Profs Takashi Kunimoto & Shurojit Chatterji

Feature on Prof Takashi Kunimoto, Associate Professor of Economics & Prof Shurojit Chatterji, Professor of Economics
School of Economics, Singapore Management University 

Enabling more robust institutional design

A fundamental theorem in mechanism design posits that the only institution that can elicit truthful information in all situations is dictatorships. Through their research, Takashi Kunimoto and Shurojit Chatterji explore novel and robust methods of institutional design that address this outlook. 

It’s an age-old question – are humans inherently selfish? For centuries, thinkers have proposed that each individual seeks to advance their own gains, compete against one another for resources and to accumulate power and possessions. One school of thought even suggests that human beings must form social contracts and governments to prevent their selfish, violent tendencies from taking over.  

We see this interesting dynamic play out in a subfield of economics known as “mechanism design”. Takashi Kunimoto, Associate Professor of Economics at the Singapore Management University (SMU), explains that mechanism design studies the construction of mechanisms that induce socially desirable outcomes in the presence of rational but selfish agents. “Essentially, it is a tool that allows us to think about institutional design,” he says.  

In mechanism design, economists act as institutional engineers, choosing an outcome and then arranging a set of market rules and conditions to achieve it. “Think about it as a group of people trying to coordinate the actions of the collective whole in certain directions, so that we can somehow achieve the optimal outcome, or maximum wellbeing, for everyone in the group.”  

Shurojit Chatterji, Professor of Economics at the SMU, elaborates that mechanism design is central to economics – which loosely speaking, is the study of institutions that are useful in allocating resources in an efficient way across society. “However, any such institution has to confront the fact that individual preferences, which efficiency is based on, are their own private information. This data has to be elicited from multiple agents. The dilemma is that once these agents understand the rules of an institution, they are liable to misreport their preferences in order to seek self-benefit,” he says.  

This brings about tension between the players, who only care about their own private utility, and the social planner, whose aim is to reach a socially desirable outcome. The underlying question, therefore, is how to incentivise these agents to cooperate, in order to reach desirable social outcomes.  

Getting around a grim theorem 

Much of the work in the realm of mechanism theorem is aimed at addressing the famous fundamental Gibbard-Satterthwaite Impossibility Theorem. Prof Chatterji explains, “This theorem states that if you do not put any a priori structure on preferences of agents, then the only institutions that will elicit truthful information about their preferences in all situations are going to be dictatorships.”    

Therefore, the economics literature that evolved from this negative result has focused on studying scenarios where not all preferences are admissible. Assoc Prof Kunimoto adds, “This is an area that I’m currently concerned with in my research as well. After all, most people would prefer not to live within dictatorships, where misrepresenting their preferences are inconsequential. We want to give them the freedom of choice while encouraging them to offer personal information that helps in institutional design. So, the big question is: What would be a good characterisation of alternative scenarios where one can move away from dictatorship and form more ideal institutions and mechanisms?”    

Finding a robust way around   

Assoc Prof Kunimoto and Prof Chatterji recently successfully clinched the Singapore Ministry of Education (MOE) Academic Research Fund (AcRF) Tier 2 grant for a joint project titled “Robust Design of Institutions: An Implementation Theory Approach”.  

Explaining that the research proposal focuses on the interaction among agents that occurs within the framework of an institution, Assoc Prof Kunimoto says, “In this case, we are assuming that the central authority, which is the social planner, has the power to commit to an institution – that is, a specific set of rules that govern what actions agents may take, and what the consequences of these actions are.”  

However, the authorities will not be able to dictate the outcomes that result, since these are determined by the actions agents take and cannot be a priori controlled. Prof Chatterji explains that hence, to achieve socially desirable outcomes, the design of the institution has to provide the agents with the right incentives.  

“This means that the type of information we want to elicit from agents should not be too demanding. We must find ways to strike a balance between asking for too much detail and making do with more fundamental basic information on preferences. This is because when we ask for too many details, people sometimes become suspicious and are more likely to willingly or inadvertently misrepresent their true preferences,” he explains.   

The technological interest of mechanism design 

In recent years, computer scientists have become increasingly interested in leveraging the concepts of mechanism design for the design of software programmes. Prof Chatterji says, “There is now feedback between literature in the two areas, and in fact, a lot of research in mechanism design these days is actually carried out by computer scientists.”  

An example would be in the case of self-driving cars. “An autonomous car is set to fly on the roads, and it is imaginable that the car may have to make a decision at some point between colliding into a baby or colliding into an elderly person. The thing is, on what basis should the autonomous vehicle decide which is a better option?” he asks.  

Inevitably, the topic now overlaps with social choice and ethics, as these ideas and choices will have to be operationalised and put inside the algorithms that are driving the machine. He adds, “Whenever you try to decentralise decision-making, you also have to provide the right incentives for making the correct decisions. But what the correct decisions are, is a matter of debate.”  

Reflecting on how the realm of technological development has become an area where ethics, social choice and philosophy are all intertwined together, Prof Chatterji says, “It will be interesting to see how they evolve together over time.”