Prof Li Jiangtao

Feature on Prof Li Jiangtao, Associate Professor of Economics
School of Economics, Singapore Management University 

More robust models for real life answers 

By making modelling mechanisms better at capturing how people act in real life, Associate Professor Li Jiangtao hopes to move mechanism design from theory to practice in designing effective systems. 

Most games, from outdoor activities like Capture the Flag, boardgames like chess or computer games like Fortnite, run on a common aim—to win. While the objectives may differ, no game is complete without a set of rules that lead to strategic behaviours as players try to best one another. 

In economics, game theory works in a similar way, says Associate Professor Li Jiangtao. He uses mathematical models to simulate how rational agents—or people who play an active role in a system—act strategically to maximize their own interest. In other words, how people try to win, given a system’s rules. 

“Game theory offers a systematic approach for us to analyse the behaviour of agents. It helps us to examine the world and make sense of why things work the way they do,” he says. “My research in mechanism design uses game theory as a building block.” 

While game theory attempts to describe how agents will act, mechanism design examines a system from the other end—asking how different rules can influence the players’ interactions with each other. The aim is not to win but to identify the most optimal set of rules for the game towards a given purpose. 

“Once we understand how agents behave, we have a systematic approach to design better institutions for them,” says Li. 

In policymaking, for example, Singapore’s government might use mechanism design to figure out the rules for distributing Certificates of Entitlement, a license to own vehicles used to manage the number of cars on the road. Mechanism design helps public officials compare find the mechanism that best promotes that policy goal. 

Mechanisms for maximal gain 

While mechanism design applies to many practical scenarios, such research is still heavily theoretical in nature. Li, however, does not define impact as changing how the world works in one stride. Instead, he is most happy when other researchers take inspiration from his models and frameworks, leading to the birth of new ideas and endeavours to address related research questions. 

“The exciting part about research is trying to discover something previously undiscovered. By doing that, we gain a unique and new perspective of looking at the same thing, even about longstanding classical economic problems.” 

He did just that in a recent study, Are simple mechanisms optimal when agents are unsophisticated?, in collaboration with Northwestern University Associate Professor Piotr Dworczak. 

In a system with simple mechanisms, agents can easily identify the optimal strategy and behave accordingly to get what they want. In contrast, complex mechanisms demand more in-depth and sophisticated reasoning that could confuse agents who whose behaviour is not fully rational. 

System designers often prefer simple mechanisms as it means they can be “confident about predicting the agents’ behaviour and about what will happen under their system,” says Li. Moreover, the agents are also more likely to participate since they understand how to ‘play’ the game. 

But Li and Dworczak flip the script in their new work, showing that there may be cases where a complex mechanism is useful even if the agents are not strategically sophisticated. Using an approach called robust dominance, where the production of a higher payoff influences the designer to prefer the complex mechanism at all times, they proved that the inability of the designer to predict the outcome of a complex mechanism does not justify the choice of employing a simple mechanism instead. 

“The designers may not need to care about whether the mechanism is simple or not, because that’s not their objective. Their objective is to maximize their own utility,” Li explains. “Some of the agents may be confused, but as long as they are willing to participate and as long as the eventual profits are higher or the system achieves their desired outcomes, then the designer could actually choose the complex mechanism.” 

Better models for broader applications 

Inevitably, mechanism design models operate on many assumptions about how people will respond to rules and what will motivate them. Assumptions make economic modelling more straightforward. One standard assumption, for example, is the level of strategic sophistication of the agents. But this comes with an obvious weakness.  

“It’s safe to say that everyday real-life people are not as rational as is assumed in typical models,” Li points out. 

“Mechanisms work well when you have identified the optimal mechanism for a particular set of assumptions, and the assumptions are true. But they fail miserably in the much more frequent cases where people act ways that make the assumptions false and unhelpful.” 

Given this dilemma, Li is steering his research towards robust mechanism designs that work on a broader set of assumptions and so predict outcomes in a larger set of situations. He hopes this approach can lend greater credence to the data models and analyses used in the field. 

“If we can relax these assumptions, we can make more practical mechanism designs and so be more helpful to real-life policymaking,” he adds. 

Empirical evidence is important to this goal, says Li. He plans to conduct experiments by implementing mechanisms in controlled environments and testing whether the results match up with the theoretical findings. 

In support of this research, Li is grateful to have twice received the Lee Kong Chian Fellowship, an annual award  for research excellence. 

“With the funding, I can meet with other researchers and initiate potential collaborations. I can also run experiments to test my theoretical findings and revise my models to coincide with actual behaviour even more,” he says. “I’m very privileged to win the award.” 

As he continues to run the numbers on his models, Li hopes to inspire fellow researchers and students alike to see the value of game theory and mechanism design in analysing real-life phenomena—understanding “what happens, why things happen, and what will happen” in the world around us.