Michael Ricks and Dr. Lars Lefgren, Economics Department

Introduction

The consumption of illicit drugs is on the rise. In 2013 the number of people in the United States who admitted that they had used an illegal drug in the last month rose to an all-time high of 24.6 million (NIDA). As drug use spreads across the nation, so do efforts and initiatives to curb its rampant negative effects, ranging from stricter law enforcement, more comprehensive treatment, and broader inoculation efforts. While it seems that few of these interventions have proven to be effective—let alone cost effective—Swensen (2015) demonstrated that more available significantly reduces the number of drug related deaths in the counties where they function. This project assessed whether the availability of drug treatment has positive spillover effects on negative behaviors like child abuse that are highly correlated with substance abuse.

Methodology

To answer this question, we collected data from three sources. First, using the U.S. Census Bureau’s County Business Patterns data, we can determine the number of drug treatment facilities in each county. The data on child abuse came from the National Data Archive on Child Abuse and Neglect at Cornell University. Using the National Child Abuse and Neglect Data System Child File data from 1998-2015, we received detailed data about all cases of child abuse reported to local and state agencies (although county information was withheld for cases in counties with fewer than 1000 annual reports). From these data we determined the yearly rate of substantiated reports of abuse and neglect per thousand minors in the county. We also used county demographic data from the Surveillance, Epidemiology, and End Result Program’s County Attributes dataset to create a set of controls for each specification or of regression analysis.

Using a regression analysis, I examined the relation between one year’s abuse rate and the number of drug treatment facilities in the county in the previous year. I did this while including demographic controls as well as county and state-by-year fixed effects (essentially drawing comparisons within each county rather than across different counties) and while weighting each observation by the county population.

Results

Table 1 presents the results of the regressions specified in the previous section. Each column presents the results of progressing regressions. The first is only uses the county and year fixed effects, then adding state-by-year fixed effects, and finally demographic controls. Although the effect size looks relatively small, it is significant and has a surprising direction—it suggests that an increase of one drug treatment facility increases the rate of child abuse by about 0.02 (0.12%) per 10000 children.

Discussion and Extensions

Because these results were counter-intuitive, I examined three potential reasons for the relationship: (1) treatment facilities increased reporting of child abuse (but not necessarily abuse), (2), child services agencies in counties with high drug abuse could bring in drug treatment facilities (i.e., reverse causality), and (3) drug treatment may help poor caregivers retain custody of their kids (leading to continued neglect). I conclusively ruled out reasons one, but couldn’t show anything definite about reasons two or three.

Conclusion

Although it makes intuitive sense that providing better assistance for substance abusers would decrease child abuse rates, we are unable to validate that intuition with the data. Our results show that on average there is a small (0.02) increase in the rate of child abuse in counties that developed a new drug treatment facility in the year before. I also showed that the effect does not operate through increased reporting—in fact, if anything, reporting decreases slightly after the introduction of another facility—however it is possible that it could operate through caregivers either with reverse causality or by helping them to become sober (while possibly failing to help them treat their children better).

References

County Attributes. Surveillance, Epidemiology, and End Results (SEER) Program.
County Business Patterns. United States Census Bureau.
National Child Abuse and Neglect Data System Child File. NDACAN.
National Survey on Drug Use and Health. 2013. National Institute on Drug Abuse.
Swensen, Isaac D. 2015. “Substance-abuse Treatment and Mortality.” Journal of Public Economics 122 . pp 13–30.