Artificial Intelligence (AI) is transforming many aspects of our everyday lives. But why should AI be vital for European fixed-income investors? Today’s low-yielding and less-liquid markets hold the key.
From Google Maps to facial recognition systems, AI is becoming omnipresent in everyday life. In factories and offices, robotics and machine learning are taking over routine tasks. In total, AI is expected to add US$13 trillion to the global economy by 2030, according to PwC.
Fixed-income investment managers generally have been slow to adopt AI. Smart managers have started to digitize investment research so that their findings can be readily referenced and evaluated. But most have yet to make the really big advances in integrating AI into research and trading.
Low Yields Transform Active Management
Now investors are facing unique challenges and opportunities that only AI can address. In the euro area, the European Central Bank’s extraordinary monetary policies have resulted in low or negative rates across the yield curve accompanied by ultra-low credit spreads for all but the riskiest corporate bonds.
In this environment, to generate worthwhile levels of returns without taking excessive risk, investors need to adopt genuinely active strategies. Buy-and-hold no longer works with German 10-year Bunds trading at –0.3% and high-yield (HY) spreads at around only 3.5%. In contrast, 10 years ago, HY was yielding 8%–9% and, irrespective of spread fluctuations, investors were still likely to end up with a positive return. Now, with the yield compressed to 3%, even a small spread change could impact the bond price and greatly increase the chance of a negative return. So investors have to act fast to avoid losing money. To generate positive returns consistently, their asset managers must identify opportunities fast, locate the right bonds in the marketplace and execute the desired trades quickly and efficiently. In today’s complex and disparate bond markets, with literally millions of data points to be evaluated daily, that task is way beyond the scope of manual processes—it takes AI to get ahead.
Liquidity Is Hugely Important
Scarce liquidity also calls for AI solutions. Since the global financial crisis (GFC), market liquidity has contracted drastically. The global bond market has doubled in size while dealer balance sheets have shrunk to around 5%–10% of their pre-crisis size. That makes it much harder for asset managers to trade their bonds, particularly in fast-moving markets.
What’s more, markets have become much more fragmented, with perhaps 15–20 different investment platforms for traders to cross-compare before trying to deal.
And finally, complexity has soared. For example, US investment-grade credit is a US$5 trillion market with 6,500 individual securities. Every day, about 40,000–45,000 trades are executed.
So, liquidity is badly needed, but is now very scarce and hard to find in a hurry.
Active Bond Management Enters the AI Era
These challenges can be addressed by integrating AI with investment processes from end to end. Starting with digitized research, the next stage is to use algorithms to source pockets of liquidity quickly across multiple platforms, and to create digital assistants (“chatbots”) that can build orders and highlight opportunities within seconds.
The first-mover advantage is considerable across both research and trading. AI-enabled managers can focus their resources more effectively, for instance, by working only on the ideas that they are certain can be executed, or by delegating smaller less important trades to their chatbot to transact directly with brokers’ bots. With superior insights into market pricing and dynamics, they can not only execute purchases more efficiently, but can also time their sales better to profit from changing market demand. And by harnessing natural language processing in their research process, they can automatically comb through thousands of corporate reports to spot wording changes that may signal possible improvements or deterioration in the quality of a bond issue.
Measuring the Performance Impact
How do these advantages feed through to performance? It’s hard to compare current AI-enabled performance with returns generated in a completely different market environment. But we can gauge improvements in execution, by measuring the difference between a security’s price when a new idea breaks and the same security’s price later in the day. That difference reflects the impact of an AI-driven information advantage.
In our experience, the greatest performance advantage arises in more volatile markets and in markets with a negative trend. For instance, in a difficult but not disastrous final quarter of 2018, US high-yield bonds declined by 5%. Prompt idea generation and execution in this tough period were particularly important.
But in very approximate terms and under average market conditions, fully integrated AI might typically add between 10 and 20 basis points on an annualized basis to the return of a fixed-income portfolio, in our view. In a business where the outperformance objective is frequently between 50 and 100 basis points, that’s a very conspicuous advantage.
AI Evolution
Today, the investment world is still in the early stages of exploring the full range of AI’s potential. But we’re certain that its capabilities and applications will continue to multiply. Smart investors should recognize the potential of AI, and make sure that their asset managers are fully equipped to integrate it comprehensively into their processes.
John Taylor is Co-Head of European Fixed Income and Director of Global Multi-Sector at AllianceBernstein.
The views expressed herein do not constitute research, investment advice or trade recommendations and do not necessarily represent the views of all AB portfolio-management teams and are subject to revision over time. AllianceBernstein Limited is authorised and regulated by the Financial Conduct Authority in the United Kingdom.