreases your chance of getting screwed, by buying a big chunk of stock on a day when the price goes way up. It also decreases your chance of a windfall, by buying a chunk on a day when prices go down. The volatility of the market is large enough that a big chunk of annual returns could be wiped out by simple bad luck.

    Gambling can be fun, for those of a certain mindset. When you buy stocks, you have to realize that, over the short term, it’s just a random walk, and you are gambling some portion of your money, just as surely as if you were at the craps table. Dollar cost averaging lessens the amount that you are gambling.

  • By Robert, October 6, 2009 @ 12:19 am

    This is a nice examination of a major pillar of PF. I use Prof. Michael Edelson’s method of Value Averaging. Basically, you set a target, growth rate, and some other parameters. His handy spreadsheet plots a “value path” that you follow over period of time that you’re shooting for that target. Details can be found on wikipedia or in his book– google “value averaging”.

    I would be interested to hear Frank’s thoughts on this approach.

  • By Wm Tanksley, October 6, 2009 @ 12:35 am

    DCA can’t be measured by talking about the ridiculously rare situation of receiving 12K of new money. Not only is that rare, but also DCA doesn’t even apply; you should be deciding where to put it, not pretending that you’re putting it slowly or quickly.

    In the normal situation, where new money to save comes from your regular income, your choice is to DCA, or to time the market. My coworker is a market timer; I’m a DCAer.

    The choice isn’t between investing “a lump sum now” or “small amounts every month”; the choice is between “investing small amounts into your (hopefully balanced) portfolio every month”, or “letting your portfolio’s cash balance fluctuate wildly”.

    -Wm

  • By aliotsy, October 6, 2009 @ 2:59 am

    Yes, what Wm Tanksley said.

    A fairer comparison may look like this: on January 1st, 2009, Peter and Paul both have zero dollars. Every month, they earn $1000 to invest. At the end of each month, Peter invests his $1000 in the stock market, while Paul sets his $1000 aside in savings. On December 30th, Paul invests all $12,000 in the stock market.

    On January 1st, 2039, is Paul really in a much better place than Peter? Does the analysis support that?

  • By KC, October 6, 2009 @ 8:51 am

    I sort of dollar cost average – I generally invest a little all the time. But when I purchase a stock I analyze it and see if its at a price I like. If it is relatively low (like a 3 year low) and my research turns up no red flags, I’m putting money in. If it’s comparatively high in price and it looks like its a spike, I’ll wait for the price to become more of a value. But I’m always investing and rarely selling – I’m just not investing in the same stocks all the time, I’m only investing when the stocks I watch are of value.

    When I invest in my index funds I treat the index like a stock. Has it had a run of late? Is the market undervalued? This is actually easier to gage cause you almost have a “gut” feeling. For instance I didn’t invest a dime from July 08 until March 09. Why? I was moving and needed significant cash and the market was in the toilet and everyone knew it. I started investing again in March 09 cause I’d completed my move, had extra cash from the sell of my home and I knew GE at $6.50 and VFINX (Vanguard 500 index) around $62/share was the deal of a lifetime. From July until the present I’m still investing but not at the “all in” approach I was a few months earlier. The market has had a good run and its going to be a bit volatile for a while. But I have complete confidence in the next 10 years and I know my investing is going to pay off.

  • By Jim, October 6, 2009 @ 9:22 am

    Frank, you’re of course correct that the expected return is higher if you invest the lump sum upfront. However, the standard deviation of the expected return will also be much higher investing the lump sum.

    For non-statisticians: assuming a rising market, people do better on average investing a lump sum as soon as possible. However, that lump sum investing also makes it more likely that you’ll have an extreme return, losing or making a lot of money. DCA greatly increases the odds of a “middle of the road” outcome.

    Especially when people have a lump sum to invest, they’re often concerned more about not losing big than about winning big.

  • By Retirement Savior, October 6, 2009 @ 9:22 am

    Frank isn’t talking about investing 401k’s; he is only talking about a lump sum or windfall.

    @aliotsy You are suggesting the same point as Frank. If expected market returns are positive, you are better off investing as soon as possible. No matter the volatility, the averages and expectancy is on your side.

    @Kelly You are right that annual volatility is much less than the sum of the absolute daily moves. But putting money in at one time isn’t gambling. If you expect the market to return 10% per year, then the odds are on your side the sooner you invest. The actual gambling is waiting to invest the money over time, as you are betting that the market will have a sideways to negative return.

  • By Dove, October 6, 2009 @ 12:24 pm

    “Put another way, the dollar cost averaging choice is between buying at today’s price and buying at the average price over the next year.”

    This is statistically incorrect. When you dollar cost average, you do not get the average price. You do slightly better. This is easy to see with a simple example.

    Suppose you invest $100 twice, at $10 a share and $20 a share. Dollar cost averaging, you buy 10 shares the first time and 5 shares the second time. Having thus bought 15 shares for $200, your average price is $13.33. This is better than the market’s average price of $15.

    In general, with a lump sum, I would rather dollar cost average–even over just a few installments and months–than buy in at once. Assuming I cannot guess the behavior of the market and ignoring transaction costs, buying it at once gets me the mean price as an expected value, while dollar cost averaging weights the price toward the low end. So over a short enough period that the market is not expected to rise too much (say, less then a decade), the expected value is better, and the risk profile is a *lot* better.

    It is not as simple as dollar cost averaging being strictly superior. Frank is right that you lose out on gains if you average over too long a window. Choosing the right window and frequency involves comparing the size of the volatility to the size of the general upwards trend, and there are transaction costs and risk tolerence to consider. And, of course, if you feel like you know what the market’s going to do in a macro sense, that plays into it as well. There are more statistics to do than appropriate for a blog comment, but suffice it to say that dollar cost averaging is sometimes the rationally correct strategy.

  • By Wm Tanksley, October 6, 2009 @ 7:09 pm

    “Frank isn’t talking about investing 401k’s; he is only talking about a lump sum or windfall.”

    Absolutely not. Frank specifically said that the entire question of dollar cost averaging boils down to a hypothetical situation of a windfall, yes; but he isn’t talking about a windfall, but about the entire question of DCA. He’s using a windfall to illustrate his point, but his point is vastly broader.

    And I don’t see how it’s even possibly right, much less useful.

    Windfalls are rare, and when you get one, would this article help you know how to invest it at all? No, it gives some broad generalities about how DCA reduces the chance of feeling bad (?). And in the common case, you don’t have a windfall, and this illustration doesn’t hold for you at all: you’re not deciding whether to invest large now or small over a span, but rather you’re deciding whether to invest small now, or save the cash to invest big later.

    I made a mistake in my last comment. When deciding how to invest a windfall, you’ve got to consider not only your portfolio’s balance, you also have to consider whether you’re making too big of an uncertain bet. I’ve never figured out how to use the Kelly criterion in real life, but the basic idea makes sense: you make a large bet on one thing only when you know something about that thing that the other bettors don’t.

    -Wm

  • By TJR, October 7, 2009 @ 4:45 pm

    Supporting Dove’s point there is a theorem from probability theory: Jensen’s inequality. Basically, since you buying an amount of stock that is proportional to one divided by the price, you can apply Jensen’s inequality to the “one divided by” function and get the conclusion your expected returns with DCA will never be less than with a lump sum purchase. Of course, this assumes you have the choice between investing in the risky portfolio (whose time-dependent price is described by the probability space in the inequality) or parking your money in a risk-free, liquid asset that performs exactly as the expectation value of the risky choice.
    Since people generally do not have access to such a risk-free investment, the theorem might be regarded as a rational justification for the risk premium observed in the real world.

    Unfortunately, I have never been able to compute exactly how much of the risk premium is due to compensating for DCA because I don’t know any realistic but tractible mathematical models for stock prices. The gain from DCA is zero for a Wiener process (and only that process) as a model for stock prices, but Gaussian distributions imply possible negative stock prices and other problems, so we can safely say DCA gives you a non-zero gain for whatever stochastic process governs stock prices.

  • By TJR, October 7, 2009 @ 4:48 pm

    Erratum: The gain is zero for any martingale that can be constructed as a time-transformed Wiener process, all of which are still Gaussian, and are therefore ruled out by “no negative stock prices”.

  • By Frank Curmudgeon, October 8, 2009 @ 1:43 pm

    Where to start? Commenters should feel free to use as advanced math and financial theory as they like. I can keep up, although most of your fellow readers might roll their eyes.

    Robert: I’m not familiar with Prof. Edelson’s work, but based on your short description, it sounds like it assumes some things about stock price movements I’m not ready to assume, in particular a strong mean-reversion in prices.

    Wm Tanksley: Having a windfall is, sadly, rarer than we’d all like. But having a sum of money to invest and wondering if you should do it all at once or not is not all that uncommon. For example, a person who got chased out of the market at the start of the year and is now feeling optimistic enough to get back in. In any case, I don’t think it makes sense to try and evaluate DCA in another context. What, then, is the alternative to DCA?

    Aliotsy: Yes, I think we can all agree that saving up your money and investing all at once at the end of the period is inferior to investing in stages over the period.

    Jim: If by standard deviation you mean volatilty, then yes, DCA has a lower standard deviation, but this is only becuase your money is (on average) half in cash. You will have the same volatility if you compare DCA from January to December with lump-sum investing in June. The monthly investments don’t really diversify in the standard sense. Consider that by April, your January, February, and March investments are perfectly correlated. All that’s going on is that you are shifting the weightings of your portfolio from cash to stocks over the course of the year.

    Dave: You are correct, DCA doesn’t give you the average per-share price, for the reasons you explain. It will, however, give you the average return over the period from each investment date to the end, i.e. the average of January to December, February to December, March to December, etc.. I was sloppy with my terms, sorry. But I hope we can agree that it is returns we really care about.

    TJR: Obviously (?) if stock price movements are a random walk or a true martingale, then how you do or don’t invest won’t have an impact on your expected returns, since cash and the market have the same expected return (either 0% or the risk-free rate) so any combination of them on any timetable leaves you just where you started. Just as obviously, if the stock return is preferable to you, either because of a higher expected return or because of superior risk profile within your portfolio, then you are better off investing all at once.

    French explains this well in the video. There is only one optimal portfolio. If your current portfolio is sub-optimal because you are under invested in the stock market by $12,000, then you should fix that by buying $12,000 in stocks today. Dribbling it out over 12 months just means you will be in 11 other sub-optimal portfolios before you get to the optimal one after a year.

    And I think lognormal models stock price movements well enough for all intents and purposes.

  • By Wm Tanksley, October 16, 2009 @ 2:50 pm

    Frank, I agree with your last post, especially the paragraph that begins “French explains this well in the video.” That explains precisely why even asking about DCA in that context is a confusion of categories — it’s not that it’s always bad to reinvest slowly, but rather that the decision process doesn’t involve deciding whether DCA is good or bad.

    “…But having a sum of money to invest and wondering if you should do it all at once or not is not all that uncommon. For example, a person who got chased out of the market at the start of the year and is now feeling optimistic enough to get back in.”

    This person changed their portfolio balance quickly, and now is wondering how quickly they should rebalance. That’s not a DCA question; it’s a rebalancing question, and would be answered by comparing how confident the person is with the transaction cost per dribble. This is a decision the person made before without talking about DCA (and they decided to sell quickly that time); there’s no reason to bring DCA into it now.

    “In any case, I don’t think it makes sense to try and evaluate DCA in another context. What, then, is the alternative to DCA?”

    The alternative shows up in the context everyone who attempts to save encounters every month: when you see your paycheck deposit and decide when and whether to buy more of your chosen portfolio.

    You can either invest immediately, and talk about how dollar cost averaging is going to affect your returns, or you can save the cash and talk about how timing the market’s going to affect your returns.

    I know which one I’m betting on.

    And again, this happens every month for most people who invest outside of a 401k. Your windfall (or massive rebalancing) happens only once in a while — and as you’ve explained, talking about DCA there doesn’t even make any sense.

    -Wm

  • By Evolution Of Wealth, October 18, 2009 @ 10:52 pm

    The comments just show the confusion surrounding DCA. It’s because every financial institution shoves it down people’s throats in their search for more money. As you point out it doesn’t apply to most situations. People mistake a periodic investment plan for DCA. They are not the same. Investing periodically because you don’t have a lump sum is NOT DCA. I think you do a great job of breaking it down simply in the 5th paragraph. Maybe you should have started and ended there?

  • By Thomas Farley, October 22, 2009 @ 8:59 pm

    “Dribbling it out over the next year will only reduce your returns if the market has a tendency to go up, and if it didn’t have a tendency to go up, why would you want to invest in it at all?”

    Whether you want to invest it all is entirely dependent on the relationship between volatility and return (as Rick Francis points out above). If volatility is high (20% annual standard deviation) and expected returns are low (3-5% over risk-free alternative), then it is very likely that “dribbling it out” will be the winning strategy.
    One way to think about it is that the “dribbler” is effectively granting high volatility options on a market that doesn’t move that much in the longer term.

  • By Patrick, February 10, 2010 @ 5:02 pm

    I’m disappointed in this post. Dollar-cost averaging is not about maximizing expected return. That’s a red herring, pure and simple.

    Splitting up one deposit into a dozen or so smaller deposits reduces risk significantly, and the expected cost is about half of one year’s worth of returns. To some rational people, that’s a fair price.

  • By Patrick, February 10, 2010 @ 5:35 pm

    @Frank: “You will have the same volatility if you compare DCA from January to December with lump-sum investing in June.” Interesting. I hadn’t thought of that.

    However, volatility isn’t the whole story regarding risk. Consider: in any given year, there is some probability of a market crash. If you invest all your money in June, you’re hosed if the crash happens in July. Not so with DCA.

    Naturally DCA would also miss out on a huge market surge in July. That’s part of the price you pay for avoiding the crash scenario.

    Also, even with DCA you’re hosed if the crash happens in December.

  • By new2thegame, March 9, 2011 @ 9:00 pm

    For funsies, I like to DCA throughout the year, with part of the monthly investment going to a money market fund. Then, whenever I feel the prices are greatly lower then average, I’ll exchange the whole lump sum that built up in that money market for the cheap stock price.

    Not sure if this works, but it makes me feel like I am DCA and buying extra stock when the price is low. I suppose it might be better to use that excess money market fund to buy the stock immediately and get it invested, but this makes me happy and what else really matters?

  • By Shree, October 9, 2012 @ 6:55 pm

    Not sure if I understand the Math.

    Scenario #1: No DCA. I invest $100 in Jan and market is at 100. Market drops to 50 in July and is backup to 100 in Dec. Therefore my return is $0 dollars

    Scenario #2: DCA. I invest $50 and market is at 100 in Jan. Market drops to 50 in July and I invest the other 50. Market is back to 100 in December. My return is 50%.

    How is DCA bad?

Other Links to this Post

  1. » Weekly Roundup – National Save for Retirement Week Edition — October 18, 2009 @ 9:28 pm

RSS feed for comments on this post. TrackBack URI

Leave a comment

WordPress Themes