Printer Friendly

The demise of hospital philanthropy.



Until the beginning of this century, most hospitals primarily served the poor and derived a major portion of their revenue from private philanthropy. The more affluent avoided hospital care and were treated in doctors' offices and at home (Starr [1982]; Rosenberg [1987]). This situation changed largely because of technological improvements such as anesthesia and asepsis, and the nonpoor began to demand care in hospitals. Private insurance plans offering the nonpoor nearly complete coverage for hospital care began in the 1930s and spread rapidly after World War II. major health legislation, Medicare and Medicaid, was enacted in 1965, extending hospital insurance to the elderly and the poor. By the late 1960s, hospitals derived nine-tenths of their revenue from insurance. Real donations for medical facility construction rose from $76 million in 1935 to a peak of $2.1 billion in 1965 and fell to $603 million by 1981 (1984 dollars).(1) In 1984, only 5 percent of the total spent on such construction was funded by philanthropy (Levit et al. [1985]).

The conventional wisdom has it that private and public hospital insurance has "crowded out" philanthropic support of hospitals. The public insurance argument is consistent with that of Abrams and Schmitz [1978] and Roberts [1984] who maintain that public welfare expenditures reduce or eliminate private giving to the poor. This argument appears to extend to private insurance as well.

This article describes the decline in donations to hospitals from conceptual as well as empirical perspectives. We first describe a simple one-period model of hospital and donor behavior. Hospitals choose utility-maximizing levels of output, while altruistic donors decide how much to contribute to hospitals and how much to spend on other goods. The model is used to analyze how insurance, various government subsidies, and other factors affect hospital and donor behavior.

The major insight from this model is that although rising insurance coverage tends to increase hospital output in the same way as a lump-sum government subsidy, the two have different effects on donations to the hospital. Growing insurance coverage may increase or decrease donations, while lump-sum subsidies to the hospital will unambiguously crowd out private donations.

The empirical work following the model incorporates empirical counterparts of the exogenous and donation variables examined in the theoretical section. Data come from (1) a time series spanning 1935-81 and (2) a single cross-section of nonprofit hospitals in 1978. The results indicate that increased insurance coverage for hospital care, especially the enactment of Medicare and Medicaid in the mid-1960s, reduced giving to hospitals. The contribution of public subsidies for construction varied, depending on whether the subsidy more closely resembled a matching or a lump-sum subsidy.


The model consists of a single not-for-profit hospital and a single representative donor.(2) The hospital chooses output (n) to maximize its utility from output,(3) while the donor chooses donations (D) to maximize utility from total hospital output and other goods (y). The donor behaves altruistically in that he does not use hospitals himself. The behavior of hospital patients is not explicitly modeled, although their presence will be felt. The output and donation decisions are interrelated since the donor likes hospital output and the level of donations affects the hospital's budge get constraint. To untangle the endogenous feedback effects between output and donations, the problem is described as a two-stage sequential game. In the first stage, the donor decides how much to donate. In the second stage, the hospital chooses how much output to produce, given the previously-determined level of donation.

Of course, the donor is acutely aware that the hospital's subsequent output decision will affect his utility. We assume that the donor correctly anticipates how the hospital will use his donation. To incorporate this assumption into the donor's behavior, we first solve the hospital's optimal output rule in the second stage as a function of exogenous factors and a predetermined level of donations. The hospital's decision rule from this stage is considered by the donor in determining his optimal level of donations. The solution yields as estimable reduced form donations equation.

The Hospital's Output Decision

The hospital's utility is denoted by V(n) where [V.sub.n] [is greater than to] 0, [V.sub.nn] [is less than to] 0. In the second stage (when donations are predetermined), the hospital chooses output to maximize utility subject to a breakeven constraint. The hospital receives its revenue from three sources: patients, lump-sum government subsidies (G), and donations (D). The hospital maximizes utility subject to a zero profit constraint:

Max V (n)

subject to

p (n,I)n + G + D - C(n) [is greater than or equal to] 0.

The inverse demand curve facing the hospital is p(*), and I is a measure of the extent of consumers' insurance coverage. Price increases as insurance coverage rises ([p.sub.I] [is greater than to] 0), and the demand curve for out-put is downward-sloping ([p.sub.n] [is less than to] 0). In addition, assume that marginal revenue ([p + p.sub.n] n) is non-increasing in n. The total cost of producing n units of output is C(n), with [C.sub.n] [is greater than to] 0 and [C.sub.nn] [is greater than or equal to] 0.

Solving the Lagrangian for (1) yields:


Since hospital utility increases with output, the constraint must be binding. At the optimal level of output (n.sup.*), marginal revenue is less than marginal cost.(4) The effects of changes in donations, government subsidies, and insurance coverage were derived using comparative statics analysis and are reported in Table I.

A one-dollar increase in donations or government subsidy relaxes the budget constraint by one dollar, allowing the hospital to produce more output. An increase in insurance coverage boosts patient willingness-to-pay for a unit of output by [p.sub.I] with the result that patient revenue rises by [p.sub.I] n; the hospital will increase output to absorb this extra revenue.

To understand how exogenous changes in government subsidies and insurance coverage affect donations, it is first necessary to determine how these variables affect the marginal product of donations on output [Mathematical Expression Omitted]. These comparative statics results are also shown in Table I. A change in D or G simply has an output effect on [Mathematical Expressions Omitted]: as D or G increases output,the marginal revenue - marginal cost differential becomes more negative, causing the marginal product of donations to fall. The effect of changes in insurance coverage is different. It too has an output effect, but a second effect of increased insurance is to boost marginal revenue.(5) This effect will increase the marginal product of donations, since one dollar of donations will produce more output when combined with higher payments from patients and their insurers. Combining the effects, insurance has an ambiguous overall effect on the marginal product of donations. [TABULAR DATA OMITTED]

The Donor's Decision

The donor decides how much to donate to the hospital and how much of a composite good (y) to consume.(6) The donor's utility is given by U(y,n), with [U.sub.y], [U.sub.n] [greater than to] 0, [U.sub.yy], [U.sub.nn] [less than] 0, and [U.sub.yn] [is greater than or equal to] 0. Output is the same for both donor and hospital.(7) The donor's budget constraint states that disposable income equals consumption plus donations. For simplicity, we assume a proportional income tax and tax-deductible donations. The donor knows that the hospital will produce [n.sup.*] units after the donations are made. Thus the donor maximizes utility as a function of composite good and hospital output subject to his budget constraint.

Max U(y,[n.sup.*] (I,G,D,))


subject to

[I.sup.D] [equal to] [tau] ([I.sup.D] -D) + y + D

where [I.sup.D] is the donor's income and tau is the marginal personal tax rate. Solving the budget constraint for y in terms of disposable income, taxes, and donations, equation (3) may be rewritten as:


The first-order condition from (3') states that the marginal utility from the composite good equals the marginal utility from donations. Using comparative statics yields the effects of exogenous variables on donations shown in Table II. Assuming that output is a normal good for the donor, donations rise with pre-tax income. (The term [-Phi] / SOC in Table II gives the effect of a one-dollar change in after-tax income.) An increase in the marginal tax rate causes the relative price of donations to fall, but also reduces the donor's after-tax income, producing an ambiguous overall effect on donations. Increased government spending decreases donations; this is the "crowding out" effect. Although negative, the effect of government subsidies on optimal donations ([D.sup.*.sub.G]) is between zero and -1, indicating that an increase in government subsidies to hospitals does not, by itself, completely crowd out an equal amount of donations.(8) Intuitively, if G increases by one dollar, the donor could lower donations by one dollar and the donor's utility would not change. That leaves, however, an additional ([1 - tau]) dollars of income which the donor would split between additional purchases of y and D.(9)


The most interesting result is that increased insurance coverage has an ambiguous effect on donations because insurance's effect on the marginal product of donations ([n.sup.*.sub.DI]) cannot be signed unambiguously. By providing a substitute source of revenue to the hospital, increased coverage crowds out donations (the first term in the result for I in Table II and part of the second term). However, it also raises patient willingness to pay for hospital output, which lowers the shortfall between marginal revenue and marginal cost and allows a dollar's worth of donations to buy more output (the rest of the second term). The patient in effect matches part of the donor's gift. If this impact is large enough, the marginal product of donations will increase with insurance (i.e., [n.sup.*.sub.DI] [is greater than to] 0) and the second term in ([D.sup.*.sub.I]) will be positive. If the second term dominates the first, increased insurance coverage would actually stimulate, or "crowd in," donations.(10)

This result shows that increased government financing of hospital insurance programs will not have the same effect on donations as equivalent government lump-sum subsidies to hospitals. Under some circumstances, in fact, expenditures which increase insurance coverage will actually increase donations, while lump-sum subsidies will reduce donations. The difference occurs because donations and lump-sum subsidies to the hospital are perfect substitutes as far as the donor is concerned. Both allow the hospital to increase output by cutting prices to all patients. In contrast, donations and increased insurance coverage are not perfect substitutes, since the latter allows patients to pay for more of their care.(11)

Extensions of the Model

The basic model can be extended to incorporate other variables which effect hospital donations. First, some of the government subsidies available to hospitals have been matching grants instead of lump-sum grants. The most important of these was the Hill-Burton program which provided matching grants for hospital construction; it generally paid one-third of construction costs on approved projects (Lave and Lave [1974]). Matching grants provide "more bang from a buck" of donations. Their impact can be incorporated into the model by changing D to (1+m)D in the hospital's budget constraint in (1), where m is the matching percentage. Total government subsidies will then equal mD + G with the size of the matching grant depending endogenously on the amount of donations. Like insurance coverage, an increase in the matching percentage has an ambiguous effect on the marginal product of donations and therefore an ambiguous effect on donations themselves. We can definitely predict, however, that a one-dollar lump-sum subsidy will crowd out more donations than a similar amount paid as a matching subsidy.

We can also analyze the impact of hospital input prices by studying how they affect the hospital's cost function in (1). The total effect of an input price (w) change is the sum of a direct effect of input prices on output ([n.sup.*.sub.w]) and the effect of the change in w on the marginal product of donations. According to the direct effect, an input price increase would unambiguously decrease output and hence increase donations. However, the effect operating through the marginal product of donations is ambig as in the case of insurance, making the overall effect of a change in [omega] on donations ambiguous.

Finally, the donor's utility can be modified so the donor receives utility directly from the size of his donation. This third utility argument, which Andreoni [1987] called the "warm-glow" effect of giving, will generally promote higher donations but will not change the qualitative effects of the exogenous variables on donations. In particular, increased government lump-sum subsidies will still crowd out donations, while insurance coverage will still have an ambiguous effect on donations.[12]


This section investigates determinants of donations to hospitals empirically. In it we estimate donations equations using two samples, a time series and a cross-section.

Time Series Analysis

The time series is the real value of philanthropic contributions to the construction of health care facilities, almost all of which was for hospitals, from 1935 through 1981. Data through 1975 come from Terenzio [1978], who assembled the series from varied sources. We updated Terenzio's series through 1981 using data on contributions from the American Hospital Association's (AHA) Sources of Funding for Construction Surveys (Mullner et al. [1981] and Mullner et al. [1983]). Values for the dependent variable for 1970-72 were interpolated. Data are also not available for 1936-39, 1941-44, and 1946-47. Given the possible instability in donations in the period around World War II, we did not interpolate values for these years. Our time series sample contains thirty-seven observations (1935, 1940, 1945, 1948-81).

The time series analysis contains empirical counterparts to these explanatory variables from our model: the percent of persons with private insurance for hospital care and real Federal expenditures on Medicare and federal-state expenditures on Medicaid are measures of insurance (I); real Federal expenditures under the Hill-Burton hospital construction program and the value of tax-exempt bonds issued on behalf of hospitals are both measures of public subsidies in millions of dollars (G); total real household wealth (I.sup.D); the marginal tax rate faced by households with taxable incomes of $100,000 in 1984 dollars ([tau]) expressed as a percentage; and a factor price, the real prime interest rate in percents (r). In addition, we included a variable for the supply of potential donors, the percent of the U.S. population over age sixty-five. Age has been used in previous empirical studies of donations (e.g., Clotfelter [1985] and Kingma [1989]). All monetarily-expressed variables are in 1984 (Consumer Price Index) dollars and pertain to the U.S. as a whole.

The Hill-Burton program, in effect from 1948-78, provided matching grants for hospital construction, mainly in smaller communities and poorer states. To be eligible for Hill-Burton funds, projects had to be in compliance with a state health plan and other requirements. This was a closed-ended program. Total funding in constant dollars peaked in 1968. Tax-exempt bond financing first became available to nonprofit hospitals in the early 1970s (Clapp and Spector [1978,295]). We measured the total value of tax-exempt bond issues on behalf of hospitals; the subsidy in the form of reduced personal income tax obligations is directly proportional to the value of bonds issued. The tax variable is for households in the $100,000 bracket because families at this income level and above are comparatively likely to give to hospitals (Clotfelter [1985,24]). The initial regressions also included a wage variable, but it was highly correlated with wealth and other explanatory variables.

Cross-Sectional Analysis

The data on donations for the cross-sectional analysis come from the American Hospital Association's 1978 Special Hospital Topics Survey. We limited this analysis to the 784 private nonprofit hospitals that both responded to this survey and provided data on pertinent explanatory variables in their response to the AHA's 1978 Annual Survey of Hospitals and the AHA's 1979 Reimbursement Survey. Of the hospital respondents in our sample, 89.2 percent received some donations during the year preceding the survey.

The dependent variable is total donations received by an individual hospital in 1978 and, alternatively, total donations per adjusted (for out-patient output) admissions. The first dependent variable is theoretically more appropriate in that admissions are an element of n, but much of the variation across hospitals reflects differences in hospital size. When total donations is the dependent variable, explanatory variables for broad hospital bedsize categories are included (bedsize categories are: 100-249, 250-399, and 400 and over with bedsize under 100 being the omitted reference category).

The explanatory variable counterparts to our model are: bad debt and charity care divided by net hospital revenue, percent of net hospital revenue from cost-based sources, and percent of net revenue subject to regulation by a state hospital rate-setting agency for the insurance variable (I); personal per capita income in the hospital's county ([I.sup.D]); state personal income taxes as a percent of state personal income (tau); and nurse payroll per full-time equivalent nurse employed by the hospital for the hospital wage rate [omega]. "Net revenue" is used in the hospital industry to refer to gross revenue less various types of discounts. The cost-based share and rate-setting variables operate through the hospital's inverse demand function p(*). Unfortunately, no hospital-specific measures of government grants were available. Except for reasons of default risk, for which we had no direct measure, interest rates do not vary among hospitals in a single time period.

By 1977, only 9 percent of the U.S. population did not have insurance for hospital care (Farley [1985]). Some hospitals operated in markets with a much higher share of uninsured. Unfortunately, data by hospital or local areas on the percent uninsured are not available for 1978. However, judging from more recent data which permit a comparison, the amount written off by the hospital as bad debt-charity care as a ratio to its revenue is highly and negatively correlated with the percentage of its patients without health insurance (Sloan et al. [1986]). Cross-sectional insurance measures are complicated by differences in the calculation of payments. Insurers that pay cost (Medicare, Medicaid, and some Blue Cross plans) rather than charges obtain higher discounts from hospitals (Ginsburg and Sloan [1984]). This is, in part, a reflection of an explicit disallowance of a payment of a return on donated capital (Long [1976]; Conrad [1984]). Likewise mandatory state rate-setting programs reduce hospitals' output price (Sloan [1981]; Morrisey et al. [1983]).

As in the time-series analysis, we include a variable for the percent of the population over sixty-five in the hospital's area. In addition, binary variables distinguish between nonprofit hospitals run by religious groups and secular nonprofits, hospitals located in SMSAs versus those which are not, and teaching (member of Council of Teaching Hospitals) versus nonteaching hospitals. The teaching variable undoubtedly embodies some of the qualitative attributes in the weighted output variable n. However, for practical reasons, it, like hospital bedsize, must be considered to be exogenous in our empirical work.

Means and Standard Deviations

Means and standard deviations of the variables in the time series and cross-section analyses are shown in Table III. [TABULAR DATA OMITTED]


Time Series Results

The time series results on real donations for hospital construction are

DON = -4189.73(a) - 32.84INS(a) - 0.047MM(a) + 1.21HB(a) - 0.056BOND

(1452.66) (9.89) (0.014) (0.33) (0.049) + 0.0028WEALTH - 8.03 [tau] - 41.70r(b) + 628.10(b) POLD [R.sup.2] = 0.87
(0.0011) (8.87) (18.85) (245.35) D-W = 1.09
 N = 37 (4)

(a) = significant at 1% (two-tail test); (b) = significant at 5% (two-tail test).

The insurance variables, the percent of population with private hospital insurance (INS) and real government expenditures on Medicare and Medicaid (MM), have negative and statistically significant impacts on private donations to hospitals in (4). The associated elasticities evaluated at the observational means are substantial: -2.0 for INS, and -0.5 for MM. Contributions to hospitals reached their peak in 1965 at $2.1 billion and by 1981 had fallen to $603 million (1984 $). According to our estimates, Medicare-Medicaid alone contributed to a $1.9 billion drop in private giving to hospital between 1965 and 1981 and the small growth in private coverage for hospital care (from 72 percent in 1965 to 81 percent in 1981) reduced giving by another $290 million. Thus, changes in insurance coverage alone more than account for the substantial drop in giving to hospitals after 1965.

Measures of government subsidies of hospital construction in (4) are the real value of tax-exempt bonds issued through state bonding authorities (BOND) and the real value of Hill-Burton subsidies (HB). The real value of tax exempt bonds has a negative impact on donations, supporting the dominance of crowding out by public subsidies, but the coefficient lacks statistical significance at conventional levels. Hill-Burton subsidies (HB) raised private donations, probably because it was a matching grant. Because HB may be considered endogenous, we reran equation (10) with a dummy variable substituted for HB. The rationale for this alternative specification of HB is that the existence of the program itself was not endogenous to donations, but the actual amount of the matching subsidy depended in part on the level of donations. The results were essentially the same: there were no sign reversals and only modest changes in the statistical significance of the estimated parameters. The coefficient on the dummy for the Hill-Burton program was positive and statistically significant at almost the 1 percent level.

As predicted (see Table II), household wealth has a positive and statistically significant effect on private donations. According to our theoretical analysis, the tax rate has an ambiguous effect on donations (Table II). Unlike other studies of contributions based on household data, we found that the Federal personal income tax rate on real income of $100,000 (tau) had essentially no effect on donations. We experimented with alternative measures of [tau]; none had a statistically significant impact.

The real interest rate (r) has a negative impact on donations. The effect of changes in input prices such as r cannot be deduced from the model, but apparently the negative terms dominate.

The coefficient on the percent of population over age sixty-five (POLD) suggests that, ceteris paribus, the aging of the U.S. population increased private donations to hospitals. This is plausible because the elderly probably have a greater taste for donations, which operates through the marginal rate of substitution between n and y in the model. In addition to this altruistic motive, they may also expect to use hospitals more often and sooner.

The Durbin-Watson value of 1.09 is based on the 1948-69 period for which annual, non-interpolated data are continuously available. The test of autocorrelation falls in the uncertain range. We did not correct for autocorrelation because autocorrelation appears to be small, and we would have had to drop a number of observations (those for noncontinuous years) to make this adjustment. We also examined the effects of interpolating values for 1970-72; the coefficient on a dummy variable for those years was insignificant and left the remaining coefficients virtually unaffected.

Cross-Sectional Results

Table IV reports the results of the 1978 cross-sectional analysis. Among the three insurance variables, only the hospital's bad debt-charity burden (INDIGENT), a proxy for lack of insurance of patients in the hospital's market area, shows a statistically significant positive influence on donations received by the hospital. This result adds support to the view that increased insurance coverage crowds out private giving.

Table : TABLE IV Cross Section Regression Results
Explanatory Donations per
Variables Total Donations Adj. Admissions
INTERCEPT -1.32E6(a) -93.82(a)
 (0.28E6) (25.27)
INDIGENT 3.92E6(a) 504.40(a)
 (1.33E6) (131.95)
CSHARE -0.49E5 -9.28
 (0.18E6) (17.86)
RATESET -0.41E5 -10.09
 (0.87E5) (8.63)
STAX 1.89E6 236.54
 (3.01E6) (298.06)
INC 119.98(a) 0.009(a)
 (24.42) (0.002)
WAGE 15.14(b) 0.0004
 (7.83) (0.0008)
POLD 2.88E6(a) 329.53(a)
 (0.92E6) (89.90)
TEACH 0.78E6(a) 29.32(a)
 (0.11E6) (9.04)
CHURCH -0.12E6(c) -8.27
 (0.07E6) (6.47)
SMSA -0.11E6 -6.94
 (0.08E6) (7.73)
BED1 5.21E3 -
 (0.14E6) (-)
BED2 0.13E6 -
 (0.14E6) (-)
BED3 0.26E6(c) -
 (0.16E6) (-)
[R.sup.2] 0.22 0.08
F 17.14 6.31
N 784 784

Note: Standard errors are shown in parentheses.

(a) Significant at the 1% level (two-tail test).

(b) Significant at the 5% level (two-tail test).

(c) Significant at the 10% level (two-tail test).

The state tax rate on personal income (STAX) has a positive impact on donations but with a high associated standard error. Personal per capita income (INC) in the hospital's market (defined as the county) has a positive influence as did wealth in the time series regression. The annual wage the hospital paid nurses (WAGE) has a positive sign, but the parameter estimate is statistically significant only in the total donations regression.(13) The interest rate had a negative effect on donations in the time series. This is not necessarily a contradiction; different inputs could have varying effects on the terms expressing the effect of an input price change on donations. The result on the percent elderly in Table IV again indicates that the elderly donate more to hospitals.

The remaining explanatory variables represent influences not explicitly incorporated in the model. They may affect the donor's marginal rate of substitution between donations and the composite good. Teaching hospitals were more likely to receive donation, probably because they had greater appeal to donors on quality grounds. Religiously-affiliated hospitals were less likely to receive donations. It is likely that donors contributed to hospitals through their churches, and the hospital reported such gifts as other income rather than as donations if and when such funds were transferred to the hospitals.

Hospitals located in metropolitan areas (SMSA) were less likely to receive much in the form of donations. One explanation is that a donor's marginal utility from increasing the output of an individual hospital is higher when there is only one (or a few) hospital in the area.(14)


There has been substantial growth in insurance coverage for hospital care, both private and public. The conventional wisdom among industry experts is that growth of insurance crowds out private giving to hospitals. This view runs parallel to results of studies of the effects of public subsidies on private giving to nonprofit organizations in general. The comparative statics analysis in this paper suggests that increased insurance coverage may or may not crowd out private donations to hospitals. The effect of insurance is qualitatively different from the effect of a lump-sum subsidy. The time series and cross-sectional results indicate that increased insurance coverage does crowd out private giving.

The effect of public subsidies of hospital capital on private donations to hospitals depends on how the subsidy is structured. While the program existed, Hill-Burton grants stimulated private donations. This result is not surprising since recipient hospitals had to match Hill-Burton subsidies with private funds. Tax-exempt bonds became available to hospitals on a widespread basis about the time the Hill-Burton program ended and the percent of the U.S. population with hospital insurance reached its peak. By providing a low cost substitute for private donations, tax-exempt bonds may have further reduced hospital donations. Our conclusion on the score is tempered by the fact that the coefficient on the tax-exempt bond variable is not statistically significant at conventional levels.

Increased affluence of the population made potential donors more willing to give to hospitals as did a higher proportion of elderly persons. However, we were unable to theoretically determine the direction of the effect of changes in marginal income tax rates. This plausibly accounts for the inconclusive findings on marginal tax rates, both in the time series and in the cross section.

Previous estimates of the price elasticity of charitable contributions, based on variations in marginal tax rates, indicate that donations rise with increases in marginal rates. (See, for example, Boskin and Feldstein [1977]; Clotfelter [1985]; Feldstein [1975]; Feldstein and Taylor [1976]; and Reece [1979]). Such work was based on microdata from donors. This paper's results by contrast do not suggest that the secular decline in marginal personal income tax rates has contributed to the demise of hospital philanthropy.

Does the demise of hospital philanthropy mean the demise of the nonprofit hospital? It appears that the nonprofit hospital continues to exist even with trivial amounts of private donations. The private nonprofit hospital remains the dominant ownership form in this sector. A substantial body of research documents the competitive advantages of nonprofit hospitals. Lave and Lave [1974], for example, demonstrate that the Hill-Burton program was somewhat effective in restricting investor-owned hospital entry into the 1946-70 market. McCarthy and Kass [1983] argue that state certificate of need laws (CON) restricted entry of investor-owned hospitals into many local markets. Sloan et al. [1987] show that the ability to offer tax-exempt debt reduced the cost of debt for nonprofit hospitals by over one percentage point, controlling for hospital risk and characteristics of the debt. The elimination of the Hill-Burton program in 1978, the expiration of federal statutes requiring states to have certificate of need laws, and recent proposals to drastically cut back on the access nonprofit hospitals have to tax-exempt debt, however, all represent threats to the private nonprofit hospital's market share.

( 1.) Data come from the time series described below in section III.

( 2.) By assuming one donor and one hospital, we do not consider the free rider problem that might arise with more than one donor. Nor do we assess entry of charities. In Rose-Ackerman [1982], charities compete for funds by expending resources on donor solicitation. Her model predicts that, in the absence of entry barriers, the fundraising share of the marginal charity approaches one.

( 3.) In previous work, some models of nonprofit hospitals have had hospital quantity and quality as arguments in the hospital's utility function (see, for example, Newhouse [1970], and Feldstein [1977]). Such hospitals maximize utility subject to a breakeven constraint. The focus of such work is on the hospital's quantity-quality tradeoff. Given our focus, the cost of added complexity of separating quantity and quality outweighs any benefit in terms of added "realism." In our model, more quantity is consistent with higher quality.

( 4.) Of course, a for-profit hospital would stop production where marginal revenue equals marginal cost. Donors are unlikely to contribute to for-profit hospitals since the donations simply increase profits without increasing output.

( 5.) Theoretically, it is possible that an increase in insurance coverage would cause patient demand to increase and marginal revenue to decrease at some levels of output. We assume that the more plausible case holds: increased insurance coverage causes patient demand to shift upward so that marginal revenue also rises at every output level.

( 6.) Our assumption differs from the assumption in typical models of altruism that the altruist receives utility from total expenditures on the public good. The assumptions are equivalent if the price and cost of the public good are constant; that is, a one-dollar increase in expenditures will always have the same marginal effect on output. In our model, where inverse demand falls and marginal cost may rise as output increases, an additional dollar of donations produces less additional output as output increases.

( 7.) Rose-Ackerman [1987] considered a "buying in" effect in addition to a substitution effect, which is part of our model of donor behavior. By "buying in," she meant that a dollar's donation may produce more donor satisfaction if the recipient institution already has a high quality or output. But even with buying in, the substitution effect dominated "buying in" in her model in equilibrium. Our substitution effect is discussed below.

( 8.) This can be seen by rearranging the two negative terms in Table II so that [MATHEMATICAL EXPRESSION OMITTED] [is equal to] -1 plus the income effect.

( 9.) Recent criticisms of models of altruism have focused on the possibility that a one-dollar increase in government spending on a public good financed by a one-dollar increase in taxes will completely crowd out one dollar of private donations for the public good (see, for example, Margolis [1982]; Roberts [1984]; Bernheim [1986]; Bergstrom et al. [1986]; and Andreoni [1987]). In the typical model of altruism, it is easily shown that if the government levies a one-dollar lump-sum tax on a donor and then spends the proceeds on the public good, the donor's contributions will decrease by a dollar. Complete crowding out occurs because the lump-sum tax and subsidy leaves the donor's budget constraint unchanged. Bernheim [1986] and Andreoni [1987] extended the argument to show that if the price of donations is distorted through tax deductions, direct government subsidies for the public good will still completely crowd out private donations if they are financed by equal taxes on the donor.

Andreoni maintained that pure altruism cannot explain charitable giving in the U.S. where massive increases in government spending have not completely crowded out private giving. He proposed a model in which donors receive utility from the act of giving itself, as well as the overall level of spending on the public good. In the presence of this "warmglow" effect, complete crowding out does not occur.

If our model is expanded to include lump-sum taxes, increases in G financed by lump-sum taxes borne only by the donor will completely crowd out donations. Because taxes are spread across both donors and nondonors and increases in government subsidies to hospitals are not necessarily accompanied by matching increases in taxes, we are primarily interested in the partial effects of G and tax rates separately.

(10.) This result has clear implications for the argument that subsidies for public goods which are financed by matching lump-sum taxes will completely crowd out private giving. Suppose that government imposes a one-dollar lump-sum tax on the donor. If y is a normal good, D will fall by less than a dollar as a result of the tax. If the government then spends one dollar to increase insurance coverage (through Medicare or Medicaid) and [MATHEMATICAL EXPRESSION OMITTED] [greater than] 0, complete crowding out will not occur.

(11.) Since donors care about total output and not simply total hospital expenditures, they may be better off if government expenditures are funneled into insurance coverage, particularly if the insurance coverage is targeted to individuals who could not otherwise afford hospital care. We assume that information problems such as adverse selection prevent donors from giving donations directly to the poor. People desiring a subsidy may just claim they are poor to elicit the subsidy. In practice, the government may be better able to identify individuals who need public insurance coverage. Rose-Ackerman [1981] and Kingma [1989] discuss crowding out issues when donations to charitable organizations and government expenditures are imperfect substitutes.

(12.) However, the degree of crowding out caused by increased G will be less complete, and increased insurance coverage will be more likely to crowd in donations.

(13). This is not an artifact of area cost of living differences. In earlier runs we adjusted all monetary variables by a state cost of living index. The results were essentially unchanged.

(14). Fama and Jensen [1983] argued that donations decisions are partly guided by the extent to which the philanthropist can have confidence that the donated funds will not be misused. The nonprofit organizational form was seen as a mechanism to avoid donor-residual claimant conflicts. Residuals can be appropriated by internal agents, however. Thus, Fama and Jensen asserted that nonprofit firms will separate management - initiation and implementation - from control - ratification and monitoring. Specifically, they hypothesize that nonprofit boards will include few, if any, internal agents as voting members.

To investigate the importance of board composition to giving, we experimented with two variables in preliminary work; a binary variable equal to one if the chief executive officer of the hospital also was chairman of the hospital board; and medical staff board members as a percent of total hospital board members. Administration (Sloan [1980]] and the medical staff (Pauly and Redisch [1973]) are often characterized as de facto hospital residual claimants. Using ordinary least squares, neither variable showed an impact on donations.


Abrams, B. A. and M. D. Schmitz. "The `Crowding-out' Effect of Government Transfers

on Private Charitable Contributions." Public Choice 33(1), 1978, 30-9.

Andreoni, J. "Private Charity, Public Goods and the Crowding Out Hypothesis."

Unpublished paper, Social Systems Research Institute, University of Wisconsin, 1987.

Bergstrom, T., L. Blume, and H. Varian. "On the Private Provision of Public Goods." Journal

of Public Economics, 29(1), February 1986, 25-49.

Bernheim, B. D. "On the Voluntary and Involuntary Provision of Public Goods." American

Economic Review, 76(4), September 1986, 789-93.

Boskin, M. J. and M. S. Feldstein. "Effects of Charitable Deductions by Low-income and

Middle-income Households: Evidence from a National Survey of Philanthropy."

Review of Economics and Statistics, August 1977, 351-54.

Clapp, D. C. and A. B. Spector. "A Study of the American Capital Market and its

Relationship to the Capital Needs of the Health Care Field," in Health Care Capital:

Competition and Control, edited by G. K. MacLeod and M. Perlman. Cambridge:

Ballinger Publishing Co., 1978, 275-304.

Clotfelter, C. T. Federal Tax Policy and Charitable Giving. Chicago: University of Chicago

Press, 1985.

Conrad, D. A. "Return on Equity to Not-for-Profit Hospitals: Theory and Implementation."

Health Services Research, April 1984, 41-64.

Fama, E. F. and M. D. Jensen. "Separation of Ownership and Control." Journal of Law and

Economics, June 1983, 301-25.

Farley, P.J. Private Insurance and Public Programs: Coverage of Health Services. NCHSR National

Health Care Expenditure Study Data Preview 20, March 1985.

Feldstein, M.S. "Quality Change and the Demand for Hospital Care." Econometrica, October

1977, 1687-1702.

___. "The Income Tax and Charitable Contributions: Part II - The Impact on Religious,

Educational and Other Organizations." National Tax Journal 28(2), June 1975, 209-26.

Feldstein, M. S. and A. Taylor. "The Income Tax and Charitable Contributions." Econometrica,

November 1976, 1201-22.

Ginsburg, P. and F. A. Sloan. "Hospital Cost Shifting." New England Journal of Medicine, 5

April 1984, 893-98.

Kingma, B. R. "An Accurate Measurement of the Crowd-out Effect, Income Effect, and

Price Effect for Charitable Contributions." Journal of Political Economy, 97(5), 1989,


Lave, J. P. and L. B. Lave. The Hospital Construction Act. Washington, D.C.: American

Enterprise Institute, 1974.

Levit, K. R., H. Lazenby, D. R. Waldo, and M. D. Lawrence. "National Health Expenditures,

1984." Health Care Financing Review, Fall 1985, 1-35.

Long, H. W. "Valuation As a Criterion in Not-for-Profit Decision Making." Health Care

Management Review, Summer 1976, 34-52.

Margolis, H. Selfishness, Altruism, and Rationality. New York: Cambridge University Press,


McCarthy, T. R. and D. I. Kass. "The Effect of Certificate of Need Regulation on

Investor-Owned Hospital Market Share." Paper presented at the Allied Social Science

Association, San Francisco, CA, December 1983.

Morrisey, M. A., F. A. Sloan, and S. A. Mitchell. "State Rate Setting: An Analysis of Some

Unresolved Issues." Health Affairs, Summer 1983, 36-47.

Mullner, R., D. Matthews, C. Byre, and H. Kubal. "Funding Aspects of Construction in

U.S. Hospitals (1973-1979)." Hospital Financial Management, November 1982, 30-4.

Mullner, R., D. Matthews, J. D. Kubal, and S. Andre. "Debt Financing: An Alternative for

Hospital Construction Funding." Health Care Financial Management, April 1983, 18-20,


Newhouse, J. P. "Toward a Theory of Nonprofit Institutions: An Economic Model of a

Hospital." American Economic Review, March 1970, 64-74.

Pauly, M. V. and M. Redisch. "The Not-for-Profit Hospital as a Physicians' Cooperative."

American Economic Review, March 1973, 87-99.

Reese, W. S. "Charitable Contributions: New Evidence on Household Behavior." American

Economic Review, March 1979, 142-51.

Roberts, R. D. "A Positive Model of Private Charity and Public Transfers." Journal of Political

Economy 92(1), 1984, 136-48.

Rose-Ackerman, S. "Ideals versus Dollars: Donors, Charity, Managers, and Government

Grants." Journal of Political Economy, August 1987, 810-23.

___. "Charitable Giving and Excessive Fundraising." The Quarterly Journal of Economics,

May 1982, 193-212.

___. "Do Government Grants to Charity Reduce Private Donations?" in Nonprofit Firms

in a Three-Sector Economy, edited by M. White. Washington, D.C.: The Urban Institute,

1981, 95-114.

Rosenberg, C. E. The Care of Strangers: The Rise of America's Hospital System. New York:

Basic Books, 1987.

Sloan, F. A. "The Internal Organization of Hospitals: A Descriptive Study." Health Services

Research, Fall 1980, 203-30.

___. "Regulation and the Rising Cost of Hospital Care." Review of Economics and Statistics,

November 1981, 478-87.

Sloan, F. A., J. Valvona, and R. Mullner, "Identifying the Issues: A Statistical Profile," in

Uncompensated Hospital Care: Rights and Responsibilities, edited by F. A. Sloan, J. F.

Blumstein, and J. M. Perrin. Baltimore: Johns Hopkins University Press, 1986, 16-53.

Sloan, F. A., M. A. Morrisey, and J. Valvona. "Capital Markets and the Growth of

Multihospital Systems," in Advances in Health Economics and Health Services Research,

edited by R. Scheffer and L. Rossiter. Greenwich: JAI Press, 1987, 83-109.

Starr, P. The Social Transformation of American Medicine. New York: Basic Books, 1982.

Terenzeo, J. V. "A Survey of the History and Current Outlook of Philanthropy as a Source

of Capital for the Needs of the Health Care Field," in Health Care Capital: Competition

and Control, edited by G.K. MacLeod and M. Perlman. Cambridge: Ballinger Publishing

Co., 1978, 239-58.
COPYRIGHT 1990 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1990 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Sloan, Frank A.; Hoerger, Thomas J.; Morrisey, Michael A.; Hassan, Mahud
Publication:Economic Inquiry
Date:Oct 1, 1990
Previous Article:Intertemporal substitution and constraints on labor supply: evidence from panel data.
Next Article:A contribution to the theory of depletable resource scarcity and its measures.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters