¶ Notes · 2026-07
The Livret A rises to 1.7%. A chance to look at where a yield really comes from
The Livret A rate rises to 1.7% on 1 August 2026. This state-guaranteed figure says something essential, by contrast, about the yields advertised by crypto and DeFi savings, and about what an honest product sheet should show.
# The Livret A rises to 1.7%. A chance to look at where a yield really comes from
On 15 July 2026, the Banque de France recommended raising the rate of the Livret A, and the government followed. As of 1 August 2026, it goes from 1.5% to 1.7%. It is the first increase since 2023, after a long series of cuts.
A quarter of a point is no revolution. But this small figure is one of the most interesting there is for anyone who builds savings products, because it is guaranteed. And once you understand what "guaranteed" means here, you look differently at every other rate, starting with the higher ones advertised by crypto-asset savings products.
Part one, what this 1.7% really says
What the Livret A is
The Livret A is the most widely held savings product in France, a passbook account available at almost every bank. Three features define it. It is state-guaranteed, meaning the capital deposited cannot be lost, at least not in euros, and we will see that the caveat matters. It is liquid, the money is available at any time, with no delay or penalty. And its interest is tax-free, exempt from income tax and social levies. Its deposit ceiling is 22,950 euros.
Why the rate moves, and how it is set
The Livret A rate is not decided by hand. It follows from a regulatory formula, revised twice a year, on 1 January and 1 August. That formula combines, to put it simply, two ingredients measured over the previous six months: inflation (the general rise in prices) and a short-term euro-area reference rate, known as the €STR (the cost at which banks borrow overnight, not only from each other but from a broad range of financial players). The Livret A rate targets, roughly, the average of the two.
The Banque de France computes the result and recommends it, the Minister of the Economy validates it. The Governor may propose to depart from the formula in exceptional circumstances, but the logic remains that of a rate tracking inflation and the price of money. If the rate rises on 1 August 2026, it is therefore mechanically because those two ingredients have stopped falling.
Where this yield actually comes from, and who pays it
The most important question remains, the one we almost always forget to ask: why does the Livret A pay anything at all? An account does not manufacture interest on its own. If it pays, it is because the money deposited does not sleep, it is put to work.
The circuit is as follows. A large share of the deposits, around 60%, is centralized at the Caisse des Dépôts, a public institution, in what is called the savings fund. It lends that money long-term, chiefly to social housing bodies and local authorities, to build and renovate. The rest of the deposits stays on the banks' balance sheets, which use it as a cheap resource to fund their own lending.
These loans and investments earn interest, and that is where the true source of the yield lies. The interest paid by borrowers, social landlords and local authorities, together with the return on investments, funds three things: your 1.7%, the commission paid to the banks that collect the savings, and the running of the scheme. What you receive is your share of what your money earned elsewhere. One mechanism is particularly elegant: loans to social housing are indexed to the Livret A rate itself, so that when your remuneration rises, the cost to those borrowers rises too, which preserves the balance of the system. That mechanism has a flip side, though: the same increase that delights the saver makes social landlords' debt more expensive, and can therefore slow down housing construction. A rising Livret A rate is good news for you, a heavier bill for social housing.
Two clarifications close the subject. First, why exactly 1.7%, and not whatever those loans actually earn? Because this rate is not the mechanical reflection of the return on the assets, it is an administered rate, a political compromise between two goals: protecting the saver against inflation, and keeping that money cheap enough to fund social housing. Second, the point that closes the loop: if those revenues were not enough to cover your remuneration, it is the state guarantee that would step in. The capital is not risk-free because the money would stay warm and idle, it truly works. It is risk-free because the state stands behind it.
What it means, concretely
For the saver, the effect is direct: from 1 August, 10,000 euros invested earn 170 euros a year instead of 150, net of all tax. The same move applies to the LDDS (the sustainable and solidarity development passbook), whose rate is identical and also rises to 1.7%. The LEP (the popular savings passbook), reserved for households under income conditions, stays at 2.5%, a rate deliberately kept above the formula to protect the most modest savers.
A word, finally, on the one real risk the Livret A does carry: inflation. Your capital is guaranteed in euros, not in purchasing power. If prices rise by more than 1.7% a year, your money is protected nominally, but it grows poorer in real terms. It is the exact counterpart of the risk we are about to see on the crypto side: where a crypto product can make you lose capital, the Livret A can make you lose, silently, purchasing power. No yield, anywhere, is ever entirely free.
Hold on to this point, because it is what makes the following comparison illuminating: the Livret A's 1.7% pays a guaranteed deposit, with no counterparty risk. You know where the yield comes from, and you know who guarantees it.
Part two, the corollary on the crypto-asset side
Against this 1.7%, part of the digital savings world advertises far higher yields. The practitioner's question is not "does it pay more", it is where does this yield come from, and what risk does it carry. It is precisely there, in the yield products in regulated finance that I have built, that the honesty of a product sheet is decided.
The stablecoin pays no interest, and that is the law
Let us start with the most common support for these products, the stablecoin. It is a crypto-asset designed to be worth, at all times, one unit of an official currency, most often a dollar or a euro. A well-built stablecoin is always worth about one euro, unlike Bitcoin whose price swings widely. A word of caution right away: "regulated stablecoin" does not mean "all stablecoins". The very largest of all, USDT, is not MiCA-compliant and has been removed from the European Union's regulated venues. Others, such as USDC or Circle's euro token EURC, have instead brought themselves into compliance. The landscape is therefore mixed, and the word "regulated" can never be assumed: it must be checked, issuer by issuer.
Here is what many people ignore: a regulated stablecoin is not allowed to pay you interest. In Europe, the MiCA regulation (Markets in Crypto-Assets, the framework governing crypto-assets in the European Union since 2024) explicitly bars the issuer from paying interest on e-money tokens, in its article 50. In the United States, the GENIUS Act passed in 2025 sets the same ban for payment stablecoins, at its section 4(a)(11).
The consequence is decisive. If a product offers you a yield "on stablecoins", that yield never comes from the stablecoin itself. It comes from a separate layer, added on top: someone takes your stablecoins and does something with them that does earn a return. It is that layer, and not the advertised rate, that you must look at.
One legal point closes the demonstration, and it is the heart of the matter: this ban targets only the stablecoin issuer and the platforms that distribute it, not third parties. Nothing stops another company, or an affiliate, from building a yield product around the stablecoin. That is exactly the space where the advertised yield lives, and it is no accident. In the United States, the banking regulator, the OCC, even proposed in September 2025 to extend the ban to third parties and affiliates: proof that the gap is real, and contested.
What "someone does something with them" means
Two broad families produce this yield.
The first is centralized finance (often shortened to CeFi), a company that holds your funds and lends them to borrowers, or invests them, to pass part of the gain back to you. You trust it like a bank, except it offers neither the state guarantee nor, most of the time, the same supervision.
The second is decentralized finance (DeFi), a set of financial services provided not by a company but by smart contracts (autonomous programs that run on a blockchain and execute the agreed rules on their own, with no bank in the middle). Your funds are, for instance, lent automatically to other users, and the interest they pay makes your yield.
So who puts your money to work, and how? The yield on a crypto product always comes from the same thing: someone takes a risk with your money. The forms change and multiply. Today it might be lending to traders who want leverage, collecting what they pay to hold their positions on futures markets, being paid in freshly minted tokens, or passing on the yield of Treasury bills. Tomorrow it will be something else. But the invariant does not move: each time, a risk is taken, and it is that risk your yield pays for. The only useful question is not the list of mechanisms, it is: which one, and what risk does it make you carry?
And this is where the parallel with the Livret A takes on its full meaning. Here too, your money is put to work. The difference is not there. It lies in who borrows your money, in the soundness of that borrower, and in the fact that no one, here, guarantees your capital.
The three risks that almost never appear in the same font size as the percentage
- Counterparty risk: the risk that whoever your funds were entrusted to, company or protocol, cannot pay you back. On a Livret A, that counterparty is the state. On a crypto product, it is a private company or a piece of software.
- Liquidity risk: the possibility of not being able to recover your funds immediately, because they are lent out, locked, or because too many people want to exit at once. The Livret A, by contrast, is available at any moment.
- Peg risk, the famous depeg: the moment a stablecoin stops being worth the parity it promises. Two cases must be told apart. Algorithmic stablecoins, not backed by real reserves, can collapse: in May 2022, TerraUSD and its sister token LUNA destroyed around 45 billion dollars of value within days, once confidence turned. It is precisely that model that MiCA and the GENIUS Act now prohibit, by requiring 100% backing in liquid reserves. But, and this is the key point, even a regulated stablecoin backed one-for-one can lose its peg: in March 2023, USDC fell to around 0.87 dollar when it emerged that part of its reserves, real though they were, was stuck in a failing US bank, Silicon Valley Bank. It regained its parity once the bank was rescued. Full 100% backing removes the risk of an algorithmic collapse, not the risk of reserves held in the wrong place, nor that of a redemption panic. The guarantee shifts the risk, it does not erase it.
None of these three risks weighs on the Livret A's 1.7%. It carries its own, inflation, but not these. All of them, however, weigh to varying degrees on a higher crypto yield. The extra rate is not a gift, it is the payment for those risks.
What an honest product sheet should show
My conviction, forged in building these products, is simple. A yield advertised without its risk is not information, it is advertising. An honest product sheet would show the risk in the same font size as the percentage: where the yield comes from, who the counterparty is, what happens if it defaults, and whether the capital is guaranteed or not.
The Livret A, with its modest, guaranteed 1.7%, is not "better" than a crypto product. It is simply legible: you know exactly what you hold. That legibility, for that matter, is not the state product's monopoly. Some regulated issuers now publish audited reserves and monthly attestations, at times clearer than the Livret A savings fund circuit is for the average saver. But it remains what you should demand, not what you observe by default. It is that level of legibility, more than the level of the rate, that digital savings must reach to deserve the trust it asks for.
And you, when you are offered a yield, are you shown where it comes from?
---
Sources
- Livret A rate raised to 1.7% as of 1 August 2026 (from 1.5% since 1 February), on the Banque de France's recommendation of 15 July 2026: service-public.gouv.fr, La finance pour tous.
- The Livret A circuit, the setting formula, around 60% centralized at the Caisse des Dépôts and the financing of social housing: economie.gouv.fr, the Livret A circuit.
- Ban on paying interest to stablecoin holders: MiCA regulation (EU) 2023/1114, article 50 for e-money tokens; GENIUS Act 2025, section 4(a)(11) for payment stablecoins in the United States. OCC proposal to extend the ban to third parties, September 2025: Federal Register.
- Collapse of the algorithmic stablecoin TerraUSD, May 2022, around 45 billion dollars of value destroyed (TerraUSD and LUNA): Bitstamp, Terra network collapse.
- USDC depeg, March 2023, around 3.3 billion dollars of reserves stuck at Silicon Valley Bank, fall to about 0.87 dollar then return to parity: CNBC, US Federal Reserve, analysis.