Identification of Benchmark and Investigation of Cost Drivers in Hospital Industry: How Inefficient are U.S. Hospitals?
MetadataShow full item record
By how much can a hospital reduce cost level while maintaining the service provided? In our problem, we estimate an input oriented measure of inefficiency and a cost function to understand the relationship between cost and number of performed procedures in the U.S. hospital industry. In addition, our model accounts for contextual variables which provide insights regarding cost drivers. For estimation, we use the method called Multivariate Bayesian Convex Regression (MBCR). Our data are composed from two databases. We use the American Hospital Association Annual Survey and the National Inpatient Sample provided by the Healthcare Cost and Utilization Program. Our cost measure is total expenditures and the output is number of procedures which is classified in four categories according to nature of service and type of operating room. The contextual variables (hospital size, region, teaching status and ownership) are selected using Bayesian Information Criterion (BIC). Many factors can impact costs level. Our results show that larger hospitals and teaching hospitals located in the Northeast are more cost inefficient. In the same way, private hospitals are less cost inefficient compared to public hospitals. Average cost inefficiency levels for an ~10% sample of all U.S. hospitals are 27%, 18% and 23% for years 2004, 2007 and 2011, respectively. Further, we found evidence that production in the U.S. hospital industry might be better characterized by the Regular Ultra-Passum Law than by a convex cost function.
Multivariate Bayesian Convex Regression
Crispim Sarmento, Marcella (2016). Identification of Benchmark and Investigation of Cost Drivers in Hospital Industry: How Inefficient are U.S. Hospitals?. Master's thesis, Texas A & M University. Available electronically from