Amex Is Cutting Credit Limits: Recession Signal or Risk Management?

It's one of those topics that reliably comes back every few years. Someone posts a screenshot where American Express has cut their credit limit from five figures to four. Dozens of similar reports pile up underneath. And within 48 hours the headline is no longer "Amex cut my limit" but "the banks know something we don't".

The logic is seductive because it isn't historically wrong. In 2008 and again in 2020, card issuers did slash credit lines at scale before unemployment numbers turned. Anyone sitting in bank risk management sees payment behaviour in real time, while official statistics arrive weeks late. So the idea that a limit cut is a leading indicator has a real basis.

Except this time, the published numbers say something else. Put side by side, what Amex itself reports, what the Federal Reserve measures, and what the ECB surveys for Europe produce a picture that is considerably less dramatic than the forum version. It's also not the all-clear that the other camp likes to declare.

What is actually documented

Let's start with what can be evidenced, and separate it cleanly from what gets made of it.

Individual cases are evidenced. In the relevant forums — myFICO in the US, Head for Points in the UK — there are 2026 reports from cardholders whose limit was reduced without warning, in some cases by more than 80 percent. Those reports are real, and for the people affected they hurt: in the US, a shrinking denominator suddenly means a much higher credit utilisation ratio, which costs score points without a single dollar of additional spending.

What is not evidenced is everything derived from that as a trend. There is no reporting documenting a systematic, portfolio-wide wave of cuts at American Express in the US or in Germany — not in the trade press, not in regulatory publications. What exists are anecdotes with high emotional reach. That distinction has to be tolerated before anyone starts extrapolating.

What Amex itself reports

On 24 July 2026, American Express published its second-quarter results. If a card issuer is turning defensive because it sees defaults coming, you find it there — in the provisions, in the write-off rate, in the tone of management.

From the official release:

• Revenue net of interest expense of 19.6 billion US dollars, up 10 percent year over year.
• Card Member spending (billed business) of 455.8 billion US dollars, up 9 percent — according to CEO Stephen Squeri, the highest rate in three years on an FX-adjusted basis.
• Provisions for credit losses of 1.1 billion US dollars, against 1.4 billion in the prior-year quarter. The decrease explicitly reflects a reserve release, compared with a reserve build a year earlier.
• Net write-off rate of 2.0 percent, flat year over year.

Squeri states in the release that the company's credit performance "further strengthened". That is investor-relations language, but the reserve release is not: releasing reserves is a balance-sheet statement that you expect fewer losses than previously assumed. An institution that sees a recession rolling toward its own book builds reserves. Amex did the opposite.

On top of that comes the guidance. Amex raised its full-year 2026 revenue growth guidance to 10 percent. That is not a move you make while considering your customer base at risk.

What the shareholder filing says

The press release is the short version. The more interesting document is the quarterly report Amex filed with the SEC the same day, because it states which macroeconomic scenarios the company actually reserves against. That is as close to an official American Express recession forecast as you will get.

Amex publishes a table of the ranges of key variables, as of 30 June 2026 against 31 December 2025. Comparing the two columns is where the story is:

• US unemployment rate, second quarter of 2026: the range was narrowed from 4 to 7 percent down to 4 to 6 percent. The pessimistic end of the near-term labour market scenario was pulled back.
• US unemployment rate, fourth quarter of 2026: unchanged at 4 to 8 percent.
• US GDP growth, fourth quarter of 2026: here it moved the other way. At the end of 2025, the scenario ran from plus 3 to plus 0.5 percent. As of 30 June 2026, it runs from plus 3 to minus 4 percent.

That matters, because it protects the story from a too-comfortable reading. Amex explicitly reserves against a scenario in which US unemployment reaches 8 percent and the economy contracts by 4 percent in the fourth quarter of 2026. The company by no means rules out a recession — it has hung the adverse branch considerably lower than it did six months earlier.

And yet what was actually booked still moved down. The report explains it in a sentence that needs no interpretation: the card balances reserve for credit losses decreased during the quarter, "primarily driven by lower delinquencies". A year earlier the same line item had increased, and there the report names the macroeconomic outlook as the reason.

The reserve movement in detail: opening reserves of 6,065 million US dollars, provisions of 1,017 million, net write-offs of 1,207 million. More was written off than was added — the balance shrinks because reality is running better than the model expected.

And the delinquency numbers underlying that judgement:

• 30+ days past due (consumer and small business): 1.2 percent, against 1.3 percent in the prior-year quarter.
• Net write-off rate, principal only: 2.0 percent, unchanged.
• In the US Consumer Services segment — the classic retail business with Green, Gold and Platinum: 30+ days past due of 1.1 percent against 1.2 percent, and a write-off rate of 1.8 percent against 1.9 percent.

What is equally notable is what the filing does not contain. In a document that has to disclose material credit risk management actions, there is not a word about reducing Card Member credit lines. The only context in which Amex writes about credit lines at all is its own unused funding facilities. That is no proof that no limit anywhere was cut — but it is strong evidence that none of it reaches a scale that would be reportable.

What the Federal Reserve measures

The second place a genuine wave of cuts would have to show up is the aggregate statistics. If issuers strip limits across the board, the sum of all credit card limits in the country falls. That number is measured.

The Federal Reserve Bank of New York published its Quarterly Report on Household Debt and Credit for the second quarter on 11 August 2026. The decisive sentence: aggregate limits on credit cards continued to rise, by 85 billion US dollars, or 1.1 percent.

That is the direct counter-test to the thesis, and the result is unambiguous. In aggregate, more card credit is being extended in the US right now, not less.

The rest of the report fits the picture:

• Total US household debt fell slightly, by 13 billion dollars (0.1 percent), to 18.8 trillion.
• Credit card balances rose by 21 billion to 1.26 trillion dollars.
• 4.7 percent of outstanding debt was in some stage of delinquency at the end of June — 0.1 percentage points below the previous quarter.
• The New York Fed describes transition rates into serious delinquency (90+ days) as largely unchanged; for credit cards, early delinquency transitions were steady as well.

Put differently: the dataset in which a wave of cuts would surface first shows expansion. That does not rule out individual issuers cutting hard in individual customer segments — it rules out that this is currently happening on a scale that would matter macroeconomically.

Where tightening genuinely is happening

This is where it gets interesting, because "all clear" is not the right answer either.

The Federal Reserve surveys senior loan officers at large banks every quarter. In the July 2026 round, a "modest net share" of banks reported tighter standards on credit card loans in the second quarter — after a quarter in which standards had been left essentially unchanged. So something is moving.

But the survey explicitly asks about individual terms as well, including credit limits, minimum required credit scores, and lending to borrowers below scoring thresholds. And there the finding is: banks left most queried terms unchanged. The tightening is happening at the front door — the question of who gets a card at all — not broadly on the limits of people already inside.

The consumer-side view comes from the New York Fed's June 2026 Credit Access Survey: application rates for credit of any kind hit their highest level since October 2021, while the overall rejection rate rose slightly to 16.1 percent — well below the 23.1 percent recorded in June 2025. More people are applying, and on average they are rejected less often than a year ago.

What survives of the tightening thesis is therefore quite precise: the edges are hardening. Subprime feels it first. Anyone holding a premium Amex portfolio sits at the other end of that distribution.

Why a limit gets cut anyway

If it isn't a macro signal, what is it? Almost always a decision about one account, triggered by patterns the issuer sees in its own data. The usual triggers:

Inactivity. A credit line that goes unused costs the issuer regulatory capital and earns nothing. It is the first candidate for a cut — and therefore often trimmed with no deterioration in creditworthiness at all.

Revolving balances. Amex has traditionally been sensitive when a customer starts carrying balances instead of paying in full. That is precisely the behaviour that precedes default in risk models.

Changes in the credit file. New accounts, rising total debt, a late payment at another institution: the issuer sees all of it through routine account review, even when nothing has ever gone wrong with them.

Sudden spending jumps. An atypically large charge can trigger a review — in the US, in the extreme, a Financial Review, where Amex requests income documentation or tax records and freezes the account until it is resolved.

In the US, an issuer must notify you when it lowers a limit based on credit report data, and must state the reasons. That letter is the most useful information you will get in such a case — and the only real starting point for an appeal. Searching a forum for a pattern instead just costs time.

If you operate on the US side, it's worth thinking about this in the context of building US credit history: a limit cut there hits not only your liquidity but, via the utilisation ratio, your score directly.

Germany works differently

This is where most translations of the US debate go wrong — and for German cardholders it's the part that actually matters.

American Express states on its own German site that there is no standardised credit limit on American Express cards ("Bei American Express Kreditkarten gilt kein standardisierter Verfügungsrahmen"). The limit is set individually and managed dynamically. Amex Germany names your current account balance and current spending behaviour, among other factors, as the basis of its assessment.

An uncomfortable consequence follows: a "limit cut" in the sense US forums talk about often never occurs in Germany as a visible event. There is no letter, because there was never a firmly committed number that could have been lowered. The available limit simply moves — and you notice at the till, not in your mailbox.

The practical consequence: before a large purchase, check the limit actively rather than relying on memory. Amex offers a function for this that many cardholders don't know about:

• In the app, via the Abrechnungen (Statements) menu item to "Verfügungsrahmen prüfen" (check available limit).
• In the online account, via Kontoverwaltung (Account Management) and Kartenverwaltung (Card Management) to the same item.
• You enter the intended amount and immediately get an answer on whether it would go through.

Two things matter here: the query is capped at three requests per card within 24 hours, and it triggers no credit inquiry — your Schufa score is untouched. If an amount is declined, the route to manual approval runs through customer service, which for pre-announced large purchases almost always works.

How far the German card system otherwise diverges from the American one is something I've written up in more detail in the US versus German credit cards comparison.

The European finding is the more serious one

And now the part the recession debate should actually be about — except it isn't happening at Amex, but in European banks' consumer credit business.

The ECB runs a quarterly Bank Lending Survey. The July 2026 round, conducted between 15 and 30 June among 159 banks with a 100 percent response rate, shows a net tightening of credit standards for consumer credit of 12 percent in the second quarter of 2026 — after 15 percent the quarter before. What matters is not the individual figure but the series: this tightening has run without interruption since the second quarter of 2022.

Banks name their own lower risk tolerance and higher risk perception as the main drivers. The share of rejected loan applications rose on a net basis — and it rose more for consumer credit than for corporate or housing loans. For the third quarter of 2026, banks expect further tightening across all loan categories.

On the consumer side, that matches what Germany's regulator BaFin describes in "Risiken im Fokus 2026": in 2024, more than ten million new instalment loan contracts were concluded for the first time, a 50 percent increase since 2020, driven mainly by small loans below 1,000 euros. The number of over-indebted consumers rose in 2025 to roughly 5.67 million — 8.16 percent of adults, and the first increase since 2018. Among consumers already over-indebted, average debt from instalment purchases grew by about 31 percent between 2019 and 2024, from 7,414 to 9,692 euros.

That is the point where I take the forum thesis seriously, just in a different place: not among premium cardholders with a Platinum or Centurion, but in the small-loan and buy-now-pay-later segment. That is where tightening is genuinely happening, where payment difficulties are rising, and where the regulator is looking.

Incidentally, the Bundesbank does not forecast a recession for Germany in 2026, but a slow and uneven recovery. You may consider that forecast too optimistic — it simply doesn't support the claim that banks are already steering against a recession they've spotted.

What this means in practice

For cardholders, the resulting to-do list is fairly unspectacular:

A cut is a message about you, not about the world economy. The first look belongs to your own behaviour over recent months — inactivity, carried balances, a new account elsewhere, an atypical spending jump.

Keep cards moving. A card left untouched for twelve months is the most likely candidate for a quiet cut. A regular small charge is enough.

Pay balances in full. With Amex this isn't only an interest question. The switch from transactor to revolver is a signal in the risk models.

Check the available limit before large purchases. In Germany via the function in the app or online account, with no credit inquiry. It replaces the number that German Amex cards never had.

Don't put everything on one card. Spreading spend across several issuers makes you considerably more robust against a single cut or a single Financial Review. How I split mine is in my credit card combination.

In the US, read the letter and appeal. Issuers must state their reasons, and appealing with current income documentation is a normal and often successful route.

Conclusion

The reports of cut Amex limits are real. The conclusion drawn from them does not hold up against the numbers.

Amex itself released reserves rather than building them in the second quarter of 2026, reports an unchanged write-off rate of 2.0 percent, lower delinquencies, and raised its full-year guidance. That the same filing carries a scenario with 8 percent unemployment and a 4 percent GDP contraction only shows that caution and a healthy customer book aren't mutually exclusive — the reserve came down anyway, and it came down because of the observed numbers, not the forecast. In the same quarter, the New York Fed measures aggregate credit card limits up by 85 billion dollars. And the Fed's bank survey states explicitly that limits as a lending term were left largely unchanged, while front-door standards tightened modestly.

What is real: the edges are hardening — in the US in subprime, and in Europe in consumer credit, where the ECB has measured four uninterrupted years of tightening and BaFin documents the first rise in over-indebtedness since 2018. That is a serious matter. It has remarkably little to do with a trimmed Platinum line.

If your limit was cut, the productive question isn't what the bank knows about the economy. It's what the bank sees on your account — and that question can actually be answered.

---

Sources: American Express, Q2 2026 Earnings Press Release, 24 July 2026 · American Express, Form 10-Q for the quarter ended 30 June 2026 (SEC, 24 July 2026), Tables 3.1 and 3.2 · Federal Reserve Bank of New York, Quarterly Report on Household Debt and Credit 2026:Q2, 11 August 2026 · Federal Reserve Bank of New York, SCE Credit Access Survey, June 2026 · Board of Governors of the Federal Reserve System, Senior Loan Officer Opinion Survey, July 2026 · European Central Bank, Euro Area Bank Lending Survey, 21 July 2026 · BaFin, Risiken im Fokus 2026 · Deutsche Bundesbank, forecast for Germany · American Express Germany, credit card limit and available-limit check.