By Allan W. Gregory and Eliane Hamel Barker, Queen’s University

Statistics Canada recently took up the difficult challenge of finding out what Canadians pay for their cannabis both medically (licensed and unlicensed) and recreationally. Currently only licensed use of medical cannabis (both dried and oil) is legal to purchase from licensed producers under Access to Cannabis for Medical Purposes Regulations (ACMPR). One reason governments are so interested in the street price of cannabis is the legalization of marijuana for recreational use due sometime this summer. The thinking is that legal marijuana prices must not be greater than those on the street; otherwise black markets will continue to flourish.

Statistics Canada is not new to the survey business and have in the past attempted to price cannabis and quantities smoked. However, in their most recent effort, a novel feature was using crowdsourcing on a web site survey to gather the data. Statistics Canada understood that there was a selectivity or participation problem in such a methodology but decided this was the best approach possible. We agree with this decision. Since cannabis was soon to be legalized for recreational use and no special personal identifiers were asked, the participating decision should not be associated with either positive (higher price) or negative (lower price) bias.

As any pollster or survey expert knows: the framing of the question is everything. Pose the question badly and the results notwithstanding participation decisions can be biased. For example, a biased question is: “Knowing that medical marijuana can be obtained at roughly $9 per gram, what do you pay for a gram?” This provides background pricing information and would almost surely impact the responses.

Statistics Canada’s lead question on their survey was:

The manner in which the information is supposed to be entered suggests that Statistics Canada is concerned about how purchasers received their quantity of marijuana and hence the relevant price. So if one usually bought 10 grams at $10 per gram, then the correct answer to this would be $10 and specify by entering 10 grams which was not one of the options and needed to be typed-in. The fact that the survey seemed to have round units in ounces rather than grams would indicate the Statistics Canada was thinking in old British measures rather than the current metric system. This assumption turned out to be right as these options were selected most often.  

Statistics Canada made their data available and we downloaded it. As of March 29, 2018 on the Statistics Canada Web site, the following prices across Canada are presented along with this the questionnaire (not shown). Despite what we said earlier Statistics Canada is asking for pricing information of participants while showing the results of the existing survey. We have downloaded the data of individuals that is kindly provided at the bottom of the site and we have seen no evidence of such bias so we might just be nitpicking.

Figure 1: Statistics Canada’s Cannabis Prices per Gram

The most striking aspect of the prices per gram are how low they are for Canada and the provinces when compared to the data of Brad Martin of CannStandard who has scraped the publicly available web sites of distributors. For instance using the August 10, 2017 of 206 webpages, the average price of cannabis was $9.68 with a median of $9. These values are substantially higher than the values in Statistics Canada survey. We know there is a natural inclination to understate the price paid since everyone wants to say they got a bargain. But this differential is much too high and more importantly would suggest that the federal government’s announcement at the end of December, 2017 of $10 per gram might be way too high given the recent survey findings.

So given this large gap it is worthwhile to review the micro survey data available. As of March 29th there are an impressive 17,514 responses to the survey. There is some evidence that some quantities in grams purchased were either too small (less than a gram) or too large (1/4 of a kilogram) to be included in the analysis and could be reasonably excluded.  So if one wanted to calculate the price per gram, the logical operation is to take amount paid divided by the number of grams purchased.

If we do that calculation and do a histogram we get the shocking result that the average price is just under $2.5 per gram and the median is a paltry 82 cents. This leads us to think that the price data has been “massaged” to give the implied price per gram directly.  If we accept this is price per gram we indeed get the exactly the same sample mean as Statistics Canada of $6.81 for Canada (median of $6.42).

The fact that this price is price per gram and not total paid for the quantity purchased is seen in the graph below: price falls with quantities purchased. The question posed does not match the data provided but if Statistics Canada did the gram conversion properly then there should be no issue.

But did Statistics Canada translate the question of price paid for quantity into price per grams correctly? The easiest way to ascertain this is to use only those individuals who indicated a single gram for their purchase since no recalculation of the price would be required. This is a robust measure since there is no reason that those answering only one gram bought would have a bias on the average price paid. The consequence of limiting the answers to only those that answered 1 gram to quantity is the sample size is reduced from 17,420 to 3,932. The average price paid jumps to $8.41 and a median of $9.00. The average price is almost a full $2.00 higher with this sample. This is also the same median as the web site download but lower than the mean of $9.68.

We redo this for each of the provinces and find the sample mean is roughly the same across the country (except for the Territories where we very few observations). This seems much more reasonable than the sharp provincial differences released by Statistics Canada where Quebec paid $5.85 per gram while Ontario paid $7.40.

The price per gram of marijuana on the black market is a key statistic in provincial pricing policies and the recent Statistics Canada release is misleading sharply underestimating the price of cannabis on the street. We think it would be wise for Statistics Canada to make this clear.