Compensation to managers of charitable organizations: an empirical study of the role of...
I. INTRODUCTION
This paper investigates empirical associations between accounting-based performance measures and compensation paid to chief executive officers (CEOs) of charitable organizations. We consider three research questions. The first is whether changes in executive compensation
Unique characteristics of charitable organizations suggest that prior evidence from the for-profit sector about the relation between executive compensation and accounting-based performance measures may not apply to charities. Donors to charities are typically not beneficiaries, and therefore, donors have less incentive to monitor the disposition of resources than do investors in private enterprise (Fama and Jensen 1983a, 1983b). Market-based performance measures, which discipline managers in for-profit firms, are absent since capital is donated and services or products are not "priced" in the conventional sense (Williamson 1983). These features of charities expose them to malfeasance by management, which may be attenuated by using accounting measures to ensure stewardship of contributed resources (Gjesdal 1981; Smith and Watts 1992; Sloan 1993).
On the other hand, philanthropic objectives are often subjective and nonfinancial, and returns to both donors and managers are more likely to be nonpecuniary, even altruistic, than returns to participants in other kinds of organizations (Andreoni 1989; Rose-Ackerman 1982; 1996; Handy and Katz 1998). In addition, financial statements do not reveal whether managers direct program expenditures toward meaningful projects, or whether spending is efficient or effective with respect to specific objectives. Moreover, charities' objectives are not clear-cut and their outputs are difficult to measure. Using accounting information for setting compensation is dysfunctional when contracting parties focus on the subset of tasks where performance can be measured objectively and de-emphasize other tasks where evaluations are subjective (Holmstrom and Milgrom 1991).
The potential for malfeasance suggests charities need external monitoring to reduce the agency costs of obtaining and distributing contributions (Akerlof 1970; Stigler 1971; Olsen 1971; Williamson 1983). Federal law requires charities to file financial information annually with the U.S. Internal Revenue Service (IRS). In addition, many charities voluntarily submit financial statements to private oversight agencies that report whether organizations comply with standards. These agencies, along with the popular press and individual donors, frequently cite the ratio of program expense to revenue (or the ratio of program expense to total expense) as a performance indicator for charities (e.g., Gold 1993; Hardman 1993; Schuman 1993; Smith 1993). (1) Evidence in Weisbrod and Dominguez (1986), Harvey and McCrohan (1988), and Khumawala and Gordon (1997) indicates that donation levels increase with program spending ratios. To investigate the relative importance of the program spending ratio, we distinguish changes in program spending that are attributable to changes in program spending ratios from those attributable to changes in the level of revenue that the charity raises. If charities benefit more from improving the program spending ratio than from increasing the level of revenue, then we expect compensation to be more closely associated with changes in spending ratios.
We use financial data from IRS Form 990 to obtain a sample of 331 charities. We find that changes in CEO compensation vary directly with both current- and prior-year changes in program spending, whether those changes are attributable to program spending ratios or to revenue raised. Moreover, we cannot convincingly reject the null hypothesis that charities reward increases in program spending that result from higher program spending ratios equally to those resulting from raising more revenue.
Next we focus on charities that are most likely to be concerned about program spending ratios reported to potential donors. We distinguish charities with relatively low program spending ratios and charities monitored by large, nationally recognized oversight agencies. Again, we cannot persuasively reject the null hypothesis that charities in these subsamples weigh changes in the program spending ratio equally with changes in revenue raised.
Evidence that compensation paid to executives of charitable organizations is associated with accounting measures parallels evidence documented for commercial firms (Rosen 1992). Specifically, we find that charities reward executives for increasing resources allocated to the charitable objective. We find no evidence that the use of program spending ratios by outsiders distorts the charities' incentives.
Section II develops empirical specifications and hypotheses. Sample selection procedures appear in Section III, and results are reported in Section IV. Concluding remarks are in Section V.
II. EMPIRICAL SPECIFICATIONS AND HYPOTHESES
We characterize charities as organizations that broker philanthropic resources from donors to beneficiaries. The brokering process consumes some fraction of the contributed capital, but the objective is to maximize spending on program activities. This characterization implies that management (1) raises funds, and (2) distributes the net of revenues less the costs of brokering contributions to program activities.
We consider measures of managers' performance on these two dimensions. Let period t program spending [PSPENDING.sub.t] = [REVENUE.sub.t] x [RATIO.sub.t], where [REVENUE.sub.t] is period t total revenue and [RATIO.sub.t] is the period t ratio of program expense to total revenue. Period t change in program spending is:
(1) [DELTA][PSPENDING.sub.t] = [[RATIO.sub.t] x [REVENUE.sub.t]] -
[[RATIO.sub.t-1] x [REVENUE.sub.t-1]]
= [[RATIO.sub.t-1] x [DELTA]
[REVENUE.sub.t]] + [[DELTA]
[RATIO.sub.t] x [REVENUE.sub.t]].
We interpret [[RATIO.sub.t-1] x [DELTA][REVENUE.sub.t]] in Equation (1) as the change in period t program spending that is explained by the change in the level of funds the charity raises. We interpret the product [[DELTA][RATIO.sub.t] x [REVENUE.sub.t]] as the change in period t program spending that is explained by the change in the average fraction of each revenue dollar the charity spends on program activities. (2)
To control for scale effects, we deflate Equation (1) by period t - 1 program spending. Thus, the empirical measures indicate percent changes, not absolute changes, in program spending. Deflating the first term in Equation (2) by [PSPENDING.sub.t-1] and then substituting [RATIO.sub.t-1] = [PSPENDING.sub.t-1]/[REVENUE.sub.t-1] yields [[DELTA][REVENUE.sub.t]/[REVENUE.sub.t-1]] = %[DELTA][REVENUE.sub.t]. That is, the period t percent change in program spending attributable to changes in available funds is simply the period t percent change in revenue. Deflating the second term in Equation (1) gives [DELTA][YIELD.sub.t] = [[DELTA][RATIO.sub.t] x [REVENUE.sub.t]]/[PSPENDING.sub.t-1], which indicates the extent to which the period t percent change in program spending is attributable to changes in the program spending ratio.
To summarize, we decompose the percent change in program spending as:
(2) [DELTA][PSPENDING.sub.t]/[PSPENDING.sub.t-1] = %[DELTA][REVENUE.sub.t] + [DELTA][YIELD.sub.t],
where the percent change in revenue (%[DELTA][REVENUE.sub.t]) indicates changes in fundraising performance, and [DELTA]YIELD indicates changes in the average cost of brokering contributed capital from contributors to beneficiaries. (3)
Next we posit that executive compensation, either explicitly or implicitly, depends in part on changes in program spending:
(3) %[DELTA][COMP.sub.t], = [[beta].sub.0] + [[beta].sub.1] %[DELTA][REVENUE.sub.t] + [[beta].sub.2] [DELTA][YIELD.sub.t] + [[epsilon].sub.t],
where %[DELTA][COMP.sub.t] is the period t - 1 to period t percent change in compensation paid to the: CEO. (4)
We expect compensation changes to be positively associated with changes in program spending, whether those changes are attributable to changes in revenue raised (%[DELTA]REVENUE) or to changes in the program spending ratio ([DELTA]YIELD). Thus, we anticipate both [[beta].sub.1] > 0 and [[beta].sub.2] > 0. In addition, widespread use of the program spending ratio by outsiders suggests that charities may reward increases in the program spending ratio ([DELTA]YIELD) more than they reward increases in revenue (%[DELTA]REVENUE). Thus, we predict [[beta].sub.1], < [[beta].sub.2].
Next we investigate cross-sectional differences in sensitivity to changes in the program spending ratio. Charities are more likely to be concerned about the cost of brokering contributions when the program spending ratio is relatively low than when the ratio is high. We therefore partition the sample according to whether the ratio is above or below the median value. We expect charities' compensation to be relatively more sensitive to [DELTA]YIELD (and relatively less sensitive to %[DELTA]REVENUE) when the program spending ratio is below the median value.
We also consider whether monitoring by oversight agencies--in contrast with monitoring by donors in general--affects CEO compensation. To do so, we partition the sample according to whether the NCIB or the PAS monitor the charity. If concern about oversight groups' use of the program spending ratio influences compensation, then we expect compensation to be relatively more sensitive to [DELTA]YIELD for charities that are monitored, than for charities that are not. (5)
Finally, if charities reward current-period performance by providing salary increases, rather than one-time bonuses, then current compensation changes are correlated with prior (not current) performance. We therefore estimate specifications that include both current and lagged performance measures:
(4) %[DELTA][COMP.sub.t] = [[gamma].sub.0] + [[gamma].sub.1]% [DELTA][REVENUE.sub.t] + [[gamma].sub.2]% [DELTA][REVENUE.sub.t-1] + [[gamma].sub.3] [DELTA][YIELD.sub.t] + [[gamma].sub.4] [DELTA][YIELD.sub.t-1] + [[epsilon].sub.t],
The foregoing reasoning predicts [[gamma].sub.j] > 0, j = {1,...,4}, and ([[gamma].sub.1] + [[gamma].sub.2]) < ([[gamma].sub.2] + [[gamma].sub.4]).
III. SAMPLE AND DATA
Table 1 provides the sample selection procedures. We identify 458 charities evaluated by the NCIB or the PAS during 1996 and 1997. We then randomly select 200 additional charities monitored by neither oversight agency from files maintained at the Charitable Organizations Division, Office of the Maryland Secretary of State. Data for 541 charities are from IRS Form 990 filings obtained primarily from the charities directly, although some filings are from the Maryland Secretary of State. We exclude 210 charities with missing or unusable information, as detailed in Table 1. We use compensation changes rather than levels, and therefore, we lose the first year of data for each organization. We also eliminate charities whose chief executives change in the current or the prior year (thus, CEO changes do not explain compensation changes). Finally, we eliminate potentially influential observations specified as those with absolute standardized residuals greater than 2.0 from a regression of year t percent change in compensation on year t percent change in program spending.
Panel B of Table 1 shows that these sample selection criteria yield 664 observations from 331 charities for the primary analysis. Panel B also shows that 484 observations are from 233 charities monitored by the PAS and/or the NCIB, and that 180 are from 98 charities that are not monitored.
Table 2 displays descriptive information about the sample organizations. Entries are means of mean values for each organization, computed using sums for all years that the organization appears in the sample. The first entry is for all organizations, the second entry is for 233 organizations monitored by the NCIB or the PAS, and the third entry is for the 98 sample charities that are not monitored. Asterisks indicate that monitored organizations differ significantly from nonmonitored organizations on that characteristic.
Compensation profiles appear in Panel A of Table 2. If Form 990 does not designate an employee as the chief executive officer (or a comparable title), then we use compensation tot the highest-paid employee. We specify compensation as the sum of current and deferred salary and benefits (Form 990, Part V, columns C and D). Including expense accounts as compensation yields comparable results. Compensation is higher for monitored charities than for charities that are not monitored, but compensation as a percent of revenue is lower.
Panel B of Table 2 displays distributions for other characteristics of the sample, including explanatory variables in the regression analysis. Total revenue includes direct and indirect contributions, government contributions, program service revenue, membership dues, investment revenue (including gains/losses on investment transactions), and both related and unrelated business income. Monitored charities have higher revenues, on average, than charities that are not monitored. Mean (median) percent program expenses to total revenues (RATIO) for all charity-year observations is 72.63 percent (74.59 percent). Thus, about 75 cents of each dollar contributed to our sample charities goes to the charitable objective. Mean (median) percent contributions to total revenue is 71.41 percent (81.13 percent) for monitored charities and 44.02 percent (40.37 percent) for charities that are not monitored. Thus, monitored organizations rely more heavily on contributions, as opposed to other revenue sources. Differences in the percent change in program spending (%[DELTA]PSPENDING) and in the measures that result from partitioning percent change of program spending (%[DELTA]REVENUE and [DELTA]YIELD) are not statistically significant, however.
Finally, analysis not tabulated indicates that 79 charity-year observations are from environmental/animal-related charities; 165 are from medical research, disease/disorder, or health-related charities; 66 serve the international community; 100 are from social service charities; 72 are from arts or educational charities; 24 are from civil rights organizations; and 158 are from organizations that pursue other objectives.
IV. RESULTS
Table 3 reports the results of estimating Equation (3). Results for the entire sample, displayed in Column A, indicate that percent changes in CEO compensation and percent changes in program spending vary directly. (6) For example, increasing revenues by 10 percent is associated with a 0.90 percent compensation increase, or about 13.2 percent of the mean compensation increase displayed in Table 2. Similarly, increasing program spending by 10 percent by reducing the average cost of acquiring and administering funds is associated with a 0.76 percent compensation increase, or about 11.2 percent of the mean increase. In contrast, the average increase in compensation not related to program spending is 5.8 percent (intercept [[beta].sub.0] = 0.058). Prior studies of compensation paid to executives of publicly traded, for-profit firms indicate similarly modest relations between compensation changes and accounting performance measures (Rosen 1992).
For the sample as a whole, we cannot reject the null hypothesis [[beta].sub.1] = [[beta].sub.2]. This result suggests that there is no significant difference in the extent to which charities reward increases in program spending attributable to revenue changes vs. increases in program spending attributable to changes in program spending ratios.
Columns B and C report results of estimating the regressions after partitioning the data according to whether the program spending ratio is above or below the median ratio. Between-sample statistical comparisons, obtained from a specification that uses dummy variables to allow the parameters [[beta].sub.i] to differ between subsamples, appear in Column D. Results for subsamples partitioned according to whether the NCIB or the PAS monitors the charity, and the corresponding comparisons of parameter estimates, appear in Columns E, F, and G. Chow tests of whether the overall structure of the specification differs between subsamples appear at the bottom of the table.
Comparisons in Columns B and C of results for charity-years with relatively high program spending ratios (RATIO > median) with results for those with relatively low program spending ratios (RATIO < median) support the premise that charities weigh increases in program spending that result from reducing average costs of brokering resources more heavily when prevailing program spending ratios are low. In particular, we reject the null hypothesis [[beta].sub.1] = [[beta].sub.2] in favor of [[beta].sub.1] > [[beta].sub.2] for observations with relatively high program spending ratios (Column B), but not for observations with low ratios (Column C). Moreover, the parameter estimate [[beta].sub.2] on [DELTA]YIELD is statistically significant for the low, but not for the high, ratio sample. Finally, the Chow test indicates statistically significant between-sample differences in the overall structure of the specification (p = 0.05). In contrast, comparisons in Columns E, F, and G indicate that relations do not differ significantly between charities that large oversight groups do or do not monitor.
Results for Equation (4), which include performance measures for both the current and the prior years, appear in Table 4. (7) These results consistently indicate significantly positive associations between current-period compensation changes and both current- and prior-year changes in program spending; however, associations are weaker for lagged than for contemporaneous relations. This investigation of relations between accounting-based performance measures for the prior year and current year compensation distinguishes the analysis from approaches used in studies of compensation paid to CEOs of publicly traded firms that typically restrict the analysis to contemporaneous relations between compensation and accounting performance measures (Gaver and Gaver 1998; Baber et al. 1999). This feature of the study is inspired by the possibility that charities may be more likely than commercial firms to use salary increases, rather than one-time bonuses, to reward accounting performance (Haber 1995).
In contrast with the result in Table 3, we cannot reject ([[gamma].sub.1] + [[gamma].sub.2]) = ([[gamma].sub.3] [[gamma].sub.4]) for the two partitions based on the program spending ratio. This suggests that executive compensation does not significantly differ between program spending changes resulting from revenue changes vs. those from changes in program spending ratios. Finally, differences in relations between subsamples (Columns B vs. C, and E vs. F) generally are not statistically significant at conventional levels. The exception is that the sensitivity of CEO compensation to the percent change in revenue in the prior year (%[DELTA][REVENUE.sub.t-1]) is greater for charities not monitored, than for charities monitored, by large oversight organizations. In sum, the results in Table 4, where we consider performance measures in the prior year, generally reinforce the Table 3 results that compensation changes vary directly with changes in program spending. They do not support a conclusion that compensation changes are differentially sensitive to revenue changes vis-a-vis changes in the program spending ratio.
We conduct additional tests to evaluate whether the primary results are robust to competing interpretations. In particular, we consider whether results can be attributed to the availability of revenue in excess of expenses. Donors can interpret undistributed surplus as evidence that the organization is overfunded, which likely discourages contributions. On the other hand, donors are likely reluctant to contribute to financially distressed charities. Thus, except perhaps for reasonable allocation for contingencies, managers of charities have incentives to report a "breakeven condition." If so, then discretionary spending--salary increases, in particular--are more likely when the residual of revenues less essential program spending is positive than when residual resources are at or less than breakeven. We investigate this possibility by including the period t ratio of total revenue to total expenses as a control variable. Parameter estimates for this ratio typically are positive and statistically significant, but relations for other variables are comparable to those reported in Tables 3 and 4.
The FASB issued standards during our sample period that could change the accounting methods some charities use to report revenue. (8) Adopting new accounting standards for revenue recognition potentially confounds empirical relations. Eliminating from the sample 25 charity-year observations for which the potential effects of accounting changes exceed 5 percent of the current-year beginning account balances yields comparable results. (9)
V. CONCLUDING REMARKS
We investigate the relations between changes in the compensation charities pay their CEOs and changes in accounting performance measured as amounts directed toward the charity's main purpose--program activities. We provide empirical evidence on competing expectations about the role of accounting performance indicators in charities' compensation practices. Because outputs of charitable enterprises are typically not priced in the conventional sense, market-based performance measures that discipline managers of commercial enterprise are absent. Accounting indicators may therefore help charities monitor managers' performance. On the other hand, charitable organizations typically pursue nonfinancial objectives where success is difficult to quantify, so emphasis on accounting performance measures may prove dysfunctional.
Our results support the relevance of accounting-based performance measures as a basis for contracting with executives of charitable organizations. We document relations between changes in CEO compensation and changes in program spending, whether changes in program spending are attributable to changes in revenue raised or to changes in the amount of each contributed dollar that is allocated to the program activity.
We also investigate whether the use of program spending ratios to monitor charities affects relations between compensation and accounting measures. We distinguish changes in program spending attributable to changes in revenue raised from those attributable to changes in program spending ratios. We find no consistent evidence that relations between compensation and these two components of program spending differ, even for charities where concern about the program spending ratio is likely to be high--charities for which the prevailing program spending ratio is relatively low, and those monitored by large, well-known charity oversight groups that use program spending ratios to assess organization effectiveness. Such results suggest that outsiders' use of specific accounting measures to monitor performance does not cause differential weighting of changes in program spending that result from changes in program spending ratios.
As with studies of executive compensation in general, we cannot observe whether charities explicitly use accounting measures for setting executive compensation. Even so, we document relations that are consistent with the premise that charitable organizations reward managers for actions that increase funds directed toward the charitable objective.
TABLE 1
Selection of a Sample of Charities with Available Compensation
and Financial Data
Panel A: Procedures Used to Identify Sample Organizations
Total requests (a) 658
Organizations not required to file 990 information (b) (57)
No response (c) (60)
Information received or obtained 541
Form 990 information unusable (d) (57)
Chief executive changes during the sample period (e) (69)
Officer compensation is not disclosed (f) (84)
Total organizations in the sample 331
Panel B: Profiles of Charity-Year Observations
Total Monitored (a)
Years of
Data Number of Number of Number of Number of
Available Charities Observations Charities Observations
One 77 77 57 57
Two 190 380 116 232
Three 49 147 45 135
Four 15 60 15 60
Total 331 664 233 484
Not Monitored (a)
Years of
Data Number of Number of
Available Charities Observations
One 20 20
Two 74 148
Three 4 12
Four 0 0
Total 98 180
(a) All organizations listed in the February 1996 and June 1997
editions of the National Charities Information Bureau's (NCIB)
Charities Index and the 1996 edition of the Annual Charity Index
(CBBB-PAS 1996), plus 200 randomly selected organizations not in
these publications, but registered in 1996 with the Charitable
Organization Division, Office of the Maryland Secretary of State.
Organizations from the indices are designated "Monitored";
organizations from the Maryland Secretary of State are designated
"Not Monitored."
(b) Includes subsidiaries of other organizations, and organizations
not required to file Form 990 (churches or organizations with total
receipts less than $25,000).
(c) Organizations (1) that fail to respond to two mail requests;
(2) for which requests are returned with no forwarding address;
(3) that respond either by stating that they do not provide Form 990s
or by omitting or obscuring compensation information; and (4) for
which usable information is not available at the Maryland Office of the
Secretary of State.
(d) Includes organizations (1) for which Form 990 contains obvious
errors; (2) that do not report either program, fundraising, or
administrative expenses; (3) for which consecutive fiscal years are
not available; and (4) eliminated in tests for influential
observations.
(e) Includes organizations for which the chief executive moves from
part-time to full-time.
(f) Either Form 990 filing does not include compensation information,
or no employee receives more than $50,000.
TABLE 2
Descriptive Statistics for CEO Compensation, Revenue, and Program
Spending Ratios in Charitable Organizations (a)
(entries are means of mean values computed for each organization
using all years the organization is represented in the sample)
Panel A: Total CEO Compensation (cash payments and benefits)
First
Mean Quartile Median
Total CEO compensation (in $1,000)
Total sample 130.77 * 75.69 113.06
Monitored charities 140.66 85.42 127.50
Charities not monitored 107.24 56.31 83.74
Total CEO compensation as a
percent of revenue
Total sample 3.48 * 0.07 1.79
Monitored charities 2.57 0.05 1.24
Charities not monitored 5.64 1.74 3.24
Change in total CEO compensation
(in $1,000)
Total sample 7.14 1.23 5.35
Monitored charities 6.80 1.71 6.68
Charities not monitored 7.96 1.05 4.15
%[DELTA]CCOMP percent change in total
CEO compensation
Total sample 6.84 0.02 5.65
Monitored charities 6.39 1.38 5.76
Charities not monitored 7.90 1.02 5.61
Panel B: Descriptive Information on Revenues and Program Spending Data
First
Mean Quartile Median
Total revenue (in $millions)
Total sample 23.05 * 2.10 7.22
Monitored charities 29.09 3.95 9.76
Charities not monitored 8.70 0.91 2.24
Program spending ratio, percent
program expenses to total revenue
(RATIO)
Total sample 72.63 64.46 74.59
Monitored charities 73.41 65.93 75.20
Charities not monitored 70.78 63.72 73.97
Percent contributions to total
revenue
Total sample 63.30 * 38.22 72.91
Monitored charities 71.41 55.33 81.13
Charities not monitored 44.02 15.18 40.37
Annual percent change in program
spending (%[DELTA]PSPENDING)
Total sample 11.69 0.72 7.69
Monitored charities 10.41 -0.72 6.00
Charities not monitored 14.73 4.20 9.81
Annual percent change in total
revenue (%[DELTA]REVENUE)
Total sample 13.87 1.04 8.63
Monitored charities 12.46 -0.35 7.70
Charities not monitored 17.22 3.34 10.90
Annual percent change in program
spending attributed to change in
the program spending ratio
([DELTA]YIELD)
Total sample -2.18 -7.82 -0.40
Monitored charities -2.05 -7.34 -0.61
Charities not monitored -2.49 -16.17 0.03
Panel A: Total CEO Compensation (cash payments and benefits)
Third Standard
Quartile Deviation
Total CEO compensation (in $1,000)
Total sample 176.62 77.48
Monitored charities 186.96 74.64
Charities not monitored 126.69 79.40
Total CEO compensation as a
percent of revenue
Total sample 4.03 4.90
Monitored charities 2.96 4.01
Charities not monitored 7.81 5.94
Change in total CEO compensation
(in $1,000)
Total sample 10.84 19.01
Monitored charities 12.63 18.86
Charities not monitored 8.84 19.40
%[DELTA]CCOMP percent change in total
CEO compensation
Total sample 10.51 13.26
Monitored charities 10.10 14.06
Charities not monitored 11.46 11.11
Panel B: Descriptive Information on Revenues and Program Spending Data
Third Standard
Quartile Deviation
Total revenue (in $millions)
Total sample 23.07 44.92
Monitored charities 31.16 50.93
Charities not monitored 8.14 19.20
Program spending ratio, percent
program expenses to total revenue
(RATIO)
Total sample 82.12 16.82
Monitored charities 82.63 16.82
Charities not monitored 81.59 16.77
Percent contributions to total
revenue
Total sample 91.92 31.39
Monitored charities 93.64 27.51
Charities not monitored 70.92 31.75
Annual percent change in program
spending (%[DELTA]PSPENDING)
Total sample 17.34 24.15
Monitored charities 15.23 25.99
Charities not monitored 18.26 18.83
Annual percent change in total
revenue (%[DELTA]REVENUE)
Total sample 20.91 26.31
Monitored charities 19.07 25.61
Charities not monitored 28.23 27.75
Annual percent change in program
spending attributed to change in
the program spending ratio
([DELTA]YIELD)
Total sample 5.37 19.66
Monitored charities 5.27 17.22
Charities not monitored 5.68 24.60
* Differences in means between charities monitored and not monitored
by the NCIB or PAS are statistically significant (one-tailed p < 0.05).
(a) The full sample has 331 charities, 233 are monitored and 98 are not
monitored by the National Charities Information Bureau (NCIB) or the
Council of Better Business Bureaus, Philanthropic Advisory Services
(PAS) during 1996.
TABLE 3
Regressions of Percent Change in Compensation (%[DELTA]COMP) on Percent
Change in Program Spending Attributable to:
(1) Percent Change in Revenue (%[DELTA]REVENUE), and
(2) Change in the Program Spending Ratio ([DELTA]YIELD)
(full sample and partitioned by the program spending ratio (RATIO) and
whether the charity is monitored by NCIB or PAS)
%[DELTA][COMP.sub.t] = [[beta].sub.0] + [[beta].sub.1]
%[DELTA][REVENUE.sub.t] + [[beta].sub.2] [DELTA][YIELD.sub.t]
+ [[epsilon].sub.t]
Sample Partitioned by Program
Spending Ratio
(A) (B) (C) (D)
All RATIO RATIO
Obser- >Median <Median Diffe-
Sample vations (a) (a) rence (c)
Intercept 0.058 0.058 0.052 0.006
[[beta].sub.0] (9.17) (6.29) (5.91) (0.47)
%[DELTA]REVENUE 0.090 0.109 0.072 0.037
[[beta].sub.1] (4.69) (3.48) (2.95) (0.93)
[DELTA]YIELD 0.076 0.035 0.107 -0.072
[[beta].sub.2] (3.38) (0.88) (3.80) (-1.48)
Adjusted [R.sup.2] 0.031 0.035 0.041 NA
Test: [[beta].sub.1] = t = 0.07 t = 2.24 t = 1.26 NA
[[beta].sub.2]
Sample size 664 332 332 664
Chow test (c) NA F = 2.61
(d.f. = 3,658) (p = 0.05)
Sample Partitioned by whether
Charity Is Monitored by
NCIB or PAS
(E) (F) (G)
Charities
Monitored Not
Sample Charities (b) Monitored (b) Difference (c)
Intercept 0.053 0.072 -0.020
[[beta].sub.0] (7.16) (5.81) (-1.37)
%[DELTA]REVENUE 0.106 0.058 0.048
[[beta].sub.1] (4.34) (1.77) (1.19)
[DELTA]YIELD 0.086 0.052 0.034
[[beta].sub.2] (2.82) (1.44) (0.72)
Adjusted [R.sup.2] 0.014 0.007 NA
Test: [[beta].sub.1] =
[[beta].sub.2] t = 0.66 t = 0.21 NA
Sample size 484 180 664
Chow test (c) F = 0.79
(d.f. = 3,658) (p = 0.50)
Entries are parameter estimates and t-statistics in parentheses.
(a) The third and fourth columns indicate regression parameter
estimates for the sample partitioned according to whether the program
spending ratio (RATIO) in the prior year exceeds the median for all
664 observations.
(b) The sixth and seventh columns indicate regression parameter
estimates for the sample partitioned according to whether either or
both the National Charities Information Bureau (NCIB) or the Council
of Better Business Bureaus, Philanthropic Advisory Services (PAS)
monitor the charity during 1996.
(c) Indicates results for between-sample statistical comparisons of
estimates. Entries for parameter estimates are differences in
parameters and corresponding t-statistics for tests of between-sample
differences. Results for Chow tests of between-sample differences in
overall statistical relations appear in the last row of the table.
Variable definitions:
%[DELTA][COMP.sub.t] = year t - 1 to year t percent change in
compensation paid to the chief executive
officer. Compensation is the sum of
current and deferred salary and benefits,
displayed on Form 990, Part V, columns C
and D, respectively;
%[DELTA][REVENUE.sub.t] = year t - 1 to year t percent change in
total revenue, Total revenue is the
amount on Form 990, Part I, line 12.
Contributions are the sum of lines 1a
and 1b, direct and indirect
contributions. Indirect contributions
are received from a central fundraising
organization such as the United Way; and
[DELTA][YIELD.sub.t] = the product of year t - 1 to year t percent
change in the ratio of program expense to
total revenue and year t total revenue,
deflated by year t - 1 program expense.
Program expenses are the sum of amounts
reported as program services on
Form 990, Part II, column B of line 44
and as payments to affiliates on Part I,
line 16.
TABLE 4
Regressions of Current Period Percent Change in Compensation
(%[DELTA]COMP) on Current and Prior Year Percent Change in Program
Spending Attributable to Percent Change in Revenue ([DELTA]AREVENUE)
and Current and Prior Year Change in the Program Spending Ratio
([DELTA]YIELD)
%[DELTA][COMP.sub.t] = [[gamma].sub.0] + [[gamma].sub.1]
%[DELTA][REVENUE.sub.t] + [[gamma].sub.2] %[DELTA][REVENUE.sub.t-1]
+ [[gamma].sub.3] [DELTA][YIELD.sub.t] + [[gamma].sub.4]
[DELTA][YIELD.sub.t-1] + [[epsilon].sub.t]
Sample Partitioned by Program
Spending Ratio
(A) (B) (C) (D)
All RATIO RATIO Diffe-
Obser- >Median <Median rence
Sample vations (a) (a) (c)
Intercept 0.034 0.040 0.027 0.013
[[gamma].sub.0] (4.67) (4.48) (2.19) (0.87)
%[DELTA][REVENUE.sub.t] 0.128 0.088 0.151 -0.064
[[gamma].sub.1] (5.09) (2.76) (3.82) (-1.21)
%[DELTA][REVENUE.sub.t-1] 0.069 0.069 0.074 -0.005
[[gamma].sub.2] (3.72) (2.23) (2.93) (-0.11)
[DELTA][YIELD.sub.t] 0.167 0.095 0.191 -0.100
[[gamma].sub.3] (6.27) (1.98) (5.36) (-1.43)
[DELTA][YIELD.sub.t-1] 0.085 0.065 0.085 -0.021
[[gamma].sub.4] (3.90) (1.62) (2.86) (-0.36)
Adjusted [R.sup.2] 0.137 0.056 0.167 NA
Test: ([[gamma].sub.1] + t = 1.44 t = 0.04 t = 0.95 NA
[[gamma].sub.2])
= ([[gamma].sub.3] +
[[gamma].sub.4])
Sample size 336 168 168 336
Chow test NA F = 0.54
(d.f. = 5,327) (p = 0.75)
Sample Partitioned by whether
Charity Is Monitored by
NCIB or PAS
(E) (F) (G)
Charities
Monitored Not Difference
Sample Charities (b) Monitored (b) (c)
Intercept 0.035 0.029 0.007
[[gamma].sub.0] (4.38) (1.53) (0.34)
%[DELTA][REVENUE.sub.t] 0.139 0.138 0.001
[[gamma].sub.1] (5.00) (2.18) (0.02)
%[DELTA][REVENUE.sub.t-1] 0.033 0.126 -0.094
[[gamma].sub.2] (1.31) (3.64) (-2.31)
[DELTA][YIELD.sub.t] 0.157 0.238 -0.081
[[gamma].sub.3] (5.51) (3.23) (-1.10)
[DELTA][YIELD.sub.t-1] 0.071 0.144 -0.072
[[gamma].sub.4] (2.43) (3.58) (-1.53)
Adjusted [R.sup.2] 0.140 0.143 NA
Test: ([[gamma].sub.1] + t = 1.15 t = 1.69 NA
[[gamma].sub.2])
= ([[gamma].sub.3] +
[[gamma].sub.4])
Sample size 251 85 336
Chow test F = 1.19
(d.f. = 5,327) (p = 0.22)
Entries are parameter estimates and t-statistics in parentheses.
(a) The third and fourth columns indicate regression parameter
estimates for the sample partitioned according to whether the program
spending ratio (RATIO) in the prior year exceeds the median for all
664 observations.
(b) The sixth and seventh columns indicate regression parameter
estimates for the sample partitioned according to whether either or
both the National Charities Information Bureau (NCIB) or the Council
of Better Business Bureaus, Philanthropic Advisory Services (PAS)
monitor the charity during 1996.
(c) Indicates results for between-sample statistical comparisons of
estimates. Entries for parameter estimates are differences in
parameters and corresponding t-statistics for tests of between-sample
differences. Results for Chow tests of between-sample differences in
overall statistical relations appear in the last row of the table.
Variables are defined in Table 3.
The authors are grateful to Holeri Faruolo of the National Charities Information Bureau, Bennett Weiner and Mark Blankenship of the Council of Better Business Bureaus, Philanthropic Advisory Services, and Frank Kobilis and Jennifer Light of the Charitable Organizations Division, Office of the Maryland Secretary of State for their time, assistance, and cooperation. Comments by Ashiq Ali, Rajiv Banker, Russ Barefield, Jeff Callen, Jim Canning (World Vision), Dan Dhaliwal, Randal Elder, Chris Jones, Sok-Hyon Kang, Krishna Kumar, Rev. Joel MacCollam (World Emergency Relief), Jim Patton, Karen Taranto, Dan Tinkelman, Kumar Visvanathan, Jerry Zimmerman, and workshop participants at The George Washington University, The University of Arizona, the University of Georgia, Virginia Commonwealth University, and the 1998 American Accounting Association Annual Meeting are greatly appreciated.
Submitted November 1999 Accepted October 2001
(1) At the time when the research was conducted, The National Charities Information Bureau (NCIB) and the Philanthropic Advisory Service of the Council of Better Business Bureaus (PAS) were the two largest private oversight agencies (the NCIB and the PAS combined into one agency during 2001). NCIB standards are (1) at least 60 percent of annual expenditures be allocated to program services; (2) fundraising expenses be reasonable over time; and (3) net assets available for use are not more than twice the greater of next year's expenses or next year's budget (NCIB 1996). PAS standards are that (1) at least 50 percent of total revenue and at least 50 percent of public contributions be spent on program services; (2) fundraising expenditures are less than 35 percent of related contributions; and (3) combined fundraising and administrative costs are less than 50 percent of total revenue (CBBB-PAS 1982). Both the NCIB and the PAS use nonfinancial as well as accounting-based performance standards.
(2) This decomposition of program spending changes is analogous to the decomposition of profits into volume and margin effects described in managerial accounting texts. Following convention, we consider the joint effect of changes in the program spending ratio and changes in revenue ([DELTA]RATIO x [DELTA]REVENUE) as part of the ratio effect (e.g., Hilton 1997, 552-553).
(3) The ratio of program expense to revenue varies inversely with the "price" effect, which is the locus of prior studies of the relation between program spending ratios and fundraising success (Weisbrod and Dominguez 1986; Posnett and Sandler 1989; Callen 1994; Tinkelman 1999).
(4) We use compensation changes, rather than levels, to control for cross-sectional differences in factors (e.g., executive age, experience, gender, tenure with the organization, organization size) that potentially affect CEO compensation. Although CEOs of large organizations tend to earn more than CEOs of small organizations (Werner and Gemeinhardt 1995; Oster 1998), annual changes in compensation are unlikely to be attributable to changes in the organization size from one year to the next. That is, annual changes in size are not likely to warrant increasing (or decreasing) salaries to continuing CEOs. We omit observations in which the CEO changes during the period, so cross-sectional differences unique to the individual (such as age or gender) are unlikely to affect compensation changes.
(5) We obtain similar results: when we partition the sample into thirds, quartiles, or quintiles based on the program spending ratio; when we simultaneously classify the observations according to both the prevailing program spending ratio and whether the organization is monitored; when we use alternative financial performance indicators, such as fundraising as a percent of contributions, contributions as a percent of total revenue, or program spending as a percentage of total spending; and when we compute the program spending ratio as a percent of total expense, rather than total revenue.
(6) The contemporaneous relations between percent changes in compensation and percent changes in program spending (designated %[DELTA]PSPENDING) is:
%[DELTA]COMP = 0.0892 + 0.0749 (%[DELTA]PSPENDING).
Respective t-statistics for parameter estimates are 9.510 and 4.775, and the adjusted [R.sup.2] = 0.318. These results confirm statistically significant relations between compensation changes and changes in program spending when program spending is not decomposed as in Equation (2).
(7) We assess the relation between current-year percent change in compensation and current- and prior-year percent change in program spending by estimating:
%[DELTA][COMP.sub.t] = 0.0313 + 0.1452 (%[DELTA][PSPENDING.sub.t]) + 0.0706 (%[DELTA][PSPENDING.sub.t-1]).
Respective t-statistics for parameter estimates are 4.464, 6.480, and 4.297, and the adjusted [R.sup.2] = 0.1359. These results confirm statistically significant relations between current-year compensation changes and changes in prior-year program spending when program spending is not decomposed as in Equation (2).
(8) Potential consequences of SFAS No.116, issued in 1994 and effective for most organizations for years beginning after December 15, 1995 (after 1994 for large organizations), are particularly relevant because the standard focuses on the recognition of both pledges to contribute and restricted donations. SFAS No. 117, issued at the same time, addresses the form and content of financial statements and is less likely to compromise the data reported on Form 990. For details, see Anthony (1995), Chase (1995), Khumawala and Gordon (1997), or Granof (1998).
(9) We compare beginning balances reported on the current year's return with ending balances reported on the prior year's returns for current asset and liability accounts affected by revenue recognition practices. Specifically, we compare accounts receivable (Form 990, item No. 47), pledges receivable (item No. 48), grants receivable (item No. 49), and deferred revenue (item No. 62). We assume that differences between these amounts are attributable to changes in accounting methods used to recognize revenue.
REFERENCES
Akerlof, G. 1970. The market for "lemons": Quality uncertainty and the market mechanism. The Quarterly Journal of Economics 84 (August): 488-500.
Andreoni, J. 1989. Giving with impure altruism: Applications to charity and Ricardian equivalence. Journal of Political Economy 97 (6): 1447-1458.
Anthony, R. 1995. The nonprofit accounting mess. Accounting Horizons 9 (2): 44-53.
Baber, W., S. Kang, and K. Kumar. 1999. The explanatory power of earnings levels vs. earnings changes in the context of executive compensation. The Accounting Review 74 (4): 459-472.
Callen, J. 1994. Money donations, volunteering and organizational efficiency. The Journal of Productivity Analysis 5: 215-228.
Chase, B. 1995. Get ready, here they come. Journal of Accountancy 179 (1): 79-82.
Council of Better Business Bureaus, Philanthropic Advisory Services (CBBB-PAS). 1982. Standards for Charitable Solicitations. Publication No. 311-25129. Arlington, VA: CBBB-PAS.
--. 1996. Annual Charity Index. Edited by I. Alexander, and C. Ragonese. Arlington, VA: CBBB-PAS.
Fama, E., and M. Jensen. 1983a. Agency problems and residual claims. The Journal of Law and Economics 26 (2): 327-350.
--, and --. 1983b. Separation of ownership and control. The Journal of Law and Economics 26 (2): 301-326.
Gaver, J., and K. Gaver. 1998. The relation between nonrecurring accounting transactions and CEO cash compensation. The Accounting Review 73 (2): 235-253.
Gjesdal, F. 1981. Accounting for stewardship. Journal of Accounting Research 19 (1): 208-231.
Gold, J. 1993. Finding gems among the rhinestones. Financial World 162 (16): 46-58.
Granof, M. 1998. Government and Not-For-Profit Accounting. New York, NY: John Wiley & Sons.
Haber, J. 1995. Bonus: Not a dirty word. Nonprofit World 13 (2): 49-53.
Handy, F., and E. Katz. 1998. The wage differential between nonprofit institutions and corporations: Getting more by paying less? Journal of Comparative Economics 26 (2): 246-261.
Hardman, A. 1993. Whom can you trust? Financial World 162 (16): 66-67.
Harvey, J., and K. McCrohan. 1988. Fundraising costs--Societal implications for philanthropies and their supporters. Business and Society 27 (1): 15-22.
Hilton, R. 1997. Managerial Accounting. 3rd edition. New York, NY: McGraw-Hill.
Holmstrom, B., and P. Milgrom. 1991. Multitask principal-agent analyses: Incentive contracts, asset ownership, and job design. The Journal of Law, Economics, and Organizations 7: 24-52.
Khumawala, S., and T. Gordon. 1997. Bridging the credibility of GAAP: Individual donors and the new accounting standards for nonprofit organizations. Accounting Horizons 11 (3): 45-68.
National Charities Information Bureau (NCIB). 1996. Standards in Philanthropy. June. New York, NY: NCIB.
Olsen, M. 1971. The Logic of Collective Action: Public Goods and the Theory of Groups. Cambridge, MA: Harvard University Press.
Oster, S. 1998. Executive compensation in the nonprofit sector. Nonprofit Management and Leadership 8 (3): 207-221.
Posnett, J., and T. Sandler. 1989. Demand for charity donations in private non-profit markets: The case of the U.K. Journal of Public Economics 40: 187-200.
Rosen, S, 1992. Contracts and the market for executives. In Contract Economics, edited by L. Wernin, and H. Wijkander, 181-211. Oxford, U.K.: Basil Blackwell.
Rose-Ackerman, S. 1982. Charitable giving and "excessive" fundraising. Quarterly Journal of Economics 97 (2): 193-212.
--. 1996. Altruism, non-profits, and economic theory. Journal of Economic Literature 34 (2): 701-728.
Schuman, M. 1993. How to choose a charity. Forbes 152 (14): 228-229.
Sloan, R. 1993. Accounting earnings and top executive compensation. Journal of Accounting and Economics 16 (1-3): 55-100.
Smith, C., and R. Watts. 1992. The investment opportunity set and corporate financing, dividend, and compensation policies. Journal of Financial Economics 32 (3): 263-292.
Smith, M. 1993. Best charities in America. Money 22 (12): 128-132.
Stigler, G. 1971. The theory of economic regulation. Bell Journal of Economics and Management Science 2 (1): 3-21.
Tinkelman, D. 1999. Factors affecting the relation between donations to not-for-profit organizations and an efficiency ratio. Research in Governmental and Nonprofit Accounting 10: 135-161.
Weisbrod, B., and N. Dominguez. 1986. Demand for collective goods in private nonprofit markets: Can fundraising expenditures help overcome free-rider behavior? Journal of Public Economics 30: 83-95.
Werner, S., and G. Gemeinhardt. 1995. Nonprofit organizations: What factors determine pay levels? Compensation and Benefits Review 27 (5): 53-60.
Williamson, O. 1983. Organization form, residual claimants, and corporate control. The Journal of Law and Economics 26 (2): 351-366.
William R. Baber Patricia L. Daniel The George Washington University Andrea A. Roberts Boston College
In addition, make sure to read these articles:
- Executive pay and the disclosure environment: Canadian evidence.
- Board remuneration, company performance, and ownership concentration: evidence from...
- The relationship between foreign vs. domestic earnings and executive compensation.
- Delta reports USD182m loss for November.
- Delta Air Lines to lower aircraft turn time.
- Delta adds flights to Guadeloupe.
- Delta adds Atlanta flight.
- Delta announces first quarter loss.
- Delta Passing Facility Costs To Passengers
- Delta Expands Business Fare Discounting
- Aviation Daily: Air France To Choose Delta, Heating UpAtlantic Alliance Competition.
- Delta Delivers Loyalty Program For Small, Midsize Biz.
- Delta Aims To Be Transatlantic Leader
- Delta adds Atlanta route.
- Delta launches new SkyBonus programme for businesses.
Email
Digg It
del.icio.us
Related Resources
Press Releases
- Veteran Employee Benefits and Executive CompensationLawyer James C. Williams Joins Proskauer Rose in LosAngeles.
- Thompson Hine Grows Employee Benefits & Executive Compensation Practice.
- Executive Compensation Report Released by Staffing Industry Analysts, Inc.; Public Staffing Companies Pay Out Less at the Top.
Business Resources
Sponsored Links
Site Map | Contact Us | FAQs | About Us | Media Kit | Reprints | RSS Directory | Sign Up for Free Newsletters | Disclosure Policy
Copyright © 1999 - 2008 AllBusiness.com, Inc. All rights reserved.
No part of this content or the data or information included therein may be reproduced,
republished or redistributed without the prior written consent of AllBusiness.com.
Use of this site is governed by our Copyright and Intellectual Property Policy, Terms of Use Agreement and Privacy Policy.
© Copyright © 2002 American Accounting Association
You may not repost, republish, reproduce, package and/or redistribute the content of this page,
in whole or in part, without the written permission of the copyright holder.
Online Business Database | Online Business Information | Email Marketing Lists | Sales and Marketing Solutions | Business Mailing Lists
AllBusiness.com, its parent company D&B, and its affiliates.
