Suresh Naidu, a Columbia University researcher using text-mining tools to study the role of politics in shaping economies, has received a 2015 Sloan Research Fellowship. The two-year $50,000 award recognizes young scholars with the potential to make substantial contributions to their field. 

An assistant professor in the School of International and Public Affairs, and a member of the Data Science Institute, Naidu uses natural language processing tools to analyze modern and historical texts to reveal political narratives and build historical data sets. In a recent study, he and his colleagues examined the ideological leanings of economists themselves.

Using campaign contributions, signed petitions and the collected academic works of 2,000 economists, Naidu and his colleagues developed an algorithm that could predict with 74 percent accuracy whether a randomly selected economist outside of their data set was liberal or conservative. They found that financial economists tend to lean right while labor economists lean left, and that economists in business schools are generally more conservative. “We’re as much in the fray of politics and policy as everyone else,” said Naidu recently from his office in Columbia’s International Affairs Building.

Naidu worked in computer graphics before graduating from Canada’s Waterloo University with a degree in mathematics. After graduating from the University of California, Berkeley, with a PhD in economics he went on for a fellowship at Harvard University.

He arrived at Columbia in 2010, and has continued to explore the many ways that politics and institutions shape the economy. In a 2011 study published in the Quarterly Journal of Economics, he and his colleagues used declassified CIA records to suggest that officials had traded on inside information about pending CIA coups in Guatemala, Iran, Chile and Cuba to make money in the stock market.

In a 2012 study for the Brookings Institution, a Washington D.C. think tank, he and his colleagues set out to learn whether Congress has become more polarized. They identified lists of “partisan phrases” in the Congressional Record and scanned more than 2 million digitized books written in English since 1873 for the same phrases. Comparing those partisan words with roll-call voting records, they found that political speech became more polarized in books in the late 1990s but that Congressional speech appeared to be more polarized through the late 19th centuries. In particular, they also found that partisan phrases often first appeared in published books before they appeared in Congressional speech.

More recently he has looked at the impact of coercive labor laws on wages. He showed in a 2013 study in the journal American Economic Review that Britain’s industrial era master-servant laws depressed wages by making it a crime for workers to switch jobs. When Britain lifted the laws in 1875, wages rose, but so did wage volatility as employers laid off workers during economic downturns. Naidu is currently studying how evolving labor policies in the Middle East are impacting guest-worker wages, as well as using NLP tools to study how judicial opinions shape the terms of labor contracts.

“A lot of the economy sits embedded in things that are political and cultural,” he said. “Natural language processing tools are a great way to quantify and measure the importance of politics, ideas, and laws in shaping the economy.”  

Naidu is also a member of Columbia’s economics department.