Workplace drug prevention programs: does zero tolerance work?

Abstract

Current drug policy in the US military mandates frequent random drug testing of service members and dismissal of those who test positive for illegal drugs. This article analyses the economic costs and benefits of this zero tolerance policy as applied in the US Navy. Program effects consist of the actual number of detected users and the predicted number of deterred potential users. Productivity losses imposed by drug users are based on reported annual workdays lost due to drug use in the Navy. The productivity losses avoided by deterring and detecting users constitute program benefits. Program costs include the cost of replacing service members who are dismissed under the zero tolerance policy. Net benefits are sensitive to three key parameters – the deterrence effect, replacement cost, and productivity losses due to drug use. The results show that net benefits are negative for most plausible values of the key parameters.

Acknowledgements

We thank Rosalie Pacula for helpful comments and Markus Hey, Mary Jane McCrae, Antonio Martinez and Melissa Potter for research assistance.

Notes

¹ A recent survey found that 50% of large US firms test current employees, and 60% test job applicants AMA, (2001). Macdonald and Wells (1994) survey eight types of drug testing programs.

² The Omnibus Transportation Employee Testing Act of 1991 requires employers in the transportation industry to test all workers who hold safety-sensitive jobs.

³ Both civilian and military workers are equally subject to state laws governing illegal substance possession and use.

⁴ Illicit drug use peaked in 1979 and started to decline after 1981 Harwood et al ., (1998).

⁵Martinez (1998) surveys the evolution of the four service's drug policies.

⁶ Note that using the all-industry civilian comparison group may not be a serious flaw. While some civilian workers face drug testing of some type, few face the frequent random testing imposed by the military, and virtually none face a zero tolerance policy Macdonald and Wells, (1994). To the extent that some members of the comparison group are tested the regression estimates will understate the deterrence effect. This is acceptable given that our goal is to create a conservative estimate of the deterrence effect.

⁷ If firms can shift the costs imposed by drug-using employees, hedonic wage studies would provide useful information on productivity losses from worker drug use. Unfortunately, a large literature that has used this approach has found widely varying results, including a positive relationship between drug use and wages French et al ., (1998 survey this literature). Note that if firms can shift the costs of drug use to workers in the form of lower wages, there is little incentive to use drug tests. Drug testing is efficient only if drug abusers impose external costs, productivity is hard to measure, or wages cannot be adjusted for individual productivity differences. Administrative wage setting is one reason drug testing may be efficient in the military.

⁸ Each incident of tardiness, leaving work early and performing below normal was weighted by 0.25 days; each absence was weighted by 1.0; and, an injury was weighted by 5.0 days.

⁹DiNardo (1994) points out that if employers pay workers the value of their marginal products, a worker's decision to miss work is fully internalized via a lower wage (so long as absenteeism does not create negative externalities). As noted earlier, administrative wage setting prevents military wages from adjusting to marginal changes in work effort and absenteeism. Moreover, absenteeism is likely to generate externalities in the military's team production environment, which increases military manpower expenditures and reduces unit production.

¹⁰ By way of comparison, Ippolito (1996) found that the wage discount for government workers with excessive absences or tardiness is about 7%, which is near the middle of our range of lost output estimates.

¹¹ For comparison purposes we use other sources to calculate alternative estimates of the value of lost productivity per drug user. In 1992 there were an estimated 11.4 million civilian illicit drug users over age 12. Adjusting for the number of users between ages 12 and 18 and the number of unemployed, there were roughly 10 million employed adult drug users. Harwood et al . (1998) estimate total annual US productivity losses of $14.2 billion in 1992, which yields a per worker loss in 1992 of $1420 ($1604 in 1999 dollars). This estimate falls near the top of our range of degradation factors. Harwood et al . also estimate the annual earnings loss per drug dependent worker to be $2352 in 1992 ($2657 in 1999 dollars), which exceeds the top of our range.

¹² The number of users detected is smaller than the number deterred because the probability of detection is fairly low. Borack (1995, 1997) demonstrates that the expected time until detection ranges from 1 year to as high as 10 years for a ‘nongaming’ user and even longer for a ‘gaming’ user.

¹³ Estimates of per worker turnover costs in private firms also vary widely. A recent survey reported 15 different estimates of turnover costs for a typical worker (earning $16 000 annually): the range in costs is from $3500 to $25 000. In private firms direct costs include some training costs, but most firms deliver only specific training. By contrast, the military delivers extensive general training, which boosts costs. (see ‘Employee Turnovers Costs, 2003’, http://www.sashacorp.com/turncost.html).

¹⁴ Data provided by Navy Bureau of Personnel, Drug Detection and Deterrence Branch.

¹⁵ About one-third of Fortune 500 companies have set up employee assistance programs (EAP) that refer drug abusing employees to treatment programs Felman and Petrini, (1988).

¹⁶Devine et al . (1989) obtained a benefit-cost ratio of 12 : 1 for the Navy's alcohol rehabilitation program, despite the fact that only about 50% of participants were successfully rehabilitated.

¹⁷ Whether the conclusions here apply to private firms is questionable. Damages imposed by drug use may be less when safety is not an overriding issue. On the other hand, private companies may face greater legal liability for damages imposed by drug-using employees.