Tesla founder, Elon Musk, caused waves in the cryptocurrency market with a single tweet and made a mockery of the traditional frameworks many still use to make sense of financial markets.

In 2021, Musk posted a photo of his dog – a Shiba Inu hunting dog native to Japan – on Twitter. With this tweet, he set off a buying frenzy for the meme cryptocurrency Shiba Inu (SHIB), giving rise to a 400% increase in the token’s price. However, he subsequently responded to a Twitter query, saying he didn’t own SHIB, and the price dropped. This rapid market response highlighted how sentiment-driven swings could lead to irrational market outcomes, frequently catching investors off-guard.

Markets are quite unpredictable at times. The problem with this is that human beings are not good at making predictions. Certainly, we get many things right, but we also get a lot wrong. The reason for this lies in the way we approach this critical task and our reliance on traditional financial models that focus on the probability of something happening.

Probability distribution-based predictions ignore the so-called ‘unknown unknowns’, the ‘black swans’ that Professor Nassim Nicholas Taleb explores so wonderfully in his book of the same name. This failure to make allowance for human quirks, human behaviour and unexpected curve balls is precisely why the financial crisis in 2008, and the ongoing Covid-19 pandemic, caught so many people wrong-footed.

Fortunately, there is a way around this problem. By using a framework such as agent-based modelling, business leaders can constantly learn and adjust in response to changes in market participants’ behaviours and market evolution. This is especially important as new entrants and younger generations – many of whom do not play by the old rules – enter the financial markets.

Getting to grips with an agent-based approach

Agent-based modelling is a simulation technique that studies a system (like a market) by assessing the dynamics of interactions between market participants in real-time, thereby providing a powerful tool for more effective market analysis and risk planning. A key benefit of this tool is its ability to describe unanticipated events – those so-called black swans – and factor them into our understanding of the future.

In short, an agent-based framework uses simulations to explore emergent behaviour. These insights are then used to understand a phenomenon – be it a pandemic, a crowd stampede or a supply chain – by examining how individual behaviours result in an outcome. For instance, consider the factors that led to the speculative mania fiasco when retail investors with little market understanding and no knowledge of valuations were fuelled by social media platforms such as Reddit to drive up stock prices for failing video game company GameStop. The result was huge losses for firms like hedge funds, which were short-selling the shares and were caught completely off-guard since they were not focusing on irrational retail investor sentiment. This was not the first time traditional business and risk models dropped the ball. Just consider the great recession of 2008.

Understanding financial markets…and crises

As 2008 illustrates only too well, traditional risk management frameworks grossly misread the economic situation. The great recession was years in the making, and there were countless warning signs from unsustainable debt build-up due to cheap credit and lax lending standards. Still, few market participants suspected that the worst crisis in nearly eight decades was about to engulf the global financial system.

The great recession was by no means unique. Currently, markets are again looking increasingly fragile. Historically, ultra-loose monetary policies and low borrowing costs have contributed to dangerous levels of debt build-up in the global economy. In fact, total global debt more than tripled over the past four decades to 350% of global GDP. The spiralling corporate debt in the non-bank sector is of particular concern.

Moreover, it is becoming evident that inflation is not transitory. Spiking energy prices and persistent supply chain bottlenecks are adding additional pressure. Policymakers are under increasing pressure to control accelerating inflation through interest rate hikes. However, rising interest rates may set the scene for a debt crisis that could be aggravated by declining credit ratings and reduced liquidity, a symptom of the lower market-making activities of banks in response to the tougher capital adequacy requirements since the 2008 financial crisis.

How might an agent-based approach assess the risks inherent in this mix?

During a crisis, liquidity falls and each market participant stops to take stock of the environment before acting in line with heuristics shaped by business models, risk appetite and culture. For example, a lender will limit the funding it provides based on the collateral it holds. If the value of the collateral falls, funding will be recalled, or more collateral demanded. This will have a chain reaction in an agent-based simulation of the situation and may result in fire sales to cover covenants.

As a result, risk in one part of the environment will become a risk in related areas and could end with a bank run. An agent-based simulation will show how an initial market shock results in forced selling by highly leveraged firms, which in turn causes less leveraged firms to sell, followed by the evaporation of liquidity due to forced selling in wider markets, resulting in more collateral demanded as prices fall and volatility increases. This will likely have a cascading effect, and the outcome can be assessed as part of an agent-based model to determine how the environment is changing and what the chain reaction will be.

It always comes back to people

An agent-based approach is preferable to the likes of the pervasive value-at-risk framework, which uses unsound assumptions about the behaviour of market participants. It assumes that all problems are solved using perfectly rational logic to optimise behaviour, but it ignores the most important aspect of any market occurrence: the human factor.

Human behaviour is notably irrational, driven by whims and widespread fear of missing out (FOMO). Failing to consider factors such as FOMO, the rise of social media, ingrained cognitive bias and hard-wired neurological responses to risk and fear means that any simulation will invariably fail to account for the panic and greed that sets in when a financial market goes into meltdown. This human response will have a ripple effect through the entire system, impacting businesses that suffer heavy losses due to being caught napping and destabilising the banking system. This was certainly the case in 2008. Any future economic contagion will also be shaped by the market participants and how they interpret the unfolding macro-economic factors and choose to act based on their heuristic shortcuts, which are shaped by business models, risk appetite and culture. Unless the sum of these moving parts is understood beforehand, the chances of steering a safe course through the fallout are slim at best.

Since change is ever-present, and human beings will inevitably surprise, it is important that organisations, businesses and countries utilise behaviour finance tools such as agent-based modelling to constantly update their strategies and then shift accordingly. If they don’t, many business leaders and risk managers will find themselves scrambling for answers in the face of market tumbles, global pandemics, shifting technological uptake, or just an unexpected change in consumer sentiment.

Agent-based simulations in action

Agent-based frameworks are widely used in various fields, including biology, economics, social sciences, game theory and network theory. They are used to simulate a range of complex occurrences, including pandemics, weather patterns, crowd stampedes and traffic jams.

Since this modelling tool provides valuable information about the dynamics of the real-world system it emulates, it is increasingly being used in the financial markets by banks and investment firms, for example. Currently, based on the insights unlocked by this simulation tool, a number of investors are preparing themselves for a new financial disruption, having been tipped off that a tipping point is being reached due to human behaviour.

It’s not just the world of finance and investing that uses agent-based frameworks. Other uses include:

  • Determining the epidemic dynamics of a virus or disease to determine issues such as the magnitude of an outbreak and the impact of possible interventions. For example, in 2014, an agent-based model was used in Liberia and Sierra Leone to investigate the Ebola outbreak.
  • Improving supply chains by identifying opportunities to improve, as was the case when the United States Air Force used agent-based modelling to streamline its complex military logistics.
  • Mitigating against a crowd stampede at a sporting event by applying the same modelling that explains the human behaviour that results in a stampede when financial markets go up or down.

Agent-based thinking is also being used by marketers to understand how the emergent behaviour of consumers, for instance, can combine to adjust and change the environment. Even subtle behavioural rules can have surprising outcomes, according to professors William Rand and Roland Trust from the University of Maryland in the United States, who explain, “For example, consumers often make buying decisions based on their friends’ advice or their social network, which affects product diffusion and influences the dominance of a brand in a market.”

Using an agent-based simulation approach, marketers can better understand how shifts in behaviour can give rise to complex and interlinked patterns that might affect marketing campaigns or branding efforts. For instance, how does the rise of planet-conscious younger consumers impact sustainability messaging? Similarly, agent-based simulations are also being undertaken in the worlds of fashion and textiles to better understand the impact of personal influences (such as social communication) on this industry. Another example is in the entertainment sector where the tool is used to better understand sophisticated subscriber behaviours.

Dr. Francois Laurens

Dr. Francois Laurens is a company board member and adjunct faculty at GIBS. He has wide experience in investment banking and asset management, having served in senior positions in the financial services sector in South Africa and the UK. He has worked in markets in Europe, the Middle East and Africa, and before this, he practised as a corporate attorney.

Related

Three Lessons for (Business) Leaders in the Age of AI

Three Lessons for (Business) Leaders in the Age of AI

Clever Co-piloting: Is A.I. a Coach or a Tool?

Clever Co-piloting: Is A.I. a Coach or a Tool?

AI. Not as bad as your current boss?

AI. Not as bad as your current boss?