Donate to Remove ads

Got a credit card? use our Credit Card & Finance Calculators

Thanks to johnstevens77,Bhoddhisatva,scotia,Anonymous,Cornytiv34, for Donating to support the site

Home Bias Paradox?

Index tracking funds and ETFs
Hariseldon58
Lemon Slice
Posts: 835
Joined: November 4th, 2016, 9:42 pm
Has thanked: 124 times
Been thanked: 513 times

Re: Home Bias Paradox?

#7971

Postby Hariseldon58 » November 22nd, 2016, 10:31 pm

One doesn't need the benefit of hindsight ... markets are not always right, a private investor doesn't need to make quarterly returns, compare oneself to a benchmark, this is an enormous advantage and you can make occasional investment judgements that allow outperformance.

For those who want an easy life then a the world tracker is great too.

GeoffF100
Lemon Quarter
Posts: 4720
Joined: November 14th, 2016, 7:33 pm
Has thanked: 178 times
Been thanked: 1363 times

Re: Home Bias Paradox?

#8026

Postby GeoffF100 » November 23rd, 2016, 7:23 am

One doesn't need the benefit of hindsight ... markets are not always right, a private investor doesn't need to make quarterly returns, compare oneself to a benchmark, this is an enormous advantage and you can make occasional investment judgements that allow outperformance.


The problem here is that there is abundant evidence that professional investors cannot beat the market in risk adjusted terms, except by chance. On average, they match the market, before costs, and there is no known way of predicting in advance who will do better. I do not know of any evidence that private investors do any better than professional investors. If private investors were winning at professional investors expense, professional investors would be under-performing the market before costs.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#8034

Postby Bubblesofearth » November 23rd, 2016, 8:29 am

GeoffF100 wrote:
The problem here is that there is abundant evidence that professional investors cannot beat the market in risk adjusted terms, except by chance. On average, they match the market, before costs, and there is no known way of predicting in advance who will do better. I do not know of any evidence that private investors do any better than professional investors. If private investors were winning at professional investors expense, professional investors would be under-performing the market before costs.


There is a fundamental difference to the strategies private and professional investors can adopt. One such strategy, familiar to HYP investors for example, is equal weight on purchase followed by long term buy and hold. It would simply not be possible to run a fund along these lines because of the continuous flow of money to and from investors. There is some evidence that this sort of strategy can outperform the market. Any such outperformance may well not show up in overall fund vs market performance data simply because the total amount of money invested in this way is so small.

BofE

GeoffF100
Lemon Quarter
Posts: 4720
Joined: November 14th, 2016, 7:33 pm
Has thanked: 178 times
Been thanked: 1363 times

Re: Home Bias Paradox?

#8073

Postby GeoffF100 » November 23rd, 2016, 10:10 am

One such strategy, familiar to HYP investors for example, is equal weight on purchase followed by long term buy and hold. It would simply not be possible to run a fund along these lines because of the continuous flow of money to and from investors. There is some evidence that this sort of strategy can outperform the market.


There would be no problem running this as a fund. You just invest / dis-invest cash flow to / from investors in proportion to the current values of the stocks.

There is no credible evidence that this strategy consistently outperforms the market in risk adjusted terms.

I have a similar portfolio that I bought when low cost trackers were not available on the British market. It has large locked in capital gains, so I cannot sell it in one go. It has done well for me, but was just good luck. I plan to gradually reduce it by thinning my largest holdings and re-investing in trackers.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#8125

Postby Bubblesofearth » November 23rd, 2016, 12:27 pm

GeoffF100 wrote:
There would be no problem running this as a fund. You just invest / dis-invest cash flow to / from investors in proportion to the current values of the stocks.


This would mean new investor money is not invested according to the equal weighting promise of the fund.

There is no credible evidence that this strategy consistently outperforms the market in risk adjusted terms.


There's not much evidence out there either way. Not surprising given funds are not run along these lines so there is little incentive to back the research. There has been one study showing the performance of the original Dow 30 stocks in which equal weighting on purchase comes out top;

http://www.moneyweb.co.za/archive/what- ... 30-stocks/

But I agree there's not much else apart from the anecdotal HYP's we are mostly familiar with.

All I'm saying is that one shouldn't rule out the possibility that private investors can beat the market based only on the evidence that professionals (read fund managers) cannot do so. Equal weight LTBH is only one difference in what strategy can be adopted. Another would be size. It's hard for large funds to allocate money the way small private investors can. Another would be research capability. It is easier for private investors to focus all their attention on, say, a number of very small cap companies and maybe find inefficiencies in pricing.

BofE

GeoffF100
Lemon Quarter
Posts: 4720
Joined: November 14th, 2016, 7:33 pm
Has thanked: 178 times
Been thanked: 1363 times

Re: Home Bias Paradox?

#8305

Postby GeoffF100 » November 23rd, 2016, 6:50 pm

The HYP is supposed to be an eternity portfolio, so it should not matter when you climb aboard. I do not believe that it is all plausible that the HYP strategy consistently beats the market in risk adjusted terms, when the best investment minds have failed. However, since the initial selection of stocks is subjective, there is no way of back-testing it. It is also worth noting that HYP fans always compare their HYP to the FTSE 100, rather than the world index. They are reducing their risk adjusted return by restricting themselves to UK stocks. They also ignore other asset classes.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#8864

Postby Bubblesofearth » November 25th, 2016, 10:04 am

GeoffF100 wrote:
I do not believe that it is all plausible that the HYP strategy consistently beats the market in risk adjusted terms, when the best investment minds have failed.


But that's the point, never mind HYP, the best investment minds have barely even looked at equal weight LTBH as a strategy let alone tried to implement it.

However, since the initial selection of stocks is subjective, there is no way of back-testing it.


Equal weight LTBH can be back-tested just like any strategy. The Dow 30 study is an example of such back-testing. Pick enough portfolios at random and then back-test them in the same manner as the link and you would increase your level of confidence in the result. To my knowledge this has not been done.

It is also worth noting that HYP fans always compare their HYP to the FTSE 100, rather than the world index. They are reducing their risk adjusted return by restricting themselves to UK stocks. They also ignore other asset classes.


Agreed re HYP but that is tangential to the debate about the merits of equal weight LTBH. Just make sure the benchmarks are appropriate.

Other asset classes reduce risk regardless of the particular strategy you adopt for your equity allocation and is, again, tangential to the debate.

BofE

GeoffF100
Lemon Quarter
Posts: 4720
Joined: November 14th, 2016, 7:33 pm
Has thanked: 178 times
Been thanked: 1363 times

Re: Home Bias Paradox?

#8920

Postby GeoffF100 » November 25th, 2016, 12:13 pm

But that's the point, never mind HYP, the best investment minds have barely even looked at equal weight LTBH as a strategy let alone tried to implement it.


How do you know?

Equal weight LTBH can be back-tested just like any strategy. The Dow 30 study is an example of such back-testing. Pick enough portfolios at random and then back-test them in the same manner as the link and you would increase your level of confidence in the result. To my knowledge this has not been done.


That is not the HYP strategy. They do not randomly select shares.

A market weighted tracker is effectively a LTBH strategy. Some shares will be equally weighted on purchase of the tracker, but nearly all of them will not. Unequal weights increase the volatility of the portfolio, which is not good. Large investors have to accept that drawback. If they all bought equal weights, the large cap stocks would become under-priced and the small cap would become stocks under-priced. If you are a small investor, you can buy equal weights without distorting the market. Nonetheless, if lots of small investors do it, they will distort the market and get unfavourable prices.

mc2fool
Lemon Half
Posts: 7812
Joined: November 4th, 2016, 11:24 am
Has thanked: 7 times
Been thanked: 3017 times

Re: Home Bias Paradox?

#8950

Postby mc2fool » November 25th, 2016, 1:02 pm

GeoffF100 wrote:It is also worth noting that HYP fans always compare their HYP to the FTSE 100, rather than the world index. They are reducing their risk adjusted return by restricting themselves to UK stocks. They also ignore other asset classes.

Actually the most common HYP metric is RPI vs the (change in) income produced. It is, after all, an income generating strategy.

When HYPers look at capital value they compare their HYPs to the FTSE 100 because "standard" HYP selects from the FTSE 100 (although some do dip into the top end of the 250). You can argue that it would be advantageous to include non-UK stocks -- and some do (although often to the side) -- but for those that do choose or want to compare only their UK stocks, a UK index is appropriate (although many have argued it should be the FTSE350HY rather than the 100).

As to ignoring other asset classes, there is nothing in HYP "canon" that says that a HYP is all you need or all you should have, and while, like in everything in life, I'm sure you can find some folks that think that, you can also frequently find HYPers posting on other boards about their international ITs, bond & gilt ETFs, etc. So the assertion that "they", as a group/stereotype, ignore other asset classes doesn't really stack up.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#9036

Postby Bubblesofearth » November 25th, 2016, 5:17 pm

GeoffF100 wrote:

How do you know?


Only by the lack of what is out there. I can't speak for any privately held research.

That is not the HYP strategy. They do not randomly select shares.


I'm not talking about testing HYP. I gave HYP as one example of equal weight LTBH. It is this latter strategy that I propose could be one that beats cap-weighted.

A market weighted tracker is effectively a LTBH strategy. Some shares will be equally weighted on purchase of the tracker, but nearly all of them will not. Unequal weights increase the volatility of the portfolio, which is not good. Large investors have to accept that drawback. If they all bought equal weights, the large cap stocks would become under-priced and the small cap would become stocks under-priced. If you are a small investor, you can buy equal weights without distorting the market. Nonetheless, if lots of small investors do it, they will distort the market and get unfavourable prices.


The drawback with unequal weights on purchase is not increased volatility, it is the likelihood of reduced returns. If markets allowed for diversification benefit then small cap expected returns should actually be slightly lower than for big caps. I realise this is counter intuitive but all it requires is looking at the risk/return profile of portfolios rather than individual shares. Basically, diversification benefit is something small private investors can tap thanks to the huge weight of money going into cap-weight.

BofE

hiriskpaul
Lemon Quarter
Posts: 3852
Joined: November 4th, 2016, 1:04 pm
Has thanked: 681 times
Been thanked: 1489 times

Re: Home Bias Paradox?

#9272

Postby hiriskpaul » November 26th, 2016, 8:06 pm

Bubblesofearth wrote:The drawback with unequal weights on purchase is not increased volatility, it is the likelihood of reduced returns.


Unequal weights do not necessarily lead to increased volatility over equal weights. As a simple counter example, there are low volatility indices and ETFs that have lower volatility than equal weighted indices/ETFs.

I would like to know the basis of your assertion that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns." As a general case I can see no logical reason for claiming that unequal weighting of assets in a portfolio is likely to produce lower returns than equal weighting. In fact, I can think of cases where, with a certain common assumption about constituent returns, unequal weighting may increase expected portfolio return.

If markets allowed for diversification benefit then small cap expected returns should actually be slightly lower than for big caps. I realise this is counter intuitive but all it requires is looking at the risk/return profile of portfolios rather than individual shares. Basically, diversification benefit is something small private investors can tap thanks to the huge weight of money going into cap-weight.


What do you mean by "diversification benefit" and why do you think this only available to private investors?

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#9319

Postby Bubblesofearth » November 27th, 2016, 7:43 am

hiriskpaul wrote:
I would like to know the basis of your assertion that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns." As a general case I can see no logical reason for claiming that unequal weighting of assets in a portfolio is likely to produce lower returns than equal weighting. In fact, I can think of cases where, with a certain common assumption about constituent returns, unequal weighting may increase expected portfolio return.


Market returns are asymmetric with the bulk of gains coming from a relatively small fraction of shares. This also tends to be true for any portfolio that is a sub-set of that market. You can, for example, see it in the evolution of HYP1. If you cap-weight a portfolio on purchase then you are effectively saying you expect greater returns from those companies that you have put the most money into. Or a lower overall risk associated with the portfolio. In reality, returns from big caps have tended to lag small caps whilst portfolio risk can be reduced by selecting smaller companies with weak correlations between them.

What do you mean by "diversification benefit" and why do you think this only available to private investors?


Diversification benefit is simply the reduction in risk provided by diversification.

Imagine there are only two shares available and both had similar risk profiles (volatility). However, one company might generate 10X the earnings of the other*. Based solely on that information you might expect the market cap of the bigger company to be 10X that of the smaller one. However, again based only on the information above, the best possible weighting an individual can give to the two companies shares if selecting them for a portfolio will be 50:50 to maximise the diversification benefit.

Clearly total market weighting between the two companies cannot be 50:50, or even close to it because, that would imply the bigger company trading on a massive discount compared to the smaller one. However, the desire for risk reduction should lead to at least some distortion away from the 10:1 cap ratio that would seem rational based only on consideration of each companies metrics. Assuming the share prices of the two companies are not perfectly correlated, diversification benefit should lead to a small, but measurable, discount for the bigger company. The market would effectively be accepting a lower return in exchange for diversification benefit. But this is not how the bulk of money in the market is invested. Instead it is allocated largely by cap weighting. This leaves something on the table for small investors prepared to equal weight on purchase.

*This is not entirely fanciful, there are many examples of companies with similar volatility but very different market caps.

BofE

hiriskpaul
Lemon Quarter
Posts: 3852
Joined: November 4th, 2016, 1:04 pm
Has thanked: 681 times
Been thanked: 1489 times

Re: Home Bias Paradox?

#9406

Postby hiriskpaul » November 27th, 2016, 3:57 pm

Bubblesofearth wrote:
hiriskpaul wrote:
I would like to know the basis of your assertion that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns." As a general case I can see no logical reason for claiming that unequal weighting of assets in a portfolio is likely to produce lower returns than equal weighting. In fact, I can think of cases where, with a certain common assumption about constituent returns, unequal weighting may increase expected portfolio return.


Market returns are asymmetric with the bulk of gains coming from a relatively small fraction of shares. This also tends to be true for any portfolio that is a sub-set of that market. You can, for example, see it in the evolution of HYP1. If you cap-weight a portfolio on purchase then you are effectively saying you expect greater returns from those companies that you have put the most money into. Or a lower overall risk associated with the portfolio. In reality, returns from big caps have tended to lag small caps whilst portfolio risk can be reduced by selecting smaller companies with weak correlations between them.


Ok, but none of this answers my question as to why you think "The drawback with unequal weights on purchase .. is the likelihood of reduced returns."

I agree that returns from stocks are skewed. 2 out of 3 stocks underperforming the market is a good rule of thumb, but that in itself is of no help in determining a weighting strategy. As for " If you cap-weight a portfolio on purchase then you are effectively saying you expect greater returns from those companies that you have put the most money into." I completely disagree. This is a flawed argument put about by active/smart beta fund managers. If you don't have any further information as to which stocks may outperform and which underperform then it makes no difference to expected portfolio return how you weight the portfolio (but does to risk). Cap weighted happens to be convenient for a number of reasons, in particular it has minimal trading costs and implementations typically have the lowest fund management charges. Absent other information, including considerations of risk, cap weighting is perfectly rational.

If you bring in "big caps have tended to lag small caps" and believe that tendency will persist, then that may be an argument to overweight smaller caps, but that does not automatically imply equal weight. For the UK market (FTSE 350), I overweight small caps by investing 50/50 FTSE 100/FTSE 250. Both the FTSE 100 and FTSE 250 ETFs are still cap weighted though so costs are kept low, but not as low as with a FTSE allshare tracker. Weighting for minimum volatility can be done relatively easily with a number of ETFs available and they certainly seem to reduce portfolio volatility, but again have higher ongoing costs.

Equal weight trackers may have some merits, especially if you think that the small cap and value premiums will persist as at each rebalance point they are buying into both smaller caps and value caps (previously big caps that have fallen out of favour into value territory), but dedicated small cap and value trackers may prove just as good and cheaper to hold. Equal weighting may do strange things to sector weighting as well. The db x-trackers FTSE 100 equal weight ETF has twice the exposure to basic materials companies as the FTSE 100 and many would consider the FTSE 100 to already be overweight the sector.

What do you mean by "diversification benefit" and why do you think this only available to private investors?


Diversification benefit is simply the reduction in risk provided by diversification.


Ok, I thought you might have something else in mind.

Imagine there are only two shares available and both had similar risk profiles (volatility). However, one company might generate 10X the earnings of the other*. Based solely on that information you might expect the market cap of the bigger company to be 10X that of the smaller one. However, again based only on the information above, the best possible weighting an individual can give to the two companies shares if selecting them for a portfolio will be 50:50 to maximise the diversification benefit.


Having the same risk profile does not imply the same volatility, but otherwise agreed. Same historic volatility, no information on earnings growth, debt, etc. Volatility is sticky (past vol is a predictor of future vol) so 50/50 will be likely to be the lowest portfolio volatility and likely to give the best risk weighted return. But purely from an expected return point of view on the information available, 50/50 is no better than cap weight, or for that matter a punt on just one of the stocks.

Clearly total market weighting between the two companies cannot be 50:50, or even close to it because, that would imply the bigger company trading on a massive discount compared to the smaller one. However, the desire for risk reduction should lead to at least some distortion away from the 10:1 cap ratio that would seem rational based only on consideration of each companies metrics.


Well no. For a start you only mentioned a single metric - earnings. Companies are not just priced on earnings, so there is no reason to expect a market cap ratio of 10:1 just because recent earnings (or estimated forward earnings) are in the ratio 10:1. But lets say all the companies metrics and prospects are considered and fair valuations happen to come up with a ratio of 10:1. Are you now saying that a "desire for risk reduction" should lead the market away from a 10:1 valuation? i.e. the ratio should be something like 9:1 or whatever due to a market "desire for risk reduction"? I suppose that might just be possible, but the point is somewhat moot because in reality thousands of stocks are available for purchase. This risk reduction factor, should it exist at all, is likely to be very small and swamped by other factors. I have never come across this and unless you can provide some contrary evidence, I would dismiss it as pure fantasy.

Assuming the share prices of the two companies are not perfectly correlated, diversification benefit should lead to a small, but measurable, discount for the bigger company. The market would effectively be accepting a lower return in exchange for diversification benefit. But this is not how the bulk of money in the market is invested. Instead it is allocated largely by cap weighting.


This does not follow at all and I think you must be misunderstanding how the market works. All of the money in the market is invested by cap weighting, not just "the bulk" of it.

This leaves something on the table for small investors prepared to equal weight on purchase.


No it does not! This "diversification" factor you are talking about is likely to be minimal and even if it does exist, is likely to be swamped by other factors such as small cap and value. Furthermore aside perhaps from the closet trackers, there are many active fund managers who completely ignore the market cap weighting of stocks within their remit. Just take a look at the enormous Woodford fund to see what weighting he gives to banks.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#9553

Postby Bubblesofearth » November 28th, 2016, 8:07 am

hiriskpaul wrote:
Having the same risk profile does not imply the same volatility, but otherwise agreed. Same historic volatility, no information on earnings growth, debt, etc. Volatility is sticky (past vol is a predictor of future vol) so 50/50 will be likely to be the lowest portfolio volatility and likely to give the best risk weighted return. But purely from an expected return point of view on the information available, 50/50 is no better than cap weight, or for that matter a punt on just one of the stocks.


Expected return is the same but risk is not. If 1/3 of stocks drive most returns and you pick a single stock then you have a 2/3 chance of under-performing and only a 1/3 chance of outperforming.

But lets say all the companies metrics and prospects are considered and fair valuations happen to come up with a ratio of 10:1. Are you now saying that a "desire for risk reduction" should lead the market away from a 10:1 valuation? i.e. the ratio should be something like 9:1 or whatever due to a market "desire for risk reduction"? I suppose that might just be possible, but the point is somewhat moot because in reality thousands of stocks are available for purchase. This risk reduction factor, should it exist at all, is likely to be very small and swamped by other factors. I have never come across this and unless you can provide some contrary evidence, I would dismiss it as pure fantasy.


Take a couple of companies we all know about, RB and RDS(A+B). Volatility for these companies is similar but RDS has around 4X the market cap of RB last i checked. Given the high dependence on the oil price and the possibility for a major disaster it would be a brave person that would claim RDS is less risky than RB on other metrics. Now lets also assume, by extrapolation of your claim that expected return is agnostic to allocation, expected return is the same for the two companies. If I decide to invest in these two companies then why should I invest 4X as much in RDS than in RB?

If equal weight on purchase does give a measurable uplift in performance then this should be visible in any long-term data. The fact that you can see it in the original Dow 30 study that I linked is one such data point. I'm not saying that, or the argument I gave above, in any way constitutes proof but it does IMO mean that it cannot be dismissed as pure fantasy. I would be interested in any comment you may have on that link by the way.

BofE

hiriskpaul
Lemon Quarter
Posts: 3852
Joined: November 4th, 2016, 1:04 pm
Has thanked: 681 times
Been thanked: 1489 times

Re: Home Bias Paradox?

#9679

Postby hiriskpaul » November 28th, 2016, 1:57 pm

Bubblesofearth wrote:
hiriskpaul wrote:
Having the same risk profile does not imply the same volatility, but otherwise agreed. Same historic volatility, no information on earnings growth, debt, etc. Volatility is sticky (past vol is a predictor of future vol) so 50/50 will be likely to be the lowest portfolio volatility and likely to give the best risk weighted return. But purely from an expected return point of view on the information available, 50/50 is no better than cap weight, or for that matter a punt on just one of the stocks.


Expected return is the same but risk is not.


Good, so we are in agreement then that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns" is not true?

If 1/3 of stocks drive most returns and you pick a single stock then you have a 2/3 chance of under-performing and only a 1/3 chance of outperforming.


Yes, to reduce unsystematic risk, diversify.

Take a couple of companies we all know about, RB and RDS(A+B). Volatility for these companies is similar but RDS has around 4X the market cap of RB last i checked. Given the high dependence on the oil price and the possibility for a major disaster it would be a brave person that would claim RDS is less risky than RB on other metrics. Now lets also assume, by extrapolation of your claim that expected return is agnostic to allocation, expected return is the same for the two companies. If I decide to invest in these two companies then why should I invest 4X as much in RDS than in RB?


I did not claim a blanket "expected return is agnostic to allocation". That only applies if assets carry broadly the same risk. More risky assets tend to have higher expected returns (but it does not always work out that way as the low volatility anomaly shows).

Volatility of RDSA/B and RB is not in the least similar. Based on daily price movements over the last year, RB has annualised volatility of 19% and RDSA 32%. Just plot price movements of both over the last year - you can easily see the difference in volatility. Ignoring that however, how you allocate between 2 securities would depend on your assessment of valuations, future prospects, what returns you want and what risks you want to take. I don't know enough about either company to make a proper assessment (I buy widely diversified trackers so I don't have to bother). I could be totally wrong as I have carried out no security analysis at all, but my uninformed view is that RB is likely to plod along, but could face upsets if say sufficient customers get fed up paying the inflated prices for some of their products. On the other hand RDS is in a much more risky environment, but the potential rewards are much higher should there be a dramatic rise in oil prices and RDS escapes Deepwater Horizon type incidents. Given this uninformed guesswork, it seems to me If you want to go for the higher risk/potential higher reward option, overweight RDS, if not overweight RB.

In my case, I prefer not to deeply research and assess either option, but just buy them both as part of a globally diversified portfolio of trackers. At the present time that would give me about 0.55% of RDSA+RDSB and 0.15% in RB. I am totally comfortable with those weights within my global equity portfolio. If in 10 years time it turns out I would have been better off with 0.15% in RDSA/B and 0.55% in RB, or better off with the weights in a low volatility tracker, then so be it.

If equal weight on purchase does give a measurable uplift in performance then this should be visible in any long-term data. The fact that you can see it in the original Dow 30 study that I linked is one such data point. I'm not saying that, or the argument I gave above, in any way constitutes proof but it does IMO mean that it cannot be dismissed as pure fantasy. I would be interested in any comment you may have on that link by the way.


Don't kid yourself. With a single data point it could easily be a fluke. Any 2 portfolios are likely to have quite different outcomes when invested over 50 years. Another explanation, other than the to be expected random behaviour, for the disparity in performance between the un-fiddled index and the Dow is that it could be payback for taking on more risk. Taking on more risk can lead to higher returns, but this is not guaranteed (otherwise it would not be a risk). I glanced through the link. My initial reaction with articles like this if I bother to read them at all is to ask myself "What is the author trying to sell?". But ignoring that, I thought this bit says quite a lot in support of the risk hypotheses:

"But this no-fiddling index would have topped out at just over 30,000 in October 2007 and would have finished 2008 at 14,600."

A 52% drop. Over the same period the Dow dropped by 36%, so whatever this no-fiddling index looked like, it carried substantially more risk than the Dow in 2007.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#9750

Postby Bubblesofearth » November 28th, 2016, 5:03 pm

hiriskpaul wrote:
Good, so we are in agreement then that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns" is not true?


If you agree there is a 2/3 chance of under-performance with extreme overweighting in the previous 3-share example then how does that not correspond to a likelihood of reduced returns?





Volatility of RDSA/B and RB is not in the least similar. Based on daily price movements over the last year, RB has annualised volatility of 19% and RDSA 32%. Just plot price movements of both over the last year - you can easily see the difference in volatility. Ignoring that however, how you allocate between 2 securities would depend on your assessment of valuations, future prospects, what returns you want and what risks you want to take. I don't know enough about either company to make a proper assessment (I buy widely diversified trackers so I don't have to bother). I could be totally wrong as I have carried out no security analysis at all, but my uninformed view is that RB is likely to plod along, but could face upsets if say sufficient customers get fed up paying the inflated prices for some of their products. On the other hand RDS is in a much more risky environment, but the potential rewards are much higher should there be a dramatic rise in oil prices and RDS escapes Deepwater Horizon type incidents. Given this uninformed guesswork, it seems to me If you want to go for the higher risk/potential higher reward option, overweight RDS, if not overweight RB.


OK, it appears I'm out of date with my understanding of the relative volatilities of RDS and RB (I'm sure they were roughly the same not that long ago). But from what you've pointed out it would appear that RDS is actually riskier than RB based on volatility. As for believing potential rewards from RDS are higher than for RB that flies in the face of the share price performance over the last couple of decades where RB has utterly trounced RDS. Far from plodding along shares in RB have gone up some 15X from their low in 1999! Not a predictor of future performance but does make believing RDS is likely to provide higher rewards hard to swallow.

But my point is not that RB is a better bet than RDS. It is that there is no rationale, based on risk/reward, for an individual allocating 4X as much money to RDS even though that is what the market as a whole does. And it is exactly what you are doing when you buy your Globally diversified portfolio of trackers. Just because you are comfortable with the weights in your equity portfolio does not mean they represent an optimal allocation.

Don't kid yourself. With a single data point it could easily be a fluke. Any 2 portfolios are likely to have quite different outcomes when invested over 50 years. Another explanation, other than the to be expected random behaviour, for the disparity in performance between the un-fiddled index and the Dow is that it could be payback for taking on more risk. Taking on more risk can lead to higher returns, but this is not guaranteed (otherwise it would not be a risk). I glanced through the link. My initial reaction with articles like this if I bother to read them at all is to ask myself "What is the author trying to sell?". But ignoring that, I thought this bit says quite a lot in support of the risk hypotheses:

"But this no-fiddling index would have topped out at just over 30,000 in October 2007 and would have finished 2008 at 14,600."

A 52% drop. Over the same period the Dow dropped by 36%, so whatever this no-fiddling index looked like, it carried substantially more risk than the Dow in 2007.


You can't assess risk by just selecting one years performance.

Can you point to a single study that shows long-term outperformance of a cap-weighted index vs an equal weight LTBH portfolio of shares taken from that index? I don't have much data but I do at least have some. Are you simply happy to accept the word of those who say you can't beat the (presumably cap-weighted) market? Do they have an agenda (ref you comment on the agenda of those presenting the Dow study)?

BofE

hiriskpaul
Lemon Quarter
Posts: 3852
Joined: November 4th, 2016, 1:04 pm
Has thanked: 681 times
Been thanked: 1489 times

Re: Home Bias Paradox?

#9853

Postby hiriskpaul » November 28th, 2016, 11:05 pm

1nv35t wrote:
hiriskpaul wrote:Good, so we are in agreement then that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns" is not true?

Compared to cap weighted, equal weight tilts more towards smaller cap, which tend to be more volatile and have a compensatory reward premium. There are indications that broadly small or equal weight both tend to outperform cap weight, but not consistently. Cap weighted is more of a outsized bet on a few stocks and individually the average stock underperforms the average. Great if the few stocks you overweight are right tailers, but a greater likelihood they wont be.


Yes this makes more sense. Weighting to small caps would have been a good tactic in the past and given that smaller caps are likely to continue to be higher risk, should be a good tactic in the future. Unless of course the market has now fully incorporated this small cap factor and priced small caps so that they do not outperform in future!

If we assume someone does want to take the risk and overweight small caps, the question then is how to do it? An equal weighted ETF is one way, but that may involve an awful lot of trading at rebalance points. Another way is the one I use, which involves buying cap weighted trackers and then supplementing by cap weighted small cap and small cap value ETFs (and some ITs). I don't know which is best, but I am fairly confident my way is cheaper in terms of both management charges and trading costs.

One advantage of the equal weighted ETF approach is that the regular rebalancing can capture more of the rebalancing bonus that you will not get with cap weight. It then becomes a tradeoff between the rebalancing bonus, which might at times be very small or even negative, and the most definitely always negative trading costs.

hiriskpaul
Lemon Quarter
Posts: 3852
Joined: November 4th, 2016, 1:04 pm
Has thanked: 681 times
Been thanked: 1489 times

Re: Home Bias Paradox?

#9971

Postby hiriskpaul » November 29th, 2016, 12:00 pm

Bubblesofearth wrote:
hiriskpaul wrote:
Good, so we are in agreement then that "The drawback with unequal weights on purchase .. is the likelihood of reduced returns" is not true?


If you agree there is a 2/3 chance of under-performance with extreme overweighting in the previous 3-share example then how does that not correspond to a likelihood of reduced returns?


I thought we had already covered that and you were in agreement! Clearly not, so I will give an illustration. This is not intuitive, so I don't blame anyone for questioning it.

Take 3 stocks A, B and C. After a certain period of time the prices will have changed by factors a, b and c respectively. If you invest with weights Wa, Wb and Wc, where Wa+Wb+Wc=1, then for each pound invested, you will end up with a*Wa+b*Wb+c*Wc. Now lets say A halves in price (a=0.5), B goes up by 50% (b=1.5) and C goes up 300% (c=4), so 2 stocks underperform the average, 1 outperforms, but you have no in advance knowledge of the differences in performance that would give you an edge. What is the optimal weighting (ignoring risk)? The answer is that weighting does not matter.

Lets say you put equal amounts into all 3 stocks. The the return will simply be the arithmetic average return (0.5+1.5+4)/3 = 2. You get £2 back for every £1 invested. Now consider what happens if you pick just one stock. You could have picked A, B, or C and you have no edge, so each pick is equally likely. Expected return is then 0.5/3 (you picked A) + 1.5/3 (you picked B) + 4/3 (you picked C). Net result 2 again. If you picked 2 stocks with equal weight, then you must either have picked (a,b), (a,c) or (c,a). Again equal 1/3 likelihood of each portfolio. The return if you had picked (a,b) is a*Wa+b*Wb = (0.5*0.5+1.5*0.5) = 1. Similarly, the return for (a,c) is (0.5*0.5+4*0.5) = 2.25 and the return for (b,c) is (1.5*0.5+4*0.5) = 2.75. Expected return is then 1/3+2.25/3+2.75/3 = 2 again.

Trying something more complicated weights of (0.2, 0.3, 0.5). With no edge, there are 6 equally likely portfolios that could have been formed with these weights (a,b,c), (a,c,b), (b,a,c), (b,c,a), (c,a,b), (c,b,a). The return from portfolio (a,b,c) is (0.5*0.2+1.5*0.3+4*0.5) = 2.55. Similarly for the others, the returns are in order 2.05, 2.45, 1.75, 1.7 and 1.5. The expected return is than simply the average of the equally likely portfolios, which comes out at 2. Whatever weights you assign to A, B and C, even negative ones indicating you have gone short, the expected return will always be 2 (provided the sum of the weights is 1). It is not hard to mathematically prove that the expected return on a portfolio of stocks is independent of the underlying distribution of returns and independent of stock weights.

When risk is brought in, you could optimise for minimum volatility, but if that is your aim by going for equal weight it will not work. Equal weight would match minimum volatility if all stocks had the same expected returns, volatilities and correlations. In the real world an equal weight portfolio can end up more volatile than the equivalent cap weighted index (you can see it with the S&P, S&P equal weighted indices). An equally weighted portfolio is just a stab in the dark and no better than cap weighted. It will not have a higher expected return, aside from small cap factor, and will require a lot more maintenance to keep in balance.

OK, it appears I'm out of date with my understanding of the relative volatilities of RDS and RB (I'm sure they were roughly the same not that long ago). But from what you've pointed out it would appear that RDS is actually riskier than RB based on volatility. As for believing potential rewards from RDS are higher than for RB that flies in the face of the share price performance over the last couple of decades where RB has utterly trounced RDS. Far from plodding along shares in RB have gone up some 15X from their low in 1999! Not a predictor of future performance but does make believing RDS is likely to provide higher rewards hard to swallow.


So you know better than the market then, based on what has happened in the past? If you are convinced you do (lots of people are) then why bother to equal weight? Optimise your portfolio based on your own convictions on risk/reward. It is very important for us tracker buyers that there are sufficient investors prepared to do that.

But my point is not that RB is a better bet than RDS. It is that there is no rationale, based on risk/reward, for an individual allocating 4X as much money to RDS even though that is what the market as a whole does. And it is exactly what you are doing when you buy your Globally diversified portfolio of trackers. Just because you are comfortable with the weights in your equity portfolio does not mean they represent an optimal allocation.


There is no rationale, based on risk/reward, for buying them equally weighted.

You can't assess risk by just selecting one years performance.


Agreed. And you cannot assess a portfolio building strategy based on a single empirical sample with no rational explanation as to how these excess returns were achieved.

Can you point to a single study that shows long-term outperformance of a cap-weighted index vs an equal weight LTBH portfolio of shares taken from that index?


No, neither can I point to a study on myriad other wacky/irrational ideas for portfolios that someone might dream up. Not any that stand up to scrutiny anyway. I can point to an equally weighted S&P 500 ETF that has beaten the S&P 500 since it was launched though - Guggenheim's RSP. It is more volatile than the S&P, dropping more in down turns and rising faster in rising markets. More trading is required to keep the ETF in balance than with a cap weighted ETF like SPY, but that has clearly not detracted too much from performance. However, smaller caps have tended to outperform larger caps since RSP was launched. If the small cap factor turns negative for a prolonged period, RSP might underperform the S&P.

Someone on the old TMF boards (LookingForClues) chose to buy a broad portfolio of stocks, equally weighted and then simply leave them to do their own thing. I have thought about that and discussed it a few times. In the end think I concluded that there was nothing particularly wrong with his approach, in fact in some ways I quite like it. The only clear downside was that the portfolio will be likely to get increasingly unbalanced/less diversified over time and specific risk will rise. Provided the portfolio was broad enough though it could take many years (longer than LFC's expected holding period) before specific risk became a significant problem.

The only (rational) reasons I can think of as to why such a portfolio might beat a cap weighted tracker are 1) random chance, 2) overweighting of smaller caps, 3) Potentially lower ongoing charges than the tracker. Ignoring 2 and 3, I can see little reason to say the portfolio should not have a 50/50 chance of beating the cap weighted tracker. With 2 and 3 as well, I reckon he has better than even odds. The only other thing that occurred to me that could be a problem is that he is getting less new blood into the portfolio compared to a tracker. This can come from takeovers, or other corporate actions when he gets cash to reinvest elsewhere. A tracker will as well buy into new companies when they enter the index. I have no evidence to suggest this might drag performance, it is just a nagging doubt.

Are you simply happy to accept the word of those who say you can't beat the (presumably cap-weighted) market? Do they have an agenda (ref you comment on the agenda of those presenting the Dow study)?


It depends what you mean. It is factually and logically incorrect to simply say "you can't beat the (cap-weighted) market". Generate a few thousand portfolios with weights randomly distributed about cap weights and I would expect half of those portfolios to beat the market. Generally though I think what people mean is that high charging/high turnover fund managers cannot beat the market over prolonged periods. Or more precisely, the percentage of high charging managers that beat the market will diminish over time. There is plenty of evidence to support that hypotheses, with more being produced every year. If active fund managers charged the same as cap weighted trackers and nothing for trading, the active managers would on aggregate perform the same as cap weighted trackers.

As to whether someone saying "you can't beat the market having an agenda", then yes of course some will.

Bubblesofearth
Lemon Quarter
Posts: 1080
Joined: November 8th, 2016, 7:32 am
Has thanked: 8 times
Been thanked: 432 times

Re: Home Bias Paradox?

#10276

Postby Bubblesofearth » November 30th, 2016, 9:38 am

hiriskpaul wrote:
Now consider what happens if you pick just one stock. You could have picked A, B, or C and you have no edge, so each pick is equally likely. Expected return is then 0.5/3 (you picked A) + 1.5/3 (you picked B) + 4/3 (you picked C). Net result 2 again. If you picked 2 stocks with equal weight, then you must either have picked (a,b), (a,c) or (c,a). Again equal 1/3 likelihood of each portfolio. The return if you had picked (a,b) is a*Wa+b*Wb = (0.5*0.5+1.5*0.5) = 1. Similarly, the return for (a,c) is (0.5*0.5+4*0.5) = 2.25 and the return for (b,c) is (1.5*0.5+4*0.5) = 2.75. Expected return is then 1/3+2.25/3+2.75/3 = 2 again.


Expected return and chance of under-performing are two different things. If I pick a single stock from the 3-stock example you give then I agree expected return will be the same regardless of the stock chosen. However my chance of under-performing this expected return is 2/3 simply because, by your maths as well as my previous assumption, one stock dominates performance. If I pick the right stock then I will outperform by a large margin but if I pick one of the other two I will under-perform.

Basically expected return is unchanged regardless of how I allocate my money across the 3 stocks (this is the bit we agree on) but my risk is dependent on that allocation. With the highest risk being associated with picking just one stock.

Overweighting to any degree needs to be justified along similar lines. If expected return is no different across stocks then it comes down to risk reduction which, I agree will be about correlation between stocks. So when building a portfolio you need to look at different business sectors to minimise this correlation. Cap weighted trackers not only allocate heavily to certain stocks they also tend to overweight certain sectors, a double whammy when it comes to risk.

BofE

hiriskpaul
Lemon Quarter
Posts: 3852
Joined: November 4th, 2016, 1:04 pm
Has thanked: 681 times
Been thanked: 1489 times

Re: Home Bias Paradox?

#10369

Postby hiriskpaul » November 30th, 2016, 12:30 pm

Bubblesofearth wrote:
hiriskpaul wrote:
Now consider what happens if you pick just one stock. You could have picked A, B, or C and you have no edge, so each pick is equally likely. Expected return is then 0.5/3 (you picked A) + 1.5/3 (you picked B) + 4/3 (you picked C). Net result 2 again. If you picked 2 stocks with equal weight, then you must either have picked (a,b), (a,c) or (c,a). Again equal 1/3 likelihood of each portfolio. The return if you had picked (a,b) is a*Wa+b*Wb = (0.5*0.5+1.5*0.5) = 1. Similarly, the return for (a,c) is (0.5*0.5+4*0.5) = 2.25 and the return for (b,c) is (1.5*0.5+4*0.5) = 2.75. Expected return is then 1/3+2.25/3+2.75/3 = 2 again.


Expected return and chance of under-performing are two different things. If I pick a single stock from the 3-stock example you give then I agree expected return will be the same regardless of the stock chosen. However my chance of under-performing this expected return is 2/3 simply because, by your maths as well as my previous assumption, one stock dominates performance. If I pick the right stock then I will outperform by a large margin but if I pick one of the other two I will under-perform.

Basically expected return is unchanged regardless of how I allocate my money across the 3 stocks (this is the bit we agree on) but my risk is dependent on that allocation. With the highest risk being associated with picking just one stock.


Yes, I agree with all of this unless you include the possibility of going short as well, in which case higher risk portfolios can be formed. That does not necessarily mean though that someone might not want to take a higher risk portfolio if their own research and convictions points towards a better risk adjusted outcome.

Overweighting to any degree needs to be justified along similar lines. If expected return is no different across stocks then it comes down to risk reduction which, I agree will be about correlation between stocks. So when building a portfolio you need to look at different business sectors to minimise this correlation. Cap weighted trackers not only allocate heavily to certain stocks they also tend to overweight certain sectors, a double whammy when it comes to risk.


Not necessarily. You can invest in a developed world cap weighted tracker in the UK from about 15bps and one that includes emerging markets from about 20bps. If you go for individual geographic trackers, and include US listed ETFs, you can drive this down even lower. My cap weighted portfolio costs less than 10bps to run and is spread across thousands of companies worldwide, with minimal trading within the funds. The moment anyone starts moving away from cap weight, both management costs and trading costs start to rise. There are trackers that have reduced volatility, but they come with higher costs. So it then becomes a tradeoff between the potentially better risk adjusted returns of the lower volatility portfolio and the lower absolute costs, but unoptimised cap weighted tracker.

Comparing the iShares developed world ETF and iShares world minimum volatility ETF, the cap weighted tracker is overweight financials (17.5%), IT (14.9%), Consumer discretionary (12.5%), health care (12%) and industrials (11.3%). The lowest 3 sectors are telecoms (3.2%), real estate (3.1%) and utilities (3.1%). But the min vol fund over and under weights sectors as well, boosting lower volatility sectors such as health care at (17.5%) and consumer staples (15.2%), at the expense of (unsurprisingly given recent history) materials (2.8%) and energy (1.6%). Aside from materials and energy, the min vol portfolio does look more diversified across sectors, but at present this is paid for in terms of higher P/E and P/B.

It is not obvious to me that the increased running costs of a volatility optimised portfolio is justified, however I do hedge my bets in this respect with a fairly significant holding of the iShares Min Vol USA ETF (USMV) in my SIPP.

As for equal weighted ETFs such as RSP, you end up overweighting according to the number of companies in a sector. Where is the logic in that? In terms of outright performance though I fully admit that the higher volatility RSP has won in capital terms over the last 5 years, with a 100% USD return compared with 91% for the cap weighted SPY and 81% for the low vol USMV.


Return to “Passive Investing”

Who is online

Users browsing this forum: No registered users and 4 guests