The Financial Impact of Hospital Closings on Surrounding Hospitals

Two competing theories lead to opposite conclusions about the expected financial impact of a hospital closure on surrounding hospitals.  On one hand, a standard market model would predict that nearby hospitals would gain financially from a nearby hospital closing.  This theory would imply that the closing hospital had a competitive disadvantage because of poor management, diseconomies of scale, insufficient negotiating power and other competitive factors.   

Conversely, if the closing hospital had an unusually large share of unprofitable patients conditional on the reimbursement rate, surrounding hospitals could experience a negative financial shock by absorbing those patients.  Insurers construct reimbursement rates based on the average cost of treating a pool of patients with an illness.  Their assumption is that hospitals will have a large enough group of patients with each condition to cross-subsidize the high cost cases with the low-cost cases.  However, patients do not sort randomly into hospitals.  Community characteristics such as socioeconomic status, culture and disease prevalence play a large role in determining where a patient falls on the cost distribution.  A hospital attracting a high share of the least profitable patients, conditional of insurer payment, may be less likely to remain solvent.  If these factors drive insolvency, surrounding hospitals will suffer by absorbing those patients.

We focused our analysis on hospitals closing in 2004 because more hospitals closed that year than any other.  Also we had a sufficient amount of data both before and after 2004 to conduct the kind of analysis described below.  

We use a two step process to evaluate the financial impact on surrounding hospitals.  In our first step, we assume that patients from a closing hospital will go to the same open hospitals as their neighbors, and developed a method of mapping patients from a closing hospital into open hospitals based on their 5-digit zip code.  To check the validity of our methodology, we used an ARIMA model on 1995 to 2004 data to predict the number of patients each hospital would have served in 2005, and compared it to the actual 2005 figures and our absorption predictions.

In the second stage of our analysis, we used a random effects model to test whether hospitals absorbing a greater share of patients from the closing hospitals were negatively or positively impacted financially.  In particular, we looked at the operating ratio, expenses divided by revenue, because that financial measure would be most likely to move if hospitals were closing because they were serving the least profitable population.  We controlled for the HHI and drew some additional conclusions based on that result.

Our main finding is that hospitals which absorbed more patients from closing hospitals had a significantly higher operating ratio (greater expenses compared to revenues) than hospitals which did not absorb such patients, after accounting for hospital size.  These results are statistically significant.