AI and machine learning in the finance and banking industries are poised to transform how organizations manage their revenue, communicate with customers, and scale their investments.
The underlying adoption of artificial intelligence across industries is predicted to drive global revenues of $12.5 billion in 2017 to $47 billion in 2021 with a compound annual growth rate (CAGR) of 55.1% from 2016 to 2021.
The term artificial intelligence was coined in 1955 by John McCarthy, a math professor at Dartmouth. Due to its evocative name, this field has produced a wide array of hype and claims. Nonetheless, data science is becoming increasingly recognized as the motive power steering the leading industries to the future.
Faster processor speeds, lower hardware costs, and better access to computing power have given rise to a growing number of FinTech companies. There has also been a rapid growth of high-quality datasets for learning and prediction owing to increased digitization and the adoption of web-based services.
In the highly competitive financial sector, artificial intelligence is at a rapidly evolving phase, with new use cases and algorithms uncovered in a matter of days rather than years. The availability of AI-powered systems lies heavily on the existing data and infrastructure, and the fundamental demands of financial regulation.
A recent study pointed out that the rise of data science among financial institutions is driven by five key factors: the general advancement of technology, factors particular to the financial sector, the potential for increased profitability, competition on the market, and regulatory compliance.
Machine learning can help companies to reduce costs by increasing productivity and making decisions based on information unfathomable to a human agent. Intelligent algorithms are able to spot anomalies and fraudulent information in a matter of seconds.
Main Application of machine learning in the finance and banking industry:
Cost reduction and process optimization
The finance industry is harnessing machine learning to lower operational costs and drive profitability. This field involves both front- and back-office activities across multiple institutions.
COST REDUCTION IN INSURANCE
Insurance companies sort through vast sets of data to identify high-risk cases and lower the risk.
Artificial intelligence is applied to functions such as underwriting and claims processing. One of the key technologies here is the application of Natural Language Processing (NLP) that improves decision-making by analyzing large volumes of text and identify key considerations affecting specific claims and actions.
Another set of factors can be included in the insurance claim evaluation process. For example, an ongoing AI-powered dialogue through bracelets, sensors, etc. leads to a more comprehensive understanding of the insured.
By collecting and analyzing additional data, insurers are able to analyze the habits of their policyholders and offer highly customized products, adapted in real-time to the needs and expectations of their clients.
OPERATIONAL COST REDUCTION IN THE BANKING INDUSTRY
To maximize their profitability, banks rely heavily on capital optimization.
Artificial intelligence algorithms can be applied to handle large quantities of data to increase the efficiency, accuracy, and speed of mathematical calculations. Using machine learning, banks can find the best combination of the initial margin reducing trades at a given time based on the degree of initial margin reduction in the past under different combinations of those trades.
Banks are also looking to apply AI algorithms to back-testing, in order to assess the overarching risk models.
Using a range of financial settings for back-testing helps to perceive unpredictable shifts in market behavior and other trends, leading to better decision-making. A similar approach is often applied to stress testing.
Technological advancements can also help financial institutions by introducing a machine learning approach to minimize the trading impact on prices and liquidity, thereby predicting the market impact of specific trades (and the best timing for such trades). This can ultimately lead to the minimized impact of trading both into and out of large market positions.
Front office activities such as credit scoring and insurance can be optimized to the extent where many financial decisions are based on wide-scale data analysis. For example, such data can help assess risks for selling and pricing insurance policies.
Historically, most financial institutions based their credit ratings on the lender’s payment history. Increasingly, banks are looking towards additional data sources, including mobile phone activity, social media usage, to capture a more accurate assessment of creditworthiness and improve the profitability of loans. Leveraging such technologies allows for faster and cheaper credit scoring and ultimately makes quality loan assessments accessible to a larger number of people.
In the past years, a number of customer-facing FinTech companies have emerged. Using an algorithmic approach, some of these companies apply data analysis to provide credit scores for individuals with ‘thin’ credit files, using alternative data sources to review loan applications rejected by lenders.
ADVANTAGES/DISADVANTAGES OF USING AI IN CREDIT SCORING MODELS
- AI allows large quantities of data to be analyzed very quickly
- Potential cost-reduction of assessing credit risks
- Increasing the number of individuals with measurable creditworthiness
- Lack of transparency for customers
- Difficult to understand the underlying factors of algorithmic decisions
- New data sources can bring bias to credit decisions
- Gender or racial discrimination based on historic data analysis
- Lack of availability or unreliability of third-party data
Algorithmic trading and risk management
Artificial Intelligence has made its way to the back offices of asset managers and trading firms. In addition to R&D, some firms now use machine learning to devise trading and investment strategies. Big data and machine learning help large trading firms to strengthen their risk management techniques by centralising the risks that arise from various parts of their businesses.
In the past years, a new generation of quant funds has appeared on the market. An AI unit is generally part of a larger team to aid the asset manager with portfolio construction. Harnessing the predictive power of data can help funds spot new trends and potentially profitable trades that are outside of the human scope of understanding. For example, Hong Kong-based Aidiya is a fully autonomous hedge fund that makes all of its stock trades using artificial intelligence).
According to an extensive 2017 study, machine learning likely only drives a minor subset of quant funds’ trades. Quant funds manage on the order of $1 trillion in assets, out of total assets under management (AUM) invested in mutual funds globally in excess of $40 trillion.
There are also a growing variety of vendors that provide Big data services for financial market participants.
Such players could scrape news and/or metadata and enable users to identify the specific features (web pages viewed, etc.) that correlate with the events their customers are interested in predicting.
However, we’re far from AI algorithms continuously outperforming human traders. In March 2018, Bloomberg reported that the index of hedge funds using AI had fallen 7.3 percent the past month, compared to a 2.4 percent decline for the broader Hedge Fund Research index.
AI-powered fraud detection
According to an Intel report, The United Nations claims that less than 1% of global illicit financial flows are frozen or seized, and that up to 5% of global GDP – $5 trillion annually – are money laundering transactions. Fraudulent claims account for $80-100 billion annually in the U.S. alone.
AI has proven extremely applicable to security and fraud detection use cases. Machine learning algorithms can analyze thousands of data points in real-time and flag suspicious or plain-right fraudulent transactions, stopping many fraudulent claims in the process.
According to Samir Hans, an advisory principal at Deloitte Transactions and Business Analytics LLP,
Mastercard recently introduced its latest pioneering security platform, Decision Intelligence. The system uses machine learning technology to make data-driven, real-time decisions tailored to the account, including defined alert and decline thresholds. By detecting anomalous shopping spending behaviors, the system can prevent thefts and fraudulent transaction claims.
Robo-advisors have brought a data-driven and partially automated approach to wealth management systems.
AI-powered tools can help traders streamline the account opening process, and advise them on scaling their portfolio. This could include developing a financial plan, advising on planned home purchases, retirement, protection needs, estate planning, etc.
The main advantage of robo-advisors is that they are low-cost alternatives to traditional advisors. In the long term, robo-advisor technologies could make financial counselling available to an increasing number of people, resulting in more informed personal finance decisions.
While current robo-advisor total assets under management (AUM) only represent $10 billion of the wealth management industry’s $4 trillion (less than 1% of all managed account assets), a Business Insider article estimates that this figure will rise to 10% by 2020. This equates to around $8 trillion AUM.
If you’re interested in learning more about robo-advisory, we recommend this report by Accenture.
Personalized offers and customer retention
Machine learning offers a wide array of custom solutions for improving the customer lifetime value and optimizing the sales of financial products.
For example, imagine a recommendation engine capable of suggesting to existing and new customers the most suitable insurance package or identifying new potential users fit for an upselling offer.
By analyzing what makes some customer segments remain loyal customers and others seek out new financial service providers, banks and other stakeholders can target the in-danger segments with motivating offers and products.
Another widely popular AI use case (also in the telecom business) are intelligent chatbots.
With some exceptions, AI-powered customer service solutions can be divided into two categories:
- Customer service communication
- Customer engagement and personalized offers
Custom-built chatbots could be used to streamline large parts of the tedious customer service process, automatically solving simple customer requests and routing others to the right department within the company.
The financial industry with its large sets of data is particularly fit for building intelligent customer service bots and systems. To be able to accurately evaluate and resolve customers’ issues, AI algorithms empowering customer communication must process a massive amount of data and interactions.
Regulatory compliance in the financial sector
New regulations have increased the need for efficient regulatory compliance, which has pushed banks to seek cost-effective means of complying with regulatory requirements. Regulatory technology (RegTech) focuses on making regulatory compliance more efficient and native to a financial institution’s core processes.
NLP could be used by asset management firms to cope with new regulations. For example, in the EU, investment managers have to comply with specific requirements in the Markets in Financial Instruments Directive (MiFID II), the Undertakings for Collective Investments in Transferrable Securities (UCITS) Directive, and the Alternative Investment Fund Managers Directive (AIFMD). To comply with these regulations, companies can apply AI-powered data analysis to build integrated risk and reporting systems. Moreover, machine learning could help trade repositories (TRs) tackle data quality issues, increasing the value of TR data to authorities and the public. Check out other natural language processing use cases applicable in the financial industry.
Conclusion and overview
Every single one of these fields of study is still in its infancy, showing promising advancements, yet far away from complete autonomy from human agents.
We recommend financial institutions take steps to introduce AI and machine learning to various processes across the company. In the long term, this will benefit the organization both in terms of increased efficiency as well as a competitive advantage.
A number of developments might impact the future adoption of a broad range of financial applications of AI and machine learning. This includes a growing number of data repositories, data quality, increasing processing power, but also new regulations and laws.