The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Volunteer Labor Sorting Across Industries Lewis M. Segal, Elizabeth Mauser and Burton A. Weisbrod Working Papers Series Macroeconomic Issues Research Department Federal Reserve Bank of Chicago December 1997 (WP-97-19) FEDERAL RESERVE BANK OF CHICAGO Volunteer Labor Sorting Across Industries Lewis M. Segal Federal Reserve Bank of Chicago Elizabeth Mauser Health Care Financing Administration Burton A. Weisbrod Northwestern University December, 1997 The views expressed in this paper are solely those of the authors and are not official positions of the Federal Reserve Bank of Chicago or the Federal Reserve System. Abstract Most prior analyses treat volunteer labor as a homogeneous activity, implicitly assuming that the marginal effects of tax changes and demographic shifts are equal across all forms of volunteering. We test the homogeneity assumption by estimating and comparing volunteer labor supply functions to three industries that make heavy use of volunteer labor - health, education, and religious organizations. The results demonstrate that individuals have systematically different preferences over the industries to which they volunteer. We find significant interindustry differences in the marginal effect of personal demographics, household composition, and tax status on the supply of volunteer labor. 1. In trod u ction The Current Population Survey of May 1989 estimates that approximately one fifth of the adult U.S. population volunteers to an organization or institution annually (Hayghe 1991). The 38 million volunteers are an important labor resource even though they receive no direct monetary compensation. Volunteering is not a homogeneous activity and preferences for different types of service may systematically attract volunteers to specific industries. For example, individuals with children might be attracted to volunteer to a school but not a hospital, suggesting that a decline in the birth rate will reduce the supply of volunteers to educational institutions without altering the supply to health organizations. The goal of this article is to measure the extent of systematic differences in volunteer labor supply across health, education, and religious organizations. The sorting process, if it exists, has implications for assessment of the effects of tax and demographic changes on the supply of volunteer labor. Researchers have examined both the factors that influence an individual’s decision to vol unteer and the extent of volunteering (Mueller 1975; Long 1977; Morgan, Dye, and Hybels 1977; Dye 1980; Clotfelter 1985; Steinberg 1990; Menchik and Weisbrod 1987). Those analy ses treat volunteering as a homogeneous commodity. Two recent studies, however, document systematic heterogeneity in the supply of volunteer labor. Sundeen (1991) considered the effect of personal goals and attitudes on the probability that a person volunteers to a given industry, and in a study of multiple sectors within a single industry Wolff, Weisbrod, and Bird (1991) found significant differences between volunteers to university-based, federal, and private nonprofit hospitals. We expand on the disaggregation theme by examining hours of volunteering in three industries that make heavy use of volunteer labor. The paper has five sections including the introduction. Section 2 presents a consumption model of the supply of volunteer labor. The model differs from earlier economic analyses by allowing for multiple simultaneous forms of volunteer labor. Section 3 translates the economic model into a statistical framework that consists of a tobit style equation for hours of volunteer service to each industry. The empirical analysis in section 4 uses data from a nationwide 1990 Gallup survey to compare the supply of volunteer labor to health, education, and religious organizations. The disaggregate estimates reveal significant interindustry differences in the 1 marginal effect of variables likely to affect the supply of volunteer labor. Differences are found in variables related to personal demography (age, education, race and sex), household com position (marital status and number of children), and tax variables (item ization status and deductibility of donations on the personal income tax). The final section summarizes the findings and provides suggestions for further research. 2. T h e T h e o r e t i c a l M o d e l There is substantial diversity among volunteers and among the organizations that they serve. In common usage, the term “volunteer” m ay evoke the im age of a person working in a hospital gift shop, organizing a school or church function, or leading a tour in a museum . W hile religious, health, and education organizations are the largest recipients of volunteer time, volunteers are employed by a much broader spectrum of industries, and perform a wide variety of tasks. Organizations using volunteers differ in goals, production m ethods, supervision, structure and costs. It is likely that they also vary in w hat they provide to their volunteers in return for service. Volunteer jobs in some industries offer substantial face-to-face contact with children or the elderly, for exam ple, while other volunteer jobs may involve record keeping, preparation of mailings, fund raising, or other non-service delivery support activities. Similarly, while the typical volunteer is a white, college educated, woman with children, volunteers are as diverse as the population itself. T hey vary in demographic characteristics as well as in paid em ploym ent status and m otivation. The heterogeneity of volunteers and industries suggests a sorting process whereby vol unteers with certain characteristics m igrate towards organizations in certain industries. Re searchers, ignoring the specific characteristics of volunteer jobs, im plicitly assum e that vol unteering is homogeneous in the supply and demand forces that affect the m atches between volunteers and organizations. We focus on the individual’s decision to volunteer and model volunteering as a m ethod of producing and consuming certain characteristics.1 Each type of volunteer service is regarded as a distinct com m odity providing a unique set of characteristics. In our simplified supply m odel, consumers allocate time between paid labor ( H ), leisure (L), and m ultiple types of volunteer service (Vj-,t = 1,...,K) to m axim ize utility that is a function of consumption ^ h e model builds upon the ideas presented in Lancaster (1966) and Becker (1975) where goods are valued for the characteristics they possess and consumption involves time as well as goods. 2 (C ), leisure, volunteer tim e, and monetary donations ( D ).2 Volunteer tim e enters the utility function directly as a proxy for the unobserved characteristics it produces. The consumer’s optim ization problem is written as ma x U ( V u . . . V k , L , D , C ) (2.1) subject to ZVi + L + H = T and tu(l - < ) ( £ K - + r ) + (l - t ) D + C < (1 - t ) ( w - T + A ) , where T and A represents the endowments of tim e and non-labor income respectively. The marginal hourly wage is w and t is the marginal tax rate on earnings.3 The implicit price for an hour of volunteer service is the marginal wage net of taxes regardless of the type of service. Equation 2.1 m otivates the three questions explored in this paper. First, do individuals have preferences over different t y p e s of volunteering? Second, do industries vary in the types of volunteer opportunities they provide? In other words, do industries vary in the types of activities for which they demand volunteers? Third, do differential preferences and opportunities, if they exist, result in interindustry differences in the marginal effect of exogenous wage, tax, or demographic changes? The following section describes a statistical m ethodology that will be useful in answering the third question and thereby shed light on the first two questions. However, we are unable to identify the separate effects of volunteer preferences and industry demand. 3. T h e E m p i r i c a l M o d e l The data for this study are from a 1990 national survey of individuals in the United States collected by the Gallup Organization for Independent Sector.4 Gallup asked a random 2In so doing, we assume that organizations are willing to utilize all of the volunteer labor that individuals wish to supply. 3The budget constraint in 2.1 corresponds to a person who itemizes deductions on his or her tax return. If the individual does not itemize, the price of donating one dollar is one dollar, rather than one minus his marginal tax rate. independent Sector is a nonprofit advocacy organization based in Washington D.C. Hodgkinson and Weitzman (1990) document the survey instrument and methodology. 3 sample of 2,727 adults a variety of questions including several regarding fourteen categories of volunteer service; appendix 1 describes the full list of categories. Respondents were asked about the number of volunteer hours provided in each category of service. The dependent variables in our analysis are the number of volunteer hours supplied to health, education, and religious institutions during the preceding month. W e limit the analysis to these three categories in order to focus on a single dimension of disaggregation, namely industrial. The selection of independent variables follows economic theory, previous studies, and the available data. As suggested by equation 2.1, volunteer hours are influenced by variables affecting preferences and variables affecting constraints. The regressors fall into four groups: individual demographics (sex, race, age, education, and time in the community), household composition (marital status, number and age distribution of children), employment variables (available wage and employment status), and tax variables (price of a one dollar donation and tax itemization status). Table 3.1 describes the complete set of regressors and table 3.2 presents descriptive statistics from the subsample of 1,683 respondents with complete data. The meaning of most of the variables is self evident. However, the employment and tax variables require further description. The opportunity cost of volunteering is proxied by the marginal wage which is captured in two survey questions. The survey recorded the average hourly market wage for those employed, and the self-estimated available wage for those not employed.5 W e estimate a total wage effect without decomposition into income and substitution effects due to the inability to separate earnings and wealth in the survey data. Full time, part time and self employment indicators are included as regressors even though they might be endogenous to the work/volunteer decision; in effect we assume that volunteer decisions are made in a two stage process, subsequent to the paid work decision. The marginal tax rate and deduction itemization status define the price of a cash do nation to an individual. Volunteering is often construed as a donation of time as opposed to money, and the tax variables allow us to consider whether each form of volunteering is a substitute or complement for monetary donations. It is theoretically and empirically possi ble to disaggregate donations across industries as we do with volunteer time. However all donations to tax deductible organizations bear the same price for a given individual, just as all volunteer labor has a single price. 5For our purposes, a measure ofthe reservation wage would have been more appropriate than the available wage, particularly for people without paid employment. The reported available wage isonly a lower bound on the marginal value of time. 4 Table 3.1: Description of Variables D escrip tio n N am e P erso n a l D em ographics Male Binary variable equal to 1 if the Binary variable equal to 1 if the White Age of respondent in years. Age Binary variable equal to 1 if the College Binary variable equal to 1 if the LessHS Community Binary variable equal to 1 if the less than 2 years. respondent is male. respondent is white. respondent is a college graduate. respondent has not completed highschool. respondent has been in the community H ousehold C o m position Married Children ChildLE5 Binary variable equal to 1 if the respondent is currently married. Number of children residing in the household. Binary variable equal to 1 if there is at least one child under the age of 6 residing in the household. E m p lo ym en t S ta tu s FullTime PartTime Wage Binary variable equal to 1 if the respondent is employed full time. Binary variable equal to 1 if the respondent is employed part time. Hourly wage if employed or self reported available wage. Tax S ta tu s Itemize DonPrice Binary Variable equal to 1 if the respondent itemized deductions on most recent tax return. Price of a one dollar donation, equal to one minus the marginal tax rate for itemizers. 5 Table 3.2: Descriptive Statistics N am e M ean S td .D e v . P e r s o n a l D e m o g r a p h ic s Male White Age College LessHS Community 0.52 0.82 44.05 0.23 0.21 0.13 0.50 0.39 16.56 0.42 0.41 0.33 H o u s e h o ld C o m p o s itio n Married Children ChildLE5 0.67 0.94 0.24 0.47 1.23 0.43 E m p lo y m e n t S ta tu s FullTime PartTime Wage 0.53 0.11 13.33 0.50 0.31 22.47 0.43 0.89 0.49 0.13 T ax S ta tu s Itemize DonPrice Observations 1,683 6 Table 3.3: Volunteer Participation and Hours During Preceding Month P a r tic ip a tio n R a te M ean H o u r s /M o n th Health 7.1 % Education 10.6 % Religous 20.1 % Total 28.8 18.2 (0.3) 13.9 (0.3) 15.1 (0.4) 18.6 (0.5) I n d u s tr y % Standard deviation in parentheses. Table 3.3 describes the participation rate and average monthly hours of service for vol unteers to health, education, and religious organizations based on the 1683 individuals for whom there are complete data.6 A majority of volunteers provide time to a religious or ganization but there is substantial participation in the health and education industries as well. Note that volunteering in one industry does not preclude doing so in others, just as paid employment does not preclude a second job. In fact, some 26 percent of volunteers provide their services in more than one of the three industries (table 3.4). If volunteering, regardless of industry or type, yielded a single set of utility increasing characteristics and if the demand for volunteers was perfectly elastic then we would expect a person to participate in at most one volunteer activity and volunteers would be indifferent to the organization or industry for which they work. Alternatively, if industries vary in what they provide to their volunteers, consumers would then attempt to equate the marginal utility of each volunteer activity, leading to participation in multiple volunteer activities. Two complications arise in the regression of volunteer hours (per month) in each industry on the independent variables listed in table 3.1. First, ordinary least squares estimation would produce biased and inconsistent estimates of the population parameters since many of the respondents (actually a majority) report no volunteering. We use a tobit model to deal with the truncation at zero hours under the assumption that the unexplained portion of volunteer hours is distributed normally. Thus, hours of volunteer service by individual t 6N ea rly 70 p e rcen t o f all v o lu n te e r s p ro v id e tim e in a t le a st o n e o f th e th ree ca te g o r ie s. W e lim it th e a n a ly sis to th e s e th ree c a te g o r ie s in order to fo c u s o n th e in d u str ia l d is a g g r e g a tio n o f v o lu n te e r la b or. 7 Table 3.4: Number of Simultaneous Industries Per Volunteer I n d u s tr ie s P e r c e n t o f S a m p le P e r c e n t o f V o lu n teers 71 21 6 2 0 1 2 3 74 21 5 Based on the health, education and religious volunteering activities of 1,683 individuals with complete data. in industry j ( V^j) is expressed as V ij = (3'X i + i f p 'X i + m j > 0 (3.1) and 0 otherwise. In the preceding equation, X* is a vector of individual specific control variables, is a vector of industry specific parameters, and is a normally distributed mean 0 error term with variance cr|. A single set of regressors determines the supply of volunteer labor to each industry, but we allow the parameter estimates to vary across industries. The second regression complication arises from the error structure. The errors are as sumed independent across individuals, but are likely correlated across industries for a single individual. The correlation stems from at least two sources. Decisions regarding all uses of time, including volunteering, are constrained by a single time budget, leading to the expecta tion that, c e te r is p a r ib u s , an individual who is indifferent between two forms of volunteering and who is more active in one industry would be less active in another.7 Additionally, pref erences may be correlated across some forms of volunteer service due to unobserved factors related to preferences or budgets. Full information maximum likelihood (FIML) estimation of the system of tobit equations requires the evaluation of up to three multiple integrals of the normal density — an untenable approach unless the errors are uncorrelated across equations.8 As an alternative to FIML, we estimate a separate maximum likelihood tobit equation for each industry, providing consistent, although inefficient, estimates of the 7T h is n eed n o t be tru e as in crea sed v o lu n te e r in g co u ld o c c u r a t th e e x p e n se o f d im in ish e d p a id w ork or leisu re. 8A te s t for in d e p e n d e n c e o f th e errors w a s p erfo rm ed u sin g a L a g ra n g e M u ltip lie r te s t a n d a s y s t e m o f p ro b it e q u a tio n s (K iefer 1 9 8 2 ). T h e p r o b it m o d e l for v o lu n te e r p a r tic ip a tio n is c o n siste n t w ith th e a ss u m p tio n s o f th e to b it m o d e l for v o lu n te e r h o u rs. A te s t s t a t is t ic co u ld b e fo r m u la te d for th e to b it m o d e l d irectly , b u t th ere is little to b e g a in e d g iv e n th e o v e r w h e lm in g r e su lts o f th e te s t for co rre la tio n in th e p r o b it m o d e ls. W e reject th e n u ll h y p o th e s is o f u n c o r r e la te d errors b a se d o n a te s t s t a t is t ic o f 110 d is tr ib u te d as a ch i sq u are w ith 3 d egrees o f freed o m . 8 parameters. The goal of this paper is to compare the marginal effect of various regressors across indus try specific tobit equations. Does marital status have the same affect on hours of volunteer service to health organizations that it has on volunteering to religious organizations? Does an increase in the tax rate have a similar effect on hours of religious service and hours of educational service? Answers to these questions require cross equation comparisons of the tobit derivatives, which are themselves a function of the tobit parameters. The comparison is a three step procedure. First, we estimate the complete three equation parameter covariance matrix.9 Next, we compute the derivative of the expected number of volunteer hours in each industry with respect to each regressor.10 The final step involves computing the covariance matrix of the tobit derivatives with respect to each independent variable.11 Finally, we use Wald tests to make cross equation comparisons. It is tempting to use the system of equations to consider the degree to which different forms of volunteering are complements or substitutes. The classification of two goods or activities as Hicksian substitutes or complements hinges on the sign of the cross price elas ticity. An hour of volunteering has a price of zero dollars and one hour of time. Although the price of volunteering varies from person to person because the value of time varies, it is constant for a given person across all industries. There would be a distinct price for each industry if there was variation in travel costs, clothing costs, or search costs. Unfortunately, the survey contains no information on such costs. Identical coefficients on all parameters across two or more industries (equations) indicate that the two forms of volunteering appeal in the same way to the aggregate population but without information on the variation in the price (cost) of volunteering across industries we cannot classify volunteer time among*I 9T h e th r e e e q u a tio n c o v a r ia n c e m a tr ix is e stim a te d fro m a q u a d ra tic form o f th e cro ss p r o d u c t o f th e sta c k e d Ja co b ia n v e c to r s o f th e lik e lih o o d fu n c tio n and a b lo ck d ia g o n a l m a tr ix o f th e in d iv id u a l h essia n mar trices (J a k u b s o n 1988 an d M roz 1 9 8 7 ). T h e te ch n iq u e is an e x te n s io n o f th e e stim a tio n o f a h e te r o sk e d a sticc o n siste n t c o v a r ia n c e m a tr ix for a sin g le eq u a tio n . I C o e f f ic ie n t s in th e to b it m o d e l do n o t h a v e th e d eriv a tiv e in te r p r e ta tio n o f th e O L S m o d e l (M c D o n a ld an d M offitt 1 9 8 0 ). T h e d e r iv a tiv e s in th e to b it m o d e l c o n sist o f th e to b it c o efficien t m u ltip lie d b y th e c o n d itio n a l p r o b a b ility o f a n o n -zero d e p e n d e n t variab le. T h u s, th e d e r iv a tiv e s d ep en d on th e e stim a te d v e c to r o f p a ra m eters (Bj, crj) as w ell as th e ch a r a c te r istic s v ecto r ( X ,). T h r o u g h o u t th is p a p er, to b it d e r iv a tiv es are e v a lu a te d a t th e m ea n v a lu e o f th e regressors. II T h e c o m p u ta tio n is str a ig h t fo rw a rd g iv e n th e full co v a ria n ce m a tr ix o f th e p a ra m eters c o m p u te d in th e first ste p an d a lin ea r T a y lo r series e x p a n sio n o f th e fo rm u la for th e to b it d e r iv a tiv e u sed in th e sec o n d ste p . 9 industries as complements or substitutes. 4. R e s u l t s Table 4.1 presents the tobit regression estimates for the number of volunteer hours supplied to each of the three industries. Overall, the results are quantitatively similar to those of aggregate studies: c e te r is p a rib u s men supply fewer volunteer hours than women, higher education implies greater volunteer activity, longer community attachment increases the level of volunteering, volunteer hours are higher for married persons than for single persons, and children in the household increase volunteer participation except when the children are young (under age 6). However, the regressions are not quantitatively the same across the three industries. We reject the hypothesis of equality of all parameters, including the intercepts and the error variances, across the three equations based on a chi square test statistic of 134 with 32 degrees of freedom. Although differences between industries were expected, we have no a priori predictions of the direction or magnitude. Table 4.2 reports the predicted change in volunteer hours due to a change in each inde pendent variable evaluated at the mean of the variables. Standard errors of the derivatives are reported in parentheses. The table highlights the interindustry differences. For example, being married increases the supply of volunteer time to an educational institution by more than one half hour per month (a 4 percent increase) but has a statistically negligible effect on volunteering to health and religious organizations. In order to summarize the results of table 4.2, we examine the cross equation similarities and differences with respect to four groups of regressors — personal demographics, household composition, employment status, and tax status. 4.1. Personal Demographics The joint test for equality across equations of the tobit derivatives with respect to sex, race, age, education, and community attachment is rejected. The chi square test statistic is 31.8 with 12 degrees of freedom. Evidently, these variables affect the supply of volunteer labor to each industry in a distinct way. It is useful to consider whether the rejection is to due to a single demographic regressor or a single pair of equations. Table 4.3 shows the probability of a false rejection (significance level) of the hypothesis that the derivative estimates are equal across each pair of equations. For example, table 4.2 indicates that being male decreases the 10 Table 4.1: Tobit Regression of Volunter Hours Per Month R e g re sso r I n d u s tr y E d u c a tio n H ealth R e lig io u s P e r s o n a l D em o g ra p h ics Male White Age College LessHS Community -11.383** (3.056) 9.506** (4.711) -0.194 (0.120) 15.130** (3.909) -21.548** (6.069) -2.803 (4.513) -5.556** (2.582) -0.249 (3.458) 0.241** (0.098) 4.541 (2.779) -15.395** (4.022) -10.180** (4.667) 8.733** (3.638) 9.303** (1.666) -15.931** (4.179) 2.631 (2.687) 2.922** (1.083) -4.768 (3.390) -0.380 (5.629) 8.962 (7.005) 0.019 (0.075) 0.871 (3.576) 13.073** (4.903) 0.046 (0.058) 2.673 (2.824) 6.532* (3.563) -0.128 (0.090) -15.793 (15.605) -100.615* (58.211) -37.058** (12.418) -138.090** (43.214) 23.405** (7.310) 45.640* (26.558) 83.881** (42.118) 31.840** (3.827) -90.428** (28.624) 33.250** (2.688) -14.953** (5.256) 4.420 (7.475) 0.062 (0.177) 7.697 (4.960) -29.779** (9.312) -27.298** (11.170) H o u seh o ld C o m p o sitio n Married Children ChildLE5 0.130 (5.498) -1.343 (2.399) -7.646 (6.750) E m p lo y m e n t S ta tu s FullTime PartTime Wage Tax S ta tu s Itemize Don Price Intercept Sigma 29.377 (59.693) 47.791** (5.862) Standard errors in parentheses. ** significant at the 5% level. * significant at the 10% level. 11 Table 4.2: Tobit Regression Derivatives Evaluated at Mean of Independent Variables R e g re sso r H ealth I n d u s tr y E d u c a tio n R e lig io u s P e r s o n a l D e m o g ra p h ic s Male White Age College LessHS Community -0.760** (0.274) 0.225 (0.380) 0.003 (0.009) 0.391 (0.252) -1.513** (0.457) -1.387** (0.546) -0.678** (0.172) 0.566** (0.273) -0.012 (0.007) 0.902** (0.248) -1.284** (0.344) -0.167 (0.268) -0.939** (0.435) -0.042 (0.584) 0.041** (0.017) 0.767 (0.469) -2.601** (0.660) -1.720** (0.784) 0.520** (0.221) 0.554** (0.103) -0.949** (0.259) 0.444 (0.454) 0.494** (0.180) -0.806 (0.570) -0.019 (0.286) 0.455 (0.352) 0.001 (0.004) 0.052 (0.213) 0.779** (0.302) 0.003 (0.003) 0.452 (0.477) 1.104* (0.603) -0.022 (0.015) -0.803 (0.798) -5.113* (2.980) -2.208** (0.723) -8.228** (2.518) 3.954** (1.240) 7.710* (4.498) H o u seh o ld C o m p o s itio n Married Children ChildLE5 0.007 (0.279) -0.068 (0.122) -0.389 (0.340) E m p lo y m e n t S ta tu s FullTime PartTime Wage Tax S ta tu s Itemize DonPrice Standard errors in parentheses. ** significant at the 5% level. * significant at the 10% level. 12 Table 4.3: Significance Level of Interindustry Equality of Tobit Regression Derivatives Eval uated at Mean of independent Variables R e g re sso r H ealth vs. E d u c a tio n In d u stry H ealth vs. E d u c a tio n vs. R elig io u s R elig io u s P e r s o n a l D em o g ra p h ics Male White Age College LessHS Community 0.70 0.67 0.04 0.45 0.14 0.72 0.55 0.32 0.00 0.78 0.06 0.05 0.13 0.00 0.16 0.37 0.01 0.51 0.87 0.76 0.80 0.83 0.45 0.69 0.36 0.35 0.15 0.41 0.61 0.12 0.19 0.42 0.00 0.01 0.00 0.00 0.79 0.40 0.19 0.13 0.68 0.04 H o u seh o ld C o m p o sitio n Married Children ChildLE5 E m p lo y m e n t S ta tu s FullTime PartTime Wage Tax S ta tu s Itemize DonPrice supply of volunteer hours to health organizations by 0.76 hours per month (compared with the overall mean of 18.2 hours) and reduces the supply to educational institutions by 0.68 hours per month. However, the first cell of table 4.3 reveals that there is a 0.79 probability that the difference between the 0.76 and 0.69 estimates is mere chance. In other words, males volunteer statistically less than females to both health and education organizations, but there is no statistically detectable difference in the male/female gap across industries. The top panel of table 4.3 describes the cross equation comparisons of the demographic variables. The cross industry demographic difference is largely due to variation in the age, education, and community attachment effects, but the interindustry difference is unrelated to the sex and race parameters. The estimates suggest significant differences in the marginal effect of age between health and religious organizations (probability value of 0.04), and between education and religious organizations (probability value below 0.01). Referring back to table 4.2, we note that age is significant only in the equation for hours of volunteering to religious organizations. Thus, an upward shift in the population age distribution, all 13 else held constant, is likely to bring an increase in the supply of volunteer time to religious organizations but have little impact on health and education organizations. Similarly, an exogenous increase in the number of college educated individuals will increase the supply of volunteers to education and religious organizations more than to health organizations. 4.2. Household Composition As with the demographic variables, the marginal effect of household composition on the supply of volunteer labor varies across industry. The overall equality of the household com position derivatives is rejected based on a chi square test statistic of 22.9 with 6 degrees of freedom. Marital status and the number of children in the house increase the supply of volunteer time to education and religious organizations but have a statistically insignificant effect on volunteering to health organizations (table 4.2). The difference across industries in the marital effect estimates are not significant; the most significant difference is between the health and education industries (probability value of 0.13). However, there is a significant difference between the health and education industries in the marginal effect of children (probability value below 0.01). Similarly, there is a significant difference between the health and religious industries. The increase in volunteering caused by children in the household is lessened by the presence of young children (age 5 and below) in the home. 4.3. Employment Status The joint test for equality of the change in volunteering to each industry due to a change in the employment variables cannot be rejected. The test statistic is only 4 with 6 degrees of freedom. This is not to say that employment status is unrelated to volunteering. As tables 4.1 and 4.2 indicate, in all three industries, part time employment is positively related to volunteering relative to being not-employed and it is statistically significant at the five percent level in two industries. People employed part time provide an additional 0.46 hours per month to health organizations, 0.78 hours per month to education organizations and 1.10 additional hours per month to religious organizations. However, we are unable to find a statistically significant difference between the estimates of 0.46, 0.78, and 1.10 hours (table 4.3). In our estimates the opportunity cost of time is measured by the actual or available wage, and its effect is not significantly different from zero in any industry. Since we do not control 14 for full income, the estimated wage may proxy for wealth as well as the opportunity cost of time. The results suggest that the substitution and income effects of an increase in the wage offset one another. 4.4. Tax Status The cross industry differences in the effect of the tax variables are perhaps the most inter esting. The variables for itemization status and the tax level have very different effects on religious volunteering relative to the other two industries. Individuals who itemize deduc tions on their tax returns supply more volunteer time to religious organizations than those who do not itemize, approximately 4 hours per month, an increase of 26 percent. Simulta neously, itemizers provide less volunteer time to education organizations than non-itemizers, approximately 2.2 hours less per month (table 4.2). The effect of itemizing is statistically insignificant for health organizations. The difference between the 4 hour increase to religious organizations and the 2.2 hour decrease to education organizations is statistically significant at better than the 1 percent level (table 4.3). The relationship between income tax rates and volunteer activity also differs across in dustries. The D o n P ric e coefficients in table 4.2 imply that a decrease in the tax rate, which raises the cost of making a monetary donation, reduces volunteering to health and education organizations but has positive effect on volunteering to religious organizations. We estimate that a one percent decrease in the marginal tax rate causes a 1 hour increase in health volunteering, a 1.4 hour increase in education volunteering, and a 4.6 hour decrease in reli gious volunteering. Thus, giving of money and time appear to be complements for education organizations but substitutes for religious organizations. Table 4.3 confirms that there is a significant difference across industries in the complementarity of donations of money and time. 4.5. Direct Evidence Of the Returns To Volunteer Service The preceding four sections demonstrate significant and substantial differences across in dustries in the marginal effects of individual demographic variables, household composition variables, and tax variables. The findings support our hypothesis that individuals systemat ically differ in their preferences for volunteering in the three industries. We attribute these differences to the unique consumption and investment opportunities provided by the volun- 15 Table 4.4: Reasons for Continuing to Volunteer R e a so n P e r c e n t o f V olu n teers R e p o r tin g E a ch R e a s o n A ll I n d u s tr ie s H ealth E d u c a tio n R e lig io u s P e r s o n a l C o n s u m p tio n 1. I thought I would enjoy doing the work; feel needed 2. Had a child, relative, or friend who was involved in the activity or would benefit from it 3. Previously benefitted from the activity 34.88 39.39 42.91 31.54 25.32 23.74 41.49 24.12 12.79 14.65 16.31 12.24 6.85 10.10 9.93 6.12 61.63 71.21 66.67 61.04 33.59 6.07 27.78 6.57 23.40 6.03 45.27 6.31 8.40 12.12 10.99 7.79 P e r s o n a l I n v e s tm e n t 4. Wanted to learn and get experience; work experience; get a job P u b lic B e n e fit 5. Wanted to do something useful; help others; do good deeds for others U n cla ssified 6. Religious concerns 7. Had a lot of free time 8. I wanted to engage in activities more fulfilling than my current job 774 Observations teer experience in each industry — for example, opportunities for human capital investment and altruistic opportunities to help various other people. The Gallup survey includes some direct evidence on interindustry variation in the re turns from volunteering. Volunteers were asked the reasons motivating them to serve as a volunteer.12 Table 4.4 lists the possible responses and their frequencies. Volunteers were asked to indicate all applicable reasons for continuing to volunteer and they provided two responses on average. We divided the possible responses into three categories — personal consumption, per sonal investment and public benefit. The three categories classify five of the eight possible responses. We observe people volunteering, thus volunteer service, to the extent that it is 12V o lu n tee r s w ere ask ed a b o u t m o tiv e s for b e g in n in g to v o lu n te e r and also for c o n tin u in g to v o lu n te e r . W e fo c u s o n th e la ter as it m o re c lo se ly co rresp o n d s to th e a c tu a l retu rn s from v o lu n te e r in g . M o tiv e s for b e g in n in g to v o lu n te e r are lik e ly to in c lu d e e x p e c ta tio n a l errors w h ic h c o m p lic a te ou r a n a ly sis b u t d eserv e fu rth er stu d y . 16 a voluntary arrangement between volunteers and organizations, and involves a benefit to both parties. However, our classification focuses on the extent to which a personal or public benefit is the primary volunteer motivation. In aggregate, a majority (more than 61%) of volunteers report that they volunteer to help others (public benefit, reason 5). Slightly fewer volunteers (about 40%) report that they volunteer because of a current or previous benefit to the volunteer or someone close to them (private consumption, reasons 2 and 3 ). Only seven percent of the volunteers report human capital acquisition (personal investment, reason 4) as a motivation for volunteering. Turning to the cross industry comparisons, two entries in the table stand out as being unusually high in one industry relative to the other industries. First, 41 percent of volunteers to education institutions report that they do so for the benefit of someone they know, as compared to 24 percent of volunteers to health and religious organizations. Second, not surprisingly, volunteering because of “religious concerns” is more frequent for volunteers to religious organizations (45%), although it is still substantial in the health (28%) and education (23%) industries. Interindustry comparisons of the raw frequencies ignore personal differences. Table 4.5 contains probit regression estimates of the probability of a volunteer reporting each of the five categorized responses conditional on the industry to which they volunteer as well as their demographic, household composition, and employment status variables.13 The first three columns of the table are the personal consumption motives, column four is the personal investment motive, and the last column is the public benefit motive.14 Our analysis centers on the top two panels, V o lu n teer P a r tic ip a tio n and S ig n ifica n ce o f I n te r in d u s tr y D iffe re n c e s , which present the regression parameters for the industry indicators and the statistical sig nificance of comparisons between industries. A striking difference across industries in what people report they receive from volunteering is the extent to which volunteering in the education industry is dominated by personal consumption motives. The coefficient on the education indicator is significant at the 1 percent level in the E n jo y m e n t and B e n e fit regressions (columns 1 and 2). Volunteers to education organizations are more likely than other volunteers to report that they volunteer for their own enjoyment and benefit. With respect to E n jo y m e n t variable, the difference between the 13T h e p r o b it r eg ressio n s in c lu d e an in te r c e p t te r m in a d d itio n to th e in d u str y in d ic a to r variab les. M u ltic o llin ea rity is n o t an issu e b e c a u se a sin g le v o lu n te e r is o fte n a c tiv e in m u ltip le in d u str ie s. 14R eg ressio n r e su lts for th e u n cla ssifie d m o tiv e s are n o t p r e se n te d b u t are a v a ila b le fro m th e a u th o rs. 17 Table 4.5: Probit Analysis of the Returns from Volunteer Service Personal Personal Consumption Investment R egressor E n jo y m e n t C u r re n t B e n e fit P r io r B e n e fit W o rk E x p e rie n c e Public Benefit H elp O th e r s V o lu n teer P a r tic ip a tio n Health 0.215 (0.135) 0.645** (0.128) 0.294** (0.134) 0.083 (0.146) 0.199 (0.143) 0.148 (0.149) 0.530** (0.173) 0.289** (0.168) 0.197 (0.175) 0.401** (0.128) 0.158 (0.123) 0.292** (0.128) 0.21 0.09 0.00 0.01 0.63 0.02 0.54 0.71 0.76 0.28 0.11 0.65 0.11 0.44 0.32 -0.018 (0.109) 0.138 (0.162) 0.003 (0.004) 0.197* (0.111) -0.147 (0.194) 0.040 (0.184) -0.262** (0.125) 0.362* (0.189) 0.001 (0.005) 0.044 (0.124) -0.392 (0.247) -0.080 (0.211) -0.203 (0.136) 0.039 (0.200) -0.003 (0.005) 0.315** (0.133) -0.449 (0.306) 0.538** (0.201) -0.015 (0.169) -0.193 (0.221) -0.016** (0.006) 0.109 (0.168) 0.004 (0.309) 0.412 (0.232) -0.016 (0.107) 0.307** (0.152) -0.006 (0.004) 0.359** (0.113) -0.060 (0.181) 0.091 (0.185) 0.013 (0.123) -0.098** (0.053) 0.262** (0.151) 0.226 (0.148) 0.325** (0.056) -0.132 (0.159) 0.012 (0.152) 0.020 (0.062) -0.249 (0.188) -0.053 (0.191) 0.129** (0.068) -0.234 (0.210) -0.028 (0.121) 0.028 (0.052) -0.098 (0.150) -0.119 (0.127) 0.066 (0.154) 0.200 (0.144) 0.152 (0.175) 0.140 (0.156) -0.164 (0.204) 0.005 (0.197) 0.246 (0.225) 0.160 (0.124) 0.216 (0.156) -0.645 (0.307) Standard errors in parentheses. -2.109 (0.362) -1.320 (0.371) -1.237 (0.416) -0.258 (0.300) Education Religious 0.092 (0.122) 0.283** (0.119) -0.147 (0.123) S ig n ifican ce o f I n te r in d u s tr y D iffe re n c e s Health vs. Education Health vs. Religious Education vs. Religious P e r s o n a l D e m o g ra p h ic s Male White Age College LessHS Community H o u seh o ld C o m p o s itio n Married Children ChildLE5 E m p lo y m e n t S ta tu s FullTime PartTime Intercept 18 education coefficient and the religion coefficient is significant at the 1 percent level, but the difference between the education and health coefficients is significant only at the 21 percent level. In contrast, volunteers in the health industry are most likely to report work experience and helping others as motives although none of the interindustry differences are statistically significant at the ten percent or better level. Volunteers to religious organizations report prior personal benefit and helping others as motives with statistically significant frequencies. 5. C o n c l u s i o n We find considerable evidence that volunteering is not a homogeneous activity, but that it provides differentiated non-pecuniary rewards across industries. Volunteering to health, education, and religious organizations provide quite different opportunities, it appears, for individuals to gain work experience, obtain information about an organization, fulfill altruis tic desires, and acquire benefits that are otherwise unavailable. By modeling volunteer labor supplied to each industry as a distinct commodity and considering whether the factors that influence the supply of volunteer labor differ across industries, we find that aggregation does obscure important interindustry differences. In particular, changes in population demogra phy, household composition, and taxation are likely to have different implications for the supply of volunteers to the health, education, and religious industries. However, we do not find differences across industries in the relationship between employment and volunteering. We interpret the interindustry variation as reflecting differences in the supply of volunteer labor under the assumption that each industry, while providing a distinct bundle of volunteer opportunities, has a perfectly elastic demand for volunteers. The nature of the demand for volunteers, its variation among industries, and its interaction with supply, however, remain important issues for research. 19 A . C a t e g o r i e s o f V o l u n t e e r S e r v i c e L i s t e d in t h e 1 9 9 0 G a l l u p S u r vey 1. Health (including mental health) - General and rehabilitation, including institutions and organizations for mental health and mental retardation and developmentally dis abled; substance abuse; diseases (research and public education, etc.); hospitals, nurs ing homes, hospices, clinics, etc., crisis counselling, hotlines, etc.; fund drives of private health associations such as American Cancer Society, American Heart Association. 2. Education/Instruction (formal and informal) - Elementary, secondary or higher edu cation (public or private, which may be church-affiliated or nonsectarian); libraries; research at educational institutions; adult education; informal education; educational fund drives for educational associations. 3. Religious Organizations - Religion-related, spiritual development (includes giving to churches, synagogues, monasteries, convents, seminaries, etc.; but not giving to churchaffiliated schools offering broad educational curricula, nursing homes, Catholic chari ties, Jewish federations, etc.). 4. Recreation (for adults) - Membership clubs in such areas as swimming, boating, skiing, aviation, rifle marksmanship, hunting. 5. Arts, Culture, and Humanities - Includes architecture, design, performing arts; culture/ethnic awareness groups, other cultural groups; historical preservation; humanistic societies; museums; art exhibits; operas; symphony orchestras; photography; theater; public television and radio. 6. Work-related Organizations - Labor unions, credit unions, professional associations (lawyers, medical personnel, engineers, etc.), Chamber of Commerce, industrial stan dard committees, etc. 7. Informal-alone - Helping a neighbor, friend or organization on an ad hoc basis; spending time caring for elderly person or babysitting children of a friend, but not part of an organized group or for pay. 8. Human Services - Day care centers; foster care services; family counselling; consumer protection; legal aid; crime and delinquency prevention; employment/jobs; food; housing/shelter; public safety, emergency preparedness and relief; recreation, sports, ath letics; Red Cross, YMCA, and other combined multi-purpose charity drives.9 9. Environment (including animals) - Environmental quality protection, beautification; animal-related activities (exhibitions, public education, animal population control); protection and welfare; humane societies, wildlife and animal sanctuaries. 20 10. Public/Society Benefit - Civil rights, community and social action, advocacy (includes minority and women’s equity issues); community improvement, community capacity planning; science; technology; technical assistance; voluntarism; philanthropy; charity, between groups, Rotary, Kiwanis, etc.; consumers organizations, advocacy organiza tions, such as nuclear freeze, antipoverty boards, etc. 11. Political Organizations - Political party clubs (Democratic, Republican, other), non partisan political or community groups, and other political causes. 12. Youth Development - Boy and Girl Scouts; Camp Fire Groups; 4-H Clubs; youth groups with religious affiliations, such as Catholic Youth Organizations, Little Leagues, and other athletic groups engaged in youth development. 13. Private and Community Foundations - Ford Foundation, Rockefeller Foundation, Carnegie Foundation, etc., San Francisco Foundation, New York Community Trust, Boston Foundation, Cleveland Foundation, Fund for the City of New York, etc. 14. International/Foreign (in U.S. and Abroad) - International education; health abroad; international peace or security; refugee-related activities; relief abroad; other social services; student exchange and aid; cultural exchange; economic development; techni cal assistance; promotion of friendly relations among nations; United Nations and its associations. 21 References [1] Becker, Gary (1975): Human Capital, Columbia University Press, New York. [2] Clotfelter, Charles T. (1985): Federal Tax Policy and Charitable Giving, University of Chicago Press, Chicago. [3] Dye, Richard F. (1980): “Contributions of Volunteer Time: Some Evidence on Income Tax Effects,” National Tax Journal 33, 89-93. [4] Hayghe, Howard V. (1991): “Volunteers in the U.S.: Who Donates the Time?’,’ Monthly Labor Review 14(2), 17-23. [5] Hodgkinson, Virginia A. and Murray S. Weitzman (1990): Giving and Volunteering in the United States, Independent Sector, Washington D.C. [6] Jakubson, George (1988): “The Sensitivity of Labor Supply Parameter Estimates to Unobserved Individual Effects and Random Effects Estimates in a NonLinear Model using Panel Data,” Journal of Labor Economics 6(3), 302-329. [7] Kiefer, Nicholas M. (1982): “Testing for Dependence in Multivariate Probit Models,” Econometrica 69(1), 161-166. [8] Lancaster, Kelvin (1966): “A New Approach to Consumer Theory,” Journal of Political Economy 74 (April), 132-157. [9] Long, Steven H. (1977): “Income Tax Effects on Donor Choice of Money and Time Contributions,” National Tax Journal 30, 207-211. [10] McDonald, John F. and Robert A. Mofitt (1980): “The Uses of Tobit Analysis,” The Review of Economics and Statistics LXII(2), 318-31. [11] Menchik, Paul L. and Burton A. Weisbrod (1987): “Volunteer Labor Supply,” Journal of P'ublic Economics 32, 159-183. [12] Morgan, J.N., R.F. Dye, and J.H. Hybels (1977): “Results from Two National Surveys of Philanthropic Activity” in Research Papers of the Filer Commission, U.S. Department of the Treasury, Washington D.C. [13] Mroz, Thomas A. (1987): “The Sensitivity of an Empirical Model of Married Women’s Hours of Work to Economic and Statistical Assumptions,” Econometrica. [14] Mueller, Marnie W. (1975): “Economic Determinants of Volunteer Work by Women,” Signs: Journal of Women and Culture in Society 1, 325-388.15 [15] Steinberg, Richard (1990): “Labor Economics and the Nonprofit Sector: A Literature Review,” Nonprofit and Voluntary Sector Quarterly 19(2), 151-169. 22 [16] Sundeen A. Richard, “Differences in Personal Goals and Attitudes Among Volunteers,” Conference Proceedings, 1991 Annual ARNOVA Conference. [17] Wolff, Nancy L., Burton A. Weisbrod and Edward Bird (1993): “The Supply of Volun teer Labor: The Case of Hospitals,” Nonprofit Management and Leadership 4(1). 23