Core Banking Systems Primer
A comprehensive primer on the evolution of core banking systems, from legacy systems to current trends in modern core banking architecture
written by Roshni Joshi, N26 and Sophie Meralli, Eight Roads Ventures
Introduction
The history of banking in the United States is as old as the history of the country itself, beginning when Alexander Hamilton established the Bank of the United States in 1791 as a depository institution for federal funds.
Given its age, the banking industry has evolved very little in some aspects and significantly in others. At the most basic level, banks continue to offer a secure source for deposit and loan products while developing long-term customer relationships. Even banks’ branch-centric experience did not begin to modernize until the late 1990s and early aughts with the exception of the advent of Automated Teller Machines (ATMs) in the 1960s/70s.
With technological advances in recent decades, customer requirements began to shift. The internet age brought in-home experiences and the mobile revolution brought on-the-go conveniences to retail, travel, education and other industries, and customers desired the same from most major industries. Entrepreneurs and investors took note of changing customer demand for online, personalized, well-designed services and the significant opportunity to expand the banking market by bringing unbanked or underserved customers into the financial system. They began to transform the front-end banking experience with the launch of internet and mobile banks, offering free or low-cost banking products with slick user interfaces to a broader population.
Now, technologists and investors are beginning to understand that the front-end customer experience can only be advanced and enriched further if back-end technologies are brought into the 21st century. The primary focus for back-end evolution in the banking industry lies with the core banking system. This article will discuss the retail bank business model, provide an overview of legacy core banking systems and illuminate legacy cores’ transformation into cloud-based, modularized and open source services.
Section I — Legacy core banking infrastructure
1. Retail bank business model
Before diving into legacy and modern core banking infrastructure, it is helpful to understand the retail bank business model that core banking systems support.
A Bank for International Settlements (BIS) report identifies three bank business models: retail-funded commercial bank (retail bank), wholesale-funded commercial bank and capital markets-oriented bank as defined and categorized by certain balance sheet characteristics of each bank business model. Per BIS, retail banks serve retail customers, as further described herein, and are “…characterized by a high share of loans on the balance sheet and high reliance on stable funding sources including deposits. Customer deposits are about two-thirds of the overall liabilities of the average bank in this group.”
BIS bank business model profiles
Retail banks provide financial products to individual and small business customers. The simplified retail bank balance sheet broadly comprises the following:
Deposit-taking products (liabilities)
- Deposit-taking products include checking, savings and money market accounts, certificate of deposit (CDs), etc.
- Deposits are a liability to the bank as the bank must make deposited funds available to customers on demand
- Banks may offer modest interest rates on deposit products to encourage customers to maintain long-term funds in their accounts that the bank uses to extend loans
Loan products (assets)
- Loan products include credit cards, mortgages, personal loans, student loans, auto loans, etc.
- Loans are an asset to the bank since the bank can borrow deposits to lend funds at higher interest rates than they are paying out on deposits
- Banks earn net interest income on loans. Net interest income is defined as interest earned on loans less interest paid on deposits
Retail banks have four primary revenue sources: interest, investment, interchange and fee revenue. Interest revenue is earned through lending, as described above. Banks may also invest deposits in non-loan assets and earn investment income. Bank investments are highly regulated to ensure financial stability and availability of deposits. Both interest and investment revenues are generated through balance sheet management activity.
Banks earn interchange and fee revenue through transaction activity. Many retail banks issue debit or credit cards that allow them to earn interchange revenue, an incentive payment facilitated by card networks such as Visa and Mastercard, whereby merchants pay the issuing banks a small fee for all purchases made using the banks’ debit or credit card to cover card payment acceptance costs. Banks also charge fees on certain products and services. The list below outlines some of the common fees that retail banks may charge:
- Account maintenance or inactivity fees
- Overdraft fees
- ATM withdrawal fees
- Card order and card replacement fees
- Check issuance fees
- Safety deposit box fees
- Payment related fees (e.g., bounced checks, stop payments, late payments)
- Application fees (for mortgages and other products)
Asset size can further classify retail banks. There are three types of retail banks in the United States based on asset size: community, regional and large/national. The table below outlines the characteristics of each type of retail bank.
Now that we understand the retail bank business model, we can better understand how core banking infrastructure supports retail bank activities.
2. Overview of legacy core banking systems
First, it is important to have a clear understanding of the purpose of core banking systems (“core”, “cores”). This article refers to the Gartner definition, which defines a core as “a back-end system that processes daily banking transactions and posts updates to accounts and other financial records. Core banking systems typically include deposit, loan and credit processing capabilities, with interfaces to general ledger systems and reporting tools.” A bank general ledger records every account of the bank and every transaction (debits and credits) taking place within those accounts.
Cores serve as the bookkeeping tool for all accounts that reside at the bank and the money movement or transactions within them; essentially, a technological replacement for a written bank general ledger. Celent research aptly describes the core as “… the primary system of record for the accounts of a bank, [that] thus forms the technological foundation on which the entire bank operates.”
Software is added-on to the core to support bank functionality that relies on the underlying account information — e.g., loan management, payments management and financial reporting. Add-ons must be compatible with the core, given the reliance on the core to support additional products and reporting. Historically, this meant that most banks were using a single enterprise software suite from their core provider or a select software suite compatible with their core for all operations and product management. The diagram below illustrates a simplified retail banking legacy tech stack.
Celent’s research provides a clear overview of legacy cores’ critical retail banking functionality (inclusive of add-ons). As demonstrated by the diagram, the legacy core comprises the general ledger functionality while the core-compatible add-ons support product management for traditional retail bank products such as deposits, loans and customer relationship management.
In addition, the table below outlines some of the legacy core and add-on offerings available in the market today, catering to retail banks across the spectrum of community, regional and national banks. The US core banking market is highly concentrated, with Finastra, FIS, Fiserv and Jack Henry holding 95% of core banking market share as recently as 2017. (Metz-Pasquier et al.)
3. Characteristics of legacy cores
While all legacy core systems are unique in their architecture and offerings, the evolution of cores can be traced from early mainframe computing in the 1960s, to server-based architecture in the 1990s, to cloud computing today, as per the timeline below:
Beginning in the late 1950s, and especially during the 1960s, banks adopted the new technology of IBM mainframes utilizing Common Business-Oriented Language (COBOL) programming language to handle back-office processing. Mainframe computing remained the industry standard through the advent of the more modern server-based architecture models. Even today, many of the world’s largest banks rely on mainframe core infrastructure. Below we outline the benefits and restrictions of mainframe and server-based legacy cores in the modern world:
Benefits of legacy systems
- Scale. Mainframes can process significantly larger amounts of data vs. servers and may make it easier to scale monolithic cores. IBM Z13 mainframes can process 2.5B transactions per day vs. ~26M transactions per day on a typical server. To date, no large bank has successfully implemented a cloud-based core at scale, which means there are no current market examples to prove out that cloud infrastructure can offer the same security and reliability as legacy cores
- Security and reliability. Mainframe and server-based cores are private and considered more secure than public cloud-based cores. In addition, downtime or outages for legacy cores is extremely rare making them highly reliable
- Regulatory and industry requirements. Legacy cores have developed over a significant period of time and have been adapted specifically for the banking industry. They have proven successful (however operationally inefficient they may be) in handling the required regulatory and product requirements of the banking industry
Limitations of legacy systems
- Conversion from mainframe to servers. Mainframe volume cannot be converted to servers due to lower processing capacity on servers, thereby restricting banks to possibly scrapping mainframes altogether if they aim to modernize their tech stack. Server-based cores may also impose higher maintenance costs than mainframes given their distributed architecture versus mainframes’ centralized structure. IBM research indicates that 80–85% of total costs are captured in mainframe IT budgets with a 1.17–1.20x multiplier for unaccounted costs, whereas server budgets only account for 50–60% of total costs with a 1.7–2.0x multiplier for unaccounted costs.
- Compatibility and modularity. Legacy cores built on mainframes and servers may require middle layer technology to be compatible with more modern, API-driven modular components provided by fintechs
- Restrictive switching costs. Transferring core banking data and all of the add-on services reliant on the existing legacy core can be prohibitively costly and have significant lead times of multiple years to implement without customer impact. A McKinsey report states, “For a medium-size[d] bank, the cost of this integration [switching cost] could exceed $50 million depending upon its complexity; for larger banks, $300 million to $400 million is not unheard of (based on estimates for traditional implementations).”
- COBOL programming language. Mainframe reliance on COBOL, a programming language first introduced in 1959, creates a significant engineering resource problem as the language is no longer taught widely by computer science programs and there are fewer engineers in the marketplace capable of handling the language. According to The Newstack, “As recently as [2012], the IT group at the Bank of New York Mellon had to tend to 112,500 different COBOL programs — 343 million lines of code…” Given retail banks’ continued reliance on COBOL based programs, the dwindling engineering population poses outsized risks
- Customer experience. Legacy cores treat each mid-layer product as a separate service. For example, loan data would only be visible to those with access to the loan module. This creates a disparate view of the customer and does not allow for the personalized customer experience that modern consumers now demand
Core banking infrastructure is modernizing as new entrants re-shape the core banking stack to truly power innovative customer experiences. In the next section, we review the advantages, characteristics, trends and players of the modern core banking tech stack.
Section II. Re-making the core banking stack
1. Core infrastructure as a competitive advantage
Until now, banks have been focusing on improving their front-end to deliver better customer experiences, gradually realizing the limitations of their back-end capabilities. As competition from fintechs is intensifying, and with the current macroeconomic environment and low interest rates placing further pressure on banking revenues, banks are now shifting their attention towards the heart of innovation: fundamental changes in core infrastructure. New fintech entrants are embedding banking through as-a-service models — a business model in which a solution is delivered on a subscription basis via the cloud, without a physical component (e.g. software-as-a-service) — to deliver better customer experiences and operate at up to 1/10th of the cost of traditional banks thanks to their lean and flexible core infrastructure. Core banking is now becoming an enabler of hyper-personalized and real-time customer experience.
70% of traditional banks are reviewing their core infrastructure, according to a McKinsey survey of 37 banking executives in May 2019. According to their survey, In terms of budget, the majority of banks have earmarked $10M or more over 2020, with around 20% planning to invest $20 to $40M. Companies such as Temenos, Mambu, Thought Machine, Finxact, and 10x Banking are at the forefront of driving this shift to modern core infrastructure.
In 2020, the pandemic placed digital transformation at the center, exacerbating the pressures on aging, complex legacy infrastructure including slow and cumbersome deployment, high maintenance costs and incompatibility with new applications. Global core banking software market size is expected to reach $16B by 2026. Built on this premise, modern core infrastructure can enable seamless customer experience, flexible updates and product launches, and all of this at a much lower cost than legacy systems — so what does the modern core banking stack look like?
2. Key characteristics of the modern core banking stack
A. Cloud-native architecture
The modern core is cloud-native, which drives cost efficiency and superior customer experience: unlike legacy systems, cloud-native infrastructure has lower operating and maintenance costs due to lower reliance on hardware, less staff training and outages, and no need to maintain bulky and lengthy code bases. Cloud-native infrastructure is modular and agile which accelerates time-to-market for new products and enables a real-time customer experience.
Transition towards cloud storage had been historically slow in the banking industry mainly due to security and risk management concerns. However, banks and regulators are now starting to recognize that cloud technology can provide a very secure environment, which is better than what most banks could achieve themselves. Recently, banks are gradually more comfortable considering moving to a cloud paradigm to support agility and fast innovation. Of the world’s 38 largest financial institutions and insurance companies, 25 have already signed up with Microsoft and are beginning to put applications in the cloud. In December 2020, Deutsche Bank announced “it will move the “Heart” of IT systems into Google’s Cloud. The two companies expect to sell some technology that they develop together to other financial services providers as white-label products and split the revenue. The contract is set to last at least 10 years and Deutsche Bank expects to make a cumulative return on investment of $1.2B through the alliance. It is the first partnership of its kind for the financial services industry. Over the coming years, an increasing number of financial institutions will certainly be turning towards the public cloud in the realization of broader opportunities it brings in comparison to on-premise operations.
Different cloud deployment models exist, with the public cloud model providing the most flexibility and scalability. Key benefits that cloud computing brings to banking include the following:
- Agility. Cloud has significantly reduced the time it takes to launch new products. Legacy systems operated in silos — with one system for customer accounts, another system for lending, another for deposits and so on — and consisted of thousands of lines of COBOL code. Any update required writing layers upon layers of scripts on top of old systems. Many companies don’t know exactly how their systems run, because rules extracted long ago are embedded in hundreds of thousands to tens of millions of lines of COBOL. In contrast, cloud’s hyper-parameterized configuration capabilities provide an agile and resilient product development environment to test and launch new products. It enables monthly releases and online software updates much faster
- Elasticity: Cloud allows companies to quickly scale processing capacity up or down, enabling banks to optimize costs and to respond quickly to rapid changes in customer demand, without the need for up-front capital investment to procure equipment, nor to incur high operational costs such as data center operations. For the Commonwealth Bank of Australia, cloud reduced the time and cost of standing up a new server from eight weeks and several thousand dollars to eight minutes and 25 cents, making the bank much more responsive to changing customer demands
- Availability. Cloud-native solutions rely on microservice architecture, a distinctive method of developing software systems that focuses on building single-function modules with well-defined interfaces and operations. In the simplest form, microservices help build an application as a suite of small services, each running its own process and is independently deployable. It allows teams to work simultaneously on different modules without the risk of overlap. In addition, server-less computing (i.e. on-demand) allows banks to build automated application workflows across multiple products (i.e. event-driven architecture). By removing silos, cloud enables banks to listen to thousands of transactions as they happen and increase straight-through processing rates. This allows customers to get instant fulfillment and 24/7 accessibility across multiple devices. One of the largest banks in Spain, Bankinter, has been recently reported to be using the cloud to run credit risk simulations in 20 minutes, down from 23 hours
- Remote collaboration. With regard to the pandemic and the shift to remote and distributed teams, the cloud embraces modern organizations’ remote-friendly culture
“At N26, we have built many of our core banking systems from the ground up, on top of a modern, cloud-native technology stack,” said Lindsey Grossman, former Director of Product US at N26 in an interview with PYMNTS.com in June 2020. “Instead of having computer hardware and software sitting physically in-house, like in a server room or data center, [the cloud has] allowed us to really be nimble[and] really allowed us to make sure we are also secure because that is table stakes to managing people’s money.”
“Cloud-based approaches allow banks to keep pace with recent mobile banking shifts”, she continued, “and their flexibility also allows financial institutions to create and test features geared toward mobile-first customers. N26’s Spaces feature allows customers to set up personalized subaccounts and work toward their savings goals, for example. She said it also developed a feature called Perks, which offers cash back rewards for debit spending, and created a “discreet mode,” which allows users to wave their hands over their phone screens to hide their personal details when checking their accounts in public.”
B. A customer-oriented and future-proof design
To answer the real-time economy’s needs, the core needs to get closer to the front-end customer experience and be modular to easily adapt to changes in business models in a fast paced and competitive environment. To ensure fit for an uncertain future and to keep up with front-end innovation, banks and fintechs are now re-thinking core banking in a simplified way: core banking — at its core — is simply a ledger of customers’ details and an accounting system.
Built on this premise, modern core infrastructure incorporates both art and technology to power truly innovative products. The key characteristics of a customer-oriented design include:
- A unified customer view. The modern core is designed with a single customer view in mind. Unlike in legacy systems where data is siloed, in the modern core customer data should be able to flow in real-time, streaming across applications and unified into one view to power hyper-personalized experiences and contextualized banking
- A data analytics driven mindset. The modern core has embedded data analytics capabilities that improve customer engagement and risk management. Machine learning could help with loan loss mitigation, transaction limits, customer lifetime value, and a range of other use cases. Core platforms need to support a range of database management systems, both enterprise and open source.
- A generic design. We do not know what banking will look like in ten years, and to be able to build truly innovative products, the core needs to be designed in a modular and atomic way, with generic design. Computational abstraction can enable banks to develop one-pattern code which can be then deployed for multiple products such as lending, investing and spending, which gives them the ability to superimpose business models. For example, a neobank can focus on a simple card product at first and could easily link several products to that card in the future.
C. An embedded and open banking architecture
In the past, banks needed to decide whether to build everything in-house or get a whole suite from a system integrator/IT vendor, being locked-in to a specific technology with multi-year contracts. In contrast, the modern core architecture is federated and allows for customer data portability, plugging into a rich ecosystem of partners while maintaining a seamless customer experience. Core platforms with API catalogs containing robust data/information APIs are better positioned to support open banking. At a high level, open banking implies that banking information is available and integrated through APIs and in some countries such as the UK, it goes one level further as regulation requires banks to surface these APIs to third parties. Data APIs which expose the underlying data model to API consumers are necessary for institutions looking to monetize access to data or allow consumer-permissioned access by third parties.
- Embedded finance. API-First architecture has made it possible to unbundle banking products into several layers (card issuing, bank deposit, payments, compliance, loans etc.) and provide financial solutions “as-a-service” to fintechs and non-fintech companies — enabling any businesses to embed financial services without having to build entire stacks by themselves. White-label solutions are now bringing the modern core much closer to the customer experience. These solutions allow the bank/fintech to own an end-to-end customer relationship and provide seamless customer experience. Synctera is one example of a white-label solution that enables banks and fintechs to gain control over the customer support experience and deal with chargebacks/remediation. Stripe Treasury is another example of a Banking-as-a-Service (BaaS) platform which enables marketplace customers to hold funds, pay bills, earn interest, and manage cash flow, through simple API integrations.
- Open source. More recently, players such as Moov Financial have driven further innovation in the field of core banking with their open source approach. Moov provides components such as ACH, wire, check and RTP as separate modules that banks and fintechs can easily build into their platforms with simple APIs. By unbundling the core into banking primitives, Moov goes one level deeper into modularity and flexibility, enabling banks and fintechs to pick and choose the building blocks of their core to outsource or keep in-house, bringing a further degree of control and flexibility to the organization. The open source approach provides further reliability as the Moov community has attracted thousands of payments engineers, including some at the top five banks who ran $4T of previously cleared transactions through Moov’s ACH library to test its fraud models. They now have a vibrant Slack community of ~1,800 members at the time of writing.
3. Decentralized Finance (Defi) — the future of financial infrastructure and core banking?
In the future, we imagine core banking will become further autonomous, efficient and transparent. A field at the intersection of core banking and automation is the recent emergence of Decentralized Finance (DeFi), which is the use of decentralized technology — blockchain and smart contracts — to disintermediate financial services such as savings, loans, trading and more.
Today there are decentralized trading venues handling over $30B in volume per month; automated lending programs making individual loans as large as $200M; and the total estimated value of funds currently locked into DeFi-related contracts recently crossed $50B.
Compared to traditional infrastructure, DeFi provides the following advantages:
- Cost reduction: DeFi cuts out middlemen and offers a nominal fee by principle and by design. In cross border payments, the current average global remittance fee is 7%. Through DeFi, remittance fees could be lowered to less than 3%. Unlike traditional systems, DeFi transaction fees are not charged on a percentage basis. Instead, they are flat fees (e.g. in the Ethereum blockchain, they are based on how complex the computation is to perform). Therefore, DeFi can provide even further cost savings for large transfers. For instance, recently an Ethereum transaction moved $49M in ETH from one wallet to another for ~$3.95 in transaction fees
- Speed: Clearing and settlement of cryptocurrency transfers through DeFi are much faster than traditional services, with self-executing smart contracts underlying the entire system. For instance, the Ethereum transaction mentioned above was settled and cleared within 30 seconds
- Flexibility: Developers can program business logic into low-cost interoperable financial services
- Transparency: Decentralized applications (DApps) are much more transparent by design with a governance model that is decentralized and more democratic
- Accessibility: It is accessible to anyone with an internet connection. Currently, 5.4% of U.S. households, about 7 million, are “unbanked” while 1.7 billion people are underbanked globally, where DeFi is an attractive potential set of solutions for increasing accessibility and transparency to the financial system
One of the common ways that users currently access DeFi for payments and remittance is using stablecoins, a type of cryptocurrency either fully or partially collateralized and that is designed to maintain a stable value, rather than experiencing significant price changes. In emerging economies such as the Philippines and Venezuela, stablecoin usage has grown specifically due to users’ need to protect savings in times when the native currency is volatile or losing value. Stablecoins can be centralized which means they are centrally issued by, for example, a central bank (as a Central Bank Digital Currency) or a company/consortium (USD Coin or Tether). They can also be decentralized. For instance, DAI is a decentralized stablecoin issued by Maker DAO that uses a decentralized collateralized debt protocol (CDP) backed by other cryptocurrencies to ensure that the value of 1 DAI is kept close to the value of 1 US Dollar. DAI allows anyone to access the stability of the US Dollar (something not always easy for those outside the US), and deploy it throughout the DeFi world.
Decentralized Exchanges (DEX), and more recently, Automated Market Makers (AMMs), have also been critical to cryptocurrency adoption in the developing world. They don’t custody any of the crypto or fiat traded on their platforms, meaning they don’t have to connect to the banking system and they face fewer regulatory hurdles. This allows them to onboard residents of developing countries more easily, many of whom are unbanked. Uniswap is a pioneer in the field and popularized the automated market making model for DEXes in the crypto space, using liquidity pools and algorithms to set trading prices instead of the order books used by centralized exchanges.
While this is a nascent field, a lot of innovation in DeFi comes as well from decentralized lending protocols. Lending protocols, such as AAVE and Compound, are enabling for the first time to borrow or lend money on a large scale between unknown participants and without any intermediaries. Those applications bring lenders and borrowers together and set interest rates automatically in accordance with supply and demand. Moreover, those protocols are truly inclusive, as anybody can interact with them at any time, from any location, and with any amount.
Given the current macroeconomic environment of low interest rates and the longer period over which inflationary monetary policies will play out, DeFi may continue to attract both interest from retail and traditional financial institutions. Nevertheless, DeFi still presents several hurdles for adoption:
- Regulatory compliance: Regulation is unclear for many DeFI applications. Checks and balances that exist to mitigate or unwind illegal market behavior such as front-running or wash trading are often not implemented in DeFi applications on the blockchain
- Scalability: Ethereum transactions currently cost a significant amount of transaction fees (referred to as “gas”) to operate, while acceptable for higher value transactions, is less suitable for lower value transactions. Scaling solutions for Ethereum or use of a more cost effective blockchain platform are possible mitigants. In addition, Ethereum can match, settle, and clear less than 30 transactions per second while the VISA network can match 24,000 transactions per second. It is noteworthy that while VISA’s matching is much faster than blockchain, VISA settlement and clearing can take 1–4 weeks. Blockchain settlement and clearing is instantaneous
- Security: Several DeFi protocols such as bZx have been hacked, causing them to lose millions of dollars
- Middleware capabilities: There is a lack of developer tools to easily adopt/implement DeFi protocols by traditional companies. Startups such as Curvegrid and ZeroHash offer a middleware layer to solve this issue
- Usability: To access services beyond what centralized crypto asset exchanges offer, users must custody their own funds. Currently this can be complex and error prone, as there is typically no “forget password” option and funds can be locked, lost or stolen much more easily than in traditional banking
On the regulation front, the field is evolving at a fast pace. Recently, Anchorage became the first federally chartered digital asset bank in US history. As a national bank it can hold, trade, lend and issue digital assets to clients. It allows clients to hold five more DeFi tokens in the security of the Anchorage platform: Aave (AAVE), Balancer (BAL), Nexus Mutual (NXM), Uniswap (UNI), and Yearn (YFI). More recently, the Office of the Comptroller of the Currency (OCC) granted another startup conditional approval to operate as a national bank: Protego joins Anchorage as the second firm that has received this type of federal approval. Protego plans on being a “vertically integrated and fully regulated bank built on blockchain.”
Conclusion
Core banking infrastructure is the skeleton of our banking system, thereby enabling a wide variety of innovations and enhanced customer experience in the fintech space. Developing modern core banking capabilities across financial services will be the key to building the foundation of an inclusive, transparent and efficient financial system.
Additional references
Metz-Pasquier, Yann, et al. US Market Expansion. HBS/MIT Independent Study Project, Fall 2017. 2017.
Acknowledgements
We would like to extend our deepest gratitude to our fintech industry reviewers for their time in providing input on this article. The Core Banking Systems Primer is undoubtedly more thoughtful and informative with your feedback. Thank you!
In alphabetical order by last name:
- Stephanie Balint, Interim General Manager-US @ N26
- Neel Ganu, Co-Founder @ Finch
- Lindsey Grossman, Director of Product, North America @ Wise
- Bryan McCarty, Director of Marketing @ Moov Financial
- Jeff Wentworth, Co-Founder @ Curvegrid
- Edward Woodford, Co-Founder @ Zero Hash
