Prediction of Starting Salaries Through Job Mismatch: The Case of Vietnamese Graduates

Throughout the job searching and recruitment process, it is very hard to evaluate the basic data given in resumes to make the right decisions regarding the starting salaries, leading to unfair salaries. Hence, this research employed data from Graduates' Employment survey conducted by Foreign Trade University to suggest the appropriate machine learning models for forecasting the starting salaries and to explore the main factors in determining the entry-level salaries in Vietnamese labor market. Furthermore, in a developing economy like Vietnam, the education-job mismatch issue experienced by university graduates during their transition from education to employment pose a critical challenge. By utilizing Latent Class Analysis, the study classifies mismatches into four distinct categories: No mismatch, Instrumental Competencies mismatch, Professional Competencies mismatch, and Horizontal mismatch. Through the use of multiple machine learning algorithms, this research studies the effect of these mismatches and also other variables like Gender, Grade, Location of work, Foreign language proficiency and Major on the starting salaries. The results demonstrate the significant impact of Grade and Mismatch types on predicting starting salaries. Moreover, the findings suggest that the Gradient Boosting Machine algorithm is the most accurate in predicting starting salaries. These findings provide valuable insights for students in making career decisions and employers in planning the recruitment strategies. Additionally, this research also offers practical implications for educational institutions, policymakers, and employers seeking to address job-skill mismatches and enhance employability outcomes.