Cannabis prices on the dark web

Drug policies around the world have been relatively strict since the adoption of the Single Convention on Narcotic Drugs in 1961. However, in recent years many countries started adopting more liberal policies towards consumption of cannabis. European countries such as Portugal, Czech Republic and Germany pursued decriminalization policies, while the Netherlands has quasi-legalized cannabis use and retail sale. In 2013, Uruguay was the first country in the world to legalize recreational cannabis use, followed by Canada in 2018. In the US, at a federal level cannabis is still illegal but several states have legalized medical cannabis use, decriminalized or even legalized recreational cannabis use. Nevertheless, recreational cannabis remains an illegal drug in many countries.

Related to the illegal nature, information and therefore studies on quality, prices and quantities of cannabis consumed or supplied are limited. Nisbet and Vakil (1972) is a pioneering work analyzing cannabis prices using self-reported data of US university students. Many studies focused on estimation of participation elasticities, i.e. the effect of prices on the extensive margin of cannabis use (see for example Pacula, Grossman, Chaloupka, O’Malley, Johnston, Farrelly, 2001, Cameron, Williams, 2001, DeSimone, Farrelly, 2003, Williams, 2004, Williams, Pacula, Chaloupka, Wechsler, 2004, Zhao, Harris, 2004, Clements, Zhao, 2009 and Clements et al., 2010). Van Ours and Williams (2007) study the price sensitivity of cannabis use dynamics among young Australians finding that low prices stimulate the uptake of cannabis while reducing quitting from cannabis use. There are also a few studies on the supply side of the cannabis market investigating the determinants of cannabis prices.

Cannabis prices vary both between and within countries. Statistics Canada for example reported that in 2017 cannabis prices ranged from a low US $5.02 in the province of Manito to US $7.18 in the territories.¹ These numbers are in line with the 2018 cannabis price index that provides cannabis prices by city, where for Canada the price range in US dollars per gram was from 6.15 in Montreal to 7.82 in Toronto.² The 2018 cannabis price index for US cities ranged from a low US $7.58 per gram in Seattle to US $18.08 per gram in Washington DC. The overall range in the 2018 cannabis price index was from US $1.34 per gram in Quito (Ecuador) to US $32.66 per gram in Tokio (Japan).³ Cannabis prices vary across countries for various reasons, i.e. differences in legal regime with respect to the consumption and/or production of cannabis, quality of the cannabis, market power of cannabis suppliers. There is also within country variation of cannabis prices for similar reasons as also within a country the legal regime may differ like for example in the US where recreational cannabis use is legalized in some states but prohibited in others.

Previous studies on the determinants of drug prices mainly focus on quantity discounts and quality effects. An early study is Brown and Siverman (1974) who analyze data of heroin prices based on purchases by US undercover narcotics agents. They find a quantity discount while purity of the heroin is found to have a positive effect on the heroin price. Caulkins and Padman (1993) analyze price data collected by US undercover narcotics agents for various illicit drugs including cannabis. They find for most drugs quantity discounts and quality premiums. Clements (2006) studies the price of cannabis using Australian data collected through undercover buys finding a quantity discount. Caulkins and Pacula (2006) use data from a US household survey to investigate the variation in cannabis prices finding a quantity discount. Lakhdar et al. (2016) analyzes cannabis prices using information from regular French cannabis users finding a significant quantity discount and significant positive quality effects. Lahaie et al. (2016) study heroine prices in France using data on heroin samples and surveys of heroin users finding significant quantity discounts and positive purity effects. Smart et al. (2017) study cannabis prices of retail transactions from Washington state where cannabis sales are legal. They find are a significant quantity discount and significant positive price effects of quality.

Online illegal drug markets with cannabis transactions as an important element are a recent phenomenon. Through the so called Dark Web, sellers and buyers of cannabis interact anonymously. Soaka and Christin (2015) present a descriptive analysis of various online anonymous market places including the first successful one, Silk Road, which was online from 2011 to 2013 when it was taken down by law enforcement and its operator was arrested. They conclude that these online market places are resilient to law enforcement take-downs as within a month after shutdown ‘a novel incarnation’ of Silk Road was online. Van Hout and Bingham (2013) conclude on the basis of anonymous online interviews among users that transactions are experienced to be safer than negotiating in a street-level drug market. Décary-Hétu et al. (2016) study the risk-taking behavior of drug sellers at the Dark Web concluding that compared to traditional drug market transactions the risk of violence is reduced as the face-to-face transactions are eliminated. Aldridge et al. (2018) argue that online illegal markets may reduce the harm of drug use through the increase in quality and safety of the drugs sold and because in the course of the transaction there is less conflict and violence. Furthermore, it is easier to build a reputation because of customer feedback. Barratt and Aldridge (2016) argue that higher drug quality may be the most important reason for buyers to prefer transactions through the Dark Web rather than via face-to-face interactions.

Aldridge and Décary-Hétu (2016) study data of the first major internet based illegal market, Silk Road 1. In their analysis of more than thousand drug-selling vendors from about 40 countries, they conclude that drug sales accounted for about a quarter of the total revenue on the market, with ecstasy-type drugs dominating the wholesale activity. Bhaskar et al. (2019) in their analysis of more than 1.5 million online drug sales focus on moral hazard, analyzing seller reputation and performance. They find that only a small proportion of on line drug deals received a bad rating from buyers. For Dark Web vendors, bad ratings lead to big reductions in sales and to market exit. The authors also note that compared to street sales of illegal drugs, Dark Web marketplaces suffer less from problems of drug adulteration and low quality. According to Bhaskar et al. (2019) a substantial share of drug users buy their drugs online through the Dark Web. Duxbury and Haynie (2018) analyze the network structure of opioid distribution through the Dark Web. They arrive to similar conclusions as Bhaskar et al. (2019), finding that vendors’ trustworthiness explains more variation in the overall network structure than the affordability of vendor products or the diversity of vendor product listings.

Our paper presents an analysis based on unique data on cannabis prices. Whereas usually information on cannabis prices is based on undercover operations or consumer surveys we use information from cannabis sellers. We exploit data from the Dark Web site called AlphaBay, a website that operated between December 2014 and July 2017. According to Christin (2017) during that time among the online drug market places it became the leading one. From an analysis of 27 scrapes from AlphaBay Christin (2017) concludes that half of the drug sellers specialized in one drug. From a descriptive analysis of the cannabis sales he also concludes that most transactions are for 1g, 5g and 10g while there is a modest volume discounting, i.e. larger quantities are sold at a lower price.

We use data collected from AlphaBay for two weeks early October 2015. At the time, AlphaBay was still in its initial stage on a steep expansion curve. Paquet-Clouston et al. (2018) for example report about 700 vendors in the drug sections of AlphaBay end of September 2015. Five months later the number of vendors was more than doubled.⁴ We have information about 500 cannabis prices from about 140 sellers in 18 countries. The nature of our data allows us to exploit the detailed information about the quality of a particular cannabis strain, measured both by its potency (active ingredient content) and popularity among users. Our main contribution to the literature on the analysis of cannabis markets is threefold. First, we study the relationship between cannabis prices, quantities and qualities whereas often quality is not part of the analysis. Second, we use information from the Dark Web to provide a quantitative analysis whereas thus far mostly descriptive studies are available. Third, because our data allow us to include seller characteristics – including their country of origin – in the analysis we can investigate to what extent seller have some influence over cannabis prices.

The remainder of our paper is organized as follows. Section 2 outlines the Dark Web in relation to information about cannabis transactions. In the spectrum of drugs available through the Dark Web, cannabis is special in the sense that the legal status of the drug varies a lot. In many countries, cannabis is an illicit drug but in other countries or parts of countries (states in the US) cannabis use is legalized or quasi-legalized. Section 3 presents our data and some summary statistics. Section 4 discusses the set-up of our empirical analysis. Section 5 presents our parameter estimates and Section 6 concludes.