Working Paper Relational goods, monitoring and non-pecuniary compensations in the nonprofit sector: the case of the Italian social services

Transcription

1 econstor Der Open-Access-Publikationsserver der ZBW Leibniz-Inormationszentrum Wirtschat The Open Access Publication Server o the ZBW Leibniz Inormation Centre or Economics Mosca, Michele; Musella, Marco; Pastore, Francesco Working Paper Relational goods, monitoring and non-pecuniary compensations in the nonproit sector: the case o the Italian social services IZA Discussion Papers, No Provided in Cooperation with: Institute or the Study o Labor (IZA) Suggested Citation: Mosca, Michele; Musella, Marco; Pastore, Francesco (2006) : Relational goods, monitoring and non-pecuniary compensations in the nonproit sector: the case o the Italian social services, IZA Discussion Papers, No This Version is available at: Nutzungsbedingungen: Die ZBW räumt Ihnen als Nutzerin/Nutzer das unentgeltliche, räumlich unbeschränkte und zeitlich au die Dauer des Schutzrechts beschränkte einache Recht ein, das ausgewählte Werk im Rahmen der unter nachzulesenden vollständigen Nutzungsbedingungen zu vervielältigen, mit denen die Nutzerin/der Nutzer sich durch die erste Nutzung einverstanden erklärt. Terms o use: The ZBW grants you, the user, the non-exclusive right to use the selected work ree o charge, territorially unrestricted and within the time limit o the term o the property rights according to the terms speciied at By the irst use o the selected work the user agrees and declares to comply with these terms o use. zbw Leibniz-Inormationszentrum Wirtschat Leibniz Inormation Centre or Economics

2 DISCUSSION PAPER SERIES IZA DP No Relational Goods, Monitoring and Non-Pecuniary Compensations in the Nonproit Sector: The Case o the Italian Social Services Michele Mosca Marco Musella Francesco Pastore August 2006 Forschungsinstitut zur Zukunt der Arbeit Institute or the Study o Labor

3 Relational Goods, Monitoring and Non-Pecuniary Compensations in the Nonproit Sector: The Case o the Italian Social Services Michele Mosca University o Naples Federico II Marco Musella University o Naples Federico II Francesco Pastore Seconda Università di Napoli and IZA Bonn Discussion Paper No August 2006 IZA P.O. Box Bonn Germany Phone: Fax: iza@iza.org Any opinions expressed here are those o the author(s) and not those o the institute. Research disseminated by IZA may include views on policy, but the institute itsel takes no institutional policy positions. The Institute or the Study o Labor (IZA) in Bonn is a local and virtual international research center and a place o communication between science, politics and business. IZA is an independent nonproit company supported by Deutsche Post World Net. The center is associated with the University o Bonn and oers a stimulating research environment through its research networks, research support, and visitors and doctoral programs. IZA engages in (i) original and internationally competitive research in all ields o labor economics, (ii) development o policy concepts, and (iii) dissemination o research results and concepts to the interested public. IZA Discussion Papers oten represent preliminary work and are circulated to encourage discussion. Citation o such a paper should account or its provisional character. A revised version may be available directly rom the author.

4 IZA Discussion Paper No August 2006 ABSTRACT Relational Goods, Monitoring and Non-Pecuniary Compensations in the Nonproit Sector: The Case o the Italian Social Services * This paper investigates the nonproit wage gap suggesting a theoretical ramework where, like in Akerlo (1984), eort correlates not only with wages, but also with non-monetary compensations. These take the orm o relational goods and services by-produced in the delivery o particular services. By paying higher non-pecuniary compensations, the nonproit sector attracts intrinsically similarly skilled, but more motivated workers, able to provide in act a higher level o eort than their counterparts in the orproit sector. On an empirical ground, the paper provides a number o econometric tests that conirm the main predictions o the model in Italy s case. It adds to the available empirical literature by introducing in the analysis direct measures o non-pecuniary compensations and job satisaction. JEL Classiication: I00, J31, L31, L84 Keywords: relational goods, job satisaction, wage determination, non-proit organisations, eiciency wages Corresponding author: Francesco Pastore Seconda Università di Napoli Palazzo Melzi Piazza Matteotti Santa Maria Capua Vetere Caserta Italy pastore@unina.it * This research has been conducted within the context o two PRIN-MIUR research projects. Previous versions o this paper have been presented in a number o scientiic meetings at the University o Naples Federico II, at the University o Trento, at the XVI AIEL Conerence. We thank all seminar participants. We would like to thank Carlo Borzaga or stimulating discussions on an earlier version o this paper and or providing us with the ISSAN data. We also thank Maurizio Carpita or discussions on speciic aspects o the paper. All remaining errors remain ours.

5 Introduction Monitoring workers eort in the personal care sector Non-pecuniary compensations and workers eort A theoretical justiication The model Testing the model Data description The evidence on wages and non-monetary compensation The econometric analysis Concluding remarks Reerences Appendix

6 2 Introduction The debate on nonproit organisations has progressively widened rom the issue o their ability to create new employment (see, or a survey, Rose- Ackermann, 1996) to that o their impact on the process o human capital accumulation o long-term unemployed and o the ormation o social capital which is necessary to the well-unctioning o market economies (Menchik and Weisbrod, 1987; and Musella, 2002). In this paper, we address the issue whether nonproit institutions are able to optimise the use o the "human capital" input compared to their state and orproit counterparts operating in similar sectors, via a peculiar structure o incentives. The provision o personal care acilities is the sector where the greatest part o nonproit organisations concentrate (Salamon and Anheier, 1996). As noted in section 1, this sector produces a speciic type o output able to aect the job relationship. In act, output has a multidimensional nature, because it generates relational goods within (among colleagues and with managers) and outside (in the producer/customer relationship) the organisation. Such relational goods provide a sort o non-pecuniary incentive to increase the workers eort. Section 2 aims to explain the nonproit wage dierential in the long run and develops a theoretical ramework where wages are negatively correlated with nonpecuniary compensations. Like in the Akerlo (1984) and the Akerlo and Yellen (1990) git-exchange approach to eiciency wages, the workers eort positively depends not only on wages, but also on the amount o non-pecuniary

7 3 compensations 1. The higher level o non-pecuniary compensations in the nonproit sector explains why it can compete with the state and the orproit sector also paying lower wages: by paying lower wages, in act, the nonproit sector attracts similarly skilled, but intrinsically more motivated workers, who provide a higher level o eort than their counterparts in the orproit sector. Handy and Katz (1998) describe the mechanism through which highly motivated workers sel-select themselves in nonproit organisations where wages are lower. We add that this is possible i the nonproit organisations provide their motivated workorce with higher non-monetary compensations. The policy implication is obvious: iscal incentives should be used to support the production o positive externalities by nonproit organisations. As noted above, in act, nonproit organisations generate a higher degree o relational goods at a lower cost compared to orproit organisations. On an empirical ground, the paper provides evidence (section 3) and various econometric tests (section 4) that conirm the main predictions o the model. It adds to the available empirical literature on the nonproit wage gap by introducing in the analysis direct measures o non-pecuniary compensations based on the ISSAN data bank on Italy s personal care services. In act, the ISSAN data provides answers to questions on job satisaction along 15 dierent dimensions, including the perceived amount o relational goods. The empirical investigation reaches three main conclusions. First, the nonproit sector pays lower wages than its state and orproit counterparts. This is 1 Preston (1989) also assumes that workers in the nonproit sector have a utility unction whose arguments include the social beneit o their work. However, in the utility unction approach it is assumed that the worker s eort is known to the employer. Nonetheless, this assumption is diicult to hold in the social service sector. For this reason, we turn to the eiciency wage approach.

8 4 not due to the lower human capital level o its workorce, as in act the nonproit organisations employ more skilled workers than other organisations. Rather, the nonproit organisations employ a skilled workorce by providing them with higher non-pecuniary compensations, as measured by the level o job satisaction. Second, the negative nonproit wage gap persists also ater controlling or human capital, demographic, regional and occupational variables, which suggests that it is not explained by investment in education as in the mincerian approach. Third, also the positive nonproit dierential in job satisaction persists ater controlling or a set o observed characteristics o workers. 1. Monitoring workers eort in the personal care sector Much evidence suggests that nonproit organisations typically operate in the provision o personal care acilities (Salamon and Anheier, 1996; and or Italy Frisanco and Ranci, 1999; Istat, 1997; 2001; CGM, 1997; IREF, 2000). Following a recent literature (Zamagni 1997; 1999; Gori e Vittadini, 1999; Gui 2000), the main peculiarity o work perormance in this sector can be ound in that it yields a multidimensional output, constituted by the service itsel say helping a disadvantaged person to eed himsel and by the relationship between the operator (as well as the organisation) and the direct beneiciary o the service say the relationship o aection which is necessary to implement this type o service. In other words, while delivering the main produce let call it a relational service the service provider generates also relational goods 2. Thereore, 2 It is important to stress that relational goods cannot be a speciic output.

9 5 relational services can be deined as those services whose production naturally requires the collaboration o those who deliver them and o those who beneit o them; they satisy the users demand or greater well-being or or lower uneasiness; and are delivered in a process, which can be oten appraised only over time. In turn, relational goods are "immaterial goods", which yield utility only i they are shared with others and hence beneit simultaneously not only the customer, but also the operator. More speciically: a) they can not be exclusively consumed by one individual only; b) their production asks or the participation o all those who enjoy it, but the terms o this participation are not negotiable; c) the ruition o them can not be separated by the need and the preerences o others since the relationship with others is constitutive o the consumption action. The peculiar nature o personal services causes three joint consequences. First, output cannot be easily identiied and hence measured. What is or example the output produced by a therapeutic community? Is the number o children welcomed in a childcare acility a satisactory measure o its output? Second, the peculiar link between output and the labour input, which depends on the worker s eort, makes it problematic to identiy the productive technique used. Being a product that is delivered through a peculiar producer/customer relationship, the quantity and quality o the output obtained is strongly dependent not only on the hours worked, but also on the content, intensity and quality o labour 3. In turn, this last depend on the human capital endowment and eort o workers. Oten the provision o personal services requires that the worker be personally involved and available to ind the practical ways to overcome obstacles 3 This does not mean that the endowment o other inputs, such as physical capital, is not important itsel.

10 6 and diiculties. A consequence o the problems now outlined is that it can be hard to ind a satisactory measure o labour productivity, which could be opportunely redeined as X Le, where L e indicates the labour eort. However, labour productivity is necessary to measure the eiciency o supplied services or the ability o their prices to signal their relative scarcity. Third, it is impossible to monitor and encourage the worker s eort according to traditional procedures, used in the case o typical material goods or measurable services. In act, in the case o personal services, the traditional process o monitoring and evaluating risks is not applicable or gives "perverse" conclusions. Think, or instance, o the use o quantitative indicators (number o recovered patients) to measure a hospital or a doctor s perormance. All the above discussion suggests that the very existence o relational goods reveals a case o market ailure. In act, as Zamagni notes (1999, pp ), relational goods are not exchanged according to rules ixed in contracts, but are transerred on the basis o a principle o reciprocity; thereore, material exchange rules can be neither used during their production nor during their consumption, which almost always coincide. The idea that relational goods enter the work relationship can be ound also in the Akerlo s (1984) git exchange hypothesis. Thereore, when analysing the production o services with a high relational nature, the workers eort importantly aect the output produced. More precisely, the quantity and quality o output depend on the ollowing actors. First, the degree o job satisaction: (s)he has to actively, intelligently and voluntarily collaborate to the production o output. Second, the individual result o the work activity oten depends on the eort o the working group. Third, a cooperative

11 7 attitude towards exchanging adequate and correct inormation among workers is necessary. Fourth, the identiication o output should be the same or workers, between operators and managers, and between operators and customers. To explain this point it can be useul to make an example: in some cases, the output o a medical therapy could not be "to cure", as the patient would expect, but to contain the harmul eects o a pathology; and it is important that, in some way, the two interested parts or the n parts, in case, or example, more experts or also the patient s relatives are involved share the same goal.

12 8 2. Non-pecuniary compensations and workers eort 2.1. A theoretical justiication The previous section has shown that in the provision o personal services, where nonproit organisations typically operate, output is not measurable due to the existence o relational goods/services and the act that output typically depends not only on the amount o labour employed, but also on the workers eort, which cannot be monitored. As a consequence, traditional indicators o economic perormance, such as labour productivity, ail their scope and the process o wage determination can be thought o as described in the eiciency wage literature, since the best way or irms o guaranteeing high productivity is via a high remuneration. This essentially implies that the quantity (and quality) o the service delivered depends only on the eort and care o the workorce and that the remuneration aects the productivity level, via increasing the worker s eort. The idea that an eiciency wage mechanism operates in the nonproit sector was irst put to the ore in the seminal work by Hansmann (1980) and Weisbrod (1977; and 1983), who claim that the very existence o nonproit organisations is due to their ability to overcome with their ideological aim the trust problem in markets dominated by the existence o asymmetric inormation in the producer/customer relationship. This section argues that the peculiar nature o the production o personal services can contribute to explain a typical eature o nonproit organisations, namely their tendency to pay lower wages compared to orproit irms (Weisbrod,

13 9 1983; Goddeeris, 1988; Preston, 1989; Frank, 1996; or a more complex picture o the nonproit wage dierential, see Leete, 2001; and Ruhm and Borkoski, 2000) and also to the organisations operating in the state sector. Such a dierence in the wage level is sometimes explained claiming that the orproit sector (FPS) and the government sector (GS) might use wages as an incentive mechanism and a screening device, according to an adverse selection mechanism: by paying higher wages, irms hire a better qualiied workorce that guaranties higher levels o production o goods/services o a higher quality. Conversely, the nonproit sector (NPS) would attract a low qualiied workorce able as such to produce a lower amount o goods/services o low quality. Challenging this view, Hansmann (1980) suggests that, in act, nonproit employers use wages as a negative screening device by oering salaries below those in the FPS. This should deter those highly motivated by monetary concerns rom seeking nonproit employment and attract those or whom love o their work dominates. This idea is ormalised in Handy and Katz (1998) assuming that the reservation wage o skilled workers devoted to the irm aims is lower than that o skilled, but indierent workers. They argue that under this assumption it is convenient or nonproit organisations to pay lower wages to hire the motivated workers with a higher probability. The above theoretical arguments have to ace two main problems. First, in the terms o Handy and Katz (1998) analysis, one should explain why skilled, but motivated workers have lower reservation wages than skilled, but indierent workers. In other words, what do nonproits oer to motivated workers that orproit irms do not? One possibility, explored in this section is that nonproit

14 10 organisations oer higher non-pecuniary remunerations. Second, how to measure workers motivation and eort across organisations? How to measure the degree o airness in the organisation? This issue will be dealt with in the next section. This section also aims to veriy whether a non-proit wage gap existing in the short run can possibly hold also in the long run. The point developed here is that Hansmann s (1980) argument holds in the long run only i intrinsically motivated employees are at least as productive as workers motivated only by monetary concerns. In other words, in order to exist in the long run the nonproit wage gap should be counterbalanced by an opposite gap in the non-monetary remuneration, in terms o greater airness in the irm. It is shown that the Shapiro and Stiglitz (1984) eiciency wage model is not suitable to explain how this can happen and one should appeal to the morale models developed in Akerlo (1984) and Akerlo and Yellen (1990). In act, ollowing the Shapiro and Stiglitz (1984) variant o the eiciency wage approach, the NPS should either equalise their wages to the level o the FPS or disappear, becoming soon less eicient, as predicted by the opponents o the NPS The model The model considers a simple economy producing a homogeneous good/service by an ininite number o orproit and nonproit irms acting in a perectly competitive scenario. This implies that the model is chiely suited to explain a pure inra-sectoral wage dierential, though it can be used also to

15 11 explain inter-sectoral wage dierentials 4. On the demand side a constant population o identical individuals consumes the service delivered. Like in Shapiro and Stiglitz (1984), the worker s eort (e) positively depends only on the remuneration (R i ), which is essentially pecuniary in nature (W i ) as described by equation [1] 5 i ( R ) h( W ) e = g = [1] i i where i = 1,2 =, n, or, in other words, the orproit and the nonproit irm. It is possible to increase the workers eort (case o workers o identical characteristics) or to hire workers endowed with a higher level o qualiication and/or specialisation (case o observationally distinct workers) using the monetary remuneration as a screening device 6. The production unction is given by ( L, e) = [ L h( )] Y =, [2] i W i where Y i is the output produced in sector i, L is the (ixed) number o employed workers. Total output depends on the number o employed workers and on the worker s eort unction, which is, in turn, inluenced by the monetary wage level. Assume irst that labour is homogeneous and that the orproit and the nonproit irms have identical technologies and produce the same service by using a workorce with the same skill level. It is easy to see that a wage premium in avour o the orproit sector ( W > W n ) cannot hold in the long run. In act, orproit organisations will register increasing productive costs and to remain 4 As Leete (2001, p. 163) notes, in some cases a wage dierential may arise in the same statistical sector, but be due to a dierent nature o output: The wage dierential need not be related to the nonproit orm o organisation per se, only to the product produced and, in many cases, the public good content o the product produced. Thus, the nonproit wage dierential estimated or these industries may spring more rom the lack o reinement in our industry classiication scheme than rom anything else. 5 In this paper, we hold the level o physical capital constant across workers and organisations. 6 Notice that the Hansmann s hypothesis that the NPS uses lower wages as a negative screening device does not hold in the Shapiro and Stiglitz ramework, since workers are only attracted by monetary actors.

16 12 competitive, ceteris paribus, they must reduce the price or the service o labour. The equilibrium condition reached in the long run would be given when there is the same level o monetary remuneration in the two sectors ( R = R ). A dierent result is achieved assuming that labour can be either skilled or unskilled. In particular, the sector oering a higher level o monetary remuneration, namely the FPS, possesses a better technology able to distinguish between the workers abilities and will attract workers with a higher level o productivity. The consequence will be that in the long run the NPS will attract only workers with lower skill and also productivity levels. This sector will register then a progressive increase in costs and will vanish, since it will not be as eicient as her orproit counterpart 7. Table 1 summarises the main conclusions achieved until now. [Table 1 about here] However, the above analysis ails to consider that the very nature o nonproit institutions is based on the existence o non-monetary compensations and on the role played by job satisaction on their morale, as proved also by the conspicuous share o voluntary work in the NPS (Weisbrod, 1977; 1983; and Hansmann, 1980). Following this line o reasoning, in what ollows, like in Akerlo (1984) and Akerlo and Yellen (1990), new elements, additional to the monetary remuneration inluence the workers eort and productivity trough the workers morale. In Akerlo and Yellen, non-monetary compensations essentially include the degree o cohesion between the worker and the working group, as well as between the worker and the management. n 7 It is easy to oresee that i eort, which is assumed to be dierent or skilled and unskilled workers, is diicult to monitor the wage level is equalised.

17 13 However, this type o eeling is typical also o orproit irms. In the case o nonproit organisations, ideological actors, such as sharing the irm aims or enjoying the relational goods by-produced in the organisation can also be part o the workers compensation. Analytically, a non non-monetary component (NMR i ) enters the eort unction 8 : i ( R ) = h W NMR ) e = g + [3] i ( i i The total production unction then becomes: i [ L h( W NMR )] Y =, +. [4] i i Equation [4] suggests that the organisation can increase its production level not only by raising the monetary remuneration (as in the previous exercise), but also by taking into account the positive eect o the non-pecuniary component o the remuneration on the worker s satisaction (trough the morale). Assume now that the orproit and nonproit sector employ heterogeneous workers. The dierences in the workorce are not related to skills, but to the intrinsic value that each worker perceives o the higher level o the non-monetary component o the remuneration that the nonproit sector ensures ( NMR < NMR n ). The existence o a wage dierential in avour o the FPS ( W > ) is thereore a necessary condition or this sector to compete with its W n nonproit counterpart. In this ramework, the nonproit wage gap can be considered a mechanism o compensating (monetary) wage dierentials. The 8 The suggested approach is similar to that o Preston (1989). She argues that as nonproit institutions have a non-distribution constraint, managers may have a certain level o discretion on where proits are channelled in the organisation. She assumes that the utility unction o workers incorporates not only wages but also the social beneit generated by the irm. Thereore, all workers who place a positive utility on contributing to social welare are willing to trade-o wages or social beneits on a given indierence curve. Based on this trade-o one can deine a labour donation unction or every individual worker. However, the model analysed here is based on the eiciency wage approach, which is justiied by the nature o output in sectors where nonproit organisations are more numerous.

18 14 remuneration level received by workers joining the FPS can be inerior, greater or equal to that perceived by their non-proit colleagues depending on the impact o > the non-monetary component o the remuneration ( R R ). This implies that: < n > W + NMR Wn + NMR < n [5] where the sign o [5] depends on the relative gap between the monetary and non-monetary remuneration in the two sectors. We expect that the wage premium in avour o the FPS is overcome in the NPS to maintain the same eiciency level by enorcing a air organisational environment able to attract highly qualiied workers that attribute a higher weight to the non-monetary remuneration. Following this result, the existence o a nonproit sector in the long run is ensured by the act that similarly skilled, but ideologically heterogeneous employees choose to join a sector only on the basis o the non-monetary component o the remuneration, i.e. comparing NMR and NMR n. This ramework provides a justiication o the undamental assumption on which the sel-sorting mechanism studied in Handy and Katz (1998) is based, namely that skilled devoted workers have a lower reservation wage compared to skilled indierent workers. This could also be re-stated saying that wages and NMR are negatively related in some individuals, who are available to trade-o wages or NMR. Table 2 summarises the main conclusions achieved until now. [Table 2 about here] Finally, this line o reasoning conirms a conclusion o section one. The nonproit wage gap does not imply lower eiciency levels, but conversely the existence o labour donations whose social externalities the market is unable to

19 15 price. The important policy implication is that the government should support nonproit institutions to produce such externalities by using iscal incentives. 3. Testing the model The remainder o this paper aims to provide evidence o wage dierentials and o the role played by non-pecuniary compensations across organisations. Moreover, the paper studies the determinants o the nonproit wage gap in the context o mincerian earnings equations to econometrically test whether the nonproit wage gap is explained by human capital actors. This section expands on the available empirical literature on the nonproit wage gap (Preston, 1989; Leete, 2000; and 2001) along two dimensions. First, it covers Italy s personal service sector and provides direct measures o non-pecuniary compensations. Second, it provides direct measures o job satisaction to measure the greater degree o airness existing in nonproit organisations. Leete (2000) pointed out the possible role o wage equity (iniquity), as measured by the degree o wage concentration (dispersion), within the organisation/sector in maintaining intrinsic motivation and organizational identiication, relating this to wage setting practices in the nonproit sector. However, there are at least two caveats to using wage dispersion as a measure o airness. First, wage equity is only one marginal aspect o airness in the morale models. Secondly, the assumption that a low degree o wage dispersion is an incentive or workers motivation and eort is hard to reconcile not only with a neoclassical approach (Lazear, 1991), but also with an eiciency wage approach à la Shapiro and Stiglitz (1984) as proven in the

20 16 previous section. Beore showing the results o the analysis, the ollowing section describes the used data Data description The empirical analysis is based on the Survey on Employment in the Social Care and Educational Services conducted by the Istituto di Studi sullo Sviluppo delle Aziende Nonproit (ISSAN) on state, orproit and nonproit organisations operating in the supply o a limited number o personal care acilities: Assistance and guardianship, Nursing/rehabilitation, Educational, Cultural, Recreational, School and school-to-work guidance, Job-search assistance and others (see or urther details Borzaga, 2000). Less than 3-year-old organisations or organisations employing less than three paid workers and with discontinuous activity were excluded rom the universe. The survey was carried out in the irst semester o 1998 in iteen Italian provinces 9, mainly concentrated in the North, where nonproit organisations are more numerous. 724 voluntary workers, 2066 (out o 9226) paid workers, 228 organisations divided in 268 units and 266 managers returned the illed questionnaires. About 61.7% o paid workers in the sample are employed in the NPS, o which 33% are in the lay nonproits, 29.3% are in the GS and 9.0% in the FPS. One o the main advantages o this data set is that it provides direct and detailed inormation on the degree o work motivation and satisaction, which the model outlined in the previous sections considers a crucial actor o the workers 9 From the North to the South, they include: Trento, Gorizia, Pordenone, Trieste, Udine, Venezia, Cuneo, Torino, Brescia, Firenze, Napoli, Salerno, Catanzaro, Reggio Calabria and Messina.

21 17 eort. In act, one o the main aims o the survey was to provide inormation on dierences in wages and work conditions as well as in competitive advantages across organisation types The evidence on wages and non-monetary compensation Table 3 reports average net monthly wages, w 10 m, log o net hourly wages, Ln(w h ), and the degree o wage dispersion, as measured by the standard error, relative to ull-time and part-time workers by organisation type. The hourly wage has been obtained dividing the declared monthly wage by the number o hours due in one month according to the work contract 11. The motley universe o nonproit organisations includes social cooperatives as well as religious and lay institutions. Social cooperatives mainly concentrate in the provision o social services and, similar to religious nonproits, but contrary to lay nonproits they are not allowed to sell out their services to customers. Furthermore, lay and religious organisations dier or the aims they pursue. These dierences within the NPS could clearly aect also the mechanism o wage determination. Thereore, the ollowing analysis groups social cooperatives and religious nonproits (NPS1), distinguishing them rom the lay nonproits (NPS2). Table 3 suggests that on average governmental organisations pay higher monthly wages compared to private irms. The unconditional wage premium o state over proit-seeking irms amounts to 9.6% when considering all workers and 10 Interviewees are asked: Could you please indicate the average net monthly wage you received in the last months (exclusive o extra-work pay, arrays and so on)? 11 Albeit available, the declared actual hours worked are not considered in this study.

22 18 to 12.4% when considering the ull-time only. The comparable igures or the nonproits are 16.2% and 13.1% respectively. Proit-seeking irms pay their parttime employees higher wages than both governmental (13.9%) and nonproits (over 16.6%). [Table 3 about here] The comparison between orproit and nonproit organisations highlights a positive wage premium (6%) in avour o the ormer when all the workers are included into the analysis. This result is much lower than the 18% reported in Preston (1989). The nonproit wage gap almost vanishes when considering the ull-time (0.6%). Similar to Leete (2001), also in the Italian case the nonproit/orproit wage gap in hourly wages is not statistically signiicant or all workers. This essentially depends on the share o part-time workers in nonproit organisations (25.3%), which is almost double that in orproit irms (14.2%) and in governmental institutions (14%) 12. Notice that the unconditional nonproit/orproit monthly wage gap goes up to 10.6% when the analysis excludes the lay organisations. In act, these last pay higher monthly wages (2.4%) than their orproit counterparts and than the rest o nonproits (13.3%). Preston (1989) and Leete (2001) ind large dierences in the wage gap across sectors in the USA, though while the ormer concludes that sectoral dierences explain only a minor part o the nonproit wage gap, the latter argues that the nonproit wage gap is in act mostly sectoral in nature and oten depends on the dierent quality o the product o the nonproits and the orproits (also) within 12 Leete (2001, tab. 2) also reports that the share o part-time is double in the nonproit (16.5%), compared to the orproit (9.8%) sector.

23 19 sectors. To check whether the wage gap in our data is only inter- or also inrasectoral, table 4 shows wage levels across organisations and sectors. In act, the three types o organisation coexist in two sectors only, namely Assistance and guardianship and Nursing/rehabilitation, which suggests that the sectoral component is important. However, within these two sectors the wage distribution by organisation is similar, which suggests that the wage gap has also an inrasectoral dimension. This is prima ace evidence that in our data the nonproit wage gap is partly inter-sectoral (possibly spurious) and partly inra-sectoral (possibly genuine). However, considering that most part o sample observations concentrate in the abovementioned sectors (table 5), the inra-sectoral component o the wage gap is important. [Table 4 about here] The simplest explanation o the inra-sectoral nonproit wage gap would be i the NPS employed less skilled workers. However, inspection o table 5 suggests that this is not the case. Workers in the NPS attained a higher educational level (12 years) than their counterparts in the government (11.5) and orproit (10.8) sector. This conclusion is even stronger when disentangling educational qualiications, since workers who have completed at least high secondary school is much higher in the NPS (59.9%) than in the government (44.7%) or in the FPS (40.6%). The years o work experience and tenure are slightly lower in the nonproit than in the orproit and the government sector, which chiely mirrors the greater job stability and the less recent oundation o governmental institutions.

24 20 What else would then explain the nonproit wage dierential within industries? The theoretical ramework laid down in the previous sections suggests that low wages in the nonproit sector could be used as a negative screening device to attract skilled devoted workers. Accordingly, one would expect that the level o job satisaction o workers in the nonproits be signiicantly higher than that o their counterparts in the other sectors. Figure 1 provides clear evidence in avour o this theoretical prediction, using one o the most valuable eatures o the ISSAN data, the direct inormation on job satisaction. In act, previous studies (Leete, 2000) used wage dispersion as a proxy o the degree o airness and hence satisaction within organisations. However, low wage dispersion could also represent a disincentive or the workers eort (Lazeer, 1991). The histograms in the igure provide direct inormation on the declared satisaction level as measured along 15 dierent dimensions. This qualitative inormation should be taken with caution, as NPS workers could tend to answer positively or ideological reasons: in act, charitable eelings are in the nature o the NPS. However, this is exactly what we are in search or: the presence o ideological dierences across organisations to prove that they can aect the workers eort. The igures suggest that workers in the NPS (4.8) are on average more satisied with their job than workers in the FPS (4.5) and in the GS (4.3). Job satisaction is a multiaceted eeling, generated by the relational goods and services by-produced with the main output. They arise as a consequence o the interaction generated: a) among workers within the organisation and b) with customers. The answers to question 2 (personal and proessional growth

25 21 perspectives), 3 (decisional and unctional autonomy), 9 (working hours), 13 (employee-employer relationship), 14 (inter-employee relationship) and 15 (relationship between paid and voluntary workers) regard type one o relational goods/services; the answers to question 4 (recognition o completed tasks), 5 (variety and creativity), 7 (useulness o the contribution to the service) regard type two o relational goods/services. In both cases, workers in the NPS have a higher level o job satisaction than their counterparts in the government and FPS, with the exception o question 7, where dierences are negligible. Noticeable is also the low level o satisaction or monetary compensation and career advancement in all organisation types 13. [Figure 1 about here] Overall, the evidence provided in this section is in line with the theoretical ramework outlined in the previous sections. Nonproit irms tend to select and hire a skilled workorce by paying low wages, but providing high levels o nonpecuniary compensation, here proxied by the declared level o job satisaction. Considering the eect o job satisaction on the workers eort, this suggests that ceteris paribus nonproit irms are not less eicient than their governmental or proit-seeking counterparts. 13 The discontent or wage compensations mirrors the low level o average wage and low returns to education in the provision o social services documented in the next section.

26 22 4. The econometric analysis The irst econometric test, reported in Table 6, is aimed at veriying whether the within sector unconditional nonproit wage gap documented in the previous section holds also ater controlling or the typical determinants o wages, namely human capital, regional, demographic and occupational controls. The analysis is carried out within the mincerian approach (see, or a survey, Card, 1999). Only the two sectors Assistance and guardianship and Nursing/rehabilitation where governmental and private organisations coexist are considered. Overall, the nonproit wage dierential is very stable, almost insensitive to dierent controls, including human capital. This suggests that other actors but those considered in the human capital investment decisions are important to explain the nonproit wage gap. This indirectly lends support to the theoretical assumption that nonpecuniary compensations are aecting the decision to work in nonproits. This observation applies to both the considered sectors, though in the Nursing/rehabilitation services the gap is statistically not signiicant. An alternative explanation discussed in Preston (1989, p. 449) would be selectivity bias: i any unobserved characteristics (e.g. lack o ambition) is positively correlated with nonproit employment, but negatively correlated with the monetary compensation, selectivity bias may account or a portion o the nonproit wage gap. The tests carried out in Preston reject this hypothesis, which is also against theoretical expectations. In the model o the previous section, in act, a negative correlation between lack o ambition and pecuniary compensation is expected to depend on the positive correlation with the non-pecuniary

27 23 component o the remuneration. This is also conirmed by the high human capital attainment o the workorce in the NPS, which is usually considered a proxy o skill. Unortunately, due to data limitation we cannot test this hypothesis. The wage dierential in avour o governmental and against orproit workers is more sensitive to the inclusion o controls: it goes down by one third when considering only human capital variables and becomes insigniicant when including other controls, suggesting that such a dierential is almost entirely due to observed actors. However, this result does not hold in the case o the Assistance and guardianship services, where orproits do pay lower wages than governmental organisations, even ater controlling or other determinants o wages. [Table 6 about here] The evidence provided in the previous section showed that wages are lower in nonproits, though these employ a more skilled workorce. However, also dierences in the basket o other characteristics, such as demographic, regional, sectoral and occupational dierences (explained component), as well as dierent remuneration methods (unexplained component) might explain the wage gap within the context o mincerian earning unctions in the three sectors. Various tests have been carried out to assess the relative importance o dierences in means and dierences in coeicients. Table 7 gives the returns to education across organisation types and sectors using as dependent variable the natural log o hourly wages. The aim is to see whether the returns to education are actually lower in the NPS than in other organisation types. The igures show that in personal care acilities, the private

28 24 return to a year o education is low (2.9), also compared to the Italian low standards (6.6 or men and 7.7 or women, according to Brunello, Comi e Luciora, 1999), mirroring the low labour productivity in the sector. The returns to education are stable when adding new control variables to the basic earnings unction, based only on absolute and squared potential work experience and tenure, which suggests that in all cases they are genuine returns to this productive actor. They also ollow a similar behaviour across organisation types, suggesting that there are little dierences in the way human capital is valued across organisations. The higher estimated returns to education in the nursing/rehabilitation sector suggests that the higher wages reported in Table 4 are actually also a consequence o the higher compensation or schooling (productivity) there. [Table 7 about here] Table 8 provides the results o augmented earnings unctions by organisation. The human capital actor is represented by educational qualiications, rather than by years o schooling, to test or non-linearity and or dierences across educational attainment levels o the workorce (the so-called sheepskin eect). Overall, the Adj-R 2 is higher in the FPS and in the government sector, suggesting that observed characteristics explain better wages in these sectors than in the NPS, though this could also mirror the lower degree o sectoral and occupational heterogeneity in this sector. The returns to education are higher or the University degree in the FPS and in the GS, but lower or other post compulsory education degrees compared to the NPS. Omitted results 14 o an Oaxaca decomposition 14 The results are omitted or lack o space, but are available rom the authors on request.

29 25 analysis o the wage gap between workers in the NPS and workers in the FPS and GS suggest that the explained component o the wage gap, due to dierences in the basket o characteristics explain on average less than 40 percent o the gap. This suggests, in turn, that most part o the wage gap stems rom dierences in compensation mechanisms, which suggests that other consideration rom the human capital ones should be taken into account. More speciically, dierences in the way o compensating potential work experience reduce the dierential, whereas the higher remuneration o orproit organisations in the sector o Assistance and guardianship increase the gap. [Table 8 about here] A inal test is carried out estimating the determinants o the average level o declared satisaction scored by individuals in their answers to all the relevant questions. The aim is to check whether the dierential in satisaction is explained by observed characteristics or rather it depends on other actors, such as the dierent compensation schemes adopted in the nonproit compared to other sectors. One should look at dierences in the coeicients o the independent variables and at the signiicance level o organisational dummies, as we already noted that dierences in sample characteristics are small across organisations. Few observed characteristics aect the average level o job satisaction as conirmed by the low Adj-R 2. The signiicant variables are similar across organisations. Not surprisingly, higher wages have a positive and a higher number o hours worked have a negative eect on the level o job satisaction. A high level o human capital tends to slightly lower the level o job satisaction, especially in the GS, which mirrors perhaps the low returns to education in the

30 26 personal care acilities as compared to other sectors in the economy. For the rest, work experience, holding a managerial or proessional position, being a teacher or a union member increase job satisaction; whereas being divorced reduces the level o job satisaction. For our purposes, what matters is that the dierences in coeicients are negligible across organisations. Moreover, these small dierences do not cancel out the signiicance level o the dierential in job satisaction across organisations in the estimate in column one. The coeicient o the dummies relative to the organisation type conirm the existence o a signiicant negative relationship between wage levels and job satisaction across organisations, which is one o the main predictions o the theoretical model discussed in the previous sections. [Table 9 about here]

31 27 Concluding remarks This paper argues that a nonproit wage gap can rule only in the short run when the remuneration structure is merely based on the monetary component as in the eiciency wage approach pioneered in Shapiro and Stiglitz (1984). However, by their very nature nonproit organisations concentrate in markets with asymmetric inormation both within the irm, due to the diiculty to measure output and monitor the workers eort, and in the producer/customer relationship. This suggests that non-pecuniary compensations, taking the orm o relational goods by-produced in the supply o relational services, play an important role in the process o wage determination, increasing the irm s output by stimulating the workers eort through the morale (like in Akerlo, 1984; and Akerlo and Yellen, 1990) and generating a negative adverse selection mechanism (Hansmann, 1980; Handy and Katz, 1998). Providing higher non-pecuniary compensations, nonproit organisations attract skilled workers with high intrinsic motivation. Moreover, the paper tests and veriies these theoretical predictions using the ISSAN data on the social service sector. The empirical analysis conirms the theoretical predictions. Nonproit organisations employ a larger share not only o voluntary, but also o part-time workers and pay lower average wages. However, they provide workers with a signiicantly higher degree o job satisaction. Both the wage gap and the dierential in job satisaction remain statistically signiicant also ater controlling or various individual and job characteristics.

32 28 Reerences Akerlo, G. A., 1984, Git exchange and eiciency wage theory: our views, American Economic Review 74 (2), Akerlo, G. A. and J. L. Yellen, 1990, The Fair Wage-Eort Hypothesis and Unemployment, The Quarterly Journal o Economics CV (May), Borzaga, C. 2000, Capitale umano e qualità del lavoro nei servizi sociali. Un analisi comparata tra modelli di gestione (Fivol, Rome). Brunello, G, S. Comi and C. Luciora 2000, The Returns to Education in Italy: a New Look at the Evidence, IZA dp n. 130, March. Card, D., The Casual Eect o Education on Earnings, in Aschenelter, O. C. and D. Card, eds, Handbook o Labor Economics, Vol. III A (North-Holland, Amsterdam) CGM, 1997, Imprenditori sociali: Secondo rapporto sulla cooperazione sociale in Italia (Centro Studi Fondazione Giovanni Agnelli, Torino). Frank, R. H., 1996, What Price the Moral High Ground? Southern Economic Journal 63, July, Frisanco, R. and C. Ranci, 1999, La solidarietà organizzata (Fondazione Italiana del Volontariato, Rome). Goddeeris, J. H., 1988, Compensating Dierentials and Sel-Selection: An Application to Lawyers, Journal o Political Economy 96 (2), Gori, E. and G. Cittadini, 1999, eds., Qualità e valutazione nei servizi di pubblica utilità (ETAS Libri, Milan). Gui, B., 2000, Beyond Transactions: On the Interpersonal Dimension o Economic Reality, Annals o Public and Cooperative Economics 71 (2), Handy, F. and E. Katz, 1998, The wage Dierential between Nonproit Institution and Corporations: Getting More by Paying Less?, Journal o Comparative Economics 26 (2), Hansmann, H., 1980, The Role o Nonproit enterprise, Yale Law Journal 89, Ire, 2000, L impresa civica. VII Rapporto sull associazionismo sociale (Edizioni Lavoro, Rome). Istat, 1997, Le organizzazioni di volontariato in Italia (Istat, Rome). Istat, 2001, Istituzioni Nonproit in Italia (Istat, Rome). Istat, 2006, Le Cooperative Sociali in Italia, 2001 (Istat, Rome). Lazear, E. P., 1991, Labour Economics and the Psychology o Organizations, Journal o Economic Perspectives 5 (2),