Tilting for yield and the bath analogy

[dealtn]

With that said, I also think though, that we should again highlight what you've already said regarding the bath analogy, which is that it's only trying to show the frictional-cost nature of companies paying out dividends and any subsequent re-investment of them, and I think you've now been happy to make clear that it was never intended to cover any 'valuation difference' between any two given baths, where completely different market valuations might be placed on each of them, and which may (as shown in the Credit Suisse report) subsequently go on to play a large role in long-term total returns from investments in each of them.

Yes but what is important is that those valuation differences change over time. You need to be buying something cheap to its relative valuation, and selling when that cheapness disappears, preferably when it has become expensive, and repeating. These changes are small, take a long time to "out", but compound. Buy and Hold, in itself, even if you identify the "cheap" at inception, isn't enough.

So strategies that identify cheapness, and rebalance as that cheapness changes have chances of success. But as has been pointed out cheapness isn't best sought through filtering by dividend yields. There are better alternative filters. But there will be overlap and positive correlation with High Dividend Yields for sure. Correlation isn't causation though.

[AsleepInYorkshire]

With that said, I also think though, that we should again highlight what you've already said regarding the bath analogy, which is that it's only trying to show the frictional-cost nature of companies paying out dividends and any subsequent re-investment of them, and I think you've now been happy to make clear that it was never intended to cover any 'valuation difference' between any two given baths, where completely different market valuations might be placed on each of them, and which may (as shown in the Credit Suisse report) subsequently go on to play a large role in long-term total returns from investments in each of them.

Yes but what is important is that those valuation differences change over time. You need to be buying something cheap to its relative valuation, and selling when that cheapness disappears, preferably when it has become expensive, and repeating. These changes are small, take a long time to "out", but compound. Buy and Hold, in itself, even if you identify the "cheap" at inception, isn't enough.

So strategies that identify cheapness, and rebalance as that cheapness changes have chances of success. But as has been pointed out cheapness isn't best sought through filtering by dividend yields. There are better alternative filters. But there will be overlap and positive correlation with High Dividend Yields for sure. Correlation isn't causation though.

I'm late to the thread and have to be honest that currently my brain isn't in the right place to take all of this in. Sorry, as I'd like to.

When I buy a share I make sure I have a reason for the purchase and an exit strategy. For me that's quite a difficult task and I genuinely find I struggle to buy. Which leaves me defaulting to funds. I have an issue with that as I can't really value the fund and looking at past performance isn't good enough for me.

Recently I purchased a small number of shares in Bellway and Gleeson. Both house builders. The Bellway purchase was based on the book value of the company which was about 0.75 and Bellway's plans for growth. Currently the forward dividend projection of the purchase is around 8% with 3x cover. There are two issue with the fundamentals which I'm monitoring before I top up or simply hold what I've got. Bellway have a legacy issue with cladding (Grenfell Tower). This is impacting on the share price and the margins. However, Bellway have grown their sales by just over 10% this year. In addition their debt has risen by £80m in a year and that makes me feel uneasy. I need to review why the debt has grown. Is it to cover for additional work in progress connected with growth or is it essentially to top up the dividend payment. Forward sales are very robust, over 7,500 at year end and they have a good land bank.

So Bellway was a book value purchase - I didn't allow the dividend to enter my decision to buy as I prefer to look at forward EPS and the companies ability to deliver on that number.

Gleeson's stock price has fallen 20% since my purchase. Two words define that fall - Liz Truss. I never saw that one coming. I'm in a position to double down on my holdings in Gleeson, which are very small, subject to the six monthly news flow. Gleeson's was growth purchase. I believe they are well positioned to grow their volume rapidly. They have no debt and reasonable cash levels. I suspect the market is punishing the share price as 75% of Gleeson's purchasers are first time buyers. Government have withdrawn support for first time buyers of new homes in October and Liz Truss managed to send the mortgage market for first time buyers into turmoil at about the same time. The fundamentals of the business remain intact and I've carried out an acid test on affordability for the Gleeson's product. And at mortgage levels of 6% their business model remains viable and first time buyers will be able to afford their product. Noting the current dividend yield is 4.5% with 5x cover (very approximate - DYOR)

I would begin reviewing the sale of the Bellway stock

1. As the book values rises above 1, subject to momentum
2. If their volume begins to exceed 16-17,000
3. The fundamentals change dramatically and the investment would be better suited elsewhere

I would begin reviewing the sale of the Gleeson stock

1. If their volume reaches 4,000 and appears to have no momentum for further growth
2. The fundamentals change

AiY(D)

[simoan]

Out of interest I have put together a "Dreman Contrarian Value" screen. The screen is based on the top three Dreman contrarian value criteria i.e. Low P/E, Low P/CF and Low P/BV and ignores dividend yield. All companies listed are in the upper tercile of each value criteria.

I have added two sanity checks to the screen: 1) to exclude companies with market caps <£100m; 2) to exclude companies where P/TB is negative or bigger than 1.2, and 2) to only select companies with positive Free Cash Flow in the trailing 12 months (TTM). The results are listed in ascending P/E order below:



The average dividend yield of the 14 shares is 6.17% despite two of the selections not paying dividends. This yield would in itself qualify as a Low P/D, or high dividend yield.

All the best, Si

[tjh290633]

The average dividend yield of the 14 shares is 6.17% despite two of the selections not paying dividends. This yield would in itself qualify as a Low P/D, or high dividend yield.

In fact 3 are not currently paying dividends. MKS might pay a final, or it might not.

At the full-year results in May 2022, we stated that the board would consider the scale and timing of a resumption of dividend payments closer to the year end. Consistent with that announcement, we have not declared a dividend at these results.

from https://www.investegate.co.uk/marks-and ... 00077607F/
Your postulated yield has no basis in fact.

TJH

[simoan]

The average dividend yield of the 14 shares is 6.17% despite two of the selections not paying dividends. This yield would in itself qualify as a Low P/D, or high dividend yield.

In fact 3 are not currently paying dividends. MKS might pay a final, or it might not.

At the full-year results in May 2022, we stated that the board would consider the scale and timing of a resumption of dividend payments closer to the year end. Consistent with that announcement, we have not declared a dividend at these results.

from https://www.investegate.co.uk/marks-and ... 00077607F/
Your postulated yield has no basis in fact.

TJH

As indicated by the name of the column, the dividend yield is calculated on a 12 month rolling basis. This is used to allow better comparison between companies which have different reporting dates. This means there is an element of forecast numbers in the rolling 12 month figure. Your reference is from the recent HY23 results but what the rolling yield is saying is that MKS is currently forecast to make a FY23 dividend payment of 5p. This is not my postulation but is data sourced from Refinitiv, part of the LSE Group. You can take it or leave it , but it doesn't change the key message - dividend yield alone is a poor way to select shares (only the 4th best method for selecting value shares according to Dreman) and these shares often have one or more other popular value criteria which makes it impossible to know if any subsequent outperformance is due to dividend yield alone. I think you know my feelings on the latter.

[mc2fool]

Out of interest I have put together a "Dreman Contrarian Value" screen.

Ok, but I'm not clear what you're suggesting be done with this.

If it's as an alternative to the mechanical DMS high yield strategy published in the CS report that this thread was created about, may I suggest you repost it as a thread of its own, add current prices to the table, and then return to it annually and tot up and report how it's done and redo and rebalance/reinvest it.

[simoan]

Out of interest I have put together a "Dreman Contrarian Value" screen.

Ok, but I'm not clear what you're suggesting be done with this.

If it's as an alternative to the mechanical DMS high yield strategy published in the CS report that this thread was created about, may I suggest you repost it as a thread of its own, add current prices to the table, and then return to it annually and tot up and report how it's done and redo and rebalance/reinvest it.

Perhaps you missed the point I made earlier in the thread i.e. shares with other value criteria.such as Low P/E, Low P/CF & Low P/B will very often have a Low P/D (as shown by the results of the screen). If the DMS high yield approach does not take account of these three factors, which have been shown by Dreman to outperform Low P/D over different time periods, then it is flawed IMHO in exploring the effect of yield alone. I have not read the paper through lack of interest, but thought it worth making that point as found by Dreman.

[spasmodicus]

6 of the stocks on that Dreman based list are financials. Two years ago I suspect that the list might have looked very different. Doesn't this imply a high rate of churn, which might wipe out a significant amount of the alleged advantage for a portfolio based on such criteria?
S

[simoan]

6 of the stocks on that Dreman based list are financials. Two years ago I suspect that the list might have looked very different. Doesn't this imply a high rate of churn, which might wipe out a significant amount of the alleged advantage for a portfolio based on such criteria?
S

I wasn’t proposing it as a portfolio. Apart from anything half the companies have falling earnings. I was just making a point that many high yielding shares also have other value criteria. In this case Low P/E, Low P/CF & Low P/B gives a very strict result. If you drop out one of these factors you see other sectors appear e.g. drop the Low P/CF filter and you see more housebuilders.

[OhNoNotimAgain]

No one wants to ruin a good argument with data, but this might:

https://factsheets.elstonsolutions.co.u ... _Index.pdf

[mc2fool]

Out of interest I have put together a "Dreman Contrarian Value" screen.

Ok, but I'm not clear what you're suggesting be done with this.

If it's as an alternative to the mechanical DMS high yield strategy published in the CS report that this thread was created about, may I suggest you repost it as a thread of its own, add current prices to the table, and then return to it annually and tot up and report how it's done and redo and rebalance/reinvest it.

Perhaps you missed the point I made earlier in the thread i.e. shares with other value criteria.such as Low P/E, Low P/CF & Low P/B will very often have a Low P/D (as shown by the results of the screen). If the DMS high yield approach does not take account of these three factors, which have been shown by Dreman to outperform Low P/D over different time periods, then it is flawed IMHO in exploring the effect of yield alone. I have not read the paper through lack of interest, but thought it worth making that point as found by Dreman.

No, I didn't miss that point but I don't see what a single snapshot screen proves, one way or the other. How are we to know that selection will outperform? How are we to know that the high(er) yielding other shares that have been filtered out by your screen won't collectively outperform? OK, so shares with a low P/E etc may often have a high yield ... well shares with a high yield may often have a low P/E etc!

Note, I'm not criticising your screen, per se, just saying that on its own it proves nothing. If, OTOH, you want to point us at a Dreman (or other) long term study that can be compared to the DMS one then that would be useful. I don't think it's fair to call the DMS study "flawed"; it simply looks at the effect of dividing the market(s) on a single factor, yield. What would be the results of doing so on just P/E? Or just P/B, etc?

However, if it, indeed, can be shown that your multi-factor screen will outperform then that's very useful, as a method kicking up just 14 shares is doable for a private investors, whereas (unless someone puts together an ETF/fund for it) the DMS half-of-the-whole-market method just isn't and must just remain as academic.

BTW, what's the data source for your screen?

[OhNoNotimAgain]

The bath analogy is, in my view, flawed but the tilt is a good point.

Forget the bath and think of the stock market, 350 odd stocks. EMH says that the bulk of your returns come from the market, i.e. beta, not stock selection, known as alpha. We all know lots of examples of fantastic stock pickers who had a great run and then blew up.

So the question is what is the best way of securing the beta, and indeed what is beta?

Jack Bogle invented index funds because he was kicked out of Wellington but wanted to stay in the game and he had the rights to run a fund. So the simplest and cheapest way was to simply replicate the S&P Index. It weights by mkt cap but Bogle, nor anyone else to my knowledge actually asked the question "What is the best way to hold all these stocks?"

Some have argued that one of each is sufficient, but that means stock splits and spin-offs introduce distortions. Others argue that weighting by mkt cap, reflects popularity more than earnings. And, as pointed out by Arnott, growth stocks are often popular and indeed become overpriced.

So the question really is this. if I have to hold 350 "things" how much of each do I hold? But each "thing or company" is very different. A large oil company is vastly different to a niche restaurant and dances to a very different tune and has completely different capital requirements.

In the end there are only five ways to measure a company:

Price
Revenue
Profits
Book value
Cash distributed.

Each measure has its own flaws but if you can get the data then you can use it to calculate a total for the market and then use each company's contribution to the total to determine the weight for each individual company. Of course some companies may have no revenue, or profits, or indeed book value or dividends. So they could get a zero weight.

Price measures popularity and has no direct connection to anything in the accounts. So using that gives a tilt to popularity.

Revenue is a simple measure of size and using that would give a tilt to large companies.

Profits introduce a measure of economic return, but can be highly volatile, subject to revision and can be simply a book profit, no cash involved.

Book value made more sense in a world of factories but has obvious flaws when dealing with companies whose main value is their IP. And we all know of companies where book value has evaporated overnight.

Cash paid out has three virtues. One it is tangible, you can verify it when you cash the cheque in a way you can't with the other measures. Secondly, ity gives you cash to reinvest as you see fit, and not necessarily in the company it came from. Thirdly it tends to favour mature companies that are often labelled as value, so it does give a tilt to value.

So, forget the bath but decide how to tilt your portfolio and what measure to use.

[simoan]

Note, I'm not criticising your screen, per se, just saying that on its own it proves nothing. If, OTOH, you want to point us at a Dreman (or other) long term study that can be compared to the DMS one then that would be useful. I don't think it's fair to call the DMS study "flawed"; it simply looks at the effect of dividing the market(s) on a single factor, yield. What would be the results of doing so on just P/E? Or just P/B, etc?

I know it proves nothing, that in a way is the point I am making i.e. the DMS study proves nothing either if it does not take account of other important value factors such as Low P/E etc. If it doesn't, then I stand by my assertion it is flawed. I think I've inferred from what others have written that it also assumes you buy, hold for a year, sell and then start over? In that case, it is not practical either, so just the usual impractical academic hand waving I got used to reading in my own area of expertise whilst working. If you want to see how something like this may work in practice look at the iShares UK Dividend ETF (IUKD) which selects high yield shares and rebalances twice a year; it has been a terrible performer. Dreman's study of the main four value criteria showed that the best outperformance was found by using Low P/E & Low P/D with smaller capitalisation stocks. We'll see how that goes because I have a number of such shares in my portfolio currently.

However, if it, indeed, can be shown that your multi-factor screen will outperform then that's very useful, as a method kicking up just 14 shares is doable for a private investors, whereas (unless someone puts together an ETF/fund for it) the DMS half-of-the-whole-market method just isn't and must just remain as academic.

Alas, this is not something I'm interested in. I have no interest in the screen and wouldn't touch any of the shares thrown up by it with a bargepole. Although I started out on TMF 23 years ago as a value investor; reading Dreman, enjoying the Value board and reading pyad's early articles etc. it just seemed to me you had to kiss a lot of frogs to make any money, and his idea of "betting the farm" was simply ludicrous from a risk perspective.

BTW, what's the data source for your screen?

Stockopedia.

[mc2fool]

Note, I'm not criticising your screen, per se, just saying that on its own it proves nothing. If, OTOH, you want to point us at a Dreman (or other) long term study that can be compared to the DMS one then that would be useful. I don't think it's fair to call the DMS study "flawed"; it simply looks at the effect of dividing the market(s) on a single factor, yield. What would be the results of doing so on just P/E? Or just P/B, etc?

I know it proves nothing, that in a way is the point I am making i.e. the DMS study proves nothing either if it does not take account of other important value factors such as Low P/E etc. If it doesn't, then I stand by my assertion it is flawed. I think I've inferred from what others have written that it also assumes you buy, hold for a year, sell and then start over? In that case, it is not practical either, so just the usual impractical academic hand waving I got used to reading in my own area of expertise whilst working. If you want to see how something like this may work in practice look at the iShares UK Dividend ETF (IUKD) which selects high yield shares and rebalances twice a year; it has been a terrible performer. Dreman's study of the main four value criteria showed that the best outperformance was found by using Low P/E & Low P/D with smaller capitalisation stocks. We'll see how that goes because I have a number of such shares in my portfolio currently.

However, if it, indeed, can be shown that your multi-factor screen will outperform then that's very useful, as a method kicking up just 14 shares is doable for a private investors, whereas (unless someone puts together an ETF/fund for it) the DMS half-of-the-whole-market method just isn't and must just remain as academic.

Alas, this is not something I'm interested in. I have no interest in the screen and wouldn't touch any of the shares thrown up by it with a bargepole. Although I started out on TMF 23 years ago as a value investor; reading Dreman, enjoying the Value board and reading pyad's early articles etc. it just seemed to me you had to kiss a lot of frogs to make any money, and his idea of "betting the farm" was simply ludicrous from a risk perspective.

BTW, what's the data source for your screen?

Stockopedia.

Uh? Well the whole raison d'etre of this thread was to discuss the DMS study but IIUC, despite having made numerous posts in the thread, you haven't actually read it, and instead you have posted a screen that you're not interested in and wouldn't touch with a bargepole?!?

Oh well....

[dealtn]

The bath analogy is, in my view, flawed but the tilt is a good point.

...

In the end there are only five ways to measure a company:

Price
Revenue
Profits
Book value
Cash distributed.
...

Cash paid out has three virtues. One it is tangible, you can verify it when you cash the cheque in a way you can't with the other measures. Secondly, ity gives you cash to reinvest as you see fit, and not necessarily in the company it came from. Thirdly it tends to favour mature companies that are often labelled as value, so it does give a tilt to value.

So, forget the bath but decide how to tilt your portfolio and what measure to use.

Cash distributed doesn't give you a way to measure the company. It tells you what has been paid out of it (what left the bath to the bucket). It might be useful as a negative measure, though. How much you need to subtract from the assets of the company when distributed.

Cash generated would be a much more useful metric for company valuation. Water leaving the taps and entering the bath.

It is tangible, yes, but it is "yours" not the company's. It gives you money to reinvest, but so what? There are other ways to realise cash to reinvest. You also don't need to reinvest it if it was never distributed - what kind of an investment advantage is that?

There is a link to value, so if value provides a long run advantage then that correlation is useful.

[Itsallaguess]

In the end there are only five ways to measure a company:

Price
Revenue
Profits
Book value
Cash distributed

Each measure has its own flaws but if you can get the data then you can use it to calculate a total for the market and then use each company's contribution to the total to determine the weight for each individual company.

And you seem to prefer looking at the 'Cash distributed' aspect, and weighting accordingly.

But then, in the other thread linked to earlier, you've said -

the BEG, Dimson and Siegel have consistently demonstrated that. Dividends, growth in dividends and reinvested dividends are the main source of equity returns over the long run.

https://www.lemonfool.co.uk/viewtopic.php?f=55&t=36639&start=60#p545804

In this thread we've been discussing a single Dimson and Siegel study that shows that *yield* could be used as an indicator of long-term out-performance, when aligned with a strict re-balancing strategy, but that is not the same as them using 'dividend amounts' as an indicator, as clearly one aspect is in direct relation to valuations and share-prices, whereas the other is simply a weighing-scale based on distribution-levels, and is likely to be highly influenced by the size of a given company.

You seem to prefer a tilt towards 'dividend amounts', but then also seem to wish to highlight data from Dimson and Siegel as a justification to that approach, and I'm struggling to reconcile those two positions with the information we've currently got.

Given that it's clear the Dimson and Siegel data represented in this thread does not give any weight to an argument purely looking at 'dividend distributions', are you please able to point to where they do?

Cheers,

Itsallaguess

[simoan]

Uh? Well the whole raison d'etre of this thread was to discuss the DMS study but IIUC, despite having made numerous posts in the thread, you haven't actually read it, and instead you have posted a screen that you're not interested in and wouldn't touch with a bargepole?!?

Oh well....

There have now been two long discussion threads on DMS and so I understand pretty much what it is and that is the usual academic study with little practical use for investors. If income investors want to use it as a means of feeling better about their approach, that's fine, but please don't tell me it is useful. Once again, you have completely missed (or chosen to ignore) the point of the screen, and I have no idea why you thought the results were of any interest to me. Where did I claim that?

[OhNoNotimAgain]

In the end there are only five ways to measure a company:

Price
Revenue
Profits
Book value
Cash distributed

Each measure has its own flaws but if you can get the data then you can use it to calculate a total for the market and then use each company's contribution to the total to determine the weight for each individual company.

And you seem to prefer looking at the 'Cash distributed' aspect, and weighting accordingly.

But then, in the other thread linked to earlier, you've said -

the BEG, Dimson and Siegel have consistently demonstrated that. Dividends, growth in dividends and reinvested dividends are the main source of equity returns over the long run.

https://www.lemonfool.co.uk/viewtopic.php?f=55&t=36639&start=60#p545804

In this thread we've been discussing a single Dimson and Siegel study that shows that *yield* could be used as an indicator of long-term out-performance, when aligned with a strict re-balancing strategy, but that is not the same as them using 'dividend amounts' as an indicator, as clearly one aspect is in direct relation to valuations and share-prices, whereas the other is simply a weighing-scale based on distribution-levels, and is likely to be highly influenced by the size of a given company.

You seem to prefer a tilt towards 'dividend amounts', but then also seem to wish to highlight data from Dimson and Siegel as a justification to that approach, and I'm struggling to reconcile those two positions with the information we've currently got.

Given that it's clear the Dimson and Siegel data represented in this thread does not give any weight to an argument purely looking at 'dividend distributions', are you please able to point to where they do?

Cheers,

Itsallaguess

Yes, the Barclays Equity Gilt Study consistently, i.e. over several decades, emphasises the importance of reinvested income in both the equity and fixed income markets. It does not refer to yield. I don't recall ever referencing yield as a factor.

The problem with yield is that it is a derived figure, a function of dividend and share price. A low share price boosts the yield but it may only be temporary as the drop in price may presage problems ahead and a possible dividend cut.

Dividends are a primary number, taken straight from the company accounts and totally ignores share price so is not subject to any emotional or popularity effects. Of course it is flawed because dividends can be cut. But the fundamental point remains that:
a) it is reflective of the underlying strength of the company to generate free cash flow
b) those dividends can be reinvested across the asset class to generate additional cash flow.

Dealing with yield, or any valuation measure, means you are trying to assess two factors, one is the factor and the other is sentiment. That leaves the investor struggling to second guess how the market will value that factor. Buffet describes it as trying to judge the judges at a beauty contest. Its not who you think is the prettiest that is important, it is assessing who you think the judges will think the prettiest. That level of complication is irrelevant if you have decided you are going to vote for all the contestants, i.e. hold all the stocks in the asset class. You just need to decide how much to hold.

The attraction of using primary data is that all that subjectivity is removed and you are only dealing with hard, well hardish, information. Once you have decided how to construct your portfolio you then populate it That is when price becomes a factor. If your model says allocate 5% to stock A but the market hates stock A and it only represents 4% of the mkt weight your portfolio will be 1% overweight. And you have a reason for that, not just a feeling in your water.

If the market subsequently decides it now likes Stock A and accords it a 6% weight then your fundamentally weighted portfolio will be 1% underweight. But it will have benefitted from the intervening price appreciation as the market reprices a stock it didn't like to one it does.

[OhNoNotimAgain]

Cash distributed doesn't give you a way to measure the company. It tells you what has been paid out of it (what left the bath to the bucket). It might be useful as a negative measure, though. How much you need to subtract from the assets of the company when distributed.

Cash generated would be a much more useful metric for company valuation. Water leaving the taps and entering the bath.

It is tangible, yes, but it is "yours" not the company's. It gives you money to reinvest, but so what? There are other ways to realise cash to reinvest. You also don't need to reinvest it if it was never distributed - what kind of an investment advantage is that?

There is a link to value, so if value provides a long run advantage then that correlation is useful.

Partially true. As an analyst I always focussed on cash flow. But it is a minefield. It works well on my old sector of mining but applying it to banks and insurance companies is neigh on impossible, as Fred Goodwin proved.

Moreover operating cash flow is fine, but you never know what the company will do with it, it might make an acquisition or commit to a new project that eats cash for years.

More importantly analysts don't publish forecasts of operating cash flow but they do kindly forecast dividends. As with generals in wars, you use what you have, not what you would like.

[Itsallaguess]

Given that it's clear the Dimson and Siegel data represented in this thread does not give any weight to an argument purely looking at 'dividend distributions', are you please able to point to where they do?

Yes, the Barclays Equity Gilt Study consistently, i.e. over several decades, emphasises the importance of reinvested income in both the equity and fixed income markets.

Thanks - I think it would be useful for a thread like this to be able to note specifically what you're referring to here, and just where it's being referred from.

Could you please pull out a short extract from one of the Barclays Equity Gilt Studies that corroborates your view, and please also provide a link to the underlying document for reference?

Cheers,

Itsallaguess