Facial Expressions as Signals of Human Capital: A 25-Year Analysis of Career Outcomes in Engineering and Business Schools

This study investigates the predictive value of early-life facial expressions on long-term career success through a 25-year longitudinal analysis. We use a baseline cohort of N=2082 graduates from four major U.S. institutions (Texas Tech University, Penn State University, University of Wisconsin-Madison and North Carolina State University) between 1999 and 2003. To maximize the contrast of our signaling effects, we constructed a targeted sample of N=388 profiles focusing on the highest and lowest smile intensities. Deploying an innovative AI triangulation methodology calculating the median score from DeepFace, Vision Transformer (ViT), and FER models, we quantify smile intensity in original yearbook photographs and correlate these signals with real-world career trajectories extracted from LinkedIn in 2026. Our findings reveal a striking divergence between professional tracks: while smile intensity is positively correlated with network visibility, mobility and success in Business, we identify an "Expert Paradox" within Engineering. In this technical domain, initial facial seriousness serves as a signal of competence and leadership, facilitating advancement to executive-level positions. To isolate the pure effect of emotional capital from social origins, we refine our final analytical sample to N=264 profiles with identifiable hometowns, controlling for socioeconomic status using the 2000 U.S. Decennial Census. Furthermore, an AIC-based Stepwise selection algorithm confirms the robustness of our results. This research challenges the universality of the "smile premium" and underscores the importance of congruency between non-verbal signals and role-specific professional expectations.