The Relationship between Nonstandard Work Schedule and Substance Use; New Evidence From NLSY97 Mona Khadem Sameni University of Wisconsin Milwaukee November 2014 The Initial Thought The shift working neighbor who always smoked outside the building… Nonstandard Work Schedule (Shiftwork) Other than: Regular Monday through Friday, nine to five schedule. U.S. Bureau of Labor Statistics (2004): One-fifth of all employed Americans Presser and Ward(2011): One-third of all dual-earner couples with children NLSY 97: • Regular evening shift • Regular night shift • Shift rotates • Split shifts • Irregular schedule or hours Why should we care? Substance Use Health hazards : • Higher probability of accidents at work and outside work (MacDonald, Wells et al. 1999) • Higher risk of HIV (Gorman and Carroll 2000) • Higher probability of major depression (Libby, Orton et al. 2005). Significant economic health costs : • Specialty treatments • Hospitals and emergency departments • Homicide and insurance administration participation • Hospitalization and premature mortality. Why should we care? (Continued) Efficiency and social costs • Lower productivity of the workforce • Higher crime rates such as drunk driving or drug possession and sale • Dependence on various pension types Potential Mechanisms Affecting Health Psychosocial mechanisms : • Related to family and social activities (Wirtz and Nachreiner 2010); (Drake, Roehrs et al. 2004) (Llena-Nozal, Lindeboom et al. 2004) Physiological mechanisms : • Circadian rhythms • Sleep patterns closely related to melatonin and growth hormone levels. (Davis, Mirick et al. 2012); (Arendt 2010); (Srivastava 2010) Previous Literature Job conditions Not a significant effect on drug use (Mensch and Kandel 1988) Working night shifts + more than eight hours/ rotating shift Higher amount of alcohol use (Trinkoff and Storr 1998) Nonstandard schedules Worse health conditions (Esp for men) (Ulker 2006) Being a shift worker Higher odds of drinking alcohol in short-term (Dorrian and Skinner 2012) Contributions to The Literature First to use a US nationally representative longitudinal survey in this field. First to investigate five separate substances. First to use survival analysis that corrects for frailty. Is Nonstandard Work Schedule Always “Bad”? • More free time (during the day) • Saving existing jobs (ILO 2004) • Helping with school • Fewer unwanted family interactions • Less tension and more relaxed pace • Accommodating “night owls”.(Finn 1981) Research Question Do shift workers indicate higher hazard ratios for the onset of substance use compared to regular day workers? People who held non self-employed jobs within the past year. Data U.S. National Longitudinal Survey of Youth, NLSY 97. Rounds 1-15 : 1997-2011 Five different categories : • Employed not shiftwork • Employed shiftwork • Unemployed • Out of the labor force • Military Reference group : Employed not shiftwork Data Continued Substances : • Alcohol • Excessive use of alcohol (Binge Drinking) • Cigarettes • Marijuana • Cocaine Methodology Duration framework using Cox (1972) Proportional Hazards Model. Why duration model? Impact of being in nonstandard work schedule on the length of time (length of spell) it takes to start using any type of previously mentioned substances. Results of Cox (1972) : The probability of starting to use any substance given that the individual has not done so yet. Model : (t) exp (β’x) • Regular control variables The Importance of Fixed Effects Estimation Besides Cox Model Previous work in this field suffer from selection bias Fixed effects : As if there were “before and after” shift work. Identical to DID estimation. Benefit of longitudinal data. Survival Plots Alcohol Binge Drinking Cigarettes Marijuana Cocaine Substance Use at Different Ages Drinkers at Different Ages Binge Drinkers at Different Ages Smokers At Different Ages Marijuana At Different Ages Cocaine At Different Ages Regression Results Pooled OLS Fixed Effects Random Effects -0.012 (3.09)** Cox without Frailty (no hazard rate) -0.090 (3.75)** Cox with frailty correction(no hazard rate) -0.066 (2.75)** Alcohol -0.003 (0.85) -0.019 (4.39)** Binge Drinking -0.001 (2.00)* -0.009 (2.01)* -0.009 (2.13)* -0.054 (2.47)* -0.057 (2.61)** Cigarettes -0.003 (0.98) -0.003 (0.88) -0.004 (1.35) -0.068 (2.89)** -0.105 (4.47)** marijuana 0.004 (1.19) -0.005 (1.67) -0.002 (0.86) -0.051 (2.33)* -0.071 (3.25)** cocaine 0.01 (5.35)** 0.004 (1.85) 0.006 (3.30)** -0.015 (0.68) -0.035 (1.58) Type of Substance Use Ruling out Reverse Causality! Shift workers use more substance or substance users do more shift work? Negative coefficients Less concern about upward bias. Pooled OLS gives smaller negative or positive coefficients Upward bias in previous literature using cross sectional data Conclusion Nonstandard schedules decrease the odds of the onset of drinking and binge drinking. Marijuana and cigarettes : Mixed results Cocaine: weakest evidence and even slightly on the opposite direction. Social aspect of work schedule Affecting teenagers and young adults more. Working at odd hours keeps them away from their circle of friends and family. Limitations and Future Work Not every nonstandard schedule has the same effect (Data limitation) Using a similar sample of adults seems appropriate. Application of exogenous variation methods.