Increase Monthly Investment Every Year. I Didn't, and That Was a Pay Cut I Gave Myself.

"One world ETF, an amount I haven't touched in months." I wrote that a month ago about my own money, and offered it as evidence of good character.

I meant it as a boast. Read it back cold and it is a confession. The amount has not moved in a great deal longer than months, and nothing in my life has stood still in that time. Prices went up. Pay went up. The transfer sat there being admired. You are supposed to increase monthly investment every year, and I had built a system whose main feature was never asking me whether I had.

This is not a clever discovery. It is division, and it has been sitting in my bank app the whole time. But it is the only input in our plan that gets worse on its own while you do nothing, and it is the one number I have been defending by not looking at it.

So here is the line I should have drawn years ago. The fund, the date, the button: set and forget, for ever, and I will argue that with anyone. The amount is different. Automate the buy. Never automate the amount.

Why you have to increase your monthly investment every year

Hold one number still and let the world move around it. That is the entire mechanism, and the size of it is worse than I would have guessed.

A contribution you set in 2020 and never touched buys about 22% less in the US today and about 23% less in the euro area. Put it the other way round, which is the version that stung: to stand still, 100 set in 2020 has to be 128.50 in the US and 129.80 in the euro area now. That is measured inflation, not a model, deflating a constant amount by BLS CPI-U and Eurostat HICP. The 2026 figure is the twelve months to August, because no 2026 annual average exists yet.

Data: US BLS CPI-U and Eurostat HICP, 2020-2026 (2026 = 12 months to August 2026, partial year).

The same thing read as a share of income is the number I actually care about. Take somebody on a clean 10% of net pay in 2020, freeze the transfer, and let their income grow at the rate pay actually grew. By 2026 they are on 7.9% in the US and 8.2% in the euro area. About a fifth of a savings rate, gone, with nobody ever deciding anything.

And 2026 is the year it stopped hiding inside good news. Euro-area inflation ran 3.2% in August, up from 2.0% a year earlier. US consumer prices rose 3.4% over the twelve months to August. Meanwhile pay growth slowed to roughly 3% in both blocs, down from about 5% at the end of 2022. When pay was climbing 5% a year, a stale contribution decayed inside a raise big enough to feel. Now prices and pay move at about the same speed, and the decay arrives with nothing to hide behind.

Which is an uncomfortable place to be standing, because I have already argued at length that the savings rate does more work than the investment returns and that the contribution is the one lever you genuinely control. I believe that more than ever. It also makes letting the lever slip by default a strange thing to have spent years doing.

Almost nobody raises the amount on purpose

Here is the statistic that made me stop writing and go and look at my own standing order.

Vanguard publishes what happened inside nearly five million workplace savings accounts. Not a survey, not a panel: records. In 2025, 45% of those savers' contributions went up, which sounds like half the population is on top of this. The breakdown says otherwise. Only 14% raised the amount themselves. The other 31% went up because an automatic annual escalation did it for them. Another 45% did not change at all, 8% chose to lower it, and 2% went to zero. The year before had the same shape, 16% voluntary against 29% automatic.

Data: Vanguard administrative data, nearly 5 million US retirement-plan savers, 2025 plan year.

One in seven chose it. More than twice as many only moved because a machine moved them — which is not a story about willpower, whatever the comment sections say. Two people set up a monthly purchase in 2020 and both forgot about it for six years, exactly as every sensible article told them to. One was inside a system with an annual step-up; the other was buying an ETF through a broker with a fixed instruction. Their contributions are nowhere near each other now, and neither made a single decision after the first one. That gap is plumbing, not character. Most brokers have no toggle for it, so you get to be your own escalation feature and edit the standing order by hand, once a year, like an animal.

Worth one aside on how different this looks depending on where you get your numbers. Allianz Life runs a quarterly sentiment study, an online survey of about a thousand US adults; last November's edition reported that 51% had stopped or reduced their retirement saving in the previous six months. The five million actual accounts record 8% voluntarily lowering it and 2% stopping. Ask people how their money is going and you get a mood. Read the records and you get behaviour. Mine looked fine as a mood, too.

The two lines that tell you whether you've drifted

This takes about ninety seconds and you need two numbers you already have.

Line one. This month's contribution divided by this month's net income. Line two. The same division, twelve months ago.

If line two is bigger than line one, the difference is the pay cut, and you can price it exactly. Take last year's ratio, apply it to this year's net income, and subtract what you are actually sending.

A worked version, with numbers chosen because they divide easily and not because they are anyone's real pay. Twelve months ago: 400 a month out of 4,000 net, so 10.0%. Today: still 400, but net income is 4,160 after a 4% raise, so 9.6%. Holding the line would mean sending 416. The gap is 16 a month, 192 a year, and it is money that never appeared as a loss anywhere, because nothing was lost. It just stopped being invested. Run the same frozen 400 across the whole 2020 to 2026 window on the measured US figures and it now represents 7.9% of income, with 514 the amount needed to stand still.

If your income bounces around, one month's ratio is noise. The JPMorganChase Institute puts a typical worker's month-to-month change in earnings at 9%, with one month in four bringing a swing of at least 21%, and pay changing in roughly seven of every ten months even when nobody changes job. The fix is boring and it works: use a trailing twelve-month average of net income as your denominator, this year and last. That instability is largely invisible once you average a year, which is why the year is the right unit.

Both lines live wherever your income history already sits: a budgeting app with a year of categorised income in it, or two years of bank statements and a calculator.

"Don't fiddle with your investments" is about the button, not the number

Now the strongest argument against this whole article, which is awkward, because it is mine. I have spent years telling people the cost of touching your investments is real and measured, and I have been my own case study.

It is real. Morningstar's Mind the Gap compares what funds returned with what the money inside them actually earned, and over the decade to the end of 2024 investors kept about 85% of their own funds' returns, giving up roughly 1.2 percentage points a year. I wrote the long version of that in the investor return gap.

Read what the gap is measuring, though. In the 2026 edition the plain large-blend category came out at a gap of precisely zero, investors matching their funds exactly. Target-date investors kept over 98%. The damage concentrates in volatile and exotic things: the least volatile fifth of funds gave up 0.4 points a year, the most volatile fifth more than two points.

Data: category gaps from Morningstar, Mind the Gap 2026 (10 years ended 31 December 2025), as reported in trade coverage of the study. The 1.2-point all-funds figure is the corroborated 2025 edition, for the decade to 31 December 2024.

That is a tax on timing and selection: on when you buy, what you swap into, and whether this week looks frightening enough to stop.

There is nothing in any edition of that study, or anywhere else I could find, that penalises the size of a recurring contribution, and arithmetically there is nothing to penalise, because raising the amount changes neither what you buy nor when. It is a payroll decision with a market side-effect of zero.

So the honest version of the orthodoxy is narrower than the way it gets repeated. The evidence says leave the button alone. It says nothing whatsoever about the number behind the button, and the number is the part that rots. I made the same distinction last month about investing at all-time highs, with the traffic going the other way: a scary valuation print belongs in the return assumption your plan runs on and nowhere near the transfer. Income belongs in the amount. Nothing about the market does.

How much to increase your monthly investment every year, and why a nudge is not it

Most of what is written on this question is not wrong so much as aimed somewhere else. Search how much should I invest every month and you get calculators and starting-point guides, dozens of them, all answering it as a decision you make once. The whole category is about beginning, not maintenance.

The closest thing to real advice I found is Curvo, who tell European readers to review the amount yearly and nudge it by €10 to €25, or 1% to 2%, after a pay rise. Directionally correct, and too small to do the job. If your pay goes up 3.4% and your contribution goes up 1.5%, your savings rate still falls. More slowly, with a better conscience.

So start from the floor rather than from a nudge. Raise the contribution by at least the percentage your net income rose. That holds the ratio. It does not improve it. Anything below it is still a pay cut, just a politer one.

Above the floor there are two defaults with something behind them. Michael Kitces argues for spending half of every raise and saving the other half; in his worked comparison of two people on identical pay, the one who banks half of each rise passes 20% savings within a decade and can stop roughly ten years earlier. The academic version has an experiment attached. Thaler and Benartzi took 207 employees who had just declined advice to save more immediately and offered them a different deal: commit now to putting part of your next pay rise into saving, so your take-home never falls. 78% said yes. Four raises later 80% were still in, and their saving had gone from 3.5% to 13.6% of pay. The 45 who refused both versions sat flat near 6%.

Same people, same advice, different timing. Ask me today and I would wince. Ask me about next March and I will sign.

That is why the useful part of this is not the percentage, it is the writing-it-down-first. Decide the split before the raise exists: a share to the standing order, a share deliberately and visibly spent, and nothing left to willpower on payday. The spent share is not a weakness in the plan, it is the reason the plan survives, and a deliberately savoured upgrade is the opposite of lifestyle creep on autopilot.

If you are certain an increase is unaffordable, the largest test anybody has run says the fear is bigger than the event. UK auto-enrolment stepped the minimum contribution from 2% of pay to 5% in April 2018 and to 8% in April 2019, with the employee's own share going from 0.8% to 4%. That landed on millions of pay slips by law. Nest Insight found little to no impact on behaviour: opt-out and cessation rates stayed very low, and the people most likely to stop were the newest savers, not the ones facing the biggest rise. Nine million people had the amount raised for them and the habit held.

When the right answer is to leave it alone

The diagnostic is for everybody. The increase is not, and I would rather say that plainly than bury it.

The Federal Reserve's household survey found only 63% of US adults could cover a $400 unexpected expense out of cash or a card they would pay off, and only 55% had three months of expenses put aside. If that is where you are, this article is the second problem. Expensive debt outranks it. A thin emergency fund outranks it. A deliberately oversized cash buffer held for a reason you can state is a decision, not a leak, and so is a year when the cost base genuinely moved: a baby, a move, a new dependent, an income that just got less certain.

And in a year with no raise, holding the ratio means holding the amount. Do nothing, and then write down that you decided to do nothing.

That sentence is the whole moral, actually. A number held on purpose is a plan. The same number held because nobody looked at it is what I had.

One date, three numbers

Pick a date that does not move and keep it. Pull three things: net income (trailing twelve months if it wobbles), the current contribution, and last year's written-down ratio. Compute, compare, and change the standing order the same day rather than soon.

Ours lands in January, next to the annual review that is allowed to change the plan, and it will take five minutes of an hour that already exists. In a long flat decade the inputs are the only thing worth maintaining. That was the whole point of the boring middle, written before I noticed one of my own inputs had been drifting through it.

What the drift actually costs, in years

Years, not currency, is the unit that made me act.

Run the standard arithmetic for how long it takes to reach 25 times your annual spending from zero, at a 5% real return. At a 10% savings rate it is about 51 years. At 8% it is about 56. That is 4.7 extra years for a drift nobody chose, and 10% sliding to 8% is precisely what the measured US numbers produced between 2020 and 2026. The same 4.7 years turns up again from 15% down to 12%. It is a projection built on stated assumptions rather than a forecast, and the shape of it does not depend on the assumptions being right.

Projection, not a forecast: years from zero to 25x annual spending, assuming a 5% real annual return, starting from zero, with spending equal to income minus savings. Computed by the author - no external data source.

Four and a half years is a long time to hand over to a division sum nobody did, and it is a cost that never shows up as an event. Nothing goes wrong. No transfer bounces. You just arrive later than the version of you who spent five minutes a year on division. The date your plan ends on is an output, and the contribution is one of the inputs behind it, which is the part people skip when they calculate their FIRE number once and never open the inputs again.

In January I have to pay myself back

I know roughly what the new number is going to be, and I know it is going to feel like too much, because it is not one year of catching up. It is every year I spent being proud of a transfer I had stopped reading.

That is the awkward thing about this whole article. The habit was never the problem. The monthly buy is the one piece of my money I have never been tempted to break, and handing that choice to a standing order is the single best bit of plumbing I own. It just turns out that automating a choice and automating a number are not the same act, and I spent years treating them as one.

So the ritual gets one more line in it, written as a rule so that next-January me cannot negotiate with it: increase monthly investment every year, on a fixed date, against net income. One number, written down so the January after that has something to compare against.

Five minutes, once a year, on two numbers I already have. That is the entire price of not quietly undoing the best decision I have ever made with money.