Artificial intelligence (AI) has the potential to fundamentally change the asset management industry. Persistently low interest rates, exponential growth in data, and increasing regulation are forcing asset managers to reconsider their traditional business approach. Currently, fewer than a fifth of institutional investors use AI technology, but there is no doubt that the fixed income industry has to catch up.
In its report, Asset Management Disrupted, Deloitte presents several use cases - amongst them automated credit scoring and self-driving portfolio optimization - that show how AI can help fixed-income managers to address key pain points and to generate a higher alpha for clients.
The asset management industry has not had it easy in recent years. Low interest rates have affected profitability, which has been further exacerbated by a drop in management revenues and rising cost pressures. At the same time, asset managers face an explosion of data: 90 percent of data worldwide was generated within the last two years alone. Current data management systems aren’t able to capture and analyze these data flows to exploit their upside potential. Moreover, with the majority of qualitative information now coming from non-traditional sources, it poses an even greater challenge to process this data without sophisticated technology.
Last but not least, fixed income managers have to cope with an ever-expanding regulatory framework. The CRA III regulation, for instance, requires asset managers to assess in-house the credit risk of externally rated assets using plausibility checks. This leads to higher personnel and expenses, whilst also increasing complexity. Already declining margins are squeezed even further.
However, recent advances in AI offer a unique opportunity to solve these challenges as the following use cases show.
Automated credit scoring to determine default risk
Asset managers have traditionally relied on rating agencies to determine the creditworthiness of companies and other bond issuers. Since the implementation of CRA III though, most asset managers also assess credit risks in-house, which is a time-consuming, manual process. Growing data volumes and the frequency of mandatory assessments further add to the workload. Moreover, including qualitative information into the analysis creates additional effort and carries the risk of bias when humans perform this task.
AI-based solutions can offer three key advantages compared to current manual approaches: Increased efficiencies, reduced human bias, and improved bases for investment decisions. However, to maximize the benefits of AI it has to be trained in advance with available data and able to learn autonomously. Especially for complex tasks such as credit scoring, it is important to support the machine-learning (ML) process with the domain knowledge of experts. Text mining and natural language processing (NLP) can further improve the results.
By training the AI to learn over time, forward-looking analyses and hence credit scoring can be improved significantly. As a result, asset managers can achieve both higher returns for their clients and higher profit margins for their services.
Self-driving portfolio optimization to generate yield
To generate alpha, portfolio managers rely on exclusive research and information. Hence, many asset management companies have large research departments. With ever increasing data volumes though, asset managers will need more sophisticated tools to optimize their portfolios.
Here, ML models can help to predict price movements and volatilities by detecting the right signals in Big Data streams. This not only applies to quantitative data, but also to qualitative data and even audio files. Using text mining or NLP models, sentiment indicators can be derived from a variety of sources, such as social media, giving portfolio managers an indication of possible future market developments.
This ability of technology to evaluate Big Data streams much more efficiently supports credit analysts in their decision making. In addition to handling a large part of the fundamental market and financial analyses, AI systems can even provide investment recommendations based on intelligent algorithms.
These use cases show only a fraction of what is possible today. Advances in NLP, the connection of domain knowledge, and “white-boxing” of intransparent machine learning algorithms have greatly increased the viability of AI applications. Now is the time for asset managers to take advantage of the new technological opportunities to gain a significant competitive advantage.
Read the full report here.