April 29, 2016
Charles J. Courtemanche and Rusty Tchernis, Georgia State University and NBER, and Benjamin David Ukert, Georgia State University
This paper aims to identify the causal effect of smoking on body mass index (BMI) using data from the Lung Health Study, a randomized trial of smoking cessation treatments. Since nicotine is a metabolic stimulant and appetite suppressant, quitting or reducing smoking could lead to weight gain. Using randomized treatment assignment to instrument for smoking, Courtemanche, Tchernis, and Ukert estimate that quitting smoking leads to an average long-run weight gain of 1.8-1.9 BMI units, or 11-12 pounds at the average height. These results imply that the drop in smoking in recent decades explains 14% of the concurrent rise in obesity. Semi-parametric models provide evidence of a diminishing marginal effect of smoking on BMI, while subsample regressions show that the impact is largest for younger individuals, females, those with no college degree, and those with healthy baseline BMI levels.
Christopher J. Ruhm, University of Virginia and NBER
In this analysis, Ruhm utilizes death certificate data from the Multiple Cause of Death (MCOD) files to better measure the specific drugs involved in drug poisoning fatalities. Statistical adjustment procedures are used to provide more accurate estimates, accounting for the understatement in death certificate reports resulting because no drug is specified in between one-fifth and one-quarter of cases. The adjustment procedures typically raise the estimates of specific types of drug involvement by 30% to 50% and emphasize the importance of the simultaneous use of multiple categories of drugs. Using these adjusted estimates, an analysis is next provided of drugs accounting for the rapid increase over time in fatal overdoses. The frequency of combination drug use introduces uncertainty into these estimates and so a distinction is made between 'any' versus 'exclusive' involvement of specific drug types. Many of the results are sensitive to the starting and ending years chosen for examination, with a key role of prescription opioids for analysis windows starting in 1999 but with other drugs, particularly heroin deaths, becoming more significant in more recent years and, again, with confirmatory evidence of the importance of simultaneous drug use.
Tom Chang, University of Southern California; Joshua S. Graff Zivin, University of California at San Diego and NBER; and Tal Gross and Matthew J. Neidell, Columbia University and NBER
Chang, Graff Zivin, Gross, and Neidell investigate the effect of pollution on worker productivity in the service sector by focusing on two call centers in China. Using precise measures of each workers daily output linked to daily measures of pollution and meteorology, they find that particulate matter pollution, which easily penetrates indoor work environments, decreases worker productivity by reducing the number of calls that workers complete each day. These results manifest themselves at relatively modest levels of pollution, suggesting that these types of effects are likely to apply in major cities throughout the developing and developed world. When decomposing these effects, the researchers find that the decreases in productivity are explained by increases in time spent on breaks rather than the duration of phone calls. To the authors' knowledge, this is the first study to demonstrate that the negative impacts of pollution on productivity extend beyond physically demanding tasks and outdoor workers to indoor, white-collar work.
Partha Deb, Hunter College and NBER, and Carmen Vargas, Hunter College
This paper reports the results of an analysis of calorie labeling laws on body mass index (BMI). Deb and Vargas use county-level variation in implementation of calorie labeling laws over time in the U.S. to identify the effects of such laws. A difference-in-difference analysis using the 2003 to 2012 waves of the Behavioral Risk Factor Surveillance System, shows a statistically insignificant change in BMI for women and a small, statistically significant negative average treatment effect for men. The researchers estimate finite mixture models and discover that the average treatment effects mask substantial heterogeneity in the effects across three classes of women and men. For both women and men, the three classes, determined within the model, can be described as a subpopulation with normal weight, a second one that is overweight on average and a third one that is obese on average. Estimates from finite mixture models for women show that the effect is largely concentrated among women with BMI distributions centered on overweight. The effects for men are statistically significant for each of the three classes and substantively large for men in the overweight and obese classes. These results suggest that overweight individuals are especially sensitive to relevant information.
Dave Marcotte, American University
Poor air quality has been shown to harm the health and development of children. Research on these relationships has focused almost exclusively on the effects of human-made pollutants, and has not fully distinguished between contemporaneous and long-run effects. This paper contributes on both of these fronts. Merging data on plant pollen, human-made pollutants and ECLS-K data on academic skills, Marcotte studies the relationship between poor air quality in the first years of life on school-readiness, and the effects of ambient air quality on achievement of young children. He finds evidence that exposure in early childhood affects school readiness at the start of kindergarten, and that the effects of air quality on the growth of cognitive skills in math and reading continue into elementary school.
Marianne Bitler, University of California at Davis and NBER, and Christopher Carpenter, Vanderbilt University and NBER
Much research has studied the health effects of expanding insurance coverage to low-income people, but there is less work on the direct provision of care to the uninsured. Bitler and Carpenter study the two largest federal programs aimed at reducing breast and cervical cancer among uninsured women in the U.S.: one that paid for cancer screenings with federal funds and one that paid for cancer treatments under state Medicaid programs. Using variation in rollout of each program across states from 1991-2005, the researchers find that funding for cancer treatment did not significantly increase most types of cancer screenings for uninsured women. In contrast, funding for cancer detection significantly increased breast and cervical cancer screenings among 40-64 year old uninsured women, with much smaller effects for insured women (who were not directly eligible). Moreover, the researchers find that these program-induced screenings significantly increased detection of the early stage pre-cancers and cancers of the breast. The results suggest that direct provision can significantly increase healthcare utilization among vulnerable populations.