Skip Navigation

In September 1979, CBO published Controlling Rising Hospital Costs, a report analyzing the Carter Administration's hospital cost containment proposal and other options to contain hospital costs. The report had been prepared at the request of the Subcommittee on Health and the Environment, House Committee on Interstate and Foreign Commerce. An important tool used to analyze the Administration's proposal was a microsimulation model of the hospital industry. This paper describes the model for a technical audience.

Paul B. Ginsburg, T. Scott Thompson, and Lawrence A. Wilson, of CBO's Human Resources and Community Development Division, prepared this report under the supervision of Nancy M. Gordon and David S. Mundel. The authors wish to thank Malcolm Curtis, Joshua Greene, Daniel Koretz, and Frank Sloan for their useful comments. Francis Pierce edited the manuscript. Special thanks go to Toni Wright who patiently and expertly prepared the paper for publication.

An earlier version of this paper was presented at the Annual Meeting of the American Economic Association, Atlanta, Georgia, December 29, 1979.

In keeping with the Congressional Budget Office's mandate to provide objective and impartial analyses, this study offers no recommendations.
 

Alice M. Rivlin
Director
February 1981
 
 

SUMMARY

CHAPTER I. INTRODUCTION

• The Hospital Cost Containment Act of 1979
• The Microsimulation Model

CHAPTER II. WHY MICROSIMULATION?

CHAPTER III. THE SIMULATION MODEL

• Aggregate Current Policy Forecast
• Creating the Simulation Base
• Simulating the Proposal
• Alternative Simulation Methods

CHAPTER IV. SAMPLE RESULTS AND SENSITIVITY ANALYSES

• Savings to Purchasers of Hospital Care
• Sensitivity Analysis

The Congressional Budget Office developed a microsimulation model of the hospital industry to analyze the Hospital Cost Containment Act of 1979 and other proposals aimed at reducing the growth in hospital expenditures. The analysis appeared in the paper Controlling Rising Hospital Costs, which was published in September 1979. This second paper describes the model, which was used to assess:

• The likely savings to the federal government and other purchasers of hospital care; and
  
  
• The distribution of these savings by type of hospital and by type of purchaser.

THE LEGISLATION

The Hospital Cost Containment Act of 1979 (H.R. 2626 and S. 570), proposed in March 1979 by the Carter Administration, would have set guidelines for increases in hospital expenditures and placed revenue controls on hospitals that failed to keep within them. The guidelines--based on the inflation rate for hospital purchases, population growth, and an intensity-of-resources factor--would have allowed hospital expenditures to increase by an estimated 14.5 percent in 1980. The controls would have limited increases in inpatient revenues per admission. Several kinds of hospitals--including small, nonmetropolitan hospitals and those in states with effective mandatory hospital cost containment programs--would have been exempt from the proposed program.
 

WHY MICROSIMULATION?

Microanalytic simulation involves forecasting the behavior of economic units--in this case individual hospitals--on the basis of a macroeconomic forecast and observed historical patterns of microeconomic behavior. With this process, data on past frequency distributions of levels and rates of change of hospital variables are used to project future distributions. As a result, values for individual hospitals sum to projected aggregate values, while the historical distributional patterns are preserved.

Microsimulation was chosen to analyze the proposal because of the complex nature of the legislation, interest in the distributional effects of the legislation, and the need for speed and flexibility in performing analyses. Computation of the guidelines and revenue limits specified in the legislation required an analysis over several years of joint distributions of several variables such as hospital type and hospital location, and rates of change of expenditures, wage rates, and admissions, a task best handled with a microsimulation model. Microsimulation also permitted speedy estimates of the effects of amendments to the legislation. The model also allowed estimation of which types of hospitals would have been most affected by the legislation.
 

THE MICROSIMULATION MODEL

The task of estimating the size and distribution of the potential savings of the hospital cost containment proposal involved several steps. Specifically, we forecast aggregate hospital expenditures and other variables, developed forecasts for these variables for each individual hospital, determined which hospitals would meet or exceed their guidelines, and, for those hospitals exceeding their guidelines, determined the savings--that is, the amount by which revenues would be limited by controls.

Aggregate Current Policy Forecast

Hospital industry expenditures, revenues, admissions, wage rates, and the labor/nonlabor factor mix were forecast with econometric time series models using quarterly data. The forecasts assumed no changes in current policies for 1980 to 1985.

Simulation Base

A file of individual hospital data covering 1972-1977 was transformed ("aged") on the basis of the aggregate current policy forecast so that it represented the population of hospitals six years later. The annual level for each variable of interest for each hospital was inflated by the ratio of the projected aggregate level for that variable to the actual aggregate level from the historical data. For example, since the increase in aggregate hospital expenditures between 1976 and 1982 was projected to be 140 percent, each hospital's 1976 expenditures were inflated by 140 percent to generate their values in 1982. As a result of this process, the ratio of a hospital's expenditures to the aggregate expenditure level for, say, the simulated 1982 survey was equal to the actual ratio in the 1976 survey.

Determining the Guidelines and Their Effects

Guidelines were determined for each hospital on the basis of formulas stipulated in the legislation. Once the guidelines were calculated, they were compared to the projected rate of increase in expenditures for each hospital. Hospitals with expenditure increases exceeding their guidelines were placed in a pool that would fall under controls the next year, while those that passed the guidelines were placed in the pool that would be reexamined the next year.

Measuring the Savings to Purchasers of Hospital Care

Limits on revenues per admission were calculated for those hospitals exceeding their guidelines the previous year. Savings were calculated by taking the difference between each hospital's projected revenues under current policies and those allowed by the legislation. For example, if a hospital's revenues were $1.00 million in 1979, and if they would increase by 15 percent to $1.15 million under current policies, a 12 percent revenue limit would result in $0.03 million savings ($1.15 million minus $1.12 million) to purchasers of hospital care. Total savings were determined by summing the individual hospital savings.

Sensitivity Analyses

Three tests were made to determine the sensitivity of the savings estimates to the aggregate hospital forecasts, to the time period used to simulate future years, and to changes in assumptions about hospital behavior in response to the legislation. First, the tests found that the savings estimates were moderately sensitive to errors in the aggregate forecasts of hospital resource intensity. For example, a one percentage-point increase in intensity growth (the difference between expenditure growth per admission and input price growth) would increase five-year savings by about 25 percent. Second, the estimates were not very sensitive to the specific years of survey data that were used. Third, the estimates were sensitive to assumptions about hospital behavior. While the main case assumed no change in hospital behavior in response to the guidelines, an alternative assumption that hospitals close to their guidelines would moderately alter their behavior would decrease five-year savings by about 20 percent.

This document is available in its entirety in PDF.