Scholarlink

<p>This study introduces a novel stochastic framework for analyzing career progression and employee retention through an integrated application of discrete-time Markov chains, survival analysis, and multivariate logistic regression. Three methodological advances distinguish our approach: first, a unified formulation capturing both hierarchical and lateral transitions within organizational structures; second, a regularized estimation procedure addressing data sparsity through bootstrap confidence intervals and Tikhonov regularization; third, policy-sensitive absorption metrics enabling quantitative impact assessment of human capital interventions. Applying this framework to longitudinal human resources data from a multinational technology corporation, we model transitions between four hierarchical states: junior, confirmed, manager, and exit; while addressing right-censoring through robust Kaplan-Meier imputation techniques. Key empirical findings establish significant structural patterns: managerial roles exhibit substantial inertia with a 75% probability of remaining unchanged between periods (95% CI 72–78%); a critical salary threshold emerges at 6500€ monthly, beyond which attrition risk decreases by 60%; survival analysis identifies temporally localized risk windows with mid-level staff facing 2.3<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\times \)</EquationSource> </InlineEquation> higher exit probabilities between two and five years tenure; and promotions demonstrate non-linear protective effects, showing 34% stronger retention impact at junior versus managerial levels. Multivariate models confirm job satisfaction (OR 0.73; 95% CI 0.67–0.80) and promotion history (OR 0.66; 95% CI 0.58–0.75) as dominant retention predictors. The framework demonstrates exceptional predictive validity (adjusted R-squared = 0.85) while conforming to parametric assumptions (Shapiro–Wilk <i>p</i>-value = 0.15). Practical applications enable organizations to simulate policy impacts on retention metrics, optimize intervention costs through ROI-calibrated targeting (333€ per percentage point of attrition reduction), and design personalized career pathways. These contributions advance theoretical foundations in organizational behavior while establishing evidence-based paradigms for strategic human capital management. Full computational implementations ensure reproducibility and facilitate scholarly extension.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modeling career mobility and attrition using Markov chains and survival analysis

  • 期刊论文

  • 发表日期：

    2025年9月9日

查看全文
  • Mohamed Yasser BOUNNITE

摘要

This study introduces a novel stochastic framework for analyzing career progression and employee retention through an integrated application of discrete-time Markov chains, survival analysis, and multivariate logistic regression. Three methodological advances distinguish our approach: first, a unified formulation capturing both hierarchical and lateral transitions within organizational structures; second, a regularized estimation procedure addressing data sparsity through bootstrap confidence intervals and Tikhonov regularization; third, policy-sensitive absorption metrics enabling quantitative impact assessment of human capital interventions. Applying this framework to longitudinal human resources data from a multinational technology corporation, we model transitions between four hierarchical states: junior, confirmed, manager, and exit; while addressing right-censoring through robust Kaplan-Meier imputation techniques. Key empirical findings establish significant structural patterns: managerial roles exhibit substantial inertia with a 75% probability of remaining unchanged between periods (95% CI 72–78%); a critical salary threshold emerges at 6500€ monthly, beyond which attrition risk decreases by 60%; survival analysis identifies temporally localized risk windows with mid-level staff facing 2.3 \(\times \) higher exit probabilities between two and five years tenure; and promotions demonstrate non-linear protective effects, showing 34% stronger retention impact at junior versus managerial levels. Multivariate models confirm job satisfaction (OR 0.73; 95% CI 0.67–0.80) and promotion history (OR 0.66; 95% CI 0.58–0.75) as dominant retention predictors. The framework demonstrates exceptional predictive validity (adjusted R-squared = 0.85) while conforming to parametric assumptions (Shapiro–Wilk p-value = 0.15). Practical applications enable organizations to simulate policy impacts on retention metrics, optimize intervention costs through ROI-calibrated targeting (333€ per percentage point of attrition reduction), and design personalized career pathways. These contributions advance theoretical foundations in organizational behavior while establishing evidence-based paradigms for strategic human capital management. Full computational implementations ensure reproducibility and facilitate scholarly extension.