History of information technologies in banks. From mail to ABS

While working as a programmer at a bank, I became interested in how banks functioned before computers came along, and how much their work changed after. This article will not go into technical details, so it will be accessible to my fellow developers, as well as any other bank employees or those who are simply interested.

Middle Ages

Marinus van Reymerswaele - — Marinus van Reymerswaele – “The Money Changer and His Wife”

Let's start with the Middle Ages. To become a bank client, a person must open an account. Nowadays, to do this, you just need to go to any bank branch or fill out an online application. After receiving an account, you can put money on it and use your funds in any convenient way: withdraw money from an ATM or through the cash desk at the branch, transfer money through the application to another client of your or another bank, spend money by paying for a purchase with a card on the Internet or through a terminal in a store. You have a centralized bank account, information about all your transactions from any source is reflected in it directly and instantly. But it was not always so.

For most of banking history, which dates back to medieval Italy, a customer's account was tied to a specific bank branch, not the bank as a whole. What does that mean?

Let's imagine that you are a medieval merchant and a client of a bank in Venice. You come to Milan and see a great investment opportunity. You need a large sum of money, but you do not have enough cash to complete the transaction. You want to withdraw money from your account by going to your bank's Milan branch. And here comes a problem. All the information about your account is maintained by your bank's Venice branch. The Milan branch does not have information about your account, and may not even know that you are its client. To get this information, the branch must contact the branch that services you in Venice, or the head office, which in turn will contact the right branch and pass on the answer. In the Middle Ages, this could only be done by mail or courier. It would take a courier on horseback about two weeks to get to Venice and back to Milan with your account information.

As for the head office, it might have information about your account, but this information was used only for analytical purposes and for making strategic decisions. This information was received with a long delay, once at a set time, for example, once a week or even less often. The entire burden of recording transactions and keeping the account balances up to date fell on the branches.

New time

The first significant change came with the advent of the electric telegraph and the code for transmitting messages, both invented by Samuel Morse. The speed of information exchange was now limited only by the speed at which telegraph operators typed and decoded messages, as well as the number of telegrams in the queue for sending. Banks were among the first to use the telegraph, and now branches could transmit information to each other not in a few days, but in a few hours. This invention also allowed banks to open branches in hard-to-reach regions of their countries, if a telegraph line was laid there.

To move on to the next invention, let's look at how branches kept track of customer transactions. Each customer had an accounting account – most often a piece of paper – that was kept using the double-entry method. This is a method of keeping accounts invented in the Middle Ages and still used today. It requires that every financial change be recorded in two accounts. Examples:

If the bank gave you money, the change would be reflected in your ledger account as a debit entry and in the bank's ledger account as a credit entry.
If you wanted to transfer money to another client, a credit entry was made in your account and a debit entry was made in your counterparty's account.

In Russian, such changes in two accounts are called simple posting.

All the accounting accounts were stitched together into a ledger – as the name suggests, a large book that contains information about the entire financial state of the bank. The word accounting can be translated as the contents of a book, and the ledger is that very book. The ledger was opened for a year, and at the end of the year a new ledger was opened, into which the balance of accounts at the time of closing the previous ledger was transferred. All subsequent changes in the accounts were reflected in the new ledger, the old ledger was preserved.

During the day, all transactions made were not recorded directly in the ledger, but in separate journals. At the end of the trading day, the data from the journals was transferred to the ledger. The end of the trading day usually occurred in the middle of the working day, for example, at 15:00, and customers had to get to the bank before this time, since after this time all operations stopped.

What did bank employees do? To put it simply, the essence of their work was adding up numbers. What did they count:

For each account, the total amount of debit transactions and credit transactions was calculated.
The sum of all debit transactions from all accounts had to be equal to the sum of all credit transactions from all accounts
If the debit and credit were not equal, then it was necessary to recalculate to find the discrepancies.

As you can imagine, all of this was done manually, on paper and with the help of abacuses. Not only could it take a huge amount of time to calculate all of the bank transactions, but there was a good chance that one of the workers would make a mistake in their calculations.

Calculations became easier with the advent of mechanical calculators. Banks began to use the Burroughs adding machine en masse. As the name suggests, the only function of this machine was the addition operation. Although adding machines had much less functionality than other mechanical calculators, it was enough for most banking tasks. Since the main job of bank employees was to add up debits and credits, the adding machine was easier to use than calculators with a larger set of functions.

Thanks to mechanical calculators, by the beginning of the 20th century, the work of bank employees had become easier: calculations were made faster, and the chance of error became smaller. But calculators are operated by people, which means there is still a chance of making a mistake by entering the wrong number, and the speed of calculations is limited by the speed of people.

First digitalization

Let's consider the main method of non-cash payment in the 20th century – the check. This method was so popular that despite all the digitalization that we will discuss below, in quantitative terms the number of check transactions grew until the early 1990s, and some particularly conservative people use checks even today.

The bank would issue its client a checkbook — a set of blank checks. When the client wanted to transfer money to a counterparty, he would fill out a check with the counterparty's name, the date the check was filled out, the amount, and his signature. The check was given to the counterparty, who could cash it in a few hours at the branch servicing the drawer's account. But most often, the check holder would take the check to a bank convenient for him and receive the money in his account within a few days. During these days, the banks would perform the following operations: check the authenticity of the check, check the sufficiency of funds in the drawer's account, exchange information between the banks, and record the transactions in the ledgers of both banks. This entire process is called check clearing and was done manually until the 1950s.

In the context of the rapidly growing post-war economy, the banking sector was under great pressure in the form of an increasing number of clients and the number of transactions carried out, a significant part of which was occupied by check clearing.

In 1959, a technological revolution occurred when Bank of America became the first bank in the world to use a computer. This computer was called ERMA (Electronic Recording Machine, Accounting) and was used to centralize account records and automate check clearing. It was a bulky device that took up half a room and cost Bank of America $10 million (over $100 million today, adjusted for inflation). To clear checks, bank branches no longer contacted each other but sent checks to ERMA for processing at the end of the trading day. By the beginning of the next trading day, the transaction was executed and reflected in the account.

Check. The numbers at the bottom were read by ERMA

In addition to the greatly reduced processing time, a radical change was that customers' checking accounts were now accounted for by a central computer rather than by branches, whose role had since become less important. Balance and transaction information was stored on the computer and was available bank-wide.

Our days

All future digital banking systems originate from ERMA, and in the 1970s the idea of accounting for checking accounts and automated payment processing by a central computer led to the emergence of a new type of hardware and software, the automated banking systems or ABS (Core Banking System). Unlike ERMA, ABS are not limited to checking operations and, having finally taken shape in the 1980s, automate all the main banking functions: deposit accounting, transactions, loans, customer accounting, bank card accounting, ledger management, and reporting to regulatory authorities. Thus, the ABS and the central computer took over all the main functions that were previously performed by branch employees.

Centralized and automated account accounting has made possible new channels of interaction between the bank and the client: by phone, by credit card, through ATMs, by computer and through an application on a smartphone. Regardless of the channel of interaction, all transactions are ultimately processed by the central computer.

Nowadays, all banks use ABS, and many banks that were the first to implement this technology still use old mainframes with ABS written in COBOL, which speaks to the reliability of these systems. Modern ABS are written in C-like languages and can run not on expensive mainframes, but on more understandable server computers or in the cloud. At the same time, most of the bank's developer staff write code not for ABS, but for channels, since this is faster and safer than making changes to the ABS.

Future

With the spread of the Internet and smartphones, financial services in general and banking in particular are becoming available to a wider number of people, which leads to an increase in the load on the ABS. Traditionally, the ABS is a monolithic application supplied with the mainframe. Despite the historically proven reliability of such a solution, it is less convenient in scaling and is subject to risks during sudden load surges. To solve these problems, the main direction of ABS development is coreless banking. Coreless banking is an ABS written using a microservice architecture, where each function is implemented as a separate application, and not part of one large one. Such an ABS is easier to expand, since changes in one part of the system do not affect others. In addition, Coreless implies deployment in the cloud, which allows you to quickly deploy and scale the ABS, and also removes the burden of maintaining its cluster of computers from banks and allows you to focus on tasks that directly benefit people and business.

Another important area is Open Banking. Open Banking implies the creation of a new channel for interaction with clients by the bank — API. In this case, clients are other companies that access the bank’s functions via API without having to obtain a banking license themselves. Such companies are called neo-banks. Unlike traditional banks, neo-banks are not burdened with the need to support and use old technologies and old channels, such as physical branches or telephone service. Thanks to this, neo-banks can test and launch new products faster, and focus only on digital channels.

Sources: