Hospital profitability in Florida: a revisitation.

Hospital financial performance and its determinants has been an important subject of empirical research and of great interest to investors, board members, creditors, rating agencies, and regulators. Sear (1991) examined hospital profitability in a sample of 50 investor-owned or for-profit (FP) and 60 not-for-profit (NP) hospitals located in Florida for the period 1982 to 1988. The legal distinction between FP and NP hospitals lies in issues related to tax exempt status--the ability to receive tax-deductible donations, and limitation in the distribution of profits and assets. His statistical results indicate that FP hospitals are more profitable than NP hospitals, and average length of stay and wages per adjusted patient day are the variables with significant explanatory power in assessing hospital profitability. Walker (1993), using a Logit regression model, found that financial variables, by themselves, are not adequate to discriminate between profitable and non-profitable hospitals. Watt, Renn, Hahn, Derzon & Schramm (1986) reported that FP hospitals had higher revenue than their NP counterparts. Herzlinger & Krasker (1987) reported that NP hospitals do not perform as well financially as FP hospitals nor do they compensate for this by returning more social benefits. However, Haddock, Arrington & Skelton (1989), Arrington & Haddock (1990), and Fitzgerald & Jacobsen (1987) disputed these findings on conceptual and methodological grounds. Based on a sample of hospitals located in Florida in 1980, Sloan & Vraciu (1983) found that FP and NP hospitals were virtually identical in terms of profitability. Forgione (1987) and Forgione, Schiff & Crumbley (1996) found that FP hospitals tend to be half the size of NP hospitals, with lower case mix, shorter length of stay, higher cost per day, lower cost per case, and greater profitability.

In this study, we reexamine issues related to hospital profitability by improving upon previous research in several directions. First, unlike most previous studies which use data from before or during the phase-in of the Medicare operating cost Prospective Payment System (PPS), the data set employed in this study is obtained from the post-PPS operating cost period and is therefore much more representative of the payment systems currently facing US hospitals. Unlike the previous cost-based reimbursement mechanism, payment under PPS is based on a predetermined payment rate. PPS operating cost payment was initiated in 1983 and was phased-in until 1987 providing hospitals with a transitional period to adjust to the new payment system. PPS capital cost payment was phased-in over the 10-year period 1991-2001. Second, we incorporate in our empirical analysis economic and financial variables (e.g. degree of competition and financial indebtedness) and also investigate the effects of conversion to FP status on hospital profitability. Third, we employ piecewise linear regression analysis which allows for a parsimonious modeling of certain types of nonlinearity in the variables. Our empirical evidence suggests that location, size, occupancy rate, profit status, competition, teaching status, and financial indebtedness continue to be statistically significant in explaining hospital profitability in the post-PPS era. Additionally, bed size and occupancy rate are found to be non-linearly related to hospital profitability.

The remainder of the paper is constructed as follows. The next section presents the data and provides descriptive summary statistics. The empirical results are presented in the following. We conclude with a summary of our main findings in the final section.

DATA AND ECONOMETRIC METHODOLOGY

The data set contains 79 FP hospitals, 70 NP hospitals, and 29 hospitals that converted from NP to FP status, covering the years 1991 and 1995. The data source is the Medicare Cost Report Data provided by HCIA, Inc., Baltimore, Maryland.

The measure of profitability in this study is the Return on Assets (ROA), a continuous financial status variable defined as net income divided by total assets. ROA reflects a type of efficiency ratio as it relates hospital output to non-labor inputs. The following regression model is estimated to explain variation in hospital profitability:

ROA = f(BEDS, OCCUPANCY, OWNERSHIP, LOS, DEBT, EMPLOYEES, TEACHING, SOLE, YEAR, CONVERT)

where,

EMPLOYEES = number of full-time equivalent employees per 100 admissions adjusted for case mix index.

OWNERSHIP = dummy variable indicating type of ownership control (equals 1 for NP status, 0 for FP status).

TEACHING = dummy variable, taking the value of 1 if the hospital provides teaching and intern training, 0 otherwise.

SOLE = dummy variable capturing the degree of competition facing a hospital (equals 1 if a hospital is the sole Medicare provider, 0 otherwise).

YEAR = dummy variable, taking the value of 1 if year is 1991 and 0 if year is 1995.

BEDS = number of beds in service, which enters the regression equation in piecewise linear fashion as follows:

* BEDS 0 to 100 = the value of the variable if BEDS < 100, = 100 if BEDS [greater than or equal to] 100.

* BEDS 100 to 500 = 0 if BEDS < 100, = the value of the variable minus 100 if 100 [less than or equal to] BEDS < 500, = 500 if BEDS [greater than or equal to] 500

* BEDS over 500 = 0 if BEDS < 500, = the value of the variable minus 500 if BEDS [greater than or equal to] 500.

DEBT = debt per bed in service. Debt is defined as bonds issued plus loans.

OCCUPANCY = percentage of beds in service occupied, entering the regression equation as follows:

* OCCUPANCY 0 to 0.10 = the value of the variable if OCCUPANCY < 0.10, = 0.10 if OCCUPANCY [greater than or equal to] 0.10.

* OCCUPANCY 0.10 to 0.50 = 0 if OCCUPANCY<0.10, = the value of the variable minus 0.10 if 0.10 [less than or equal to] OCCUPANCY < 0.50, = 0.50 if OCCUPANCY [greater than or equal to] 0.50.

* OCCUPANCY over 0.50 = 0 if OCCUPANCY < 0.50, = the value of the variable minus 0.50 if OCCUPANCY [greater than or equal to] 0.5.

CONVERT = dummy variable, taking the value of 1 if a hospital converted from NP status to FP status between 1991 and 1995, and 0 otherwise.

We employ a piecewise linear regression allowing for two changes in the slope coefficients for BEDS and OCCUPANCY. This relaxes the assumption of linearity in the relationship, which we found was not supported by the data, and constitutes a relatively parsimonious way to describe the patterns in the data. The estimation methodology used is ordinary least squares (OLS) with heteroscedasticity adjustment for standard errors according to White (1980).

Exhibit 1 reports descriptive statistics for the independent and dependent variables of the sample hospitals for the years 1991 and 1995. The profitability of our sample hospitals increased between years 1991 and 1995. Such a pattern was exhibited by both FP and NP hospitals. The enhanced financial performance of hospitals relates to improvements in collection and electronic payments, as well as hospital management learning to operate effectively under the new regime of operating cost PPS, by shortening patient LOS and raising inpatient case mix.

Hospitals show an increase in their size between years 1991 and 1995. Occupancy rates declined for the years 1991 to 1995. Declines in occupancy rates were observed for both FP and NP hospitals; however, the occupancy rate for NP hospitals was higher in both the 1991 and 1995 years. However, hospitals converting from NP to FP status showed steeper declines in occupancy rates than non-converting hospitals.

EMPIRICAL EVIDENCE

Exhibit 2 reports the empirical results for our regression model. A notable controversy in the healthcare field centers on the effects of ownership control on the economic performance of hospitals. The indicator variable of ownership status enters significantly in the regression equation. The negative sign of its coefficient estimate suggests that FP hospitals exhibit superior economic performance compared to NP hospitals, based on the variables examined. Consequently, the hypothesis that the type of ownership has no effect on profitability is rejected. FP hospitals have a higher ROA than NP hospitals. Such a finding is consistent with the results in Sear (1991, 1992), Forgione (1987), Herzlinger & Krasker (1987), Forgione, Schiff & Crumbley (1996), and others.

The teaching status variable has a negative coefficient suggesting inferior economic performance of teaching hospitals as compared to non-teaching hospitals. Teaching hospitals provide training for interns and residents. Usually teaching hospitals have an affiliation with medical schools and maintain a charitable role in the community. The significant difference between teaching and non-teaching hospitals is the scope of services provided by the teaching hospitals. Teaching hospitals tend to be larger and located in urban, economically depressed areas with a high charity care burden (HCIA, 1997). Consequently, teaching hospitals provide access for the indigent population in the surrounding area, with limited, or no, compensation. Teaching hospitals thus exhibit inferior economic performance relative to non-teaching hospitals, which emanates from the costs associated with training, teaching, and charitable services.

In a similar manner, hospitals with less than 100 beds in service are more profitable than hospitals with more than 500 beds. Small hospitals, usually located in rural areas were characterized with no, or low, penetration of managed care. Furthermore, rural hospitals usually face less competition and provide a wide range of services. Larger hospitals usually operate in urban locations with more local competition and a higher penetration of managed care--which puts substantial downward pressure on revenues, and shifts a greater uncompensated charity care burden onto the hospitals (ProPAC, 1997).

The higher the number of full-time employees adjusted for case mix, the lower the profitability, holding all other variables constant. Case mix index (CMI) is analogous to product mix in a manufacturing context. It is a measure of the mix of patient illness types treated in the hospital, relative to the national average, and proxies for relative resource consumption. Thus, a hospital with an above-average CMI is expected to consume more resources than a hospital with a lower CMI. Employee full-time-equivalents (FTEs) are divided by the CMI to provide an adjusted (standardized) FTE measure. An FTE employee is a good proxy for variable cost. However, EMPLOYEES has a low coefficient value with a low significance level. This suggests that hospitals may be operating on an optimal number of employees, and any reduction in the number of employees would not lead to significant improvement in profitability, however, the level of significance does not warrant strong conclusions.

Traditionally, the measure of performance of the hospital industry has relied on financial ratios derived from the hospitals' financial statements (income statement and balance sheet). The financial ratios measure the hospital's historic performance. Banks, creditors and rating agencies use these ratios to help predict the hospital's future performance and credit worthiness. However, this study demonstrates that additional, important measures should be used in evaluating hospital performance, such as occupancy rate, staffing ratio, and total expense per adjusted discharge. These measures tend to explain some of the underlying factors that produce favorable or unfavorable financial performance.

For example, since the implementation of operating cost PPS, and in the era of declining use of inpatient services vis a vis outpatient treatments, occupancy rate is a key predictor of financial performance. A declining trend in occupancy rate has an adverse effect on average cost, profitability, and liquidity. At a lower rate of occupancy, operating expense per adjusted discharge is greater, which will hinder ability to operate with financial efficiency.

The performance of the hospital industry is multidimensional. Major differences exist across hospitals in terms of their location, scope of services provided, size, ownership, organizational structure, and amount of graduate medical education provided. Moreover, associated with these structural and location differences, are variables such as inpatient and payer mix, government regulation, and several other non-financial factors, over which a hospital may have relatively little control, and which should be considered carefully. Furthermore, the use of the regression model must be considered in comparison with other quantitative and qualitative presentations. For example, in model (l), the type of ownership did have a significant effect on hospital profitability measured by the ROA. However, the median of ROA of NP hospitals increased from 4.075 percent in 1991 to 4.995 percent in 1995 (a 20 percent increase over a five year period); meanwhile, the ratio for FP hospitals increased from 5.535 percent to 11.80 percent for the same period (a 60 percent increase). The empirical work showed that the affiliation with a School of Medicine and location in urban versus suburban, or Urban versus rural areas had considerable influence on hospital profitability, services, and access for indigent populations (HCIA, Inc., 1997). The sample of hospitals that converted from NP to FP status was relatively small (133 hospitals). The conversion in ownership status did not show a major effect on a hospital's behavior, possibly due to antecedent culture of the management, and/or due to constraining regulation in the healthcare industry.

These findings raise the following two issues: (1) whether the reimbursement rate under PPS should equitably incorporate other characteristics and local factors; and (2) Due to government regulation and other factors beside the type of ownership, hospital managers in general may not be able to make dramatic changes in hospital levels of profitability.

CONCLUSIONS AND FUTURE DIRECTIONS

The results of this research demonstrate that operating efficiency is not uniform across ownership control categories. Furthermore, the effect of the type of ownership on healthcare services for the indigent populations needs further examination. Due to rapid changes in the healthcare system, current models may not produce stable results over time.

As discussed in Morck et al. (1988), insider ownership may reduce the probability of mergers and takeovers in non-healthcare industries. Research related to the prediction of hospitals' mergers and takeovers is therefore suggested. Prediction of hospital bankruptcy is another area for future research. There might be a strong correlation between bankruptcy and payment system, debt service load, access to health care, mergers, and acquisitions. The effect of efficiency on quality of service, the premium for receiving health services and organizational structure of hospitals also warrant future research. The trend in NP conversions to FP status is attracting the attention and opposition of the regulators and public citizen groups (Forgione, 1999). Further research on this phenomenon is also recommended.

EXHIBIT 1
DESCRIPTIVE STATISTICS FOR DEPENDENT AND INDEPENDENT VARIABLES

                               Conversion
         Variable                Status          1991           1995

Return on assets (ROA)        NP to FP              4.34           5.32
                              FP                   -0.85          10.55
                              NP                   -7.24          -4.76

Ownership status              NP to FP             29             29
(OWNERSHIP)                   FP                   79             79
(Number of hospitals)         NP                   70             70

Teaching status (TEACHING)    NP to FP              5              6
(Number of hospitals)         FP                   13             12
                              NP                   15             15

Conversion of ownership       From NP to FP        29             29
status during 1991-1995
(CONVERT)
(Number of hospitals)

Full-time equivalent          NP to FP              5.63           4.79
employees per 100             FP                    5.02           4.43
discharges, adjusted for      NP                    5.65           5.60
case mix (EMPLOYEES)

Length of stay per adjusted   NP to FP              4.75           3.77
acute case mix (LOS)          FP                    4.50           3.58
                              NP                    4.25           3.495

Debt per bed (DEBT)           NP to FP        134,095.75     156,6695
                              FP              112,808.75      81,562.09
                              NP              178,088.36     226,665.34

Sole community provider       NP to FP              0              1
(SOLE)                        FP                    2              1
(Number of hospitals)         NP                    3              3

Number of beds in service     NP to FP            259            263
(BEDS)                        FP                  208            213
                              NP                  324            335

Occupancy rate (OCCUPANCY)    NP to FP             50.80          48.12
                              FP                   50.23          46.43
                              NP                   55.75          50.46

Note: FP denotes for-profit hospitals and NP denotes not-for-profit
hospitals.

Source: Medicare Cost Report data, provided by HCIA, Inc., Baltimore,
Maryland.

EXHIBIT 2
PIECEWISE LINEAR ORDINARY LEAST SQUARES REGRESSION OF
RETURN ON ASSETS ON HOSPITAL CHARACTERISTICS
FOR YEARS 1991 AND 1995

          Independent Variables             Coefficient    t-values
                                             estimates

Constant                                      4.8274       0.80
OWNERSHIP                                     5.7153       4.54 ***
TEACHING                                     -3.8813      -2.76***
CONVERT                                     -12.33613     -5.39***
EMPLOYEES                                    -0.0503      -0.25
LOS                                          -1.3847      -2.10 **
DEBT                                         -0.0002      -4.79 ***
SOLE                                          3.1271       0.86
YEAR                                          7.2830       5.71 ***
BEDS 0 to 100                                -0.1447      -2.82***
BEDS 100 to 500                               0.0038       0.92
BEDS over 500                                -0.0018      -0.28
OCCUPANCY 0 to 10                             0.6640       0.28
OCCUPANCY 10 to 50                            0.9650       1.35
OCCUPANCY over 50                             1.1125       1.62
F statistic (marginal significance level)    10.97         0.0001 ***
Adjusted [R.sup.2]                            0.32

Notes:
The sample consists of 178 hospitals in 1991 (79 FP and 99 NP) and 178
hospitals in 1995 (79 FP, 29 FP that were NP in 1991 [converted from
NP to FP] and 70 NP). The standard errors are adjusted for
heteroscedasticity according to White (1980).

The F statistic corresponds to the null hypothesis that the coefficient
estimates for all variables included in the regression equation are
jointly zero.

*** Indicates statistical significance at the 0.01 level

** Indicates statistical significance at the 0.05 level

REFERENCES

Arrington, B., & Haddock, C. (1990). Who really profits from not-for-profits? Health Service Research, 25(2), 291-304.

Fitzgerald, J., & Jacobson, B. (1987). Study fails to prove for-profits superiority. Health Progress, 32-37.

Forgione, D.A. (1987). Incentives and Performance in the Health Care Industry: The Case of For/Non-Profit Multihospital Systems, Doctoral Dissertation, School of Management, University of Massachusetts at Amherst, Amherst, MA.

Forgione, D.A. (1999). Those conversion blues. Journal of Health Care Finance, 25(4), 38-46.

Forgione, D.A., Schiff, A.D., & Crumbley, D.L. (1996). Assessing hospital performance: an inventory of financial and nonfinancial metrics. International Association of Management Journal, Forum on Research in Healthcare Financial Management, 8(2), 65-83.

Haddock, C., Arrington, B., & Skelton, A. (1989). Who profit from not-for-profit: A reconsideration. Health Service Management Research, 2(2), 82-103.

HCIA, Inc. (1997). The Comparative Performance of U.S. Hospitals: The Sourcebook. Deloitte & Touche LLP, Chicago, IL.

Herzlinger, R., & Krasker, W. (1987). Who profits from non-profits? Harvard Business Review, 93-105.

Morck, R., Shleifer, A., & Vishny, R.W. (1988). Management ownership and market valuation: an empirical analysis. Journal of Financial Economics, 20(1), 293-316.

Needleman, J., Choilet, D., & Lamphere, J. (1997). Hospital conversion trends. Health Affairs, 16(2), 187-195.

ProPAC (1997). Medicare and the American Health Care System, Report to The Congress, June 1997. Prospective Payment Assessment Commission. Washington, DC, 94.

Sear, A.M. (1992). Operating characteristics and comparative performance of investor owned multihospital systems. Hospital and Health Services Administration, 37(3), 403-415.

Sear, A.M. (1991). Comparison of efficiency and profitability of investor-owned multihospital systems with not-for-profit hospitals. Health Care Management Review, 16(2),31-37.

Sloan, F.A., & Vraciu, R.A. (1983). Investor-owned and not-for-profit hospitals: addressing some issues. Health Affairs, 2(1), 25-37.

Walker, C. (1993). A cross-sectional analysis of hospital profitability. Journal of Hospital Marketing, 7(2), 121-138.

Watt, J., Renn, S., Hahn, J., Derzon, R., & Schramm, C. (1986). The effects of ownership and multihospital system membership on hospital functional strategies and economic performance. For-profit Enterprise in Health Care, National Academy Press, Washington, DC, 260-290.

White, H. (1980). A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity. Econometrica, 48, 817-838.

Mustafa Younis
Jackson State University (USA)

Dana A. Forgione
Florida International University (USA)

Mahmud Khan
Tulane University (USA)

John Barkoulas
University of Tennessee (USA)

The authors would like to thank Mr. George Pillari, Chairman and CEO, CBCA Inc. Oakland, CA for his generous support for this research, and Jacqueline Pugh, MD, Director, Veteran Evidence-Based Research Dissemination and Implementation Center, VA Health Services Research Center of Excellence, and Professor of Medicine at The University of Texas Health Science Center at San Antonio, for her valuable comments.

Address for correspondence: Mustafa Younis, School of Allied Health, The Medical Mall, 350 West Woodrow Wilson Avenue, Suite 3450, Jackson State University, Jackson, MS 39213 USA, [e-mail].