One particular concern with cannabis legalization or drug legalization in general is that it may reduce the amount of productivity. The classic story goes like this: people use drugs, particularly cannabis which has more sedative effects than the other drugs up for consideration, and they become addicted to them, become lazier, somtimes leaving their jobs, going on welfare, etc. and getting sucked into a life of drugs. Perhaps this is a more extreme version of the argument. If so, I am also disagreeing with the moderate version. Anyways, does it hold up?
One way we could look at the effects of cannabis legalization on productivity is simply looking at a broad measure like labor force participation. This method is particularly susceptible to error because it doesn’t take into account if people work less hours and, more broadly, the effects of cannabis legalization on creating new businesses, which seems apparent. We will get to the former later, but we can start with this as one broad, rough measure of productivity.
A very new article by Chakraborty et al. (2020) found the amount of cannabis access in an area (different counties allowed cannabis sales and other didn’t) predicted greater labor force participation and lower unemployment rates. The authors controlled for multiple variables such as population, early-adoption of liberal cannabis policies, etc. Not all of the jobs were in services. For example, there were increases in employment for construction jobs and manufacturing. The former is due to new businesses (to sell cannabis) being built. Of course, some industries didn’t face any significant increase in employment, eg. natural resources and mining. Most of the increase was due to manufacturing, which is of course fairly important to America right now. The usage of county-level variation is primarily of interest because it will serve the argument more reliably than, say, state variation.
This county analysis is very good. We can also use random samples which are nice for checking dose-dependent relationships. There was a study done by Davis (2019) which found that disability patients actually ended up being slightly more active in the labor force and were less likely to be unemployed following being prescribed medical cannabis. Another study followed teenagers through high school in Washington D.C. following decriminalization. They found that working teenagers were more likely to use cannabis than non-working teenagers before and after decriminalization and the effect of decriminalization on cannabis usage was positive and stronger in the working group (Graves et al., 2019). This study is, of course, of less interest than the others primarily because the authors argue this is mainly a testament to the worse health of working teenagers.
Light et al. (2016) found cannabis legalization was responsible for the creation of over 18,000 jobs and that “because the industry is wholly confined within Colorado, spending on marijuana creates more output and employment per dollar spent than 90 percent of Colorado industries.” Of course these effects would dissipate with federal legalization but some gains would probably remain. Some studies find negative correlations between cannabis usage and employment (DeSimone, 2002; French et al., 2001). These results however are based on cross-sectional data rather than causal observational designs which will create synthetic controls in order to determine causality without conducting an experiment. They may also be strongly confounded by specific genetic variables. One interesting study by Register and Williams (1992) found cannabis usage in general was negatively associated with labor force participation but cannabis usage on the job had the opposite effect. Popovici and French (2014) found in a cross-sectional design that cannabis usage was negatively correlated with employment status for men and for women, but when a fixed-effects technique is used, the relationship becomes statistically insignificant.
The data presented before was based on cross-sectional data and this is problematic. Instead we could use longitudinal data which is slightly better for determining causality (Cole and Maxwell, 2003). Kaestner (1994) specifically tested the association between cannabis usage and labor force participation in the NLSY in both a cross-sectional model and a longitudinal design. They found that in the cross-sectional design, there was a negative correlation between the two variables, but this disappeared in the longitudinal test. Furthermore, Ullman (2017) found that medical cannabis legalization decreased the amount of job absences by 8.4-8.7% among young men and Anderson et al. (2018) even found medical cannabis legalization was associated with decreased risk of workplace fatality. However, we should be cautious in interpretation of the second paper as it is a) on medical cannabis and b) contradicted by a study on recreational cannabis (Fardhosseini and Esmaeili, 2016).
We can also look at the amount of hours per week people are working. Sabia and Nguyen (2018) found no effect of medical cannabis legalization on labor force participation or the average amount of hours worked per week. Still, this is purely on medical marijuana legalization. There are a number of differences between medical marijuana legalization and recreational marijuana legalization, making generalization a problem, eg. THC level differences, availability with/without prescription. In a cross-sectional design, French et al. (2007) found little effect of cannabis usage in general on amount of hours worked by young men in the US. Unfortunately, I couldn’t find many other sources on this, but I will update this article if I find anything in the future.
Irons et al. (2014) found cannabis use was associated with greater laziness, but this study is not representative of standard American adults. The authors found no baseline differences in cannabis use among physically inactive people and their evidence relied on people who were cannabis-dependent and trying to quit, rather than average cannabis users (though of course an increase in the latter will generally lead to an increase in the former). These results seem typical of anyone struggling with some sort of serious addiction and attempting to quit – lack of motivation and general unwillingness to be physically active. It is improbable we can draw an important conclusion from these results.
The effects of cannabis on labor force participation have also been hypothesized to be related to the shift in drug-dependent citizens from harder drugs to cannabis. Some examples come from Bradford et al. (2018) and Vigil et al. (2017) who found medical cannabis legalization was associated with lower levels of opioid usage and prescription (perhaps, this is why pharmaceutical companies fund so many anti-cannabis studies). Jacobsen et al. (2017) use a differences-in-differences analysis to test the effect of medical cannabis legalization on opioid abuse, using opioid treament admissions and opioid overdoses as proxies. They find that when a state legalizes dispensaries as part of medical cannabis legalization, there is a significant decrease in opioid abuse (also see Dickerson, 2018 for a review). Opioid addiction continuously becomes a major problem in the United States and the results thus far in treating it with cannabis have been mixed.
Labor market productivity is simply the “average annual real wage in a state” but it is telling as to how cannabis will affect other economic outcomes. Voronetskyy (2019) used data from the fifty states and a fixed effects design to test the effect of cannabis usage, rather than simply legalization, on labor market productivity. This study is particularly useful for a couple reasons. For one, they are measuring dose-dependent relationships between the variables. Second of all, it allows for a much larger sample to be used. Using data from legal and non-legal states with actual usage of cannabis by state is going to serve this data well. The author originally found a significant, positive effect of cannabis usage on labor productivity, but this went away in the fixed effects design. The fixed effects design is more useful than the cross-sectional because it is meant to determine causality with observational data.
While wages are not perfect accounts of productivity, the evidence is generally in favor of a small effect or no effect from cannabis to wages (Kaestner, 1994; Kaester, 1991; Chakraborty et al., 2020; van Ours, 2005). The issue with drawing causation from these estimates is that tobacco usage, which has very little psychoactive effect aside from a short buzz while smoking, is also related to a reduction in wages. So, there might be some genetic underlying variables which predict all of these things. Additionally, alcohol was found to have a positive effect on wages, which seems to make it even more clear that the psychoactivity of a drug is probably not related to wages (see van Ours, 2004).
Some other measures of productivity are not purely economic measures but rather different psychological variables which may diminish the potential for a specific labor market. These are primarily depression, anxiety, and intelligence. It is more likely depression and anxiety or other health conditions will negatively affect labor market outcomes. A large study by Banerjee et al. (2013) found that depression and anxiety were related to overall worse labor market outcomes. They argue that improved mental health of the people most in need in the population would result in employment for 3.2 million individuals and would produce about 19 billion dollars for the market. As Chakraborty et al. wrote,
If the effects of cannabis on health are negative, increased use should manifest in worse labor market outcomes, including decreased labor force participation, higher unemployment, and/or lower wages, as a result of lower productivity.
I’ll start with intelligence because there has been some recent controversy in regards to the topic of cannabis and IQ. Basically, one paper got really big a few years ago because it used a large sample of teenagers over a span of 30 years and they showed that frequency of cannabis usage and usage in general was associated with a decrease in up to ten IQ points (Meier et al., 2012). Some critics pointed out that “the study failed to rule out other potential explanations for the decline in IQ, such as a teen’s family environment or whether they dropped out of school”. Others (Daly, 2013) argued this is confounded by underlying personality factors. On the other hand, an observational study by Mokrysz et al. (2016) found no effect of cannabis usage on IQ or educational performance.
Regardless, longitudinal studies are great but if we want to conduct a good study on this, we should use a twin design. This is exactly what Jackson et al. (2016) did and they found no effect of cannabis usage, nor frequency of cannabis usage on adult IQ. Overall, neither author tested if the effect was on g or not, but regardless it is safe to say there is no effect of cannabis usage on your IQ score which is the best predictor of things such as job performance which is massively important for productivity (Hunter and Hunter, 1984).
One study by Cuttler et al. (2018) found that medical cannabis usage was associated with extreme reductions (>50% of sample) in depression and anxiety symptoms. Though, the data is not that reliable. It comes from a pro-cannabis app which allowed medical marijuana patients to track their progress after beginning usage of medical marijuana. This probably leads to an overestimation of results for recreational users and in general due to the potential bias of the sample. Without a proper placebo group, the study is very difficult to really use in good faith. Additionally, they note that strains with the most CBD, rather than THC which is far more psychoactive, were the most effective in treating depression. This further emphasizes the importance of the fact that the sample was composed of medical marijuana users. Unfortunately, this area of research is very limited so not much evidence is available. Even checking pro-cannabis and anti-cannabis blogs (Leafly vs. MayoClinic) showed very little promising evidence in either direction. Sarris et al. (2020) seems to be the newest review and they find, once again, the evidence is potentially promising but far too limited to draw any conclusions.
Many studies found that increasing frequency of cannabis usage was associated with increased depression (Horwood et al., 2012; Lev-Ran et al., 2013; Degenhardt et al., 2002). The issue is of course, determining causality. People with depression can regularly fall into addictive symptoms. Buckner et al. (2007) found that social anxiety disorder served as a major risk factor for using cannabis. However, when it comes to treating people with depression, some evidence exists in the opposite direction. On the other hand, Harder et al. (2006) used data from the NLSY and found that after adjusting for other group differences, the odds ratio for self-reported depression and cannabis usage was statistically insignificant. Braga et al. (2012) found that depression was associated with better cognitive functioning for people with bipolar depressive disorder. Degenhardt et al. (2001) found no significant association between cannabis usage and anxiety and depression after controlling for risk factors such as neuroticism and usage of other drugs. Feingold et al. (2016) found no association between cannabis usage and social anxiety disorders.
As I stated earlier, the primary effect on the economy is going to be through treating people with extreme depressive symptoms. So, what is the effect of cannabis on treatment-resistant depression? Once again, little research has been done. Cannabis excites the endocannabinoid receptors in the brain which is useful for reducing overall pain and increasing feelings of well being. A primary study on the usage of cannabis for treating depression symptoms was done on rats and they found generally positive effects (Haj-Dahmane and Shen, 2014). Anderson et al. (2014) found medical cannabis legalization was associated with a decreased risk of suicide in men.
All of these previous studies are fine, but the newest paper on this topic is the real deal. The primary issue with much of the other research, aside from the cross-sectional designs, is it simply correlates cannabis consumption with depression. Gukasyan and Strain (2020) used data from the National Survey on Drug Use and Health from the years 2012-2017. The authors first look at cannabis usage and depression, verifying the finding that people who are more depressed are also more likely to use cannabis at all. But of those who use it, those who use it more end up having significantly lower rates of extreme depression. This study seems to take down the hypothesis that cannabis will make you more depressed. Rather, it would appear that more depressed people use cannabis. Of those people who do use cannabis, it seems the ones who use it the most will end up having the least amount of depression.
Corey-Bloom et al. (2012) found cannabis usage was associated with a 30 percent decrease in spasticity among patients with multiple sclerosis. Li et al. (2019) found medical cannabis usage was associated with lower levels of reported physical pain from men. The same was shown by Nicholas and Maclean (2018). Cannabidiol was found to reduce anxiety among patients with social anxiety disorder in both animal studies, though this is not THC and so not of the primary focus (de Mello Schier et al., 2012). (Webb and Webb (2014) found medical cannabis usage was associated with a 64 percent relative decrease in average chronic pain, ~50 percent decrease in stress/anxiety, and 45 percent decrease in insomnia. Doremus et al. (2019) used over-the-counter sleep aid sales as a proxy for insomnia in Colorado. The authors found that cannabis was associated with lower rates of insomnia. Using ADD Health data, Wang (2013) found insomnia was related to lower employment for men. The effect of cannabis on the endocannabinoid receptors has also been proven by Bossong et al. (2013) who found cannabis was related to different reactions to negative imagery. This may lead to some potential usage in the future of using cannabis to treat PTSD symptoms, which I think may be more severe for the market than, say, many instances of general depression.
Overall, I think the general conclusion from this article should be that recreational cannabis doesn’t seem to lower productivity in the marketplace in any significant way and in the ways that it might, there are confounding variables and they are outweighed by the positives. The evidence on depression and anxiety is very mixed and it is difficult for any conclusions to be made, but some evidence is very promising especially for more extreme cases. My prescription from this is that we should just treat cannabis like we treat alcohol. It can be dangerous to work while on it of course, but it is generally just a good way to unwind at the end of the day. Instead of consuming alcohol which is related to thousands of deaths per year, one can just consume a cannabis edible or smoke some cannabis and relax with no general health reduction.