Big Data to Infer Beliefs, Information and Unobserved Costs
Paper Session
Friday, Jan. 6, 2023 10:15 AM - 12:15 PM (CST)
- Chair: Laura Veldkamp, Columbia University
Air Quality, Avoidance Behavior and Welfare
Abstract
The cost of damages caused by air pollution runs into billions of dollars annually. In measuring costs, researchers hardly ever include the cost of changes in behavior as a result of air pollution. Because these costs represent a loss in surplus and avoidance behavior affects the severity of health effects caused by air pollution, estimates of the effects of air pollution that do not take avoidance behavior into account are most likely biased downwards. To obtain unbiased cost estimates, some measure of avoidance behavior is needed in cost calculations for the effects of air pollution. It is important for policy makers to know the full costs of air pollution so that they can make decisions on whether to focus environmental policy on air pollution reduction or on encouraging behavior that reduces exposure to air pollution.In this paper, I study changes in people’s movement patterns in response to changes in air quality. Using data on people’s mobility patterns from SafeGraph and historical air quality data from the EPA, I measure how much people change the time they spend at home, the time they spend away from home, and the distance they travel from home in response to the EPA’s air quality index.
Using a fixed effects OLS model, I find that people are on average expected to spend 4 more minutes at home and travel 195 meters less away from home on days of unhealthy air quality than on days of good air quality. This is an indication that people change their movement patterns or behaviors in response to air pollution, thus estimates from my analyses can be used as a measure of avoidance behavior.
Bank Monitoring with On-Site Inspections
Abstract
Although a seminal body of theoretical literature asserts that a key advantage of banks is their ability to monitor borrowers, we are the first paper to empirically examine the theoretical determinants of bank monitoring within non-syndicated loans. Using a proprietary set of transaction-level database of nearly 30,000 multiple-draw construction loans, we observe the frequency at which the bank monitors these loans by conducting on-site inspections, the contents of the reports, and the borrower actions over the life of the loan. Consistent with theoretical predictions, we find evidence that banks tradeoff monitoring with many types of favorable loan origination terms. Monitoring is also less frequent for loans where the bank has a prior relationship with either the borrower or project contractor, suggesting that banks may be transferring information across projects. Furthermore, we show that negative on-site inspection reports are associated with a greater likelihood of banks denying draw requests, indicating that the information that banks collect during the monitoring process is important to their decision-making. In subsequent analysis, we provide a comprehensive analysis on the determinants of construction loan default and show that more monitoring ultimately decreases loan default within an instrumental variable framework.Discussant(s)
Kristoffer P. Nimark
,
Cornell University
Galina Hale
,
University of California-Santa Cruz
Yiming Ma
,
Columbia University
JEL Classifications
- D8 - Information, Knowledge, and Uncertainty
- Q5 - Environmental Economics