91 Pages Posted: 20 Mar 2014 Last revised: 22 Jul 2014
Date Written: June 27, 2014
Executive Order 13,563 mandates that administrative agencies empirically evaluate the costs and benefits of proposed regulation, but the tendency to enact rules in response to prior crises leads to regulation that lags behind future market-disrupting events. I argue that policymakers should utilize statistical prediction with social trends as an “early warning” system to ascertain where regulation will be most beneficial in the future. As an empirical demonstration, I identify newly popular topics in the 468 billion words of the Google Ngram corpus, a sample of all English-language books published in the United States, and link these emerging trends to specific agencies by regulatory mandate. Two leading causes of the 2008 financial crisis --- subprime lending and credit default swaps --- were within the top 10 credit-related topics requiring the Federal Reserve’s attention from 2003 to 2005. Speculative real estate activity known as “flipping” properties is identified in the mid-2000s. False positives are suppressed by detecting only those trends related to each agency's regulatory mandate. I validate the predictive model by showing that social trends systematically predict future agency activity, suggesting that predictive data can both guide agencies on future interventions as well as shorten the policy lag in market regulation. While some “black swans” will remain unforeseeable, this retrospective study demonstrates the need for a real-time system with live, streaming data to ensure that regulatory policy proactively adapts to an evolving society.
Keywords: cost-benefit analysis, regulation, big data, administrative law
Suggested Citation: Suggested Citation