Under the George W. Bush U.S. presidential administration, the federal government pushed for greater accountability among institutions of higher education for educational outcomes. Graduation rate is a key performance indicator of institutional accountability. Previous researchers of student attrition focused primarily on the effects of student level factors on student persistence/withdrawal behavior. Recently, researchers put more focus on the effects of institutional characteristics on graduation rates, but most of these studies were exploratory and based on multiple regression models. No institutional model has existed to synthesize their results within a theoretical framework. Such an institutional model is needed to explain the process of student persistence at the institutional level. The purpose of this study was to develop a model of institutional performance in graduation rate for four-year, public and private not-for-profit, Title IV institutions in the United States. This study validated the institutional model based on the IPEDS dataset using the structural equation modeling (SEM) technique. Further group comparison analyses are conducted by fitting the same SEM model to several subgroup datasets based on grouping variables such as control, geographical region and state. Benchmarking analyses were conducted to demonstrate how administrators and policy-makers can use the institutional model to compare the performance of an institution with its peers and what policy changes can they pursue to improve graduation rates.