Lump sum vs. Dollar cost averaging

[DcifrThs]

so this came up in another thread but it is interesting and i'd like to get some informed opinions on which literature is best on the subject.

the overall market view i think seems to lean towards DCA whereas the academic literature leads to Lump sum.

are there any studies that show the distribution of the differences between these two so one can see (taking EV aside) how it looks on a frequency basis.

EV wise, the argument for DCA seems to be that if you invest all your money at market highs, you'll lose such a big chunck that you ma ynot make it up w/ the risk premia for risky assets over time.

conversely, Lump sum seems to imply that the win, win, win, win*, LOSE BIG for a while, win, win win type payoffs of the many markets favors investing ASAP and not worry about the smaller probability of investing precisely at win*. an argument can easily be seen for this by looking at the 1990s where the mkt continually crushed highs where many thought the mkt would be at the top. lump sum at any point would have pretty much crushed DCA.

i wasn't in the industry back then but was DCA in favor durin gthat time period?

conversely, 1987 & 2000 show examples of when DCA would be better if you invested near or at the highs since it would take some time to get back to the level at which you came in.

it seems though that the historical arguments that academics discuss may not apply going forward. what kind of logic should we think about that might prove some of the academic studies right/wrong one way or the other (DCA vs. Lump sum).

at my former job, we were huge lump sumers. no DCA at all and none of the (massssive) clients cared about slamming a whole bunch at us at any one point in time (i.e. i've never heard DCA brought up once). i trust my former employer but i'd like to think about it logically for myself as well as see some (top / best) studies.

thanks for the help & discussion.

Barron

[DesertCat]

DCA is useful for most individual investors, who add to their investments from regular savings, so they get the benefits of DCA from their normal savings patterns. It's also useful if you don't know anything about the markets, using DCA prevents you from getting in at the worst times, but usually costs you some EV, like an insurance policy.

If you are professional investor, it's your job to buy things when their prices are compelling, i.e. you understand how to figure out when a stock is cheap. DCA has no place for you.

[DcifrThs]

[ QUOTE ]
DCA is useful for most individual investors, who add to their investments from regular savings, so they get the benefits of DCA from their normal savings patterns. It's also useful if you don't know anything about the markets, using DCA prevents you from getting in at the worst times, but usually costs you some EV, like an insurance policy.

If you are professional investor, it's your job to buy things when their prices are compelling, i.e. you understand how to figure out when a stock is cheap. DCA has no place for you.

[/ QUOTE ]

but that gets back to making active management decisions with passive allocations.

i'm assuming we are talking about nothing but beta seeking investments here. since both pension funds & individuals have the same goals in terms of return maximization (roughly speaking...i.e. save liability immunization), shouldn't the same be true for both long term investors.

i.e. a 20 y/o who knows nothing about investing and a pension fund should both use the same entry method, right?

what about the following question:

what is the difference for a 20 y/o investing at win* vs. right after a loss about 40 or so years later? how big is the difference?

if it isn't very large, then shouldn't all investors w/ that time horizon use lump sum vs. DCA. (i.e. the benefit of the insurance policy is small)?

thanks,
barron

[RicoTubbs]

Given a pot of money right now, lump sum investing makes more sense than DCA. With dollar cost averaging, you are making a market-timing bet that the spread-out purchases over the next X periods will be at better prices than the current price. In other words, you're betting that prices will go down over your horizon, which runs against the obvious upward direction of the market.

The only reason DCA makes sense is that it coincides with the fact that people tend to have savings patterns that build up cash at regular intervals (e.g., with their periodic paychecks), rather than having big piles of money sitting around, or getting enormous but sporadic windfalls.

Certainly there have been times where it would have been better to DCA (i.e., the market was way overvalued), but if you are smart/skilled enough to identify those times, you are smart enough to avoid investing altogether instead of engaging in mechanical purchases at regular intervals.

[DcifrThs]

[ QUOTE ]
The only reason DCA makes sense is that it coincides with the fact that people tend to have savings patterns that build up cash at regular intervals (e.g., with their periodic paychecks), rather than having big piles of money sitting around, or getting enormous but sporadic windfalls.

[/ QUOTE ]

in this case, DCA isn't a choice, it is the only way these investments get made w/ sporadic paychecks...otherwise it is a delayed lump sum (i.e. hold your paychecks to invest when prices fall).

Barron

[RicoTubbs]

[ QUOTE ]
in this case, DCA isn't a choice, it is the only way these investments get made w/ sporadic paychecks...otherwise it is a delayed lump sum (i.e. hold your paychecks to invest when prices fall).

[/ QUOTE ]

That's right. What I should have said is that DCA seems like a convenient after-the-fact pseudo-benefit that people ascribe to the observation that most people save and invest periodicially. If it were common practice for people to be paid once a year rather than monthly or more frequently, I don't think anyone would have come up with a strategy of "wait a minute, instead of investing all your available cash, why don't you hold onto it and slowly invest it 1/12 per month".

[Scorpion Man]

I haven't read the whole thread and I am not an academic but noone has talked about the reason I have used DCA at times when putting significant sums to work -- it's emotional. I will just be so pissed if I pick a bad time that I prefer the averaging.

[Nomad84]

It seems obvious to me that investing at win* just before an x% fall will result in x% less money in 40 years than investing just after the fall. To simplify things, at any given time, we know that there is probability p of an x% fall during a particular time period. Obviously, the question is whether or not the lost expected returns associated with DCA exceeds p*x. More importantly, since the scholars tell us that they do, and therefore that lump sum has the higher EV, what we really want to know is:

(1) How much difference is there in EV after n years?
(2) What is the standard deviation of the results of each strategy after n years?

Would it be reasonable to pick an integer n and a time frame for the DCA and plot the (backtested?) results of the two strategies and analyze the differences statistically? I would probably prefer to see a string of results for a number of n-year periods using as much data as is available, then analyze that return stream like any other investment to determine the risk/reward tradeoff between DCA and lump sum. I would expect DCA to have lower standard deviation of the returns, but lump sum to have the higher EV. I suppose it should be relatively simple at that point to calculate the Sharpe ratio of each strategy (lump sum vs. DCA) and determine which has the better risk-adjusted returns. Ideally, this could be layered on top of data supporting your particular investment strategy, but it should be an interesting exercise even if just done on the S&P500 or something along those lines. It would also be interesting to perform this analysis for a variety of choices of n to see if the time frame makes a difference. My hunch is that once the DCA money is fully invested, the time frame after that point may not make a difference. Obviously the time over which the money is being invested should matter.

I haven't read any of the studies that make this comparison, but I'm under the impression that the studies supporting lump sum investing point to the higher EV. I'm not sure I've heard them mention comparisons of risk adjusted returns of lump sum vs. DCA.

This should be an interesting discussion.

[RikaKazak]

This is how I do it:

If I like a particular business or sector...but not following it daily etc. (just generally like it and generally like the price) I DCA.

If I'm following daily, AND I have a "good reason" to suspect "win, win, win" type pattern...then I lump sum.

Basically default is DCA, with "lump summing" if I have a specific reason.