Portfolio Part 3: A dividend-growth plan

Where the first two parts left off

Part one was the pandemic basket: one position that returned twelvefold on a thesis I could half-justify, and four more built on similar reasoning that came to nothing. Part two was the damage that did to the shape of the account - individual picks at 62% of the portfolio against the 30-40% I wanted - and the four years in which I barely paid anything in.

Both parts end at the same place. I need a way of deciding what to buy that does not rely on me correctly intuiting, from news coverage, which distressed business is going to survive. This part is about where I found one, the significant way I am bending it, and what the data says about whether it is working yet.

The model I’m borrowing, and where I’m bending it

I have been reading Dividendology’s dividend-growth portfolio, which sets out a simple goal: grow dividend income by 7-10% a year from durable, quality businesses with a record of raising their payouts, then let compounding work over a full market cycle rather than a single year. That is a far better description of what I want from the individual-company side of this account than “buy something cheap after a crash and hope”.

There is a second reason it appeals to me, which follows directly from the Rolls-Royce post-mortem in part one. Whether a company has raised its payout for ten consecutive years is a fact I can look up rather than a judgement I have to form, and it is a reasonable proxy for the durability I was trying to intuit from news coverage in 2020: a business that keeps raising its dividend through a recession has already demonstrated the thing I was guessing at. It lets me keep reasoning about long-term trends, the part I enjoy and am not bad at, while outsourcing the balance-sheet question to a track record rather than my own analysis.

Where I want to bend the model is geography. That approach, and most of the dividend-growth writing I have found, is built almost entirely around US large caps. Given how unpredictable US trade and tax policy has become, I am not comfortable putting new money into a strategy that assumes American companies and American dividend policy will keep behaving as they have for the past decade. I would rather build the same discipline - quality, durability, a growing payout - around businesses and funds in Switzerland, the rest of Europe and the wider world, with the US present through broad global ETFs rather than concentrated single-company bets.

What I keep from the model	What I change
7-10% a year of dividend growth as the target	Geography: Swiss, European and global rather than US-centric
Quality businesses with a record of raising payouts	US exposure arrives through broad ETFs, not single names
Judging progress on income rather than price	ETFs carry more of the load; fewer individual bets
A full market cycle as the horizon	A news-led thesis can start an idea, but a payout record has to confirm it

Looked at that way, the portfolio is already partly there, mostly by accident. The Swiss dividend and mid-cap ETFs exist because I live and am paid here. Novo Nordisk, ASML, SAP and STMicroelectronics are Danish, Dutch, German and Franco-Italian businesses bought for what they do, not for the flag. The Eastern-Europe-ex-Russia fund and a European listed-property ETF were chosen specifically because I wanted European exposure that excluded Russia, which a broad emerging-markets tracker would not have given me. TSMC provides Asian exposure rather than a bet on a US chipmaker - albeit through an American depositary receipt, because a US-listed instrument is the only way my broker offers it.

• Europe (ex-CH, ex-UK)32.7%
• United Kingdom24.7%
• Global funds20.4%
• Switzerland16.0%
• Asia6.2%

Global funds are worldwide trackers, not a single-country bet.

About a third sits in continental Europe, a fifth in globally diversified funds and a sixth in Switzerland. The quarter sitting in the UK is almost entirely Rolls-Royce - a legacy holding, not a template. So the problem to solve is not “too many individual shares”; it is “too much of one country riding on one company”, which is narrower and far more fixable.

Is it working?

The fairest test of a dividend-growth plan is not the portfolio value, which mostly reflects one share price and one decision I made six years ago. It is the income.

CHF, converted at the rate on each payment date. 2026 covers January to September.

For the first five years this account paid me almost nothing: CHF 24 in 2020, rising to a grand total of CHF 162 across the whole of 2024. Then it moves - CHF 774 in 2025, and CHF 1,752 in the first nine months of 2026 alone. That is roughly a tenfold increase in annual income in two years.

Some of that shape is payment schedule rather than growth. SAP and UBS SMIM pay once a year, Novo Nordisk and Rolls-Royce twice, and the Vanguard funds, TSMC and STMicroelectronics quarterly - so a holding bought partway through its own cycle can look like it paid nothing for months and then jump. The 2026 figure is still a floor rather than a final number for the same reason: the quarterly payers have a fourth payment still to come, and Rolls-Royce’s autumn interim - due 18 September on its own published timetable - lands right around when I am writing this, so it may not even be in the export the rest of this post is built from.

I want to be careful about what the tenfold jump proves, because most of it is simply more capital. You cannot deposit CHF 60,000 into dividend-paying assets and not collect more dividends. The real test of dividend growth is narrower: are the payouts per share rising, regardless of how much I bought?

That question has an answer, and it does not depend on my deposits at all.

Last two complete calendar years, in each holding's own currency. Dashed line is the 7% target floor.

Across the ten holdings that paid in both 2024 and 2025 - about three-quarters of the portfolio by value - per-share dividends rose by a value-weighted 7.5%. That lands just inside Dividendology’s 7-10% target, which I did not expect when I started checking.

The spread matters more than the average, though. TSMC raised its payout 36%, the Swiss dividend ETF 20% and Novo Nordisk 18% - the last one while its share price was falling, which is exactly the argument for watching income rather than quotes. The broad global trackers did what broad global trackers do, adding 2-3%. And UBS SMIM cut its distribution by 21%, the single worst result in the portfolio on this measure.

That last one needs a caveat, because on its own it is misleading. SMIM is not a dividend fund and was never bought as one. It tracks the thirty largest Swiss mid-caps, and its distribution is simply whatever those thirty companies happen to pay in a given year, plus the noise of index reconstitution as constituents enter and leave. A 21% swing in that number says very little about the quality of the fund.

Strip out that one cut and the weighted figure becomes 10.8%. I am not going to do that, because choosing which holdings to count after seeing the results is exactly how people fool themselves. 7.5% is the number that counts, even if one of its inputs is measuring something the fund never promised.

Dividends are only half of the scoreboard

Writing all of that made me notice I had started grading the portfolio on one axis. A rising payout is not the point on its own - if it were, the correct move would be to buy the highest yields I could find and watch the capital erode underneath them. What I actually want is both: a business whose payout grows and whose price appreciates, because those two things together are total return, and total return is the only number that eventually buys anything.

So both deserve plotting on the same chart.

Dividend growth is 2024 to 2025 per share. Price return is against my break-even price, so it reflects when I bought.

The top-right quadrant is where the model says everything should be: the payout growing at 7% or better and the price up as well. TSMC sits there most convincingly, and the Swiss dividend ETF and STMicroelectronics keep it company. ASML is off to the right on price and below the line on dividend growth, which is exactly what a compounding business that reinvests heavily tends to look like - and I am not about to complain about it.

The two quadrants on the left are the interesting ones, because they disagree with each other. Novo Nordisk is down 23% on price while raising its dividend 18%, and SAP and the European property fund are in the same shape to a milder degree. That is either the thesis breaking or the market being impatient, and I cannot yet tell which. This is the situation dividend-growth investing exists to help with: the payout is a fact the company controls, while the price is an opinion that can stay wrong for years - though “the market is wrong” is also what everybody says right before they turn out to have been wrong themselves.

And SMIM, alone in the bottom right, is up 11% on price with a falling distribution - the same index-turnover noise flagged above, doing a diversification job rather than a dividend-growth one.

This chart is not a like-for-like comparison, because the price returns run from whenever I happened to buy: ASML’s 146% covers a position built during the April 2025 tariff sell-off, on a deliberate bet on the AI capex cycle and its lithography monopoly, while most of the rest were bought within the last eighteen months. It also leaves out Rolls-Royce entirely, which paid no dividend across 2024 and 2025 and so has no growth figure at all - a 1,000% price return with nothing on the vertical axis would have flattened every other point into the left-hand edge anyway.

That absence is a useful reminder in itself. My single best holding by a wide margin would not appear on a dividend-growth scorecard until last year, which is a fair summary of the gap between what I have actually done and what I say I am now trying to do.

One year is a thin basis for any of this, so I went back further. Taking each holding’s declared dividends per share in 2021 and in 2025, the value-weighted four-year dividend CAGR across 73% of the portfolio is 9.4%. That is a more convincing number than the single-year 7.5%, and it sits comfortably in the 7-10% band.

Holding	Dividend per share, 2021	2025	4-year CAGR
Novo Nordisk	DKK 4.67	11.65	+25.6%
ASML	EUR 3.35	6.56	+18.3%
STMicroelectronics	EUR 0.19	0.31	+12.8%
TSMC	USD 1.87	2.68	+9.3%
iShares Swiss Dividend	CHF 4.78	6.26	+7.0%
Vanguard All-World	USD 1.56	2.01	+6.5%
SAP	EUR 1.85	2.35	+6.2%
Vanguard High Div Yld	USD 1.72	1.98	+3.7%
iShares Eur Property	EUR 0.90	0.88	−0.5%
UBS SMIM	CHF 6.57	4.42	−9.4%

ASML and Novo Nordisk carried that average past the two broad trackers plodding along at 4-6%. SMIM’s negative figure is the same index-turnover noise as before, not a quality signal, and belongs in the table for completeness rather than as a verdict.

I should be clear that I did not own most of these holdings for those four years - I bought the bulk of them in 2025 and 2026. The CAGR describes the companies’ records, not my returns from them. That is deliberate: a dividend record is the evidence I use to decide what to buy, and I wanted to check the businesses I picked actually have one.

There is one more payout worth a closer look, and it is the clearest illustration of yield on cost I own. Rolls-Royce paid its last pre-pandemic dividend in October 2019 and then nothing at all for over five years, before reinstating it in April 2025. It has since declared 10.5p per share for 2025 and 11p for 2026 - already above its pre-pandemic 4p. At today’s price of around £14.50 that is a yield of 0.7%, which looks like nothing. Against my average cost of about £1.04 a share it is a yield on cost of 10.1%. Same dividend, same company, same day - the only difference is when the shares were bought. The company that grew into a quarter of my portfolio is now also paying me for the privilege, which was no part of the original bet.

What all this shows is that the mix has changed character. In 2020 this was a basket of recovery bets, none of which paid anything. Today the income arrives from Novo Nordisk, three dividend ETFs, Rolls-Royce’s restored payout, ASML, SAP and TSMC - thirty separate payments in the last twelve months, worth about CHF 1,870 gross. The current holdings yield roughly 1.3% on today’s value, or about 2.1% on what I paid - modest figures, but moving in the direction I want.

What changes next

The next phase is less about finding another Rolls-Royce and more about changing the portfolio’s shape, because the concentration is now the main risk I carry. Concretely:

• New individual positions need a durable business and a growing dividend, not just a beaten-down share price and a good story. I still expect to find ideas the way I found this one - by reading the news and thinking about where a trend leads - but a dividend record now has to corroborate the narrative before I buy.
• New money defaults to the ETF core and to non-US dividend growers, unless a specific company earns an exception.
• Rolls-Royce: no more buying at these levels, and no plans to sell either. The tax and timing questions there deserve their own post.
• UBS SMIM stays. I nearly wrote that it was on notice after the distribution cut, which would have been a mistake: it is up around 11% and doing the diversification job I bought it for. The fix is to grade it on total return, not on a dividend record it never promised.

Rebalancing by adding, not selling

The obvious way to fix a 24.5% position is to sell some of it. I do not intend to, and the reason is mostly Swiss tax: private capital gains are not taxed here, but that treatment depends on looking like a private investor rather than someone trading professionally. A large, deliberate disposal to rebalance is not something I want to do casually, and the dividends I would give up are taxed as income either way - doing nothing costs me nothing, while selling would cost real tax.

So the plan is dilution. Every franc of new money goes to the underweight side, and the concentration falls because the denominator grows rather than the numerator shrinks. The arithmetic is unsentimental about how long that takes.

Goal	New money required	Years at recent pace
ETF core back to 50%	33,000	~1
Rolls-Royce under 20%	32,000	~1
ETF core to 60%	78,000	~2.5
Rolls-Royce under 15%	91,000	~3

Assuming roughly CHF 30,000 a year and no change in prices - a heroic assumption, since dilution only works if the concentrated position behaves itself, and mine has spent six years not doing that. If Rolls-Royce doubles again, I will be further from the target after another year of full deposits than I am today. At some point that becomes an argument for selling anyway, and I would rather decide where that line is now, calmly, than discover it during a crash.

The reassuring part is that this is a funding problem rather than a strategy problem. I do not need a new idea; I need to keep doing the dull thing for about three years. The rhythm described in part two is what makes that plausible - the monthly drip and two predictable annual lumps are, between them, roughly the CHF 30,000 the table assumes.

The number I will report against next time is per-share dividend growth, not portfolio value. Value is mostly a story about one share price, and about a ceiling I did not predict. Growth in what the holdings actually pay is the part I can influence, and it is the only figure that will tell me whether the plan is working or whether I have just been buying things that happen to go up.

There is one shameless exception to all of this. In June I bought a very small holding in Manchester United, which pays no dividend, has not been a well-run business in years and meets none of my criteria. It is about CHF 270 - a rounding error I am keeping for entirely sentimental reasons. I mention it because a series about becoming disciplined should probably admit where the discipline stops.

Taken together, the three parts describe a portfolio that grew into its current shape rather than one built to a plan, and a plan assembled afterwards to explain it. The pandemic bets created real wealth and real losses in roughly equal measure, the four quiet years cost more than any of the individual mistakes did, and the deliberate buying since 2025 has already started building the core I actually want. Rebalancing isn’t about disowning the decisions that got me here - it’s about making sure the next chapter depends on a plan rather than on how one country’s aero-engine maker happens to perform.