Long before ISO 20022, and even before SWIFT, banks moved money across borders over telex. For several decades it was the backbone of international payment instructions, and its limitations are much of the reason the messaging standards that followed look the way they do.
What telex was
Telex, short for “teleprinter exchange,” was a switched network of teleprinters — typewriter-like machines that sent text over telephone-grade lines. It emerged in the 1930s and spread internationally through the 1950s and 1960s. Each terminal had its own telex number, dialled like a phone call, and an answerback code: a short identifier the machine returned automatically so the sender could confirm it had reached the right party. A message arrived as printed text on a roll of paper at the other end. Transmission was slow — on the order of six or seven characters a second — and confined to a restricted set of capital letters, digits, and basic punctuation, so a telex was plain uppercase text and nothing more.
Why banks relied on it
International banking runs on correspondent relationships: a bank holds accounts with banks in other countries and instructs them to make and receive payments on its behalf. Before electronic networks, those instructions travelled by post, cable, or telephone — all slow, or hard to document. Telex was faster than mail, produced a printed record at both ends, and reached almost any bank in the world. For cross-border payments, confirmations, and account statements, it became the default.
How a telex payment worked, and the role of test keys
A telex payment instruction was free-form text: the sender typed who to pay, how much, in what currency, and from which account. The hard part was trust. Anyone could send a telex, so how did the receiving bank know the instruction was genuine and the amount unaltered?
The answer was the test key. Two banks that dealt with each other agreed in advance, and in secret, on a set of test tables. From the key details of a message — typically the amount, currency, date, and a running sequence number — the sender calculated a test number using those tables and included it in the telex. The receiving bank ran the same calculation; if the numbers matched, the message was authentic and the amount intact. It was, in effect, a manual message authentication code, maintained bilaterally between every pair of banks. Because the test was derived from the message’s own figures, altering the amount in transit would break it; to forge an instruction convincingly, a criminal needed the secret tables themselves.
Where telex fell short
The system worked, but it scaled badly. Test keys had to be agreed and maintained between every pair of correspondents, and replaced whenever one was compromised. The messages themselves had no structure: a clerk at the receiving end read the free text and re-keyed it into the bank’s systems, which was slow and error-prone. Transcription mistakes, ambiguous formatting, and the manual test-key process all created risk — including fraud, since a leaked test table could authenticate a forged instruction. Volume compounded the problem: as a bank’s network of correspondents grew, so did the number of test-key relationships it had to administer by hand.
The shift to SWIFT
These were industry-wide problems, and the industry’s answer was a shared utility. In 1973, 239 banks from 15 countries founded SWIFT to build a common, secure messaging network. When it went live in 1977 it introduced two things telex lacked: standardised MT (Message Type) messages with defined fields, and network-level authentication that replaced thousands of bilateral test keys with a single framework. Payments could now be formatted consistently and processed with far less manual handling. Within a few years thousands of institutions had joined, and MT formats such as the MT103 customer payment became the common language of cross-border banking.
The long tail
Telex did not vanish overnight. Banks without SWIFT access, smaller institutions, and certain trade-finance and correspondent flows kept using it well into the 1990s, and national telex services were wound down only gradually over the decades that followed. For a long time a telex line remained the fallback when the modern network was unavailable.
What telex left behind
The weaknesses of telex map almost exactly onto the features of every standard that followed. Free-form text gave way to structured fields; manual re-keying gave way to straight-through processing; bilateral test keys gave way to centralised authentication. SWIFT’s MT messages addressed the first round of these problems, and ISO 20022, today’s standard, pushes further with richer, schema-defined data. (For more, see our guide to ISO 20022.) The history is worth knowing because the problem has not changed: moving money between institutions is, and always was, a question of speed, structure, and trust.