Electrified Paper Is Still Paper

By 1968, daily volume on the New York Stock Exchange had roughly quadrupled in under a decade, and nearly every trade still ended in a physical delivery. Certificates moved by hand and by messenger, and back offices that could not keep up let failed deliveries pile into a backlog measured in billions: over four billion dollars in undelivered securities at the December 1968 peak.

The industry solved that problem, and it solved it in the way that mattered least. It did not abolish the certificate. It decided to stop the certificate from moving.

Everything built since has rested on that decision, and every attempt to get out from under it has had to argue with it. Germany's attempt lost its public scoreboard this February. On 10 February 2026, Section 20 of the Electronic Securities Act was repealed, the provision that had required every issuer of a crypto security to publish it and notify BaFin, which kept a public list of all of them.

The Standortfördergesetz, dated 4 February and published in the Federal Law Gazette on 9 February, struck the section out in Article 26: "§ 20 wird gestrichen." The same law also removed the fines attached to it, and the duty expired the following day.

This is not a story about a law that failed. It is a story about what the law was asked to replace, and why replacing it is harder than writing a statute.

The year the paper won

Back to the New York Stock Exchange in 1968, where the certificates were piling up in the back offices. The exchange's own response shows how bad it was. From Wednesday 12 June 1968, it stopped trading on one day a week so that clerks could work through the paperwork. A market shortened its own week because the filing had beaten it.

By the end of 1970, more than 160 member firms had ceased to exist as independent firms, most of them through mergers. A falling market did part of that. Paperwork they could not process did the rest.

The fix started work in the same month as the four-day week. On 21 June 1968 the exchange's clearing subsidiary began operating the Central Certificate Service, with four issues, and the idea behind it was simple: if the certificates all sat in one vault, a trade no longer had to move one. Ownership would change by an entry in a book, and the paper would stay exactly where it was.

In 1973 that service became the Depository Trust Company. In 1975 Congress ratified the logic, adding Section 17A to the Securities Exchange Act and directing the SEC to "end the physical movement of securities certificates in connection with the settlement among brokers and dealers". End the movement, between dealers. Not the certificate.

That instruction is fifty-one years old this year.

Immobilisation is not dematerialisation

The industry treats these two words as a slow and a fast version of the same thing, and they are not. Immobilisation puts the certificates into one vault and changes ownership by an entry in a book, so the certificate still exists but stops moving. Dematerialisation removes it altogether, and the security exists only as an accounting record.

The first difference is what each one asks of the foundation. Immobilisation fits the architecture that already exists, which is why an exchange's clearing subsidiary could start it in 1968, seven years before Congress wrote anything down. Dematerialisation does not fit: where the law ties a security to a document, the law has to be rewritten before the document can go.

In 1968 the United States took the route the foundation allowed, and Germany had taken the same one long before. The certificate stopped travelling, and every process built around it stayed as it was. Germany's version is the Globalurkunde: a single global certificate for an entire issue, deposited at the central securities depository in Frankfurt, with investors holding book-entry co-ownership shares in it. For many German listed shares, the chain of ownership still ends in a document in a vault, one that nobody ever looks at but that still has to exist.

The American vault is not empty either. In September 2020, DTCC made its own case for finishing the job. Less than one percent of the assets it services remained in physical form, valued at roughly 780 billion dollars, across 188,849 issues and 491,844 certificates, and even after cutting the vault by a third since 2012 it still took in around 80,000 physical deposits a year. It called 98 percent and above of all physical stock certificates a realistic target within three years. Three years from 2020 was 2023. No federal law abolishes the share certificate, though listing rules have required direct-registration eligibility since 2008, and the certificates are still there.

The second difference matters more, and it is the one both routes share. Removing the certificate does not remove what the certificate defined: the unit. A share is a piece of a company that is issued, counted, held and transferred as a piece, and the whole system is built in those pieces. Prices are quoted per share, orders are sized in shares, custody chains reconcile shares, and a dividend is an amount per share. Even a fractional share is a fraction of the piece. An accounting record that books 100 shares is a certificate for 100 shares without the paper, and the unit stays as analog as the paper it came from.

Film went through the same sequence. VHS gave way to DVD and DVD to Blu-ray, and from the DVD onward the film inside was already digital. Yet each format was still a physical unit that had to be pressed, shipped, stocked and handed over, and the film followed the rules of the object that carried it. Only streaming freed the film from that object and from its constraints. Book entries and blockchain tokens are the Blu-ray of capital markets: digital on the inside, still a unit on the outside.

Public markets will become digital when value can be expressed digitally without the unit, so that an investor holds and moves an amount of value rather than a count of pieces. This is what we are working on currently at Digital Marks: digital-marks.com.

What Europe actually legislated

A claim I hear regularly in European fintech is that the Central Securities Depositories Regulation abolished paper securities, with a final deadline in January 2025.

Article 3 of that regulation says something narrower. Its first paragraph requires issuers of transferable securities admitted to trading or traded on trading venues to have them represented in book-entry form, and names two ways to get there: "as immobilisation or subsequent to a direct issuance in dematerialised form". That paragraph applied to new issues from 1 January 2023 and to all of them from 1 January 2025, and the requirement is real, but it is also satisfied by putting the certificate in a vault.

Europe did not outlaw the certificate but accepted the vault as compliance.

The regulation's recitals settle the question of intent, because they say it should not impose one particular method, immobilisation or immediate dematerialisation. The 2025 deadline was a deadline for taking the certificate out of circulation, not for removing it: the 1968 answer, applied across the continent fifty-seven years later, and even the dematerialised route it allowed would still book every holding in units.

The shape, after the object

European cities still turn where medieval field boundaries used to run. The fields are gone, the walls are gone, the reason is gone. The streets remain, because everything built afterwards was built to fit them.

Securities processing has the same property, and it shows up in places that look like pure computing.

Record dates began with companies closing their transfer books so they could count who held the shares, and some cutoff is permanent: a dividend declared for a date needs a list of holders on that date, whatever the ledger. What paper left behind is the lag. In Europe the record date still falls a business day after the ex-date, because the holder list is only reliable once trades have settled, and a separate process of market claims exists to repair the entitlements of everyone caught in between. The United States closed that gap in May 2024, when its move to next-day settlement put the two dates on the same day.

Corporate actions still begin as announcements that someone reads, re-keys, reconciles between intermediaries, and frequently corrects. Standards exist for all of it, ISO 20022 among them, and the Eurosystem has written harmonised corporate-action standards for its own collateral platform. Adoption is uneven, and the announcement still starts life as a document that somebody has to read.

Custody runs in chains. A beneficial owner holds through a broker, which may hold through a custodian, which may hold through a global custodian, which holds at the depository. Cross-border, that can be four links, and the chain began with somebody holding a certificate on somebody else's behalf, and it survives today for better reasons: market access, collateral, credit. What paper left behind is that every link keeps its own record of the same asset, and the records are reconciled against each other rather than read from one place.

Settlement was batched first to move less paper, and it stays batched to move less money. The first reason has gone, while the second is real and I come back to it below. The settlement-cycle argument itself has its own piece: "24-Hour Trading Is the Wrong Answer to the Right Problem".

Then the test that settles the question: France abolished the paper certificate by statute on 3 November 1984, and its record dates and custody chains today look almost exactly like Germany's, where the certificate still sits in a vault. A statute removed the object and the shape stayed, because the shape was never written in law but built into the systems. The distinction between immobilisation and dematerialisation explains how the market got here, but not what it runs on now, which is electrified paper, and electrified paper is still paper.

None of this is a failure of engineering: every one of these processes was a correct solution to a problem that was real when it was designed. They were designed around an object and kept its constraints, and when the object left, the unit it defined stayed behind and carried those constraints forward.

The escape attempts, scored

Germany wrote the most serious law of any major European market. The Electronic Securities Act came into force on 10 June 2021 and created two routes to issue a security with no certificate at all: a central register security, or a crypto security recorded in a forgery-proof recording system, in practice a blockchain, kept by a licensed registrar. It was radical for German law, and it launched without shares: bearer bonds and fund units only. The Zukunftsfinanzierungsgesetz added shares from 15 December 2023, two and a half years later, and bearer shares only through the central register.

Then the adoption numbers. BaFin's list held 253 instruments when the duty ended, and it covered crypto securities alone, not central register securities, and anyone quoting it as the total count of German electronic securities is quoting it wrong. It is still a small number for four and a half years of the most advanced framework in Europe.

The EU's DLT Pilot Regime is the other serious attempt. It allows market infrastructures to operate under exemptions from rules that assume conventional structures. As of ESMA's list dated 28 January 2026, the latest it has published, six have been authorised: CSD Prague, 21X, 360X, Axiology, LISE, and Securitize Europe. Six, across five member states.

And the detail that makes the point better than the count: four of the six operate under an explicit exemption from the CSDR Article 3 book-entry requirement and from the regulation's definition of dematerialised form. To build a market without the old architecture, you need written permission not to comply with the rule that wrote it down.

The European Central Bank went live with Pontes on 21 September 2026, letting DLT platforms settle in central bank money, and has begun preparatory work on using it for its own-funds portfolio. That is the most significant institutional endorsement the field has received. And it settles against the existing legal concept of a security, which is the correct conservative choice and also the limit of what it changes.

Each of these attempts changes where the security is recorded, from vault to register to ledger, and each still issues, holds and settles it in units, which makes the electronic market better without making it digital.

Not the paper, but the unit

Physical paper is not what holds market infrastructure back today. The Giovannini Group reported in 2001 that the vast majority of EU securities were already immobilised or dematerialised, and none of the fifteen barriers it identified was the certificate. When the European Post Trade Forum looked again in 2017, only five of the fifteen had been removed. Withholding tax relief at source blocks cross-border settlement more than any certificate does, and its EU fix applies only from 2030.

A new ledger does not fix much on its own either. The Australian Securities Exchange spent years replacing CHESS, a fully electronic system with no paper left in it, with a distributed ledger. It paused the project on 17 November 2022 and wrote off capitalised software reported at roughly a quarter of a billion Australian dollars, citing solution design, governance and delivery. Faster settlement has its own price as well. Multilateral netting removes the overwhelming majority of the payment value that would otherwise move between participants, and instant gross settlement removes the netting with it, converting credit risk into intraday liquidity risk rather than eliminating it.

What paper left behind is not an obstacle but a blueprint: the unit and the processes built around it, on which half a century of systems now rest, from record dates set days after the trade to custody chains in which every link keeps its own copy of the same holding.

Stopped counting

Germany counted its crypto securities one at a time and then stopped counting. The list was a notification duty, its repeal changes nothing about the legality of electronic securities, and the registrars keep their own records, so the count was never the goal, but it was the only public number we had.

1968 bought the industry half a century with one decision, and it was the right one, because freezing the paper was faster and cheaper than abolishing it and needed far less new law. It worked so well that nobody revisited the premise, and a premise nobody revisits for half a century stops looking like a choice.

Every attempt since has changed how the pieces are held and moved and left the pieces themselves alone. A vault, a register, a ledger or a shorter settlement cycle each makes the electronic market faster or cheaper, the way each new disc improved on the last one. But as long as value is counted in the pieces a certificate once represented, the system underneath stays analog, however digital the processes on top of it become.

Anyone designing public markets from nothing today would not reach for the unit, because no paper would be on their mind to suggest it. They would start from value rather than from pieces of it, and that is the decision the industry still has not made since it stopped moving the certificate in 1968.