Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
User avatar
tadamsmar
Posts: 9014
Joined: Mon May 07, 2007 12:33 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by tadamsmar » Fri Jun 26, 2020 7:50 am

HornedToad wrote:
Wed Jun 24, 2020 5:52 pm
Regret theory says people regret and "feel" an actual loss much more than missing out on a potential gain. No one ever panic sells and leaves the market for 5 years for not making as much profit, but they sure do if they invest everything they own in the stock market on one really bad day and then look a week, month, 3 months later and say they are cursed, can't handle it and pull everything out.
Larry fears loss if he lump sums in new money.

Curly has the same amount already in and faces the same potential loss, but somehow the magic of old money make the fear go away.

Kind of reminds me of Sartre saying "Man is condemned to be free" because he thinks people are in denial about their freedom.

Bogleheads are condemned to be day traders. That is, they are condemned to make a daily decision to stay invested and live with the consequences. The consequences has exactly the same or greater than the consequences of investing a lump sum immediately.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Fri Jun 26, 2020 8:25 am

EnjoyIt wrote:
Wed Jun 24, 2020 6:25 pm
nigel_ht wrote:
Wed Jun 24, 2020 2:43 pm
EnjoyIt wrote:
Wed Jun 24, 2020 4:42 am
Beehave wrote:
Tue Jun 23, 2020 9:43 am
Uncorrelated wrote:
Tue Jun 23, 2020 8:02 am


I missed the part where you explain that it is rational to have a risk tolerance that support DCA, but not LSI. Even though it is possible to choose a LSI with the exact same risk as DCA, but higher expected return.

Don't try to pass off DCA as rational. Humans make mistakes. Embrace your inner irrationality and proclaim proudly: I choose a suboptimal approach just because I feel like it.

In response, 2 things.

(1) My assertion is that "never DCA" is wrong. That does not, as you seem to think, imply that my position must be that "never LSI" is therefore correct.
You make a simple logical error here. The negation of No A are B is Some A are B. It is not All A are B. My position is that DCA can be correct in some circumstances and LSI in others.

(2) DCA can be perfectly rational. You try to pass off financial planning as the rational kept void of any emotional or irrational considerations. My position is that financial planning is comprised of rational assessment of rational and irrational components. This is clear in setting asset allocation targets. Two people in with exactly the same financial conditions, beliefs, and life conditions who differ only in risk tolerance can rationally come up with different asset allocation strategies - - for example, one with a 50/50 allocation target and the other with 60/40.

Moreover, suppose you were advising a friend who just came into a windfall. The friend wants to DCA to some fixed allocation, say 60/40. But they have an irrational fear of LSI and want to DCA instead. You spend hours convincing them rationally that LSI is a more rational outcome relative to DCA, but their gut strongly objects. The outcome will either be that they choose (what you believe) is the rational choice and LSI straight to 60/40 but have emotional discomfort, or they bow down to their irrational fear and DCA in thereby feeling conflicted for their irrational, self-harming action.

Suppose I were advising that same person. I would explain how time in the market is important and that LSI is construed by many to be superior to the DCA they plan to do. But I understand how uncomfortable they are with LSI. So I would propose a trade-off. DCA the windfall as planned. The odds say by doing this you may miss out on some performance as a trade-off for the peace of mind you will have. In return, raise your long-term target allocation from 60/40 to 61/39, or 62/38 to make up for it. That's much less of an emotional stretch, and in the long run odds are you will have better results because of more money in the market. They do the DCA with raised target allocation and are proud of using their rationality to come up with a plan that is financially optimal (relative to your financial plan), more likely to be followed, and emotionally satisfying rather than disquieting.

You are attempting to stuff something down someone's craw and they experience discomfort either way they choose. I am giving palatable advice that makes the person feel good about themselves and their choice and gets (by the value of time and amount of money in the market way of thinking) and is easily computed and tuned to get equal or better results.

If someone takes account of their irrationality when planning they can do a better job than if they try to deny this very real part of human nature.
So which one of us is "rational?" The one who accounts for the reality of what it is to be human and works with it or the one who fights human nature?
If I was advising someone who is uncomfortable lump summing in to 60/40 now that it is very likely 60/40 is the wrong asset allocation for them. There is nothing different between 60/40 now and 60/40 six months from now.
Given the discussions on whether 60/40 is dead I wonder how comfortable anyone should be with any past AA “rules of thumb”.
What does past rule of thumb AA have to do with anything. AA is a personal thing and not a rule of thumb thing.
Also "dead" what the hell does that even mean? Plenty of investors are 60/40. In retirement I will be 60/40. I personally know people who are 60/40. I guess they did not get that memo.

The point is that DCA into an AA that the investor was uncomfortable going into is a plan for failure.
Because in times of uncertainty the requirement for certainty doesn't work too well.

It is also very weird for some folks to complain that DCA is irrational and based on emotionality because it has a sub-optimal EV and then insist that selection of personal AA based on emotionality is the rational first step to BH investing.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Fri Jun 26, 2020 8:32 am

tadamsmar wrote:
Fri Jun 26, 2020 7:50 am
HornedToad wrote:
Wed Jun 24, 2020 5:52 pm
Regret theory says people regret and "feel" an actual loss much more than missing out on a potential gain. No one ever panic sells and leaves the market for 5 years for not making as much profit, but they sure do if they invest everything they own in the stock market on one really bad day and then look a week, month, 3 months later and say they are cursed, can't handle it and pull everything out.
Larry fears loss if he lump sums in new money.

Curly has the same amount already in and faces the same potential loss, but somehow the magic of old money make the fear go away.
Larry and Curly are often two different people with two different scenarios. Something that you continue to ignore.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Fri Jun 26, 2020 9:02 am

nigel_ht wrote:
Fri Jun 26, 2020 8:25 am

Because in times of uncertainty the requirement for certainty doesn't work too well.

It is also very weird for some folks to complain that DCA is irrational and based on emotionality because it has a sub-optimal EV and then insist that selection of personal AA based on emotionality is the rational first step to BH investing.
Personal AA isn't emotional. It's based on your personal risk tolerance and subjective discount rate, which are based on your specific goals.

See https://en.wikipedia.org/wiki/Merton%27 ... io_problem for details.

Every set of goals has a different asset allocation, therefore everybody can rationally have a different asset allocation. But within this framework (more generally, for any well-behaved utility function that is risk-averse), DCA is never the optimal solution.

Sure you can say "I prefer to take more risk for the same expected return", and then declare DCA rational under this utility function. But under the same utility function, it would be rational (i.e. equivalent to DCA) to choose lump sum investment and flush some amount of money down the toilet. So, uhm, yeah...

EnjoyIt
Posts: 4183
Joined: Sun Dec 29, 2013 8:06 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by EnjoyIt » Fri Jun 26, 2020 9:35 am

nigel_ht wrote:
Fri Jun 26, 2020 8:25 am
EnjoyIt wrote:
Wed Jun 24, 2020 6:25 pm
nigel_ht wrote:
Wed Jun 24, 2020 2:43 pm
EnjoyIt wrote:
Wed Jun 24, 2020 4:42 am
Beehave wrote:
Tue Jun 23, 2020 9:43 am



In response, 2 things.

(1) My assertion is that "never DCA" is wrong. That does not, as you seem to think, imply that my position must be that "never LSI" is therefore correct.
You make a simple logical error here. The negation of No A are B is Some A are B. It is not All A are B. My position is that DCA can be correct in some circumstances and LSI in others.

(2) DCA can be perfectly rational. You try to pass off financial planning as the rational kept void of any emotional or irrational considerations. My position is that financial planning is comprised of rational assessment of rational and irrational components. This is clear in setting asset allocation targets. Two people in with exactly the same financial conditions, beliefs, and life conditions who differ only in risk tolerance can rationally come up with different asset allocation strategies - - for example, one with a 50/50 allocation target and the other with 60/40.

Moreover, suppose you were advising a friend who just came into a windfall. The friend wants to DCA to some fixed allocation, say 60/40. But they have an irrational fear of LSI and want to DCA instead. You spend hours convincing them rationally that LSI is a more rational outcome relative to DCA, but their gut strongly objects. The outcome will either be that they choose (what you believe) is the rational choice and LSI straight to 60/40 but have emotional discomfort, or they bow down to their irrational fear and DCA in thereby feeling conflicted for their irrational, self-harming action.

Suppose I were advising that same person. I would explain how time in the market is important and that LSI is construed by many to be superior to the DCA they plan to do. But I understand how uncomfortable they are with LSI. So I would propose a trade-off. DCA the windfall as planned. The odds say by doing this you may miss out on some performance as a trade-off for the peace of mind you will have. In return, raise your long-term target allocation from 60/40 to 61/39, or 62/38 to make up for it. That's much less of an emotional stretch, and in the long run odds are you will have better results because of more money in the market. They do the DCA with raised target allocation and are proud of using their rationality to come up with a plan that is financially optimal (relative to your financial plan), more likely to be followed, and emotionally satisfying rather than disquieting.

You are attempting to stuff something down someone's craw and they experience discomfort either way they choose. I am giving palatable advice that makes the person feel good about themselves and their choice and gets (by the value of time and amount of money in the market way of thinking) and is easily computed and tuned to get equal or better results.

If someone takes account of their irrationality when planning they can do a better job than if they try to deny this very real part of human nature.
So which one of us is "rational?" The one who accounts for the reality of what it is to be human and works with it or the one who fights human nature?
If I was advising someone who is uncomfortable lump summing in to 60/40 now that it is very likely 60/40 is the wrong asset allocation for them. There is nothing different between 60/40 now and 60/40 six months from now.
Given the discussions on whether 60/40 is dead I wonder how comfortable anyone should be with any past AA “rules of thumb”.
What does past rule of thumb AA have to do with anything. AA is a personal thing and not a rule of thumb thing.
Also "dead" what the hell does that even mean? Plenty of investors are 60/40. In retirement I will be 60/40. I personally know people who are 60/40. I guess they did not get that memo.

The point is that DCA into an AA that the investor was uncomfortable going into is a plan for failure.
Because in times of uncertainty the requirement for certainty doesn't work too well.

It is also very weird for some folks to complain that DCA is irrational and based on emotionality because it has a sub-optimal EV and then insist that selection of personal AA based on emotionality is the rational first step to BH investing.
I never said DCA is irrational. I agree that DCA can be a very good option for someone as they try and understand their risk tolerance.

My position is that to blindly DCA into an originally undesirable AA is not wise.
A time to EVALUATE your jitters. | https://www.bogleheads.org/forum/viewtopic.php?f=10&t=79939&start=400#p5275418

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Fri Jun 26, 2020 10:37 am

Uncorrelated wrote:
Fri Jun 26, 2020 9:02 am
nigel_ht wrote:
Fri Jun 26, 2020 8:25 am

Because in times of uncertainty the requirement for certainty doesn't work too well.

It is also very weird for some folks to complain that DCA is irrational and based on emotionality because it has a sub-optimal EV and then insist that selection of personal AA based on emotionality is the rational first step to BH investing.
Personal AA isn't emotional. It's based on your personal risk tolerance and subjective discount rate, which are based on your specific goals.
Personal risk tolerance isn't emotional? Mkay.
Every set of goals has a different asset allocation, therefore everybody can rationally have a different asset allocation. But within this framework (more generally, for any well-behaved utility function that is risk-averse), DCA is never the optimal solution.

Sure you can say "I prefer to take more risk for the same expected return", and then declare DCA rational under this utility function. But under the same utility function, it would be rational (i.e. equivalent to DCA) to choose lump sum investment and flush some amount of money down the toilet. So, uhm, yeah...
If your equivalent risk AA is different than your final target AA then it requires two steps to your target AA...which is just a weird form of DCA and may have tax implications when you step to your final AA.

In any case, I'm curious how the equivalent AA actually behaves in comparison to DCA in left tail scenarios. I guess that depends on when the event occurs during the DCA.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Fri Jun 26, 2020 11:58 am

nigel_ht wrote:
Fri Jun 26, 2020 10:37 am
Uncorrelated wrote:
Fri Jun 26, 2020 9:02 am
nigel_ht wrote:
Fri Jun 26, 2020 8:25 am

Because in times of uncertainty the requirement for certainty doesn't work too well.

It is also very weird for some folks to complain that DCA is irrational and based on emotionality because it has a sub-optimal EV and then insist that selection of personal AA based on emotionality is the rational first step to BH investing.
Personal AA isn't emotional. It's based on your personal risk tolerance and subjective discount rate, which are based on your specific goals.
Personal risk tolerance isn't emotional? Mkay.
personal risk tolerance has a mathematical definition.
Every set of goals has a different asset allocation, therefore everybody can rationally have a different asset allocation. But within this framework (more generally, for any well-behaved utility function that is risk-averse), DCA is never the optimal solution.

Sure you can say "I prefer to take more risk for the same expected return", and then declare DCA rational under this utility function. But under the same utility function, it would be rational (i.e. equivalent to DCA) to choose lump sum investment and flush some amount of money down the toilet. So, uhm, yeah...
If your equivalent risk AA is different than your final target AA then it requires two steps to your target AA...which is just a weird form of DCA and may have tax implications when you step to your final AA.

In any case, I'm curious how the equivalent AA actually behaves in comparison to DCA in left tail scenarios. I guess that depends on when the event occurs during the DCA.
The equivalent risk AA is a way to construct an asset allocation that is definitely better (i.e. has higher expected utility) than any DCA AA. I used that to show that DCA is never optimal and that the optimal solution must be in the form of LSI. I never stated that the equivalent risk AA is necessarily the same as the optimal allocation.

"tail risk" is just a way to make things more complicated than they actually are. But supposing you're actually interested in an answer, see 12.4.1 on page 120 of an introduction to computational finance without agonizing pain, which proves that DCA is worse than LSI for one class of non-normal distrubutions. There is at least one other proof that states that DCA is suboptimal for any distribution of outcomes.

Again all of this assumes you have a well-defined utility function. If you don't have a well-defined utility function, you can argue for anything. Including literally flushing money down the toilet.

tbone555
Posts: 144
Joined: Thu Apr 13, 2017 1:28 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by tbone555 » Fri Jun 26, 2020 12:05 pm

https://www.google.com/url?sa=t&source= ... doERwmxOaJ

This is a very recent paper by Ben Felix that is worth reading. His final paragraph is:

"Given the data supporting lump sum investing we believe that there is a strong statistical argument to avoid dollar cost averaging unless it is absolutely necessary from a psychological perspective, and if that is the case, we believe that the long-term asset allocation may need to be revised toward a more conservative portfolio."

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Fri Jun 26, 2020 12:09 pm

Uncorrelated wrote:
Fri Jun 26, 2020 11:58 am
nigel_ht wrote:
Fri Jun 26, 2020 10:37 am
Uncorrelated wrote:
Fri Jun 26, 2020 9:02 am
nigel_ht wrote:
Fri Jun 26, 2020 8:25 am

Because in times of uncertainty the requirement for certainty doesn't work too well.

It is also very weird for some folks to complain that DCA is irrational and based on emotionality because it has a sub-optimal EV and then insist that selection of personal AA based on emotionality is the rational first step to BH investing.
Personal AA isn't emotional. It's based on your personal risk tolerance and subjective discount rate, which are based on your specific goals.
Personal risk tolerance isn't emotional? Mkay.
personal risk tolerance has a mathematical definition.
Every set of goals has a different asset allocation, therefore everybody can rationally have a different asset allocation. But within this framework (more generally, for any well-behaved utility function that is risk-averse), DCA is never the optimal solution.

Sure you can say "I prefer to take more risk for the same expected return", and then declare DCA rational under this utility function. But under the same utility function, it would be rational (i.e. equivalent to DCA) to choose lump sum investment and flush some amount of money down the toilet. So, uhm, yeah...
If your equivalent risk AA is different than your final target AA then it requires two steps to your target AA...which is just a weird form of DCA and may have tax implications when you step to your final AA.

In any case, I'm curious how the equivalent AA actually behaves in comparison to DCA in left tail scenarios. I guess that depends on when the event occurs during the DCA.
The equivalent risk AA is a way to construct an asset allocation that is definitely better (i.e. has higher expected utility) than any DCA AA. I used that to show that DCA is never optimal and that the optimal solution must be in the form of LSI. I never stated that the equivalent risk AA is necessarily the same as the optimal allocation.

"tail risk" is just a way to make things more complicated than they actually are. But supposing you're actually interested in an answer, see 12.4.1 on page 120 of an introduction to computational finance without agonizing pain, which proves that DCA is worse than LSI for one class of non-normal distrubutions. There is at least one other proof that states that DCA is suboptimal for any distribution of outcomes.

Again all of this assumes you have a well-defined utility function. If you don't have a well-defined utility function, you can argue for anything. Including literally flushing money down the toilet.
Your reference says DCA is bad because glide paths are bad and Target Date Funds suck.. Given that this is all the document says in 12.4.2 why don't you successfully defend that position first before demanding we accept the conclusion?
"12.4 Target Date: Ineffectiveness of glide path strategies

It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."
"12.4.2 Dollar cost averaging
Suppose you have a lump sum W0 to invest. You have a target asset mix of pˆ fraction in the risky asset. Is it better to gradually buy stocks? This is often suggested, since this way you are dollar cost averaging.

Let t∗ be the period over which you dollar cost average, and T be the target date of your investment. Then, dollar cost averaging amounts to this deterministic strategy

􏰇 <formula that won't render via a cut and paste> (12.93)

However, from Theorem 1, this strategy cannot be better than buying p∗W0 right away (p∗ defined in equation (12.73))"


We'll wait for that thread to come to consensus on BH before accepting this as definitive...

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Fri Jun 26, 2020 12:24 pm

nigel_ht wrote:
Fri Jun 26, 2020 12:09 pm
Your reference says DCA is bad because glide paths are bad and Target Date Funds suck.. Given that this is all the document says in 12.4.2 why don't you successfully defend that position first before demanding we accept the conclusion?
"12.4 Target Date: Ineffectiveness of glide path strategies

It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."
"12.4.2 Dollar cost averaging
Suppose you have a lump sum W0 to invest. You have a target asset mix of pˆ fraction in the risky asset. Is it better to gradually buy stocks? This is often suggested, since this way you are dollar cost averaging.

Let t∗ be the period over which you dollar cost average, and T be the target date of your investment. Then, dollar cost averaging amounts to this deterministic strategy

􏰇 <formula that won't render via a cut and paste> (12.93)

However, from Theorem 1, this strategy cannot be better than buying p∗W0 right away (p∗ defined in equation (12.73))"


We'll wait for that thread to come to consensus on BH before accepting this as definitive...
Is there anything in particular you think there is no consensus on? AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.

User avatar
vineviz
Posts: 6757
Joined: Tue May 15, 2018 1:55 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by vineviz » Fri Jun 26, 2020 12:52 pm

Uncorrelated wrote:
Fri Jun 26, 2020 12:24 pm
AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.
I don't think this is "known", much less "widely known".

An unanticipated windfall will possibly change the shape of an investor's optimal glide path, but it doesn't necessarily make the use of such a glide path "bad" (which is an entirely subjective evaluation to begin with).
"Far more money has been lost by investors preparing for corrections than has been lost in corrections themselves." ~~ Peter Lynch

HornedToad
Posts: 1029
Joined: Wed May 21, 2008 12:36 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by HornedToad » Fri Jun 26, 2020 1:13 pm

tadamsmar wrote:
Fri Jun 26, 2020 7:50 am
HornedToad wrote:
Wed Jun 24, 2020 5:52 pm
Regret theory says people regret and "feel" an actual loss much more than missing out on a potential gain. No one ever panic sells and leaves the market for 5 years for not making as much profit, but they sure do if they invest everything they own in the stock market on one really bad day and then look a week, month, 3 months later and say they are cursed, can't handle it and pull everything out.
Larry fears loss if he lump sums in new money.

Curly has the same amount already in and faces the same potential loss, but somehow the magic of old money make the fear go away.

Kind of reminds me of Sartre saying "Man is condemned to be free" because he thinks people are in denial about their freedom.

Bogleheads are condemned to be day traders. That is, they are condemned to make a daily decision to stay invested and live with the consequences. The consequences has exactly the same or greater than the consequences of investing a lump sum immediately.

Familiarity breeds contempt.

It is not uncommon for people to try to time when to deposit a lump sum to pick the best day to deposit, low point, etc. and once they deposit they watch it closely for a few days/weeks to see if they made a good decision. But after awhile they are now used to the fluctuations

User avatar
Steve Reading
Posts: 1884
Joined: Fri Nov 16, 2018 10:20 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Steve Reading » Fri Jun 26, 2020 1:19 pm

Uncorrelated wrote:
Fri Jun 26, 2020 12:24 pm
AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.
It's not widely known, and it's not always true either. Here's some examples where it's not true:
1) If humans show changing RRA as they age. In fact, the only time an arbitrary glide path isn't optimal is in the singular case that RRA does not change with age (it's constant). To the extent you know RRA will decrease with age, then a TDF might get closer to the optimal portfolio for someone over a constant allocation, even for a lump sum with no additional savings or expenses.
2) If stocks show mean-reversion (i.e: are not a random walk). Which they historically have. Which is why Siegel argues a higher allocation to stocks makes sense as the time horizon increases. Campbell has a chapter on this in his book by the way. Interesting stuff!

Personally, Siegel's argument is very compelling as I really don't believe a 90% drop in stocks is as likely after an 80% drop as it is after a 200% gain. Just my 2 cents.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Fri Jun 26, 2020 2:15 pm

Uncorrelated wrote:
Fri Jun 26, 2020 12:24 pm
nigel_ht wrote:
Fri Jun 26, 2020 12:09 pm
Your reference says DCA is bad because glide paths are bad and Target Date Funds suck.. Given that this is all the document says in 12.4.2 why don't you successfully defend that position first before demanding we accept the conclusion?
"12.4 Target Date: Ineffectiveness of glide path strategies

It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."
"12.4.2 Dollar cost averaging
Suppose you have a lump sum W0 to invest. You have a target asset mix of pˆ fraction in the risky asset. Is it better to gradually buy stocks? This is often suggested, since this way you are dollar cost averaging.

Let t∗ be the period over which you dollar cost average, and T be the target date of your investment. Then, dollar cost averaging amounts to this deterministic strategy

􏰇 <formula that won't render via a cut and paste> (12.93)

However, from Theorem 1, this strategy cannot be better than buying p∗W0 right away (p∗ defined in equation (12.73))"


We'll wait for that thread to come to consensus on BH before accepting this as definitive...
Is there anything in particular you think there is no consensus on? AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.
It doesn’t really differentiate between LS and DCA wrt TDF...just that glide paths and TDF, which are both recommended on BH, are bad.

Why not start a thread debating the merits of glide path and target date funds using that paper as the basis and see where it goes?

Don’t just wave your hands that it is the common understanding here when both are commonly recommended.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Fri Jun 26, 2020 3:18 pm

nigel_ht wrote:
Fri Jun 26, 2020 2:15 pm
Uncorrelated wrote:
Fri Jun 26, 2020 12:24 pm
nigel_ht wrote:
Fri Jun 26, 2020 12:09 pm
Your reference says DCA is bad because glide paths are bad and Target Date Funds suck.. Given that this is all the document says in 12.4.2 why don't you successfully defend that position first before demanding we accept the conclusion?
"12.4 Target Date: Ineffectiveness of glide path strategies

It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."
"12.4.2 Dollar cost averaging
Suppose you have a lump sum W0 to invest. You have a target asset mix of pˆ fraction in the risky asset. Is it better to gradually buy stocks? This is often suggested, since this way you are dollar cost averaging.

Let t∗ be the period over which you dollar cost average, and T be the target date of your investment. Then, dollar cost averaging amounts to this deterministic strategy

􏰇 <formula that won't render via a cut and paste> (12.93)

However, from Theorem 1, this strategy cannot be better than buying p∗W0 right away (p∗ defined in equation (12.73))"


We'll wait for that thread to come to consensus on BH before accepting this as definitive...
Is there anything in particular you think there is no consensus on? AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.
It doesn’t really differentiate between LS and DCA wrt TDF...just that glide paths and TDF, which are both recommended on BH, are bad.
In the terminology of that paper, DCA, glide paths and TDF are the same thing.

BH doesn't recommend TDF's to individuals that received all their lifetime wealth up-front. The correct use case for a TDF is when you have a large amount of human capital remaining. In this case, a TDF approximates to a constant asset allocation. Which, as has been mentioned dozens of times, is optimal for any i.i.d. distribution of possible outcomes.

You don't have to take my word for it, you can read this paper: https://papers.ssrn.com/sol3/papers.cfm ... id=1149340
The recommendation from the Samuelson (1969) and Merton (1969, 1971) life-cycle investment models is to invest a constant fraction of wealth in stocks. The mistake in translating this theory into practice is that young people invest only a fraction of their current savings, not their discounted lifetime savings. For someone in their 30's, investing even 100% of current savings is still likely to be less than 10% of their lifetime savings or less than 1/6 th of what the person should be holding in equities if, as is typical, their risk aversion would have led them to invest at least 60% of their lifetime savings in stocks.
Or if you prefer a more mathematical analysis, review chapter 6 of https://faculty.fuqua.duke.edu/~charvey ... iceira.pdf, or if you prefer empirical analysis, review https://papers.ssrn.com/sol3/papers.cfm ... id=3016824


I'd prefer it if you would respond with arguments instead of complaining about hand waving or consensus. I don't care about subjective measurements of consensus and neither should you. I care about arguments and math.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Fri Jun 26, 2020 5:16 pm

Uncorrelated wrote:
Fri Jun 26, 2020 3:18 pm
nigel_ht wrote:
Fri Jun 26, 2020 2:15 pm
Uncorrelated wrote:
Fri Jun 26, 2020 12:24 pm
nigel_ht wrote:
Fri Jun 26, 2020 12:09 pm
Your reference says DCA is bad because glide paths are bad and Target Date Funds suck.. Given that this is all the document says in 12.4.2 why don't you successfully defend that position first before demanding we accept the conclusion?
"12.4 Target Date: Ineffectiveness of glide path strategies

It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."
"12.4.2 Dollar cost averaging
Suppose you have a lump sum W0 to invest. You have a target asset mix of pˆ fraction in the risky asset. Is it better to gradually buy stocks? This is often suggested, since this way you are dollar cost averaging.

Let t∗ be the period over which you dollar cost average, and T be the target date of your investment. Then, dollar cost averaging amounts to this deterministic strategy

􏰇 <formula that won't render via a cut and paste> (12.93)

However, from Theorem 1, this strategy cannot be better than buying p∗W0 right away (p∗ defined in equation (12.73))"


We'll wait for that thread to come to consensus on BH before accepting this as definitive...
Is there anything in particular you think there is no consensus on? AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.
It doesn’t really differentiate between LS and DCA wrt TDF...just that glide paths and TDF, which are both recommended on BH, are bad.
In the terminology of that paper, DCA, glide paths and TDF are the same thing.
Doesn't say anything about Lump Sum being a starting criteria.

"12.4 Target Date: Ineffectiveness of glide path strategies
It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."

See any text in that introduction about lump sum vs investment over time? Nope. If he states this elsewhere in his 127 page PDF then you should highlight that.
BH doesn't recommend TDF's to individuals that received all their lifetime wealth up-front.
Citation needed.
The correct use case for a TDF is when you have a large amount of human capital remaining. In this case, a TDF approximates to a constant asset allocation. Which, as has been mentioned dozens of times, is optimal for any i.i.d. distribution of possible outcomes.
No, human capital can be zero. What matters for TDF is the time horizon for when you need the money as you go from high risk to low risk as the time horizon approaches and not remaining human capital.

The paper you cited not only says DCA is bad but also commonly recommended BH practices like glide path to retirement and TDFs.
You don't have to take my word for it, you can read this paper: https://papers.ssrn.com/sol3/papers.cfm ... id=1149340
...
I'd prefer it if you would respond with arguments instead of complaining about hand waving or consensus. I don't care about subjective measurements of consensus and neither should you. I care about arguments and math.
Title is "Life-Cycle Investing and Leverage: Buying Stock on Margin Can Reduce Retirement Risk".

Nope, not going to read random papers behind a paywall/registration. Nor am I going to read a 200 page PDF and guess at what you are referring to.

You are arguing that X is well known and accepted on BH so it is up to you to show that X actually IS the consensus on BH. If you don't you're simply handwaving.

Beehave
Posts: 727
Joined: Mon Jun 19, 2017 12:46 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Beehave » Fri Jun 26, 2020 10:16 pm

The articles cited do not impress.

They provide allegedly precise predictive models. Then, when they attempt to use empirical data to serve as evidence that the real world behaves in a way that resembles their model, they list all the simplifications they need to make. The simplifications are telling.

1. They model stocks and cash. No bonds. Then they say you need lots of stock. The simplification is self-serving.

2. They model annual rebalancing. They purport that frequent rebalancing performs better. And that's what typical life strategy funds do, they rebalance daily. The typical holder of a stock index and bond index and money market fund rebalances far less frequently. Again, self-serving.

3. They model simple, discrete lottery choices, not nested sequences (Von Neumann-Morgensstern). But the receipt of a windfall triggers intensely complex nested decisions about how much of the windfall to allocate to each of many competing needs and wants. Childrens' education, vs. retirement savings, vs. pay down mortgage, vs. home improvement, vs. new car vs. pay down student loan, vs. needed vacation vs. kids' summer camp vs. vs. vs. And there are complex interrationships and time dependencies. The DCA choice is intimately affected by these nested factors in many ways, not the least of which are (a) to compromise between acting and keeping options open.

4. Claim they are modeling Social Security as an income stream only, and then persist in attributing a decreasing net present value to it as we age as a basis for calculating comparative stock allocations (Irlan)

5. They start the modeling of target funds as if age 50 as a begin date to compare with index funds makes sense. A fair comparison would start at age 35 (or earlier) or 40, to show a reasonable glide path. The one modeled by Forsyth is way truncated and self-serving. It looks like the real world data plugged in for an earlier target date fund start would show his analytics to be totally unsupported by real-world behavior. He barely has a case with age 50 (way too high for a start date), annual rebalancing (way too long for a target fund) and cash and stock with no bonds.

The oversimplifications are not only largely very self-serving, they also point out the complexity of formally or informally modeling complex financial decisions - - including whether to DCA or lump sum, and whether or not new money is the same as an equal amount of old money. The complexity itself can explain many DCA choices - - I want to start but I want to take my time because there are lots of moving parts and lots of unknowns and many complex interrationships. So move, but take it slow.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sat Jun 27, 2020 4:09 am

nigel_ht wrote:
Fri Jun 26, 2020 5:16 pm
Doesn't say anything about Lump Sum being a starting criteria.

"12.4 Target Date: Ineffectiveness of glide path strategies
It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."

See any text in that introduction about lump sum vs investment over time? Nope. If he states this elsewhere in his 127 page PDF then you should highlight that.
He implicitly assumes that by not modelling any additions and withdrawals to the portfolio. Under these circumstances the paper is correct that TDF's are bad.

BH doesn't recommend TDF's to individuals that received all their lifetime wealth up-front.
Citation needed.
I'm sorry, I don't remember the last time a trust fund baby came to bogleheads to ask for advice. But I can offer these academic articles:

exhibit 1: see the quote from the life-cycle investing paper in my last post.

exhibit 2: See the first line and the first equation after the "solution" header, and please tell me which part of the equation depends on the time horizon: https://en.wikipedia.org/wiki/Merton%27 ... io_problem

exhibit 3:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3016824 wrote:
I have been considering portfolios that are typical of those that might be expected for average individuals who work for a living and live in an advanced economy such as the U.S. [...]

These results do not apply to investment advice for high net worth individuals. Consider for instance an individual that receives a $10m trust fund at age 20, and chooses never to work. Their human capital is zero, and so ignoring taxes their optimal equity asset allocation is either a fixed 34% or 67% depending upon their risk aversion.
The correct use case for a TDF is when you have a large amount of human capital remaining. In this case, a TDF approximates to a constant asset allocation. Which, as has been mentioned dozens of times, is optimal for any i.i.d. distribution of possible outcomes.
No, human capital can be zero. What matters for TDF is the time horizon for when you need the money as you go from high risk to low risk as the time horizon approaches and not remaining human capital.

The paper you cited not only says DCA is bad but also commonly recommended BH practices like glide path to retirement and TDFs.
Please see exhibit 1, 2, 3.
You don't have to take my word for it, you can read this paper: https://papers.ssrn.com/sol3/papers.cfm ... id=1149340
...
I'd prefer it if you would respond with arguments instead of complaining about hand waving or consensus. I don't care about subjective measurements of consensus and neither should you. I care about arguments and math.
Title is "Life-Cycle Investing and Leverage: Buying Stock on Margin Can Reduce Retirement Risk".

Nope, not going to read random papers behind a paywall/registration. Nor am I going to read a 200 page PDF and guess at what you are referring to.

You are arguing that X is well known and accepted on BH so it is up to you to show that X actually IS the consensus on BH. If you don't you're simply handwaving.
This is probably one of the top 3 most discussed papers on bogleheads. You can download the paper by clicking on the hidden button on the bottom right of the page. You can also read this topic: viewtopic.php?f=10&t=274390, but frankly you should read the paper instead because the vast majority of people on the first 10 pages have no clue what they are talking about.

Consensus doesn't mean anything. Arguments mean something. I'm not interested in having a discussion where each side claims there is consensus for their viewpoint. This isn't subjective, this is math. If you disagree with the arguments, please tell me which part of the cited papers contains a mathematical mistake or which part of the mathematical model is wrong, and then propose a better model.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Sat Jun 27, 2020 6:19 am

Uncorrelated wrote:
Sat Jun 27, 2020 4:09 am
nigel_ht wrote:
Fri Jun 26, 2020 5:16 pm
BH doesn't recommend TDF's to individuals that received all their lifetime wealth up-front.
Citation needed.
I'm sorry, I don't remember the last time a trust fund baby came to bogleheads to ask for advice. But I can offer these academic articles:
You didn’t say a few academic articles suggest that TDF is suboptimal but that BH doesn’t recommend TDF to individuals for lump sum.

In fact there are a few threads on what to do with windfalls where posters do recommend TDF.

The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
This is probably one of the top 3 most discusses papers on bogleheads.
Citation needed. I’d guess Trinity is top three. Same for Markowitz’s Modern Portfolio Theory. Schiller would be another easy contender for top 3. I see Pfau get mentioned a lot but maybe not top #3.

I seriously doubt this paper is #3.

It’s kinda odd how you demand folks discuss stuff “objectively” using math in one paragraph and then get upset when called on these kinds of unsupported assertions in another.
Consensus doesn't mean anything. Arguments mean something. I'm not interested in having a discussion where each side claims there is consensus for their viewpoint. This isn't subjective, this is math. If you disagree with the arguments, please tell me which part of the cited papers contains a mathematical mistake or which part of the mathematical model is wrong, and then propose a better model.
You are the one claiming that BH recommends this or that. As you can see in the DCA wiki discussion that wiki page edits are based on consensus.

If you don’t want to talk about consensus then I suggest you stop making assertions like “BH doesn’t recommend X”.

As far as math not being subjective I’ve peered reviewed too many papers to believe that. While generally the math itself should hold up its the limitations and assumptions that are contentious. Just because something is published does not mean it is accepted or correct...

Beehave pointed some of his issues with the papers but I notice that you don’t address them.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sat Jun 27, 2020 6:41 am

nigel_ht wrote:
Sat Jun 27, 2020 6:19 am
The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
That would be wrong. I linked four papers that explicitly claim that asset allocation does not depend on the time horizon. It doesn't matter if your time horizon is 1 year or 20 years, the optimal AA for a given risk tolerance is the same.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sat Jun 27, 2020 7:01 am

Beehave wrote:
Fri Jun 26, 2020 10:16 pm
The articles cited do not impress.

They provide allegedly precise predictive models. Then, when they attempt to use empirical data to serve as evidence that the real world behaves in a way that resembles their model, they list all the simplifications they need to make. The simplifications are telling.

1. They model stocks and cash. No bonds. Then they say you need lots of stock. The simplification is self-serving.
They do this for simplicity. In other papers by Irlam, he investigates cash/stock/bond/annuity mixtures. You can read the rest of his papers here: https://www.aacalc.com/about
2. They model annual rebalancing. They purport that frequent rebalancing performs better. And that's what typical life strategy funds do, they rebalance daily. The typical holder of a stock index and bond index and money market fund rebalances far less frequently. Again, self-serving.
Irlam takes an empirical analysis. Many other papers take a mathematical analysis with continuous rebalancing, but find the same basic conclusions. Irlam actually claims annual rebalancing performs better than frequent rebalancing in real-world conditions, although the conclusions are obviously dependent on how likely you think it is that historical mean-reverting behavior persists in the future: https://www.aacalc.com/docs/when_to_rebalance.
3. They model simple, discrete lottery choices, not nested sequences (Von Neumann-Morgensstern). But the receipt of a windfall triggers intensely complex nested decisions about how much of the windfall to allocate to each of many competing needs and wants. Childrens' education, vs. retirement savings, vs. pay down mortgage, vs. home improvement, vs. new car vs. pay down student loan, vs. needed vacation vs. kids' summer camp vs. vs. vs. And there are complex interrationships and time dependencies. The DCA choice is intimately affected by these nested factors in many ways, not the least of which are (a) to compromise between acting and keeping options open.
This does not affect the basic conclusion that the optimal allocation is a lump sum investment. For example, look at the graph here: https://www.aacalc.com/docs/variable_wi ... tock_heavy. Imagine that you are at age 80 with $500k in capital which gives you a balanced asset allocation. Suppose that you receive a windfall of $400k. Now the optimal allocation is to switch immediately to the optimal allocation according to that chart.

If you disagree that this chart is relevant for you, you should adjust model parameters until they fit your specific scenario. However, the basic conclusion that there is one optimal allocation for each unique combination of age and net worth does not change. This basic conclusion indicates that DCA is never optimal.

4. Claim they are modeling Social Security as an income stream only, and then persist in attributing a decreasing net present value to it as we age as a basis for calculating comparative stock allocations (Irlan)
That is not an explicit assumption, that is the result of the optimizer.

5. They start the modeling of target funds as if age 50 as a begin date to compare with index funds makes sense. A fair comparison would start at age 35 (or earlier) or 40, to show a reasonable glide path. The one modeled by Forsyth is way truncated and self-serving. It looks like the real world data plugged in for an earlier target date fund start would show his analytics to be totally unsupported by real-world behavior. He barely has a case with age 50 (way too high for a start date), annual rebalancing (way too long for a target fund) and cash and stock with no bonds.
I recommend reading https://papers.ssrn.com/sol3/papers.cfm ... id=1149340 if you're interested in optimal asset allocation for the savings phase. I only refer to the Forsyth paper because it contains a direct proof that DCA is suboptimal.
The oversimplifications are not only largely very self-serving, they also point out the complexity of formally or informally modeling complex financial decisions - - including whether to DCA or lump sum, and whether or not new money is the same as an equal amount of old money. The complexity itself can explain many DCA choices - - I want to start but I want to take my time because there are lots of moving parts and lots of unknowns and many complex interrationships. So move, but take it slow.
tl;dr life is complex, so use this asset allocation that has been proven to be suboptimal at least a dozen of times in this thread.... No, just no.

ValuationsMatter
Posts: 341
Joined: Mon Mar 09, 2020 2:12 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by ValuationsMatter » Sat Jun 27, 2020 8:00 am

Uncorrelated wrote:
Fri Jun 26, 2020 3:18 pm
nigel_ht wrote:
Fri Jun 26, 2020 2:15 pm
Uncorrelated wrote:
Fri Jun 26, 2020 12:24 pm
nigel_ht wrote:
Fri Jun 26, 2020 12:09 pm
Your reference says DCA is bad because glide paths are bad and Target Date Funds suck.. Given that this is all the document says in 12.4.2 why don't you successfully defend that position first before demanding we accept the conclusion?
"12.4 Target Date: Ineffectiveness of glide path strategies

It is often suggested that investors should purchase Target Date (Lifecycle) funds. These funds are for investors saving for retirement in 20 − 30 years. The idea here is to start off with high proportion in equities (i.e. 100%), and then gradually rebalance to a large proportion in bonds (i.e. 80%) as the target date is approached. We will show here that this strategy does not make much sense."
"12.4.2 Dollar cost averaging
Suppose you have a lump sum W0 to invest. You have a target asset mix of pˆ fraction in the risky asset. Is it better to gradually buy stocks? This is often suggested, since this way you are dollar cost averaging.

Let t∗ be the period over which you dollar cost average, and T be the target date of your investment. Then, dollar cost averaging amounts to this deterministic strategy

􏰇 <formula that won't render via a cut and paste> (12.93)

However, from Theorem 1, this strategy cannot be better than buying p∗W0 right away (p∗ defined in equation (12.73))"


We'll wait for that thread to come to consensus on BH before accepting this as definitive...
Is there anything in particular you think there is no consensus on? AFAIK it is widely known that glidepaths and target date funds are bad if you have a lump sum to invest.
It doesn’t really differentiate between LS and DCA wrt TDF...just that glide paths and TDF, which are both recommended on BH, are bad.
In the terminology of that paper, DCA, glide paths and TDF are the same thing.

BH doesn't recommend TDF's to individuals that received all their lifetime wealth up-front. The correct use case for a TDF is when you have a large amount of human capital remaining. In this case, a TDF approximates to a constant asset allocation. Which, as has been mentioned dozens of times, is optimal for any i.i.d. distribution of possible outcomes.

You don't have to take my word for it, you can read this paper: https://papers.ssrn.com/sol3/papers.cfm ... id=1149340
The recommendation from the Samuelson (1969) and Merton (1969, 1971) life-cycle investment models is to invest a constant fraction of wealth in stocks. The mistake in translating this theory into practice is that young people invest only a fraction of their current savings, not their discounted lifetime savings. For someone in their 30's, investing even 100% of current savings is still likely to be less than 10% of their lifetime savings or less than 1/6 th of what the person should be holding in equities if, as is typical, their risk aversion would have led them to invest at least 60% of their lifetime savings in stocks.
Or if you prefer a more mathematical analysis, review chapter 6 of https://faculty.fuqua.duke.edu/~charvey ... iceira.pdf, or if you prefer empirical analysis, review https://papers.ssrn.com/sol3/papers.cfm ... id=3016824


I'd prefer it if you would respond with arguments instead of complaining about hand waving or consensus. I don't care about subjective measurements of consensus and neither should you. I care about arguments and math.
Just wanted to jump in to compliment you on a nice post, and to say that I'm still around, trying to keep up and enjoying each side's points.

It still seems to me that the lower p* has the effect of spreading the variance reduction over the lifetime of the investment, rather than concentrating the variance reduction up front. I do thing TDFs are way off the mark as when I consider what to do with a large amount to invest, my DCA period is over the timeframe of weeks/months/days, and not years or decades. I am confident that as time lengthens, EV dominates variance.

User avatar
tadamsmar
Posts: 9014
Joined: Mon May 07, 2007 12:33 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by tadamsmar » Sat Jun 27, 2020 8:34 am

I think I see the argument in favor of the idea that DCA of a LS is rational.

1. It is rational for investors to set their AA based on their risk tolerance even when there may be no objective basis for that risk tolerance.
2. An investor's risk tolerance may vary over time.
3. Therefore it is rational to change your AA for no objective reason.
4. DCAing a lump sum is an example of #3. Therefore it is rational.

The first 2 are axioms in the deduction. #1 is a Boglehead axiom. #2 ranks as an empirical claim based on evidence like the observation that some investors who are paralyzed by fear of regret at the thought of immediately investing a lump sum can nevertheless lose that fear by DCAing over a period of one year. (I presume there are such observations, never seen it myself.)
Last edited by tadamsmar on Sat Jun 27, 2020 7:27 pm, edited 1 time in total.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Sat Jun 27, 2020 10:08 am

Since it seems that it’s okay to just post excerpts from papers to “prove” a point:

“ Samuel-son (1989a) shows that age effects can arise under non-constant relative risk aversion. Samuelson (1991) showed that the conventionally advised age effect arise when the risky asset has a two-point distribution and returns are nega-tively serially correlated. Bodie et al. (1992) showed that age effects arise when younger people have more flexibility in responding to portfolio realizations byaltering labor supply. See also Samuelson (1989b, 1990) and Jagannathan andKocherlakota (1996) for discussions of these issues. On the other hand, Rozeadd
(1994) showed that linear dollar-cost averaging is mean variance inefficient if returns are independent through time.The present paper follows Rozeff in considering the issue of mean-variance efficiency, but allows for serial dependence of returns. Existence of serial correlation in rates of return has been carefully documented by Poterba and Summers(1988) and Fama and French (1988) who
find that, for yearly observations of stock returns, rates of return are significantly negatively autocorrelated. This phenomenon reflects a process of mean reversion (or, more accurately, trend reversion) in stock prices.”

“ The paper showed the conditions under which mean-invariance efficiency is relevant, and demonstrated analytically that relative holdings of the risky asset should indeed be diminished as retirement approaches if returns are ARMA (1, 1) and are negatively correlated at all lags; in this case dollar-cost averaging as the risky asset is first acquired is also optimal, provided the horizon is sufficiently distant. The conventional age effect is also substantially confirmed,numerically, for more complicated dynamics in which returns autocorrelations are positive at short lags and negative at longer lags. These results suggest that investment advisors implicitly predicate their recommendations on the assumption that asset prices are indeed mean reverting.”

(Note: not all cut and paste error may have been corrected)

https://www.academia.edu/19150468/Effic ... portfolios

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sat Jun 27, 2020 11:56 am

tadamsmar wrote:
Sat Jun 27, 2020 8:34 am
I think I see the argument in favor of the idea that DCA of a LS is rational.

1. It is rational for investors to set their AA based on their risk tolerance even when there may be no objective basis for that risk tolerance.
2. An investors risk tolerance may vary over time.
3. Therefore it is rational to change your AA for no objective reason.
4. DCAing a lump sum is an example of #3. Therefore it is rational.

The first 2 are axioms in the deduction. #1 is a Boglehead axiom. #2 ranks as an empirical claim based on evidence like the observation that some investors who are paralyzed by fear of regret at the thought of immediately investing a lump sum can nevertheless lose that fear by DCAing over a period of one year. (I presume there are such observations, never seen it myself.)
I don't necessarily agree with #1, #2 and #3. But supposing those are true for a second, #4 doesn't follow unless you knew exactly in which way your risk tolerance would change throughout time. But if that's the case, the resulting optimal asset allocation is very unlikely to resemble a linear glide path (I would expect to see an asset allocation something like in Irlam's work).
nigel_ht wrote:
Sat Jun 27, 2020 10:08 am
Since it seems that it’s okay to just post excerpts from papers to “prove” a point:

“ Samuel-son (1989a) shows that age effects can arise under non-constant relative risk aversion. Samuelson (1991) showed that the conventionally advised age effect arise when the risky asset has a two-point distribution and returns are nega-tively serially correlated. Bodie et al. (1992) showed that age effects arise when younger people have more flexibility in responding to portfolio realizations byaltering labor supply. See also Samuelson (1989b, 1990) and Jagannathan andKocherlakota (1996) for discussions of these issues. On the other hand, Rozeadd
(1994) showed that linear dollar-cost averaging is mean variance inefficient if returns are independent through time.The present paper follows Rozeff in considering the issue of mean-variance efficiency, but allows for serial dependence of returns. Existence of serial correlation in rates of return has been carefully documented by Poterba and Summers(1988) and Fama and French (1988) who
find that, for yearly observations of stock returns, rates of return are significantly negatively autocorrelated. This phenomenon reflects a process of mean reversion (or, more accurately, trend reversion) in stock prices.”

“ The paper showed the conditions under which mean-invariance efficiency is relevant, and demonstrated analytically that relative holdings of the risky asset should indeed be diminished as retirement approaches if returns are ARMA (1, 1) and are negatively correlated at all lags; in this case dollar-cost averaging as the risky asset is first acquired is also optimal, provided the horizon is sufficiently distant. The conventional age effect is also substantially confirmed,numerically, for more complicated dynamics in which returns autocorrelations are positive at short lags and negative at longer lags. These results suggest that investment advisors implicitly predicate their recommendations on the assumption that asset prices are indeed mean reverting.”

(Note: not all cut and paste error may have been corrected)

https://www.academia.edu/19150468/Effic ... portfolios
The papers I linked assume a CRRA utility and i.i.d. returns. If you relax those assumptions, it's possible to obtain an asset allocation that varies throughout time. Specifically if you choose a different utility function you get an unique asset allocation for each combination of net worth and time horizon (this is not a glidepath. The correct asset allocation depends on random events that are not known in advance. Therefore, it is impossible that a pre-determined glidepath/DCA/TDF is the optimal solution). To my knowledge Irlam, whose work can be observed at https://www.aacalc.com/about, is the only researcher that offered practical advice for this scenario. I mentioned this in passing.

If you relax the assumption if i.i.d. returns, then the asset allocation depends on the state of the market (such as P/E ratio's.) This is market timing. To my knowledge no attempts have been done to find the optimal asset allocation in realistic scenario's. (afaik all papers that claim to have found a solution, only present a solution for the equilibrium case). As a funny anecdote to illustrate how sensitive the results are to the estimated parameters, Campbell (2001, Strategic asset allocation: Portfolio choice for long-term investors. A paper that sparked much BH discussion) investigated optimal bond allocations for investors with infinite time horizons based on mean-reverting bond markets. He found that in the time period 1952-1999, it is optimal to hold mostly short term bonds. On the time period 1983-1999, it is optimal to hold only long term bonds. I didn't mention this because nobody appears to understand this material.

If I'm reading the paper you cite right, it says that linear dollar cost averaging is never optimal, but in some specific scenario's the optimal (equilibrium case) solution is a glidepath with very non-obvious curvy lines. I don't call that DCA.

User avatar
tadamsmar
Posts: 9014
Joined: Mon May 07, 2007 12:33 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by tadamsmar » Sat Jun 27, 2020 12:40 pm

Uncorrelated wrote:
Sat Jun 27, 2020 11:56 am
tadamsmar wrote:
Sat Jun 27, 2020 8:34 am
I think I see the argument in favor of the idea that DCA of a LS is rational.

1. It is rational for investors to set their AA based on their risk tolerance even when there may be no objective basis for that risk tolerance.
2. An investors risk tolerance may vary over time.
3. Therefore it is rational to change your AA for no objective reason.
4. DCAing a lump sum is an example of #3. Therefore it is rational.

The first 2 are axioms in the deduction. #1 is a Boglehead axiom. #2 ranks as an empirical claim based on evidence like the observation that some investors who are paralyzed by fear of regret at the thought of immediately investing a lump sum can nevertheless lose that fear by DCAing over a period of one year. (I presume there are such observations, never seen it myself.)
I don't necessarily agree with #1, #2 and #3. But supposing those are true for a second, #4 doesn't follow unless you knew exactly in which way your risk tolerance would change throughout time. But if that's the case, the resulting optimal asset allocation is very unlikely to resemble a linear glide path (I would expect to see an asset allocation something like in Irlam's work).
#4 is a heuristic. Like lots of heuristics it does not hold up to detailed scrutiny. It's viewed as a simple mostly harmless plan to get an investor invested, with is supposedly better than sitting on the sideline with a boatload of cash. I guess it must work to get some investors invested, since the idea has been around for so long. But I am a bit concerned that it is seems to be pushed by stakeholders in the investment industry who profit from getting investors invested.

Not sure that #1 is actually a Boglehead axiom. The axiom may be that it is merely acceptable, not necessarily rational.

When I had a lump sum I just invested it immediately. There are more interesting issues with lump sums. The investment was my first taxable investment. I'd tell people to learn something about taxable to prepare for a lump sum. And a lump sum can cause a big change in what you need to do the meet your goals. DCA vs LS is, IMO, less important.

ChrisBenn
Posts: 329
Joined: Mon Aug 05, 2019 7:56 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by ChrisBenn » Sat Jun 27, 2020 12:51 pm

Uncorrelated wrote:
Sat Jun 27, 2020 6:41 am
nigel_ht wrote:
Sat Jun 27, 2020 6:19 am
The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
That would be wrong. I linked four papers that explicitly claim that asset allocation does not depend on the time horizon. It doesn't matter if your time horizon is 1 year or 20 years, the optimal AA for a given risk tolerance is the same.
Is this not effectively deferring the issue of time horizon i.e. embedding it in the risk tolerance?

If a person had two different situations, but was otherwise identical - (a) 100k to invest for their child who would start college in 5 years, and (b) 100k to invest for retirement in an account (~ illiquid) for 30 years it would seem like their tolerance for volatility would be different, thus their risk tolerance would be different (so the optimal AA would be different) -- or am I missing something obvious?

I also acknowledge this is somewhat orthogonal to the TDF/glidepath discussion since either scenarios still have a fixed optimal AA; my issue is I'm not intuiting (when applied to real life situations) how time horizon would not impact AA (or I misinterpreted your last sentence)

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sat Jun 27, 2020 2:18 pm

ChrisBenn wrote:
Sat Jun 27, 2020 12:51 pm
Uncorrelated wrote:
Sat Jun 27, 2020 6:41 am
nigel_ht wrote:
Sat Jun 27, 2020 6:19 am
The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
That would be wrong. I linked four papers that explicitly claim that asset allocation does not depend on the time horizon. It doesn't matter if your time horizon is 1 year or 20 years, the optimal AA for a given risk tolerance is the same.
Is this not effectively deferring the issue of time horizon i.e. embedding it in the risk tolerance?

If a person had two different situations, but was otherwise identical - (a) 100k to invest for their child who would start college in 5 years, and (b) 100k to invest for retirement in an account (~ illiquid) for 30 years it would seem like their tolerance for volatility would be different, thus their risk tolerance would be different (so the optimal AA would be different) -- or am I missing something obvious?

I also acknowledge this is somewhat orthogonal to the TDF/glidepath discussion since either scenarios still have a fixed optimal AA; my issue is I'm not intuiting (when applied to real life situations) how time horizon would not impact AA (or I misinterpreted your last sentence)
Risk tolerance and time horizon are orthogonal concepts. The risk tolerance says: If I get to spend/leave to my child $100, I get 1 amount of happiness. And if I get to spend $200 (double), I get 1 + X amount of happiness. Then assuming your utility function has some nice properties and markets are i.i.d., it follows that investing Y amount into stocks is optimal, regardless of the time horizon.

If your utility function does not have nice properties, the time horizon can matter. For example, if you desperately want your child to inherit at least $100 by age 18 or else you gain minus negative utility. At some point (It's the day before her birthday and you have only $1) it becomes rational to borrow money and try your luck in a casino. But then you would need an asset allocation that depends on both your current wealth as well as the time horizon. If markets are i.i.d. It is not possible to beat a constant allocation with an asset allocation that depends only on time (DCA or a glidepath).

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Sun Jun 28, 2020 4:29 am

Uncorrelated wrote:
Sat Jun 27, 2020 2:18 pm
ChrisBenn wrote:
Sat Jun 27, 2020 12:51 pm
Uncorrelated wrote:
Sat Jun 27, 2020 6:41 am
nigel_ht wrote:
Sat Jun 27, 2020 6:19 am
The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
That would be wrong. I linked four papers that explicitly claim that asset allocation does not depend on the time horizon. It doesn't matter if your time horizon is 1 year or 20 years, the optimal AA for a given risk tolerance is the same.
Is this not effectively deferring the issue of time horizon i.e. embedding it in the risk tolerance?

If a person had two different situations, but was otherwise identical - (a) 100k to invest for their child who would start college in 5 years, and (b) 100k to invest for retirement in an account (~ illiquid) for 30 years it would seem like their tolerance for volatility would be different, thus their risk tolerance would be different (so the optimal AA would be different) -- or am I missing something obvious?

I also acknowledge this is somewhat orthogonal to the TDF/glidepath discussion since either scenarios still have a fixed optimal AA; my issue is I'm not intuiting (when applied to real life situations) how time horizon would not impact AA (or I misinterpreted your last sentence)
Risk tolerance and time horizon are orthogonal concepts. The risk tolerance says: If I get to spend/leave to my child $100, I get 1 amount of happiness. And if I get to spend $200 (double), I get 1 + X amount of happiness. Then assuming your utility function has some nice properties and markets are i.i.d., it follows that investing Y amount into stocks is optimal, regardless of the time horizon.

If your utility function does not have nice properties, the time horizon can matter. For example, if you desperately want your child to inherit at least $100 by age 18 or else you gain minus negative utility. At some point (It's the day before her birthday and you have only $1) it becomes rational to borrow money and try your luck in a casino. But then you would need an asset allocation that depends on both your current wealth as well as the time horizon. If markets are i.i.d. It is not possible to beat a constant allocation with an asset allocation that depends only on time (DCA or a glidepath).
You mean like a utility function where you’d like 25X expenses by retirement?

The reality is that most financial planning is bounded by a time horizon...whether it is to pay for college, buy a house or to retire.

Risk tolerance decreases as you approach those deadlines which is why glide path is used in retirement planning.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sun Jun 28, 2020 5:13 am

nigel_ht wrote:
Sun Jun 28, 2020 4:29 am
You mean like a utility function where you’d like 25X expenses by retirement?
With that utility function, it is impossible to do better than a constant allocation by using a glidepath that depends only on time. It is possible to do better if you have an asset allocation that depends on both your current net worth and time, as I showed here. As far as I'm aware of, this is the only attempt ever done to optimize for this specific utility function.

Risk tolerance decreases as you approach those deadlines which is why glide path is used in retirement planning.
That is a false assumption. If you already know your risk tolerance will decrease as you approach retirement, you should optimize for that decreased risk tolerance.

Why are you still making these false statements? I have linked several papers that assert that decreasing equity allocation is a result of human capital. If human capital is taken into account, these are technically constant allocations. If you are saving for retirement with a risk tolerance of 2, and later it turns out that your actual risk tolerance is 3, then you have chosen a suboptimal solution.

Beehave
Posts: 727
Joined: Mon Jun 19, 2017 12:46 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Beehave » Sun Jun 28, 2020 10:16 am

Uncorrelated wrote:
Sat Jun 27, 2020 7:01 am
Beehave wrote:
Fri Jun 26, 2020 10:16 pm
The articles cited do not impress.

They provide allegedly precise predictive models. Then, when they attempt to use empirical data to serve as evidence that the real world behaves in a way that resembles their model, they list all the simplifications they need to make. The simplifications are telling.

1. They model stocks and cash. No bonds. Then they say you need lots of stock. The simplification is self-serving.
They do this for simplicity. In other papers by Irlam, he investigates cash/stock/bond/annuity mixtures. You can read the rest of his papers here: https://www.aacalc.com/about
2. They model annual rebalancing. They purport that frequent rebalancing performs better. And that's what typical life strategy funds do, they rebalance daily. The typical holder of a stock index and bond index and money market fund rebalances far less frequently. Again, self-serving.
Irlam takes an empirical analysis. Many other papers take a mathematical analysis with continuous rebalancing, but find the same basic conclusions. Irlam actually claims annual rebalancing performs better than frequent rebalancing in real-world conditions, although the conclusions are obviously dependent on how likely you think it is that historical mean-reverting behavior persists in the future: https://www.aacalc.com/docs/when_to_rebalance.
3. They model simple, discrete lottery choices, not nested sequences (Von Neumann-Morgensstern). But the receipt of a windfall triggers intensely complex nested decisions about how much of the windfall to allocate to each of many competing needs and wants. Childrens' education, vs. retirement savings, vs. pay down mortgage, vs. home improvement, vs. new car vs. pay down student loan, vs. needed vacation vs. kids' summer camp vs. vs. vs. And there are complex interrationships and time dependencies. The DCA choice is intimately affected by these nested factors in many ways, not the least of which are (a) to compromise between acting and keeping options open.
This does not affect the basic conclusion that the optimal allocation is a lump sum investment. For example, look at the graph here: https://www.aacalc.com/docs/variable_wi ... tock_heavy. Imagine that you are at age 80 with $500k in capital which gives you a balanced asset allocation. Suppose that you receive a windfall of $400k. Now the optimal allocation is to switch immediately to the optimal allocation according to that chart.

If you disagree that this chart is relevant for you, you should adjust model parameters until they fit your specific scenario. However, the basic conclusion that there is one optimal allocation for each unique combination of age and net worth does not change. This basic conclusion indicates that DCA is never optimal.

4. Claim they are modeling Social Security as an income stream only, and then persist in attributing a decreasing net present value to it as we age as a basis for calculating comparative stock allocations (Irlan)
That is not an explicit assumption, that is the result of the optimizer.

5. They start the modeling of target funds as if age 50 as a begin date to compare with index funds makes sense. A fair comparison would start at age 35 (or earlier) or 40, to show a reasonable glide path. The one modeled by Forsyth is way truncated and self-serving. It looks like the real world data plugged in for an earlier target date fund start would show his analytics to be totally unsupported by real-world behavior. He barely has a case with age 50 (way too high for a start date), annual rebalancing (way too long for a target fund) and cash and stock with no bonds.
I recommend reading https://papers.ssrn.com/sol3/papers.cfm ... id=1149340 if you're interested in optimal asset allocation for the savings phase. I only refer to the Forsyth paper because it contains a direct proof that DCA is suboptimal.
The oversimplifications are not only largely very self-serving, they also point out the complexity of formally or informally modeling complex financial decisions - - including whether to DCA or lump sum, and whether or not new money is the same as an equal amount of old money. The complexity itself can explain many DCA choices - - I want to start but I want to take my time because there are lots of moving parts and lots of unknowns and many complex interrationships. So move, but take it slow.
tl;dr life is complex, so use this asset allocation that has been proven to be suboptimal at least a dozen of times in this thread.... No, just no.
Suboptimal in one context can be optimal in another. Racing slicks are great, unless you're concerned about rain. Contexts matter. That's the concern with simplifications that enable complete analyses of otherwise intractably complex systems. These logically complete analyses are of interest. They contribute to rational understanding. But, because of the simplifications that make their internal logic complete, they become less complete in larger contexts. Life requires all-weather tires, not racing slicks for most people.

The Von Neumann-Morgenstern system is logically complete at the cost of ruling out consideration of nested probabilities. But life is filled with nested probability issues. The investment I make in my child's education affects my retirement in many, interrelated ways, and I have more than one child, and they have different interests and capabilities. If I receive a windfall, I need to prioritize. Nested probabilities make this difficult. So it can be the rational choice is to compromise between optimization and flexibility by things like DCA, even if one believes LS is more effective.

Regarding Forsyth, I stand by my previous observations. He has stacked his modeling cases to "prove" his point, but more realistic model parameters would disprove it.

Regarding the Ales and Nalebuff article intended to correct Forsyth, it seems to agree that target date and fixed allocation up to 70/30 are pretty much equivalent over time, and that the really best policy would be to diversify over your lifespan's investment and intellectual capital assets by leveraging stock holdings early in career and tamping down on that as your wages and investments increase but your intellectual value grows shorter in duration and approaches a period of declining value.

So for Ales and Nalebuff, a 100% stock allocation is not good enough, they suggest amplification by taking out loans, buying options, or taking on margin early in one's career. They say:

"Of course, borrowing on margin creates a risk that the savings will be entirely lost.
That risk is related to the extent of leverage. If portfolios were leveraged 20 to 1, as we
do with real estate, this risk would be significant. We propose a maximum leverage of
2:1. It is worth emphasizing that we are only proposing this amount of leverage at an
early stage of life. Thus, investors only face the risk of wiping out their current
investments when they are still young and will have a chance to rebuild. Present savings
might be extinguished, but the present value of future savings will never be. Our
simulations account for this possibility and even so, we find that the minimum return
under the strategies with initially leveraged positions would be substantially higher
compared to the minimum under traditional investment strategies."
(From: LIFE-CYCLE INVESTING AND LEVERAGE:
BUYING STOCK ON MARGIN CAN REDUCE RETIREMENT RISK)

I'll simply say - they're probably right about what is likely to maximize investment results over time. But if it's me, I do not want to run the risk that at the very time the market plummets, my employment gets interrupted and it ends up that my optimization strategy has in a larger context put me at risk of being in a deep, deep hole of nested disoptimization into which I've fallen.

ValuationsMatter
Posts: 341
Joined: Mon Mar 09, 2020 2:12 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by ValuationsMatter » Sun Jun 28, 2020 11:40 am

Very nice post, Beehave.

Uncorrelated, you've stressed the importance of i.i.d. in your posts, sometimes several times within the same post. I'm almost certain the market is not i.i.d. Aren't you? We often model things with simplifying assumptions so that we may gain insight or intuition, even when we know the assumptions to be false. For example, physics models often disregard gravity, aerodynamic drag, etc... in order to gain understanding and intuition of kinetics, but we don't then run out apply those models to answer important real life questions.

Here's a paper that suggests that modeling markets on normal distributions is flawed: https://statistik.econ.kit.edu/download ... nation.pdf

I admit being in over my head here in regards to the implications of relaxing/removing the i.i.d. assumption. However, it's worth exploring. Again, thanks for your posts. The learning curve in this thread has been steep.

User avatar
Steve Reading
Posts: 1884
Joined: Fri Nov 16, 2018 10:20 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Steve Reading » Sun Jun 28, 2020 12:01 pm

Beehave wrote:
Sun Jun 28, 2020 10:16 am
Regarding the Ales and Nalebuff article intended to correct Forsyth, it seems to agree that target date and fixed allocation up to 70/30 are pretty much equivalent over time
That's not quite right, but I can see why you'd think that. The paper doesn't explain that well. A TDF is superior to a constant 70/30 in terms of risk-adjusted returns over your accumulation. And that's the case for the exact same reasons their leveraged approach is the best: to have a more even exposure to stocks as a % of your assets, you have to invest a higher proportion of your savings in stocks when young, than when old. A TDF does this but constant 70/30 doesn't.

The problem they find with a TDF is that it is, overall, just way too conservative in terms of stock exposure overall. So it leads to accumulations way under 70/30.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sun Jun 28, 2020 1:19 pm

ValuationsMatter wrote:
Sun Jun 28, 2020 11:40 am
Very nice post, Beehave.

Uncorrelated, you've stressed the importance of i.i.d. in your posts, sometimes several times within the same post. I'm almost certain the market is not i.i.d. Aren't you? We often model things with simplifying assumptions so that we may gain insight or intuition, even when we know the assumptions to be false. For example, physics models often disregard gravity, aerodynamic drag, etc... in order to gain understanding and intuition of kinetics, but we don't then run out apply those models to answer important real life questions.

Here's a paper that suggests that modeling markets on normal distributions is flawed: https://statistik.econ.kit.edu/download ... nation.pdf

I admit being in over my head here in regards to the implications of relaxing/removing the i.i.d. assumption. However, it's worth exploring. Again, thanks for your posts. The learning curve in this thread has been steep.
The superiority of LSI is true for any independent distribution, not just the normal distribution. You can prove this with Bellman's principe of optimality. It is true that daily returns are not normal, but monthly or annual returns are pretty normal.



Why assume independent distributions? If we don't do that, then the optimal solution will involve market timing. For instance, there is something called volatility clustering. Empirically, when volatility is high this week, it is very likely that volatility is also high next week. If volatility is low, it it very likely that the volatility next week is also low. There is a lot of evidence for this, nobody doubts that the effect exists. VIX allows investors to bet on future volatility. What can you do with this information? One approach is to use a high equity allocation when you expect future (next week) volatility to be low, and use low equity allocation when you expect future volatility to be high. But now you have a problem: this is definitely market timing. Market timing isn't supposed to be possible, so clearly the market is really inefficient or there is something we missed that prevents this from working. Probably the latter.

The easiest way to avoid this problem is to ban all forms of market timing. We can accomplish that by assuming returns are independently distributed. Without that assumption, everything becomes a question about market timing.

There are also problems with computational complexity. With the software I have, I can solve most asset allocation questions with an arbitrary utility function, dependent only on time and net worth, in a few minutes. If I were to use the same technique to solve a problem for non-i.i.d. markets, it would require more complicated software and orders of magnitude more computing power.

ChrisBenn
Posts: 329
Joined: Mon Aug 05, 2019 7:56 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by ChrisBenn » Sun Jun 28, 2020 1:37 pm

Uncorrelated wrote:
Sat Jun 27, 2020 2:18 pm
ChrisBenn wrote:
Sat Jun 27, 2020 12:51 pm
Uncorrelated wrote:
Sat Jun 27, 2020 6:41 am
nigel_ht wrote:
Sat Jun 27, 2020 6:19 am
The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
That would be wrong. I linked four papers that explicitly claim that asset allocation does not depend on the time horizon. It doesn't matter if your time horizon is 1 year or 20 years, the optimal AA for a given risk tolerance is the same.
Is this not effectively deferring the issue of time horizon i.e. embedding it in the risk tolerance?

If a person had two different situations, but was otherwise identical - (a) 100k to invest for their child who would start college in 5 years, and (b) 100k to invest for retirement in an account (~ illiquid) for 30 years it would seem like their tolerance for volatility would be different, thus their risk tolerance would be different (so the optimal AA would be different) -- or am I missing something obvious?

I also acknowledge this is somewhat orthogonal to the TDF/glidepath discussion since either scenarios still have a fixed optimal AA; my issue is I'm not intuiting (when applied to real life situations) how time horizon would not impact AA (or I misinterpreted your last sentence)
Risk tolerance and time horizon are orthogonal concepts. The risk tolerance says: If I get to spend/leave to my child $100, I get 1 amount of happiness. And if I get to spend $200 (double), I get 1 + X amount of happiness. Then assuming your utility function has some nice properties and markets are i.i.d., it follows that investing Y amount into stocks is optimal, regardless of the time horizon.

If your utility function does not have nice properties, the time horizon can matter. For example, if you desperately want your child to inherit at least $100 by age 18 or else you gain minus negative utility. At some point (It's the day before her birthday and you have only $1) it becomes rational to borrow money and try your luck in a casino. But then you would need an asset allocation that depends on both your current wealth as well as the time horizon. If markets are i.i.d. It is not possible to beat a constant allocation with an asset allocation that depends only on time (DCA or a glidepath).
Thanks for the reply. So If I'm understanding you right different time horizons would be embedded in the utility function that is being maximized. So for my scenario (college in five years, + retirement) the function would be non linear in that it might assign a fixed amount of happiness in being able to satisfy what I expect the base cost of college to be, and a decreasingly lower amount of happiness for each increment above that. The same thing might be true for the retirement portion of the component (i.e. incremental happiness per unit additional wealth would drop past a certain value).

If I'm understanding that correctly then it makes sense - but I also would despair at being able to correctly use that model with multiple expenses at different time horizons. My concern would be that the error (due to the estimations involved in all the components for each time horizon) would cause the inherent error to be a non trivial portion of the solution space. Caveat (and acknowledged) that this is just my intuition - def. not any sort of expert opinion.

My inclination would still (for large, critical expenses - i.e. child college, retirement as in the example) be optimizing each independently. I would just make a starting assumption of relative utility (i.e. how much of my capital I allocate to each) - but I feel that's more likely to be "accurate" than trying to do that by proxy by crafting a multi horizon utility function. This also seems good in that, while my current wealth is fungible between different time horizons my future income stream wouldn't be, so some "partitioning" between them doesn't seem incorrect (or maybe the utility function would have to account for that also)?

To me multiple time horizons seems to be a "thar be dragons" type situation for this type of optimization; it's not wrong, but I'm dubious about GIGO factor in it being able to be implemented correctly. For single time horizons I definitely get it (I think).

Is this reasonable, or have I gone off on some tangent and I'm missing the forest for the trees?

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sun Jun 28, 2020 2:50 pm

ChrisBenn wrote:
Sun Jun 28, 2020 1:37 pm
Uncorrelated wrote:
Sat Jun 27, 2020 2:18 pm
ChrisBenn wrote:
Sat Jun 27, 2020 12:51 pm
Uncorrelated wrote:
Sat Jun 27, 2020 6:41 am
nigel_ht wrote:
Sat Jun 27, 2020 6:19 am
The reason you wouldn’t generally recommend TDF to folks with trust funds is because their time horizon is 0 since the presumption is they will begin to withdraw immediately. So they should be in a withdrawal AA.

Not because of human capital.

If I lump sum into my newborn’s 529 then putting it into a TDF is perfectly reasonable even if I never put another dime in. The reason is they don’t need the money for 18 years.
That would be wrong. I linked four papers that explicitly claim that asset allocation does not depend on the time horizon. It doesn't matter if your time horizon is 1 year or 20 years, the optimal AA for a given risk tolerance is the same.
Is this not effectively deferring the issue of time horizon i.e. embedding it in the risk tolerance?

If a person had two different situations, but was otherwise identical - (a) 100k to invest for their child who would start college in 5 years, and (b) 100k to invest for retirement in an account (~ illiquid) for 30 years it would seem like their tolerance for volatility would be different, thus their risk tolerance would be different (so the optimal AA would be different) -- or am I missing something obvious?

I also acknowledge this is somewhat orthogonal to the TDF/glidepath discussion since either scenarios still have a fixed optimal AA; my issue is I'm not intuiting (when applied to real life situations) how time horizon would not impact AA (or I misinterpreted your last sentence)
Risk tolerance and time horizon are orthogonal concepts. The risk tolerance says: If I get to spend/leave to my child $100, I get 1 amount of happiness. And if I get to spend $200 (double), I get 1 + X amount of happiness. Then assuming your utility function has some nice properties and markets are i.i.d., it follows that investing Y amount into stocks is optimal, regardless of the time horizon.

If your utility function does not have nice properties, the time horizon can matter. For example, if you desperately want your child to inherit at least $100 by age 18 or else you gain minus negative utility. At some point (It's the day before her birthday and you have only $1) it becomes rational to borrow money and try your luck in a casino. But then you would need an asset allocation that depends on both your current wealth as well as the time horizon. If markets are i.i.d. It is not possible to beat a constant allocation with an asset allocation that depends only on time (DCA or a glidepath).
Thanks for the reply. So If I'm understanding you right different time horizons would be embedded in the utility function that is being maximized. So for my scenario (college in five years, + retirement) the function would be non linear in that it might assign a fixed amount of happiness in being able to satisfy what I expect the base cost of college to be, and a decreasingly lower amount of happiness for each increment above that. The same thing might be true for the retirement portion of the component (i.e. incremental happiness per unit additional wealth would drop past a certain value).
Yes that would be an utility function that doesn't have nice properties.

But in this specific case, there is a hack we can do to solve it. if you assume that the college expenditure is guaranteed, you can model your college expenditure as a negative zero-coupon bond. Then you can do the following:

Optimal proportion to stocks: use the formula from merton's portfolio problem. i.e. 60% stocks.
net worth: current liquid net worth - the value of the college zero-coupon bond + remaining human capital
current allocation to stocks = 60% * net worth. This may or may not be higher than 100%.

If you do this, you will preserve all properties of the original retirement utility function.
If I'm understanding that correctly then it makes sense - but I also would despair at being able to correctly use that model with multiple expenses at different time horizons. My concern would be that the error (due to the estimations involved in all the components for each time horizon) would cause the inherent error to be a non trivial portion of the solution space. Caveat (and acknowledged) that this is just my intuition - def. not any sort of expert opinion.

My inclination would still (for large, critical expenses - i.e. child college, retirement as in the example) be optimizing each independently. I would just make a starting assumption of relative utility (i.e. how much of my capital I allocate to each) - but I feel that's more likely to be "accurate" than trying to do that by proxy by crafting a multi horizon utility function. This also seems good in that, while my current wealth is fungible between different time horizons my future income stream wouldn't be, so some "partitioning" between them doesn't seem incorrect (or maybe the utility function would have to account for that also)?

To me multiple time horizons seems to be a "thar be dragons" type situation for this type of optimization; it's not wrong, but I'm dubious about GIGO factor in it being able to be implemented correctly. For single time horizons I definitely get it (I think).

Is this reasonable, or have I gone off on some tangent and I'm missing the forest for the trees?
You're on the right track.

There is a problem with optimizing each result independently: if you choose an efficient portfolio for each of your goals and then average the portfolio together, there is no guarantee that the resulting portfolio is efficient. Optimizing for multiple goals at the same time is definitely difficult.

Gordon irlam at https://www.aacalc.com/about has framework that allows you to solve for arbitrary utility functions. For example, he uses this to solve the the asset allocation or a retiree that has a very low risk tolerance for required spending and a higher risk tolerance for extra spending. In theory, the same mathematical methods can be used to solve for a different utility function at each time step. The main difficulty is coming up with utility functions that accurately reflect investor risk preferences. I don't even know what my own utility function is, I just picked a number that looked reasonable.

User avatar
Steve Reading
Posts: 1884
Joined: Fri Nov 16, 2018 10:20 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Steve Reading » Sun Jun 28, 2020 2:53 pm

Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Sun Jun 28, 2020 5:11 pm

Steve Reading wrote:
Sun Jun 28, 2020 2:53 pm
Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.
I choose γ ≈ 2.5, which results in an asset allocation of approximately 90% equities and 10% bonds with my return assumptions.

I probably don't have a CRRA, but I don't know what else would be appropriate. As usual in optimization, the main problem is an unclear definition of utility.

nigel_ht
Posts: 890
Joined: Tue Jan 01, 2019 10:14 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by nigel_ht » Sun Jun 28, 2020 5:20 pm

Uncorrelated wrote:
Sun Jun 28, 2020 5:11 pm
Steve Reading wrote:
Sun Jun 28, 2020 2:53 pm
Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.
I choose γ ≈ 2.5, which results in an asset allocation of approximately 90% equities and 10% bonds with my return assumptions.

I probably don't have a CRRA, but I don't know what else would be appropriate. As usual in optimization, the main problem is an unclear definition of utility.
Great, the math proves that constant AA of X is optimal for a given utility function while DCA is never optimal therefore DCA, glide paths and TDF are irrational behavior.

But how did you determine X for yourself? I pulled a number out of my ass for γ that seems right and lead to an AA that looked good.

:oops:

ValuationsMatter
Posts: 341
Joined: Mon Mar 09, 2020 2:12 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by ValuationsMatter » Sun Jun 28, 2020 5:22 pm

Uncorrelated wrote:
Sun Jun 28, 2020 1:19 pm
But now you have a problem: this is definitely market timing. Market timing isn't supposed to be possible, so clearly the market is really inefficient or there is something we missed that prevents this from working. Probably the latter.
That's actually not a problem for me. Markets are inefficient. Timing probably is possible. I don't know about that specific strategy, but I don't buy that markets *cannot* be timed. That's just boglehead dogma, in my opinion, though I don't intend to further distract from premise here, but nevertheless the assumption of independence, and homoscedasticity is almost definitely wrong. Perhaps, it's just necessary for modeling assumptions.
There are also problems with computational complexity. With the software I have, I can solve most asset allocation questions with an arbitrary utility function, dependent only on time and net worth, in a few minutes. If I were to use the same technique to solve a problem for non-i.i.d. markets, it would require more complicated software and orders of magnitude more computing power.
I wish I could discuss this more in-depth than time allows. Ah well, I plan to be around the forum for a while, so we may still have the time yet.

User avatar
Steve Reading
Posts: 1884
Joined: Fri Nov 16, 2018 10:20 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Steve Reading » Sun Jun 28, 2020 5:26 pm

Uncorrelated wrote:
Sun Jun 28, 2020 5:11 pm
Steve Reading wrote:
Sun Jun 28, 2020 2:53 pm
Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.
I choose γ ≈ 2.5, which results in an asset allocation of approximately 90% equities and 10% bonds with my return assumptions.

I probably don't have a CRRA, but I don't know what else would be appropriate. As usual in optimization, the main problem is an unclear definition of utility.
I don't know how much your dutch social security would cover, but assuming it's at least 10% of your overall wealth by the time you retire, does that imply you'll be 100% in stocks as a retiree? I know you abstain from leverage.

User avatar
Steve Reading
Posts: 1884
Joined: Fri Nov 16, 2018 10:20 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Steve Reading » Sun Jun 28, 2020 5:37 pm

nigel_ht wrote:
Sun Jun 28, 2020 5:20 pm
Uncorrelated wrote:
Sun Jun 28, 2020 5:11 pm
Steve Reading wrote:
Sun Jun 28, 2020 2:53 pm
Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.
I choose γ ≈ 2.5, which results in an asset allocation of approximately 90% equities and 10% bonds with my return assumptions.

I probably don't have a CRRA, but I don't know what else would be appropriate. As usual in optimization, the main problem is an unclear definition of utility.
Great, the math proves that constant AA of X is optimal for a given utility function while DCA is never optimal therefore DCA, glide paths and TDF are irrational behavior.

But how did you determine X for yourself? I pulled a number out of my ass for γ that seems right and lead to an AA that looked good.

:oops:
The two methods are entirely different:
Uncorrelated made an arbitrary decision as to his own personal risk tolerance, and then optimally invested based on that assumption.

DCA, OTOH, is suboptimal because no matter what arbitrary decision of your own risk tolerance you make, it's always the inferior way to implement the portfolio.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Mon Jun 29, 2020 2:28 am

nigel_ht wrote:
Sun Jun 28, 2020 5:20 pm
Uncorrelated wrote:
Sun Jun 28, 2020 5:11 pm
Steve Reading wrote:
Sun Jun 28, 2020 2:53 pm
Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.
I choose γ ≈ 2.5, which results in an asset allocation of approximately 90% equities and 10% bonds with my return assumptions.

I probably don't have a CRRA, but I don't know what else would be appropriate. As usual in optimization, the main problem is an unclear definition of utility.
Great, the math proves that constant AA of X is optimal for a given utility function while DCA is never optimal therefore DCA, glide paths and TDF are irrational behavior.

But how did you determine X for yourself? I pulled a number out of my ass for γ that seems right and lead to an AA that looked good.

:oops:
It's much easier to point out that things are irrational, than to find the optimal solution.

For example, flushing money down the toilet is definitely irrational. And there is some combination of DCA and flushing money down the toilet that is equivalent (in terms of expected return and risk) to LSI. Therefore if my axioms are correct, that definitely means that DCA is irrational.

But that knowledge doesn't help me to determine what lump sum investment I should use. Or how I should vary my asset allocation over time. Or how I should plan my consumption over time. Those are all very difficult problems.

User avatar
Uncorrelated
Posts: 739
Joined: Sun Oct 13, 2019 3:16 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by Uncorrelated » Mon Jun 29, 2020 9:10 am

ValuationsMatter wrote:
Sun Jun 28, 2020 5:22 pm
Uncorrelated wrote:
Sun Jun 28, 2020 1:19 pm
But now you have a problem: this is definitely market timing. Market timing isn't supposed to be possible, so clearly the market is really inefficient or there is something we missed that prevents this from working. Probably the latter.
That's actually not a problem for me. Markets are inefficient. Timing probably is possible. I don't know about that specific strategy, but I don't buy that markets *cannot* be timed. That's just boglehead dogma, in my opinion, though I don't intend to further distract from premise here, but nevertheless the assumption of independence, and homoscedasticity is almost definitely wrong. Perhaps, it's just necessary for modeling assumptions.
There are also problems with computational complexity. With the software I have, I can solve most asset allocation questions with an arbitrary utility function, dependent only on time and net worth, in a few minutes. If I were to use the same technique to solve a problem for non-i.i.d. markets, it would require more complicated software and orders of magnitude more computing power.
I wish I could discuss this more in-depth than time allows. Ah well, I plan to be around the forum for a while, so we may still have the time yet.
Market timing is more than a bogleheads dogma. There is no real evidence markets are inefficient on a scale that can be exploited. There are some market timing approaches that are promising because they do not appear to depend on inefficient markets, but they are very difficult to analyze.

Unless you're doing this for personal enjoyment, there are probably better ways to spend your time.
Steve Reading wrote:
Sun Jun 28, 2020 5:26 pm
Uncorrelated wrote:
Sun Jun 28, 2020 5:11 pm
Steve Reading wrote:
Sun Jun 28, 2020 2:53 pm
Uncorrelated wrote:
Sun Jun 28, 2020 2:50 pm
I don't even know what my own utility function is, I just picked a number that looked reasonable.
If you're talking about CRRA, what number did you personally pick that looked reasonable to you? Just curious.
I choose γ ≈ 2.5, which results in an asset allocation of approximately 90% equities and 10% bonds with my return assumptions.

I probably don't have a CRRA, but I don't know what else would be appropriate. As usual in optimization, the main problem is an unclear definition of utility.
I don't know how much your dutch social security would cover, but assuming it's at least 10% of your overall wealth by the time you retire, does that imply you'll be 100% in stocks as a retiree? I know you abstain from leverage.
I think it's likely that I will end up with a 100% stock allocation in retirement, but future tax law, social security, expected net worth at retirement and personal goals are too uncertain to analyze this scenario right now.

ValuationsMatter
Posts: 341
Joined: Mon Mar 09, 2020 2:12 am

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by ValuationsMatter » Mon Jun 29, 2020 10:35 am

Uncorrelated wrote:
Mon Jun 29, 2020 9:10 am
Unless you're doing this for personal enjoyment, there are probably better ways to spend your time.
Yes. Personal enjoyment (I'm a nerd) & self-education. I'd be interested to know your thoughts on which timing strategies you believe are promising. Perhaps you can link to any previous posts you've made on the subject, papers, or shoot a pm with your thoughts, since it's tangent to the thread?

TIAX
Posts: 1326
Joined: Sat Jan 11, 2014 12:19 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used

Post by TIAX » Mon Jun 29, 2020 11:16 am

EnjoyIt wrote:
Mon Jun 01, 2020 11:07 pm
After all, delaying an investment by a few months is trivial in the course of an investing lifetime.
I might agree if it were a one-time event. But generally, people who DCA, DCA all the time. For example, instead of maxing out their 401K as early in the year as possible, they will spread out their contributions over the full year every year. So, yes, I do think delaying "by a few months" your entire life will make a substantial difference.

User avatar
FiveK
Posts: 9320
Joined: Sun Mar 16, 2014 2:43 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used

Post by FiveK » Mon Jun 29, 2020 12:54 pm

TIAX wrote:
Mon Jun 29, 2020 11:16 am
EnjoyIt wrote:
Mon Jun 01, 2020 11:07 pm
After all, delaying an investment by a few months is trivial in the course of an investing lifetime.
I might agree if it were a one-time event. But generally, people who DCA, DCA all the time. For example, instead of maxing out their 401K as early in the year as possible, they will spread out their contributions over the full year every year. So, yes, I do think delaying "by a few months" your entire life will make a substantial difference.
For those with an employer that does not do a 401k True-Up, spreading the contributions over the full year is the only way to get the full company match.

ChrisBenn
Posts: 329
Joined: Mon Aug 05, 2019 7:56 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used

Post by ChrisBenn » Mon Jun 29, 2020 1:28 pm

FiveK wrote:
Mon Jun 29, 2020 12:54 pm
TIAX wrote:
Mon Jun 29, 2020 11:16 am
EnjoyIt wrote:
Mon Jun 01, 2020 11:07 pm
After all, delaying an investment by a few months is trivial in the course of an investing lifetime.
I might agree if it were a one-time event. But generally, people who DCA, DCA all the time. For example, instead of maxing out their 401K as early in the year as possible, they will spread out their contributions over the full year every year. So, yes, I do think delaying "by a few months" your entire life will make a substantial difference.
For those with an employer that does not do a 401k True-Up, spreading the contributions over the full year is the only way to get the full company match.
In addition to the point above (re: 401k contribution frontloading), there is also the fact that it would require adjusting my withholdings part-way through the year, which is annoying (vs never touching it except when limits are increased). I conceede it's non optimal (to not front load it) -- I just value the convenience more. From a convenience POV lump sum is, conversely, ahead :) So I don't think the analogy quite holds.

User avatar
Steve Reading
Posts: 1884
Joined: Fri Nov 16, 2018 10:20 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used

Post by Steve Reading » Mon Jun 29, 2020 1:37 pm

FiveK wrote:
Mon Jun 29, 2020 12:54 pm
TIAX wrote:
Mon Jun 29, 2020 11:16 am
EnjoyIt wrote:
Mon Jun 01, 2020 11:07 pm
After all, delaying an investment by a few months is trivial in the course of an investing lifetime.
I might agree if it were a one-time event. But generally, people who DCA, DCA all the time. For example, instead of maxing out their 401K as early in the year as possible, they will spread out their contributions over the full year every year. So, yes, I do think delaying "by a few months" your entire life will make a substantial difference.
For those with an employer that does not do a 401k True-Up, spreading the contributions over the full year is the only way to get the full company match.
Even those who don't have company matches might still want to spread it out over the year just in case they get fired and they pick up a job at some other place that does have a company match.

User avatar
tadamsmar
Posts: 9014
Joined: Mon May 07, 2007 12:33 pm

Re: Why Dollar Cost Averaging (DCA) is a mental accounting bias and should never be used [vs. lump sum]

Post by tadamsmar » Tue Jun 30, 2020 8:14 am

I was thinking about asking this question on another thread. But I hate the idea of starting another DCA vs LS thread. The question is:

Under what circumstances would you dollar cost average a lump sum into your asset allocation?

I think the question may be a little unclear since I don't think DCA has the same meaning for all. By DCA I mean commit to and follow a rigid schedule, for instance invest 1/12 of the LS each month for 12 months.

But maybe others would do something similar. Please be crystal clear about what you mean by DCA.

Here's my answer: I would never DCA a LS. But I can see myself hesitating to invest a LS immediately in some circumstances. The wiki has a page on managing a windfall: https://www.bogleheads.org/wiki/Managing_a_windfall

Post Reply