Articles

Information Systems Professor’s Research on Bond Market Dynamics Says Market Make-Up Could Spell Trouble

By Keith Morelli

Don Berndt

TAMPA (May 18, 2017) -- Research led by a University of South Florida information systems professor suggests that the expanding corporate bond market could spell trouble in the event of a market reset, due to the growing influence of mutual funds in an era of increasing risk.

"Considering its current evolution, the bond market ranks fairly high on the list of potential risks to financial stability," said USF Muma College of Business professor and lead researcher Don Berndt, "prompting regulators and industry participants to question its resilience under stress."

The models and simulations designed and run by Berndt and others are funded by the U.S. Department of the Treasury's Office of Financial Research. Berndt has visited Washington, D.C., several times while working on financial market models and simulations. Treasury department researchers share a common interest in agent-based approaches, he said.

He and fellow USF researchers James McCart and Saurav Chakraborty, along with David Boogers of Finametrics.com, have implemented an agent-based model of the U.S. corporate bond market, with classes of agents that include mutual funds, insurance companies and hedge funds, he said. An agent-based model simulates the actions and interactions of individual autonomous agents to better understand the emerging system-wide dynamics of such complex systems.

"The goal now is to use our agent-based model to assess various regulatory policies aimed at reducing these risks," Berndt said. The research has been ongoing for a couple of years, funded by a $300,000 grant from the Office of Financial Research and administered through the National Science Foundation.

The aim: to come up with computing-oriented approaches for a better understanding financial systemic risk, an area of interest following the financial crisis that occurred nearly a decade ago. It was that crisis in 2008 that sparked renewed interest in the dynamics of crises and introduced structural market changes that motivated Berndt and his colleagues to formulate their models and simulations.

"Basically, we can use these models to test regulatory policies," Berndt said. "Imagine the government steps in and buys bonds from mutual funds during a crisis.  Could that avoid a crash? These models may be able to answer that question."

Recognizing the fundamental changes in the U.S. corporate bond market following Great Recession in the late 2000s, the USF team prioritized this area as the first application domain for their agent-based modeling efforts. Corporate bonds – transferable debt securities issued by companies – are an important means by which companies fund their business operations and expansion and an integral part of the market.

Though it is unlikely the same factors that caused the recession nearly a decade ago will occur again, new scenarios could emerge that have similar results.

"Crises tend to originate in new areas," Berndt said, "and rarely copy historical patterns."

So the research looks at different models that could predict systemic failure of specific markets.

"We have been developing simulations that can help us better understand the nature of systemic risk in the financial services sector," Berndt said. "While some traditional approaches focus on the resilience of individual agents to specific shocks, they fail to address the broader question of how market stress might be transmitted among firms through the various dynamics."

He said their work could capture the subsequent effects of interactions as it tracks reaction – or behavior – of individual agents. It is particularly well suited to the analysis of the high-order effects or feedback loops that exacerbate the "fire sale" dynamics as assets are quickly bought or sold on the market.

In the years following the 2008 financial crisis, bond markets saw extraordinary growth. Analysts say the overall U.S. corporate bond market expanded from $5.2 trillion at the end of 2007 to more than $8.5 trillion in the third quarter of 2016.

With the increasing size and relative importance of the corporate bond sector, concerns around the proper functioning of its secondary market and where bonds are traded by investors after they are issued, have taken center stage, Berndt said. Central to those concerns is the deterioration of liquidity, which is the buying or selling of corporate bonds without triggering a large impact on price.

Well-functioning bond markets – the mechanisms that connect bond issuers with investors and enable trading between investors – are deemed essential for economic growth. Historically, low-interest rates and ever-tightening credit spreads, drove corporations to issue bonds in record amounts. The expansion of the bond market was enabled by strong demand.

During this period of market expansion, dealers – who perform an essential function in bond markets by intermediating virtually all transactions between buyers and sellers – a potential imbalance arose between market size and asset pricing.

Other concerns about the bond market include a lack of price transparency, reduced liquidity, significant declines in dealer intermediation capacity and increased "herding" behavior.

"So," Berndt said, "one of the first experiments pursued by the team assessed the potential systemic risk caused by the interaction between impaired market liquidity and the growth of mutual funds."

Adding to the risk of market systemic failure are changes in the bond market microstructure that have led to reduced or nonexistent market liquidity under conditions of stress and the growing percentage of corporate bonds and mutual funds in the market.

"Mutual funds contain redemption features that make them susceptible to runs," Berndt said, "potentially triggering fire sales and price swings like those associated with the activities of participants who rely on short-term debt financing."

The team's initial experiments focus on the impact of increasing mutual fund market shares in the corporate bond market.  Over the course of a few decades, mutual funds have grown from about a 3 percent market share to nearly 20 percent. The simulations included economic shocks that cause the re-pricing of bonds, along with increasing mutual fund market shares (15 percent, 25 percent, and 35 percent). As the market share grew, the simulation results showed initial price drops and protracted rebounding periods as expected.

The initial interest rate shock drops prices, with secondary feedback-driven price declines, followed by gradual stabilization and recovery, the research showed. However, at a 35 percent mutual fund market share, the model becomes unstable and the initial shock-induced price drops followed by feedback-driven price drops overwhelm broker/dealer capacity. And that, according to the simulation, means a prolonged market crisis and the dramatic destruction of wealth.

"The experiment's results confirm that growing mutual fund market shares are a concern," Berndt said. "So far, the research does show that agent-based modeling and simulation is a promising tool for understanding the dynamic nature of financial systems.

"The team is already working on their next model of the interbank-loan network that is also of interest to federal researchers," he said. "The eventual goal is to plug these component models together, allowing us to model larger financial systems."