2 Ways to value dividend-paying shares

In my last blog post, I mentioned a few tweaks I'd made to my investment strategy during the summer.

I also covered the changes I’d made to help me focus on what I call “dividend hero stocks”, which are stocks with at least a 20-year track record of growing or maintaining their dividends.

In this post, I’ll explain why I've switched from perpetual growth valuations to exit multiple valuations.

Table of contents

• Method 1: Valuing shares using perpetual growth dividend models
• Method 2: Valuing shares using dividend models with exit multiples

Method 1: Valuing shares using perpetual growth dividend models

From 2011 to 2021, I valued dividend-paying shares using fairly standard ratios. These were PE10 and PD10, which are the ratios of price to ten-year average earnings and dividends respectively. I still use them today to roughly value companies I haven’t looked at in detail, but since 2021, my detailed valuations have been based on discounted dividend models.

If you’re not familiar with discounted dividend models, here’s a good summary from the master of good summaries:

“The value of any stock, bond or business today is determined by the cash inflows and outflows (discounted at an appropriate interest rate) that can be expected to occur during the remaining life of the asset.” - Warren Buffett

In other words, the value of a stock is directly related to the dividends it's expected to pay in future, and that applies even if the company isn't currently paying a dividend.

For a more detailed overview, here are a couple of articles I wrote for my old website in 2021 when I first started using these models:

• How to value shares with a dividend discount model
• The pros and cons of discounted dividend models

For most of the last few years, I’ve used a two-stage perpetual growth dividend model. These have a first stage, where you try to come up with a realistic but conservative estimate of how the company’s earnings and dividends might progress over the medium term, and a second stage where you make a realistic but conservative assumption about the company’s long-term growth rate (and by "long-term" I mean into perpetuity).

There are many ways to build these models, but my approach is based on a simplified model of how companies grow:

1. Companies use capital (raised through equity, borrowing and leasing) to fund factories, warehouses, machinery, computers, inventory and so on.
2. Companies employ that capital in productive activities to generate a return on capital that we call earnings or profit.
3. Some of those earnings are paid to shareholders as dividends or buybacks. The ratio between earnings and dividends is called dividend cover.
4. Earnings not paid out are reinvested into the business, growing the capital base of factories, computers, inventory and so on. 
5. If the now larger capital base produces the same return on capital and if dividend cover remains the same, the company’s earnings and dividends will grow.
6. Repeat those steps, hopefully forever.

To show you what one of these models looks like in practice, I’ll use Schroders as an example because it’s a holding in the UK Dividend Stocks Portfolio.

To set the scene, here’s a chart of Schroders’ growth across equity, revenues, earnings and dividends over the last ten years (note that I focus on equity growth rather than capital growth because companies can grow their capital by taking on more borrowings, and that isn’t the sort of growth I’m looking for).

Data from SharePad

As you can see from the chart, Schroders has grown over the last ten years at a fairly steady rate, although its earnings have struggled to progress since 2016 (some of which is down to the pandemic, but how much is hard to say). On average, Schroders’ grew its dividend by 5% per year over the last ten years, which is good for a mature dividend-paying company.

That's what Schroders' past looks like. As for the future, here’s the perpetual growth dividend model I came up with after Schroders’ 2023 results were published earlier this year.

To the uninitiated that may look like a giant wall of numbers, so I’ll explain the model row by row:

1. Capital: This starts at 293p per share in 2024 and grows over time as earnings are reinvested to expand the capital base of productive assets
2. Net return on capital: In reality, there will be good years and bad years, but we can't know which will be which in advance, so it's easier to assume that returns remain at a historically average level, which for Schroders is 11%
3. Earnings: As capital per share grows, the company invests in buying more computers and other productive assets, as well as acquiring other businesses, so earnings grow at the same rate as capital
4. Dividend cover: This remains steady at 1.5
5. Dividends: With growing capital and earnings and steady dividend cover, the dividend grows progressively from its 2023 level of 21.5p
6. Dividend growth rates: The dividend grows by 3.7% over the medium-term (2024 to 2032). This seems like a sustainable growth rate, so I assume the long-term perpetual growth rate is also 3.7%
7. Fair value and discount: If the above assumptions are 100% correct (which they won’t be), Schroders's shares would generate a fair 7% annualised return (the UK stock market’s long-term average return) if purchased in 2024 at £6.12 and held for the remaining lifetime of the business. This is the estimated fair value. At the time of writing, Schroders' share price was £3.73, so it was trading at a healthy 39% discount to that fair value estimate.

Of course, there are no guarantees that my estimates for net return on capital, dividend cover and fair value are even remotely correct because the future will always be uncertain. Even so, we still have to act in the face of uncertainty, and I think that building dividend models is a sensible way to try to remove at least some of that inherent uncertainty.

Over the years I've gradually made my models more conservative, but recent events at one of my holdings (Hargreaves Lansdown, to be precise) have shown me that any attempt to value stocks based on the assumption of perpetual dividend growth is, to put it politely, a tad optimistic. That’s because, in the real world, few investors will ever hold anything forever, whether they want to or not.

Problems with perpetual growth dividend models

Hargreaves Lansdown, in case you aren’t aware, is in the process of being taken over by a private equity consortium.

HL is, by any reasonable stretch of the imagination, a spectacularly successful company and it more than meets my requirements for steady dividend growth, high returns on capital, a strong balance sheet and an enduring core business.

However, despite those elite-level capabilities and a ten-year average growth rate of almost 10%, HL is being taken private at a PE ratio of just 18, compared to its ten-year average PE of 27. The takeover dividend yield is also just 4%, even though the dividend has grown at an annualised rate of 8% per year since 2014.

I voted against the takeover but, as a minority shareholder, I’m at the mercy of far larger shareholders who (for whatever reason) voted in favour of the takeover, so it will now almost certainly go ahead in the first quarter of 2025.

The takeover price is at the lower end of the stock's ten-year range, so I'll be forced to crystallise a small loss, and this has reinforced my belief that my early valuation models were too optimistic.

Even more importantly, the fact that I'm effectively being forced to sell a company I very much like, at a price I still think is attractive, has made me realise that perpetual growth dividend models are too detached from reality.

While the assumption of perpetual dividend growth is a useful simplification, in reality, there’s a chance you’ll be forced to sell your shares at a price that has little to do with the company's long-term prospects, either because you need the cash (to pay for a divorce, boomerang kids or a Lamborghini) or because institutional investors decide to sell out to a private equity firm just to get an underperforming stock out of their portfolios.

While I was pondering what to do about this during the summer, I remembered some comments from two of my favourite investors:

The Nick Train valuation method

I don’t have the direct quotes to hand, but the first comment came from the previously much-loved and now very much out-of-favour Nick Train. Train essentially said that when his team are valuing a company, they estimate what its earnings might be five or ten years from now and then base their valuation on what they think is a fair PE multiple of those future earnings.

The Ian Lance & Nick Purves valuation method

Around the same time, I heard a similar comment from Ian Lance, one-half of a high-profile old-school UK value investing team. Lance’s approach to valuation is similar to Train’s. Lance essentially says that he works out what the company’s normalised earnings could be (ie with the company operating normally in a normal economic environment) perhaps five years from now, and then he estimates its fair value based on a fair (ie historically average) PE for that business or its peers.

In both cases, these investors are projecting earnings out over the medium term (as I do with my dividend models), but they’re not estimating earnings growth into perpetuity. Instead, they’re using what’s known as an exit multiple to work out what a reasonable share price would be at the end of the medium term. Having done that, they can then use a spreadsheet or calculator to work out if today's share price is attractive or not.

Having looked into the exit multiple approach in a bit more detail, I’m confident that in most cases it does produce more conservative valuations than the perpetual growth method, and it also has several other advantages, which means the exit multiple approach is what I’m going to use from now on.

Method 2: Valuing shares using dividend models with exit multiples

Let’s have another look at my valuation of Schroders, but this time using an exit multiple rather than an assumption of perpetual dividend growth.

The return on capital and dividend cover assumptions are the same as before, so the size and growth of the dividend are identical, but this time I assume the shares are sold after ten years at a historically normal PE ratio (the ten-year average PE), which for Schroders is 17.

In this model, instead of valuing the company based on the assumption that its dividend grows by 3.7% forever, the assumption is that Schroders will be sold in 2033 at a historically average PE (17), giving a 2033 sale price of £7.15.

Discounting those dividends and the exit price back to today using a fair 7% discount rate produces a slightly more conservative estimate of fair value, at £5.23 compared to the previous model's estimate of £6.12. That makes the exit multiple approach 15% more conservative, so the discount to fair value is now smaller at 29% versus the previous model's 39%. Also, having updated all of the dividend models for all of my holdings, I know that’s a fairly typical reduction.

Although I’ve only just started using exit multiple dividend models, I can already see several advantages:

First, as I've just mentioned, exit multiple models tend to be more conservative, and that's good because it will help me avoid overpaying for quality stocks, as I did with Hargreaves Lansdown.

Second, I no longer have to pontificate about what a company’s very long-term growth rate might be. That’s good because it makes my life easier, and anything that makes my life easier is good.

Third, the exit multiple models are far less sensitive than the perpetual growth models. In other words, a small change in the perpetual growth rate assumption can make a big difference to the resulting fair value estimate, and that isn’t the case with exit multiples.

For example, if I change the perpetual growth rate in the old Schroders model, from almost 4% to exactly 5%, its fair value increases from £6.12 to £9.23. That’s a 50% increase in fair value for a mere one percentage point change in the estimated long-term growth rate. In contrast, with the exit multiple model, I’d have to almost double the exit PE from 17 to 29 to increase the fair value estimate by 50%. This is good because it immunises the model, at least to some extent, from any irrational exuberance I may feel towards a company.

As an additional barrier against overly enthusiastic valuations, I’ve brought in a new rule of thumb:

• Rule of thumb: If the model's forward yield is below 2% at fair value, your fair value estimate is too high

This rule is a useful sanity check when you’re assessing a fast-growing company. To use the rule, I’ll build a model using the default assumptions (historically average returns on capital, dividend cover and PE), and if the forward yield at fair value (ie the yield from next year's dividend when the stock is trading at fair value) is below 2%, I’ll make the model more pessimistic until the forward yield is below or equal to 2% at fair value.

In the Schroders model above, the forward yield at fair value is 4.1%, so this isn’t a problem (hence the green highlighting). That doesn't mean the model is correct; it just means it isn't blatantly over-optimistic.

The fourth advantage of exit multiple models is that they’re conceptually easier to deal with.

With perpetual growth models, I always found it hard to swallow the idea that I was making assumptions about a company’s growth rate decades into the future. With exit multiple models, I only have to estimate returns on capital and dividend cover for the next ten years, which is far less daunting, especially as I'm mostly valuing relatively stable progressive dividend stocks. I then assume the shares are sold after ten years at a historically average PE, and that is far easier to wrap my head around than the unrealistic assumption of steady eternal growth.

It’s early days and I’m sure I have much more to learn about building and using these models, but so far I'm confident that exit multiple models are better (at least for me) than perpetual growth models.

If you’d like to try your hand at building an exit multiple model for a company, have a look at the Company Review Checklist as it contains links to a dividend model spreadsheet that you can copy.

In the final instalment of this three-part series, I’ll explain some tweaks I’ve made to my position sizing policy, where I’m attempting to strike a reasonable balance between effort, concentration and diversity.