by Jackie Swift
Scientists have turned to artificial intelligence (AI) models during the COVID-19 pandemic to predict the increase, decrease, and spread of infection. Typically, the models depend on fixed assumptions such as an externally given transmission rate for the virus and a specific pattern to human movements that supposes two people will meet with a given frequency. While this approach can shed light on the situation, it is missing a key component, says Lin William Cong, Graduate School of Management.
“We know that the spread of the disease and the movement of people are driven by economic incentives,” says Cong (pronounced Ts-óh-ng). “If confirmed cases rise, I will stay home because I worry about the risk of catching COVID. But if I lost my job for three months, I have to go out and make a living despite the pandemic. There are some very good AI predictive models that look at the dynamics of COVID and predict how employment will evolve, but they typically look at these issues separately. What if we allow economic incentives to be taken into consideration as part of the COVID model, too?”
To explore this, Cong recently joined with Beijing-based researchers Ke Tang, at Tsinghua University, and Jingyuan Wang and Bing Wang, both at Beihang University, to design a pandemic AI model that includes economic factors. Their model combines a Google community mobility measure that summarizes the movements of people; an employment model based on the assumption that if the employment rate is low, the incentive to work is high; and an epidemiology model on the spread of the disease.
“They all interact with each other, and we use that to predict the evolution of the pandemic and also the employment rate,” Cong explains. “This is one way economic reasoning can help traditional science disciplines.”
Delving into Financial Technology
Extending the tools and framework of economics across disciplines like this is one of Cong’s interests. While his research encompasses entrepreneurship and information economics, as well as new financial systems and economic policies in emerging economies, especially China, his main area of focus is the up-and-coming realm of financial technology — or FinTech. FinTech covers technologies like mobile banking and cryptocurrencies, as well as economic big data, applications of AI, analytics for structured data, digital platforms, and ecommerce.
Economists have a lot to contribute to FinTech, Cong maintains. To begin with, they can tailor the technical tools that originated from science and engineering so that they are relevant to the world of constantly evolving business and economic issues. Cong points out that in computer science, for example, researchers typically make set assumptions about agents’ behavior. “Take a digital network: in computer science, people might assume that 30 percent of the nodes [individual computers or agents that are part of the network] are faulty and aren’t going to function,” he says. “Or that 10 percent are malicious; they’ll attack certain storage or computation.”
“What if we allow economic incentives to be taken into consideration as part of the COVID model, too?”
Economists, however, make assumptions about people’s preferences. They may assume people like more leisure or more money, for instance. “Economists determine behaviors by looking at the incentives that push agents to take certain actions,” Cong says. “And as we go into these bigger digital networks, these distributed systems like cryptocurrency blockchains, we have to think about incentives. We’re in a very large, open system, so we can’t assume people are going to behave like machines.”
Also, as technology advances, more questions around financial inclusion and algorithmic discrimination emerge, and economists have expertise to contribute in that area, as well, Cong says. “We don’t really know exactly how AI works,” he points out. “It’s a black box. Scientists and engineers are starting to look at issues related to the socioeconomic implications of technology, but we social scientists have been thinking about them for a long time. What are the inequality implications of a new technology, or an innovation, or more or less competition? What are the regulatory implications? If we really want to understand what’s going on with AI, develop relevant policy, and increase social welfare, the way forward is for economists to work together with computer scientists, statisticians, and engineers.”
To further this synergy, Cong serves as the founding faculty director of the FinTech Initiative at Cornell University. “Our goal is to be a portal where we can unite researchers at Cornell and around the globe who come from all kinds of backgrounds — business, computer science, statistics, information systems — to work on these interdisciplinary topics,” he says.
Pooled Cryptocurrency Mining
Continuing his work in FinTech and economic data science, Cong has also looked at the phenomenon of pooled cryptocurrency mining, the verifying of individual blocks of data in a cryptocurrency’s blockchain by competing nodes that band together into groups to solve puzzles. Pooled mining aggregates computation power and increases the odds that the group will win the mining competition and share a payoff. At the same time, conventional wisdom says that pooled mining will result in ever bigger pools, which goes against the blockchain ideology of decentralized finance.
Cong and his collaborators, however, found that the need for nodes to spread risk by diversifying across pools, combined with the tendency for successful pools to increase the fee for joining them meant that pools never grew disproportionally large. “It’s not so straightforward that mining pools lead to overconcentration,” he says.
By contrast, the researchers found a connection between mining pools and increased environmental damage connected to the huge amounts of electricity needed to run the computations. “When mining pools grow, people devote more power to mining because the pools lower personal risk,” Cong says. “That’s causing greater environmental damage.”
Ecommerce and the Entrepreneur Gender Gap
In another project, Cong investigated the effects of ecommerce on entrepreneurship in China. He and his collaborators found that ecommerce promoted the growth of entrepreneurship generally, but specifically it narrowed the gender gap between male and female entrepreneur-led startups.
“In many countries, including China, women often take more care and responsibility for children and elderly family members,” he explains. “That makes it difficult for women to have regular career advancement and even more difficult for them to start a company. The digital platforms give them the flexibility to become online merchants and web influencers.”
The online startups are typically small. They might design T-shirts or custom-make mugs. “We’re not talking about a Bill Gates,” Cong says. “Individually these small startups aren’t comparable to the big, superstar companies, but in aggregate, as part of China’s gross domestic product, they are just as big as Facebook or Google. In that sense, digital platforms are really transforming the landscape of entrepreneurship.”
New Disruption Brings Opportunity
Reflecting on his discipline of finance and economics, Cong sees now as a time for economists to be particularly relevant in their research contributions to technological innovation. “Economists missed out on the internet revolution 30 years ago,” he says. “Computer scientists were moving really fast, having concrete impacts, and now the internet is in every aspect of our lives.
“That makes me feel that I want to be relevant in my work — and not only my own work but my fellow economists’ work, too, be it fundamental research or applied studies,” he continues. “I want them to get involved and engaged. We can make a big impact. We are at a juncture of the revolution — a new disruption from AI, from blockchains, from distributed systems — and we have something to add.”