The Relationship Between Hours Worked and Alcohol Use Disorders

The relationship between an individual=s annual hours worked and various measures of alcohol use disorders is estimated using data from the National Longitudinal Survey of Youth. Previous research on wages, hours worked, and employment, using this data has focused on the relationship between these productivity measures and frequency and quantity of alcohol use variables. An emerging contention among some researchers is that while alcohol consumption measures may not affect labor productivity, measures such as alcohol use disorders that reflect problematic drinking do. A time series cross-sectional model is used to estimate the relationship between annual hours worked of individuals reported in two time periods and standard labor supply determining characteristics as well as measures of alcohol use disorders constructed in the spirit of criteria used in the Diagnostic and Statistical Manual of Mental Disorders, third and fourth edition. Also, the relationship between hours worked and other indicators shown in the literature to be related to alcohol dependence or abuse, such as Abinge@ drinking is investigated. Results are compared to those obtained using alcohol consumption measures. Key Terms: NLSY, Alcohol Disorders, Labor Supply, Time Series Cross-Sectional Model Background and Significance Issues related to alcohol and/or illicit drug use has continued to be the focus of many researchers as well as national organizations on health and social policy. According to the National Institute on Alcohol Abuse and Alcoholism (NIAAA) web site, alcohol is second only to tobacco as the most abused substances in the United States. Fiscal year appropriations for NIAAA activities in 2001 were more than $340 million and estimated at $384 million for fiscal year 2002. Projections of 1992 estimates put the 1998 economic cost of alcohol abuse at $184.6 billion, almost 73 percent of which is attributed to lost productivity (Harwood, 2000). The major studies on the effect of alcohol use or alcohol use disorders on labor supply, or labor force status carried out in recent years are discussed below. A number of the studies utilized the data from the National Longitudinal Survey of Youth (NLSY) compiled by the Center of Human Resources Research at The Ohio State University. The common model used, was the cross-sectional model. Only the studies by Bryant et al. (1996 and 2001) used longitudinal estimates in addition to cross-sectional estimates. Using NLSY data, Kenkel and Ribar (1994) investigate the relationship between measures of “problem drinking“ including a constructed measure of alcohol abuse and alcohol dependence, and annual earnings, annual hours worked, and martial status. Their ordinary least squares (OLS) estimates show that while controlling for family background, ability, and schooling, alcohol dependence reduce annual male earning by about six percent while female earnings are increased by about 12 percent. Neither alcohol dependence nor alcohol abuse was found to be significantly related to male or female annual hours worked. Kenkel and Ribar also used an instrumental variable (IV) approach to control for unobserved heterogeneity and found alcohol abuse was associated with about a 30 percent decrease in earnings of men and a decrease of about 28 percent in earnings of women. Further, alcohol dependence was associated with about a 31 percent decrease in male earnings. Consistent with OLS results, the IV approach did not find significant relationships between alcohol abuse or dependence and male hours worked. In the female hours equation, however, the IV approach identified a positive relationship between hours and alcohol abuse or dependence A study by Mullahy and Sindelar (1996) analyzed the relationship between various measures of “problem drinking,” including alcohol abuse or dependence, on labor force status: employed or unemployed. Using an OLS approach they found that males with an abuse/dependence problem were more likely to be unemployed and less likely to be employed than those not classified with an abuse/dependence problem. For females, abuse/dependence was associated with a greater probability of being unemployed. When they control for unobserved heterogeneity through an IV approach they found much larger detrimental impacts even though the parameter estimates were insignificant at conventional standards for all cases. Bryant, et al., (1996) addressed the issue of the relationship between alcohol/drug use and employment. Both cross-sectional data and longitudinal data were used in the study to model the relationship between current and past substance use on the probability of employment. The model allowed for a direct relationship and for a relationship working through human capital variables. Separate equations were estimated for men and for women. For women, both past use and current use were associated with a decrease in the probability of being employed. For the male sample, current substance use was associated with an increase in employment probability, whereas past use was associated with a decrease. One drawback associated with this study is that the employment variable used was determined by the respondent’s employment status only in the week prior to the 1905 ASA Social Statistics Section