Deconstructing Executive Pay: A Systematic Variance Decomposition of Target vs. Realized Compensation Using SEC Pay Versus Performance Disclosures

Abstract This study provides an objective empirical decomposition of the structural drivers of executive compensation across large-cap public corporations. Utilizing recently mandated SEC Item 402(v) Pay Versus Performance (PVP) panel disclosures under Regulation S-K, we execute an order-independent Shapley variance decomposition framework to isolate the relative contributions of corporate scale and market-driven performance vectors. Analyzing a rigorously cleaned baseline panel of 2,362 CEO-year observations from 2021 to 2025, which drops minimally to a finalized regression sample of 2,357 observations due to localized sector-index tracking return gaps, we test parallel econometric models contrasting ex-ante targets against ex-post realizations. Our baseline panel estimations reveal highly compressed total explanatory power (R² = 0.0027 for targets; R² = 0.0105 for realizations), indicating that standard operational vectors capture only a small fraction of overall executive pay variation. However, within this tightly constrained explained variance boundary, game-theoretic attributions isolate an overwhelming footprint for performance and market return parameters. Performance and market velocity account for 94.20% of the model's limited explained variance for ex-ante targets (Summary Compensation Table totals) and expand to a definitive 97.82% share when evaluating ex-post realized compensation (Compensation Actually Paid). Conversely, corporate scale metrics are heavily minimized, explaining a mere 5.80% and 2.18% of the explained model variance, respectively. These results demonstrate that while standard structural drivers explain only a tiny fraction of total compensation variance, performance and market dynamics account for nearly the entirety of the model's limited explained variance, overriding absolute company size effects once equity contracts are exposed to market revaluations. This decomposition provides a neutral, transparent open-science baseline framework for corporate boards, institutional investors, and proxy advisors evaluating executive asset alignment. Open-Science Replication Archive This replication archive bundles the research manuscript alongside all primary dataset layers and processing files: sec_s_and_p_500_master_panel.csvFinalized operational data matrix tracking 2,362 baseline summary rows and 2,357 verified regression rows. sec_s_and_p_500_company_facts_cache.csvLocalized corporate asset size cache metrics streamed via the native SEC Company Facts financial API. sec_edgartools_exec_comp_cache.csviXBRL operational extraction ledger aligning chronological baseline targets and executive timelines. Execution Code Notebook (.ipynb)Complete end-to-end Python script processing pipeline for cluster-robust ordinary least squares and 2K Shapley subset model permutations.