What can be said is that the war on drugs has been a failure if the objective was to reduce drug use.
In regard to the rest, there are much less straightforward answers. The complexity of the topic begins with the fundamental concept of decriminalization itself. It is important, in fact, to distinguish concepts such as depenalization, decriminalization and legalization. Furthermore, different states have implemented different policies, which can have different outcomes. It is not so evident to compare the results. And how do we conceptualize and define drug ‘abuse’?
Portugal pursued decriminalisation and made drugs a public health issue rather than a criminal justice one. Uruguay was the first country to legalize marijuana, and Canada is the second one more recently. It is too early to assess Canada’s new policy, and as far as I am aware not much can be said about Uruguay neither, for which data for many indicators were scarce according to the UNODC in 2016.
There are debates on the actual impact of Portugal’s decriminalisation of drugs in 2001 and whether their success has been overstated. For example, Greenwald is known for having argued for the resounding success of Portugal’s approach to drug regulation, however Laqueur does appropriately point out the difficulties in assessing the outcomes, such as pointing how Portugal dealt with drugs in practice before 2001. The key word for this entire reply, is that it is a complex topic to study. “Nonetheless, as the Portugal case illustrates, legislative change can be practically small but generate significant symbolic import, and this, in turn, may produce dramatic change“.
The thing is, laws are not the best way to affect behavior. For example, the threat of incarceration and severe punishment are not effective at deterrence. The problem is that wars on drugs and the criminalization of drugs cause harm to society and to individuals, and contribute to the problem rather than produce positive outcomes (i.e. prisons are less effective than other approaches for rehabilitation). Rates of drug use are a function of many things, as with many other behaviors, including social norms, attitudes and beliefs regarding the behavior (and its outcomes), availability, etc. See Japan for quite a different scenario.
As far as the EMCDDA is concerned regarding cannabis legalisation, the results on whether policy impacts the use rates of cannabis are mixed:
The legal impact hypothesis, in its simplest form, predicts that increased penalties will decrease drug use and reduced penalties will increase drug use. However, in the original analysis, and an updated version (Figure 4), no simple association can be found between legal changes and the prevalence of cannabis use.
There are studies in the United States that suggest that use rates of marijuana are higher in states that have liberalized marijuana, and some authors have suggested a causal relationship, especially for young people. Anderson et al.‘s examined the “relationship between MMLs and marijuana consumption among high school students using data from the national and state Youth Risky Behavior Surveys (YRBS) for the years 1993–2011”:
In fact, estimates from our preferred specification are small, consistently negative, and are never statistically distinguishable from zero. Using the 95 % confidence interval around these estimates suggests that the impact of legalizing medical marijuana on the probability of marijuana use in the past 30 days is no larger than 1.5 percentage-points, and the impact of legalization on the probability of frequent marijuana use in the past 30 days is no larger than 0.8 percentage-points. In comparison, based on nationally representative data from Monitoring the Future, marijuana use in the past 30 days among 12th graders increased by 4.3 percentage-points from 2006 to 2011 (Johnston et al., 2011); based on national YRBS data, marijuana use among high school students increased by 3.4 percentage-points from 2007 to 2011.
In addition to the YRBS analysis, we examine data from the National Longitudinal Survey of Youth 1997 (NLSY97) and the TEDS. The NLSY97 allows us to follow survey respondents over time, while the TEDS data allow us to examine marijuana use among a high-risk population. Consistent with the results of Pacula et al. (2001), we find little evidence that marijuana use is related to the legalization of medical marijuana in either of these data sources.
It may be more fruitful to look at specific approaches and programs, and evaluate their effects. For example, Switzerland is known for their programmes prescribing heroin under controlled conditions as part of their four pillars policy: policing, prevention of drug use, treatment of drug use, and harm reduction. These programmes were piloted experimentally. On the practical impacts of heroin prescription, Aebi and Killias concluded:
The Swiss heroin prescription program was targeted at hard-core drug users with very well established heroin habits. These people were heavily engaged in both drug dealing and other forms of crime.
They also served as a link between importers, few of whom were Swiss, and the — primarily Swiss — users. As these hard-core users found a steady, legal means for addressing their addiction, they reduced their illicit drug use. This reduced their need to deal in heroin and engage in other criminal activities. Thus, the program had three effects on the drug market:
It substantially reduced the consumption among the heaviest users, and this reduction in demand affected the viability of the market.
It reduced levels of other criminal activity associated with the market.
By removing local addicts and dealers, Swiss casual users found it difficult to make contact with sellers.