top of page
logo
Accepted to the IARMEA Conference, June 2026

Minimum Wage Policy and Youth Labor Dynamics in Canada: A Machine Learning Approach to Causal Identification

Authors: Jinlin Liu & Tian Jin | Status: Pending Publication

FULL RESEARCH PDF

Machine Learning Algorithms

The two main algorithms used within the study is the random forest and the MLP neural network. Both of which allowed for the computation of large data sets, which was a necessary feature the algorithm must possess. The research conducted utilized the statistics collected by Statistics Canada for 11 different factors across each province from 1980 to 2025(or as close as possible to those two dates).

​

The focus of the study was to investigate the indirect relationships between minimum wage and youth employment rates in Canada. In order to account for the numerous other indirect factors, 9 different mediators are studied to calculate the weight and direction of each one. The mediators include the education level of immigrants, tariffs, market power, inflation, automation, education level of applicants, work hours of employees, unionization, and GDP.

Objective

Quantitatively compute the indirect relationship, with respect to mediators, between minimum wage and youth employment rate.

  1. Capable of handling 11 dimensions of data.

  2. High validation reliability

  3. Accounts for missing, inconsistent, and large datasets

Findings

Amongst the mediators studied, the following three were some of the most impactful and controllable mediators

Market Power

Market power, which was measured by the total industry share the largest firm holds within each singular province's most prominent youth-employing sector, is the singular most important mediator and holds the largest negative impact with a mediator weight of 0.2988.

 

Considering that the majority of young Canadians are employed in the retail trade sector, limiting the market power of loblaws is of utmost importance. However, since loblaws lacked necessary market power to be detrimental to the youth employment as a mediator. The emphasis, then, should be given to limit the monopoly Tim Hortons currently has within certain provinces as they hold, currently, 42% of the total share of the food & accomodations industry.

Education

Education is a crucial mediator, regardless of whether it is measuring the applicant's or immigrant's education level. The share of secondary degree holders among the applicants is the fourth most important mediator overall and the second most impactful positive mediator. Meanwhile, the share of immigrants with secondary degrees is not only the second most impactful mediator overall, but also the second most impactful negative mediator, surpassed only by the market power factor. Unlike other mediators mentioned previously, the share of secondary degree holders among applicants and immigrants alike is more easily influenced.

Tariffs

Being the third most impactful mediator overall and the most impactful positive mediator, tariffs' effect is rather unexpected. The tariffs mediator is measured by the tariffs placed on Canada and measured as if they were a tax. However, instead of acting as a negative mediator, which would dictate that an increase in minimum wage would result in greater unemployment as tariffs increase, it acted as a positive mediator. However, it should not be treated automatically that tariffs are intrinsically beneficial as a mediator, but rather the reactive measures as a result of receiving foreign tariffs. This includes, but is not limited to, public fervor and increased purchase of domestic goods, subsidies from the government, as well as retaliatory tariffs.

Summary

  1. This study demonstrates the importance of tariffs, education, and market power for future government policy makers.

  2. This study quantitively validated the practicality and reliability of using deep-learning neural networks in economic data analysis

bottom of page