This work investigates fairness and bias in LLM-driven academic recommendations, examining how demographic, geographic, and economic attributes of users influence the universities and programs recommended to them. Using 360 simulated user profiles and over 25,000 recommendations generated by LLaMA-3.1-8B, Gemma-7B, and Mistral-7B, the study introduces a multidimensional evaluation framework to measure representation, diversity, and disparities in recommendations. The analysis reveals systematic preferences for institutions in the Global North, reinforcement of gender stereotypes, and substantial repetition of recommended institutions, highlighting how seemingly neutral educational recommendation systems can reproduce existing societal inequalities