Cannabis legalization: do banking transactions reflect a shift away from the illegal market?

ABSTRACT

Using data on banking transactions, we analyse how people changed their spending behaviour when recreational cannabis became legal in Canada. In the data set, we identify people with purchases in a cannabis store after legalization. We apply a difference-in-difference approach to compare their spending before and after legalization. People with cannabis purchases after legalization reduced significantly and persistently their cash spending and electronic transfers, indicating a shift from the black to the legal cannabis market.

KEYWORDS:

• Cannabis legalization
• recreational cannabis
• black market
• marijuana

JEL CLASSIFICATION:

• I18
• K32
• K42

Acknowledgements

We would like to thank two anonymous reviewers, David Peel (Editor), and the AI Labs Team of ATB Financial, particularly, Ian Hargreaves, Mark Sebestyen, April Hoffort, Dhruv Mayank, Kelly Keenan, and Heli Gong. Financial support by Mitacs through the Accelerate Program and by the Social Sciences and Humanities Research Council of Canada through Insight Grant 435-2018-0049 are gratefully acknowledged.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ ‘Illegal cannabis markets have moved online where there is less oversight and manpower to shut them down as forcibly as they could when they had raid-able storefronts. Transactions for black market marijuana are done in the form of e-Transfers which are incredibly difficult to track’. Quote from Cannabis Wiki: https://cannabis.wiki/politics/the-evolution-of-black-market-cannabis-in-canada.

² See Figure A6 in the Canadian Payment Methods and Trends Report (2022): https://payments.ca/sites/default/files/PaymentsCanada_Canadian_Payment_Methods_and_Trends_Report_2022_En_0.pdf.

³ Less than $2\mathrm{\%}$ of the people in the data set identified as nonbinary or preferred not to disclose their gender. Their data were removed, as a statistical analysis would not be meaningful given the low number.

⁴ The issue of correlation over time that can occur in difference-in-difference analyses (Bertrand, Duflo, and Mullainathan 2004) is not a problem here because we compare in each analysis only two time periods: before vs. after legalization. Moreover, there is no issue with variation in treatment time (Goodman-Bacon 2021).

Additional information

Funding

The work was supported by the Mitacs [IT12983]; Social Sciences and Humanities Research Council of Canada [IG 435-2018-0049]