Introduction

Are private welfare services better than those provided by the public sector? One way to address this question is to ask users of public, for-profit, and nonprofit welfare services to evaluate services. User satisfaction is an important topic as it may influence general support for social policy and thus has importance for welfare state legitimacy. However, as social policy research has so far focused more on cash benefits than services, we have limited knowledge about how, for example, service providers influence service quality and user satisfaction (e.g. Jensen 2011). Thus, we need more knowledge about user satisfaction. In this chapter, we shed light on this important yet little studied topic, based on recently collected survey data from Norway within childcare, education and elderly care (DIFI 2011 2013a, b, 2015a, b; Kumlin et al. 2016). This supplements the book, which, so far, has discussed changes in the regulation and provision of nursing homes and schools at the national level in Denmark , Sweden and Norway as well as municipal-level case studies. We have included kindergartens as a third service in this chapter because of the high proportion of private suppliers in Norway. Substantively, this service is also relevant as it has increasingly been considered part of the educational pathway in Norway, most explicitly manifested in 2005 when the responsibility for kindergartens was moved from the Ministry of Children and Family Affairs to the Ministry of Education and Research.

It is important to distinguish between two feedback mechanisms when it comes to rating services and providers. One is ‘objective’, consisting of so-called ‘hard indicators’ such as resources and outputs. These indicators have traditionally constituted the main, if not the only, sources for monitoring public sector performance (Bouckart and Van de Walle 2003). It is only recently that ‘soft indicators’ measuring user satisfaction have become more important as a means to evaluate public services. This change of focus has gone hand in hand with an increased interest in accountability (Bouckart and Van de Walle 2003). This approach originated in the United States during the 1990s and has since spread to Europe. In fact, the Norwegian Citizen Survey is based on the American Customer Satisfaction Index (ACSI) (Christensen et al. 2011). This shift from hard to soft indicators has provided a new set of tools for evaluating public services (Van Ryzin 2004). However, the accuracy of these tools has been a source of concern (Van De Walle and Ryzin 2011). First, it appears that questions about specific services yield higher overall satisfaction than questions dealing with the general evaluation of public services. Second, studies from the United States suggest huge variation in user satisfaction across services (Miller and Miller 1991). Third, even citizens who have no experience with certain public services appear to have strong opinions about them. Fourth, there is no one-to-one relationship between user satisfaction and the quality of public services (Rolland 2003). In addition to their experiences with the services as such, evaluation is also based on users’ predispositions and expectations (James 2009). Thus, citizens’ satisfaction with a particular service will depend on a long list of factors, and there is no straightforward relationship between service satisfaction and the quality of the services.

In the introduction to this volume, Sivesind and Trætteberg asked whether public, nonprofit or for-profit providers are most conducive to active citizenship. Active citizenship implies user choice, empowerment , and participation understood as the actual ability for users to express views and argue for change. In this chapter, we contribute to the book’s overall topic by discussing user satisfaction. We are mainly concerned with the dimension of active citizenship Sivesind and Trætteberg referred to as choice , which includes two empirical indicators: ‘Promoting a broader range of services where more users obtain services that cater to their interests’ and ‘Formal and real exit opportunities give power to users’. Thus, choice could generate higher user satisfaction through at least two mechanisms: (1) improved service quality and (2) empowerment.

The first mechanism—improved service quality—rests on the idea that a well-functioning market, where users can choose freely between different service providers, will lead to higher satisfaction. Welfare systems combining public financing of services with for-profit or nonprofit service providers are referred to as a ‘quasi-market’ (Le Grand 1997, 151). As will be empirically described below, although public suppliers are dominant in the Norwegian welfare state , nursing homes, kindergartens , and schools resemble quasi-markets to some extent. According to Le Grand (1997, 159), competition could result in efficiency and responsiveness: ‘Schools will be more sensitive to parents, for fear that they will otherwise take their child away—or not apply in the first place—and the school budget will suffer. And they too will have an incentive to be more efficient’.

The other mechanism—empowerment —rests on a great corpus of psychological research that has demonstrated that choice is positively related to satisfaction, even if the outcome is incongruent with previously stated preferences (see e.g. Botti and Iyengar 2004). By introducing the possibility of opting out and choosing a different provider, users may feel empowered as active citizens, a feeling which by itself may lead to increased satisfaction irrespective of actual service improvement.

We draw on two strong data sources to illuminate our research questions: the biennial Norwegian Citizen Survey (Innbyggerundersøkelsen) conducted by the Norwegian Agency for Public Management and eGovernment (DIFI) in 2011, 2013 and 2015, and a survey on social capital and welfare attitudes (SuppA) conducted by the Institute for Social Research (ISF) in 2014 and 2015 (Kumlin et al. 2016). The DIFI data provide comprehensive information about satisfaction with public and private welfare providers and has one major benefit compared with other data sources: It distinguishes between actual users with service experience and respondents evaluating services based on other sources of information. Unfortunately, the DIFI data does not distinguish nonprofit from for-profit providers. To be able to fully discuss the importance of the service provider for user satisfaction, we thus introduce a second data source. The SuppA survey not only differentiates between users and non-users, but also includes questions for distinguishing between nonprofit, for-profit providers, and public providers.

The rest of the chapter is structured as follows. First, we give a short overview of existing research on welfare services, followed by a brief presentation of characteristics of the kindergarten sector, schools, and nursing homes . We then present our data sources and describe characteristics of users of private and public services in terms of social background, education, and income. Next, we analyse results on user satisfaction from two angles: differences in user satisfaction between public and private providers in general based on the DIFI data, and then we further elaborate on user satisfaction between different groups based on the SuppA survey for distinguishing nonprofit from for-profit providers. We discuss the general high-level of user satisfaction with Norwegian welfare services as well as differences between the three welfare services with respect to the room needed to create distinctive services. The dependent variable is user satisfaction, and the predictor we are mainly interested in is who is delivering the service. The last section concludes our discussion.

Welfare Services and User Choice

Privatisation is a disputed issue in Norway . While the previous centre–left government largely resisted privatisation, the conservative government that took office in 2013 stated explicitly in their political manifesto that they ‘will promote private and voluntary initiatives and allow for the participation of a wider range of actors, including in the provision of welfare services’. 1 By focusing on kindergartens, primary and lower secondary schools, and nursing homes, we included three important sectors of the welfare state . As mentioned by Sivesind in Chap. 2, the role of commercial and non-commercial service providers may vary substantially between sectors. While around 50% of Norwegian children attending day care institutions use a private service, some 3% attend a private primary or lower secondary school, and some 10% of nursing home residents live in privately run institutions. The three sectors vary with regard to the room for creating distinctive services, which is relevant for active citizenship . Overall, analysing important welfare services extends the limited existing knowledge on the consequences of welfare providers for user satisfaction and active citizenship.

As described above, in addition to supposedly increased cost-effectiveness, one of the main arguments in favour of the privatisation of public services is that choice and competition enhance service quality. In academic debates, this has been most boldly stated by Julian Le Grand, who argued that if policies are appropriately designed, extending choice and competition among providers will enhance service quality in most areas of public welfare (Le Grand 1997, 2007). Other studies, however, question this claim. A large meta-study on elderly care in the US has, for instance, suggested that public providers in most cases perform better than commercial providers (Comondore et al. 2009). And in 2015, two leading scholars provided a very critical overview of the results of New Public Management reforms in the UK (Hood and Dixon 2015).

Research conducted in Norway and other Nordic countries on the welfare provision of kindergartens , primary and lower secondary schools, and nursing home services is limited and ambiguous. According to Børhaug and Lotsberg (2012, 30), we have little knowledge of the relationship between competition and quality in general, and with regard to kindergartens in particular. Gulbrandsen and Eliassen (2013) found few differences in quality between public and private providers in Norway. Børhaug and Lotsberg (2012, 31), however, referred to some recent studies which showed that parents with children in private kindergartens are more pleased than parents with children in public institutions. There is also little research on the quality of public and private primary and lower secondary schools in Norway . Swedish research (e.g. Hartman 2011; Böhlmark and Lindahl 2012; Skolverket 2012) has provided both positive and negative lessons which, however, are not necessarily directly transferable, as conditions vary (e.g. with respect to school profiles and opportunities to yield a profit). Moreover, as many as 27% of Danish and 14% of Swedish pupils attend a non-public lower secondary school, compared to 3% in Norway (Udir 2015b, 32; see Chap. 2 in this volume and Sivesind 2016 for more information). Research on quality in public and private nursing homes is also scarce and inconclusive, and as Trætteberg and Sivesind (2015, 16) have stated, this literature is mostly concerned with differences between public and private providers and hardly distinguishes between for-profit and nonprofit. Vabø et al. (2013) gave an overview of existing research on elderly care and concluded that there is no clear answer to the question of whether privatisation improves quality. Bogen (2011), Gautun et al. (2013), and Petersen et al. (2014) reached similar conclusions. A large Swedish study of nursing homes summarised the mixed nature in this research field by suggesting that commercial and public suppliers outperform each other on different quality indicators. Commercial suppliers perform better on certain service indicators, such as user participation in the formulation of care plans, while public suppliers perform better on structural quality (hard indicators), such as the number of employees per resident (Stolt, Blomqvist and Winblad 2011). As this short overview of existing research reveals, it is difficult to develop clear expectations on user satisfaction with public and private providers based on findings from the Norwegian and Nordic context.

Scope and Characteristics of Public and Private Kindergartens, Schools and Nursing Homes in Norway

Table 7.1 displays the level of public, nonprofit and for-profit private providers in kindergartens , schools, and nursing homes in Norway. 2 Within childcare, Norway has a substantial level of private providers: 53% of kindergartens are private (Statistics Norway 2015). Unfortunately, this statistic does not distinguish between for-profit and nonprofit providers. Even though kindergarten coverage is now more or less universal, the choice of provider is not an option for everyone as private kindergartens are not available everywhere. Moreover, there is a fixed maximum fee irrespective of the service provider, and public demand for kindergartens may in some areas exceed the supply, forcing parents to choose the first available offer. Nevertheless, when it comes to the room for creating distinctive services, private providers do have flexibility (Børhaug et al. 2011, 183; Børhaug and Lotsberg 2012). There are many minimum requirements, but the provider decides, for instance, priority areas, admission requirements, opening hours, and the number of staff beyond minimum requirements of pedagogical leaders (NOU 2012; Kindergarten Act 2005).

Table 7.1 Share of public, nonprofit and for-profit providers (institutions)
Full size table

Approximately 6% of primary and lower secondary schools are private, and as few as 3% of Norwegian pupils attend a private primary and lower secondary school (Udir 2013, 2015a). Private schools in Norway receiving public subsidies can only be nonprofit. With a few exceptions, they have to offer a ‘religious or pedagogic alternative’ to be approved—the Montessori schools being one example. Of the 208 private schools in Norway in 2014–2015, 74 were approved on a religious basis and 99 as a pedagogic alternative (Udir 2015b, 35). Private schools often have few pupils. Sixty percent of new private schools are established after a public school has been closed, typically in remote areas. However, this does not suggest that this happens very often: only 10% of the closed public schools are replaced by new, private schools (Udir 2013, 3). We find most of the private schools in or around large cities (Udir 2015b, 36). This means that private schools in Norway often have clear alternative profiles, but they also sometimes serve as substitutes for public schools that have been closed.

In contrast to the school sector, living in private or public nursing homes is usually not a conscious choice, as demand is much larger than available institutions can offer. However, in some large cities, Oslo for example, private nursing homes constitute a large share of the total number of institutions, allowing some degree of choice. Nationally, around 90% of nursing homes are run by public providers, while private for-profit and nonprofit homes are responsible for some 5% each (St. Meld 2012–2013, 71). Providers’ scope of action is limited due to tight municipal regulations (see Feltenius’ chapter), but while it is illegal for service providers to distribute profits from primary and lower secondary schools, it is allowed in private kindergartens and nursing home services.

As the literature review in the previous section suggested, it is difficult to establish clear hypotheses concerning variations in user satisfaction between different providers. Nevertheless, as this overview of the present state of the scope and characteristics of kindergartens, schools and nursing homes in Norway has shown, we may propose some expectations. First, based on the observation that real user choice is most evident in the school sector, we expect higher levels of user satisfaction among those intentionally choosing a private (nonprofit) school. This may be the result of quality differences between providers; through well-functioning quasi-markets, private providers have strong incentives to enhance service quality. Alternatively, satisfaction with private alternatives may increase by the act of choosing, which itself can generate feelings of empowerment . Second, as nursing homes are a scarce commodity which barely meets the demands of the public, we expect the least variation in user satisfaction in this sector. The case studies presented in Chaps. 4 and 6 in this volume have shown that there are few differences between nursing homes across the municipalities, implying that variation in user satisfaction should also be limited. Following this logic, we expect the level of variation in satisfaction among users of kindergartens to be located somewhere in between the other two.

Data and Measurement

The data originate from two sources: The first is the Norwegian Citizen survey, consisting of over 30,000 answers from individuals who received the survey in 2010, 2013 and 2015 (response rates were around 42%). The survey was conducted by the Norwegian Agency for Public Management and eGovernment (DIFI) by means of postal questionnaires and with the option to answer online. The survey was introduced in 2010 and was intended to assess citizens’ satisfaction not only with public services, but also included several questions about political participation and attitudes towards politicians. In each of the surveys, respondents were asked whether they had used the services they were asked to evaluate. Respondents answering ‘yes’ to this question received a second (and much shorter) questionnaire that focused on various aspects of that service. These are the datasets used in this study, and the final analysis includes 1998 kindergarten users, 2264 primary and lower secondary school users, and 1622 nursing home users.

The dependent variable in the citizen survey was derived from a single item (here exemplified in the kindergarten survey): ‘Think back on the experiences you have had with your child’s kindergarten . Overall, how satisfied or dissatisfied are you with this kindergarten?’ The items had (in addition to ‘Do not know’) seven response categories ranging from very dissatisfied (−3) to very satisfied (+3). In addition, the three user surveys also asked respondents if the institutions they used were public or private. Hence, the citizen survey does not allow us to differentiate between for-profit and nonprofit welfare institutions. Because of the limited number of private schools and nursing homes, we are left with few users of private institutions in those policy fields (3.3% in primary and lower secondary schools and 4.1% in nursing homes); but when it comes to kindergartens , 40% had their child in a private institution. Thus, the analysis is exploratory in nature.

The second dataset comes from a survey on trust and welfare attitudes carried out in Norway in two waves (2014 and 2015) as part of the project ‘Support for the affluent welfare state ’ (SuppA), financed by the Norwegian Research Council (Kumlin et al. 2016). The data were collected by means of a web survey. Most respondents were recruited from TNS Gallup’s access panel, which is a pre-recruited, web-based pool of respondents who have been randomly sampled from the Norwegian population (recruited by means of telephone). As people with an immigrant background are underrepresented in the panel, additional respondents were recruited to the panel from the National Register. Both samples were stratified with an overrepresentation of respondents from the four largest cities. A total of more than 10,000 interviews were carried out in 2014 and 2015, of which about one half of the respondents answered the survey in both waves. In the data analysed for this chapter, we included all respondents who participated in 2014 (n = 5420) and all newly recruited respondents in 2015 (n = 2161), for a total of 7581 respondents. Response rates from the pre-recruited panel were 50–60% (Kumlin et al. 2016).

The advantage of the SuppA dataset is that it allows us to differentiate between public, for-profit and nonprofit providers of kindergartens and nursing homes, in addition to distinguishing between public and private (nonprofit) providers of primary and lower secondary schools. To the best of our knowledge, this is the first Norwegian survey to differentiate between for-profit and nonprofit providers. The dependent variable relies on a single item: ‘How satisfied or dissatisfied are you with the following services in your municipality/country?’ followed by a list of different services (e.g. a kindergarten operated by a nonprofit organisation or as a family cooperative). Responses were given on a 5-point scale, from ‘Very dissatisfied’ to ‘Very satisfied’ (in addition to ‘Do not know’). The number of respondents reporting experience with the different services varied between 166 (nonprofit kindergartens ) and 1905 (public primary and lower secondary schools).

In both surveys, it is important to emphasise that ‘users’ typically refer to relatives; parents usually answered the survey for kindergartens and schools, while both residents and relatives may have completed the surveys for nursing homes. Thus, the actual users of the two services aimed at children were their parents, while users of nursing homes may have been either residents or their relatives. This warrants caution, as earlier research has shown that relatives are often more critical than actual users (cf. Christensen and Midtbø 2016). However, all respondents had some kind of user experience, either personally or as relatives, which is different from many other surveys where some respondents have had no experience at all.

The two surveys operated with somewhat different structures. In the citizen survey, each respondent only evaluated one service provider within a maximum of five welfare sectors. If respondents had experience with more than one provider, they were instructed to evaluate only one of them based on a fixed criterion (e.g. respondents evaluating kindergartens were asked to only think of experiences with their oldest child). In the SuppA survey, no such criterion was stated, and respondents could evaluate every service provider they had some sort of experience with.

We will first look at the results from the citizen survey. We specify two models for each dependent variable—with and without control variables—but since the two datasets have different structures, the empirical strategy is somewhat different. In the analysis based on the citizen survey, we analysed user satisfaction with the three welfare services, controlling for the provider (public/private). Public welfare institutions are the reference category in the regression. In the analysis based on the SuppA dataset, we analysed user satisfaction independently for each service/provider. In order to evaluate the difference between each provider, we reported 95% confidence intervals for the constant in each regression model.

It is important to mention that differences in user satisfaction can be related to characteristics of those who choose alternatives to public services and not just qualities of the services as such. Still, in the second regression models, we controlled for some indicators associated with the use of different types of institutions. In the analysis, we included two social background variables that may be associated with the use of different service providers or service satisfaction: Education (1 = University/College) and gender (1 = Male). Education is important as an indicator of social resources. One expectation is that the more social resources people have, the more likely they are to take advantage of the possibility to make active user choices . We also included the respondents’ political party choice in the previous parliamentary election (1 = Right-wing parties [Conservative Party/Progress Party] and 1 = Christian People`s Party). The latter was included because nonprofit welfare institutions in Norway can be attached to different religions , ideologies or alternative ways of life. Finally, we controlled for the time trend in service satisfaction using 2010 (citizen survey) and 2014 (SuppA) as reference categories. Since respondents from the four largest cities were overrepresented in the SuppA survey, in the analysis based on this dataset, we also included a control variable indicating whether the respondent lived in one of the four largest cities in Norway (1 = Large city).

Results

Are users of private welfare institutions more satisfied with the services than users of public services? We start by presenting the results from the three citizen surveys (Table 7.2). We present two models—one with only the private/public indicator (Model I) and one with controls (Model II).

Table 7.2 User satisfaction by public or private service providers: Kindergartens , primary and secondary schools, and nursing homes . Unstandardised coefficients (OLS)
Full size table

Starting with the relationship between service satisfaction and service provider, the table illustrates that using a private welfare institution is associated with higher satisfaction (see Model I for all three services). This goes for all three services, but the effect is weak and not statistically distinguishable from zero when it comes to nursing homes. This is in line with the expectation that due to a supply shortage of nursing homes, we would find less variation in user satisfaction in this sector. The two services oriented towards children, kindergartens, and primary and lower secondary schools, have more satisfied users among those using private service providers compared to those using public sector institutions. Also in line with our expectation based on the possibility of real user choice, the difference is biggest among the few respondents in private schools (0.76). Having children in a private school increases satisfaction by 0.80 points on the 7-point scale. The corresponding increase in satisfaction among users of private kindergartens is 0.24 on the 7-point scale. For both kindergartens and primary and lower secondary schools, the coefficients are significant at the 0.01 level.

Adding controls does not change the results substantially (Model II for all three services). The coefficients measuring the type of service provider are more or less identical after adding control variables. Note also that none of the controls turns out to be significant when it comes to satisfaction with kindergartens , but male respondents are less satisfied with schools than female respondents. Regarding schools, it is also worth noting that satisfaction increases over time. Compared to 2010, satisfaction with primary and lower secondary schools among actual users increased by 0.21 on the 7-point scale by 2015. Turning to nursing homes , our controls have substantial influence over service satisfaction. Male respondents, those with a university education, and respondents voting for right-wing parties are significantly less satisfied with the services provided by nursing homes. Respondents voting for the Christian People’s Party are significantly more satisfied with nursing homes , and satisfaction increased over time compared to the first survey conducted in 2010.

In addition, we carried out supplementary analyses (not shown), including variables such as income, type of user experience, and age, and the coefficients for the type of service provider remained unchanged (but the number of missing values increased). It should be noted that general service satisfaction with services related to children (kindergartens and schools) is high compared to services for the elderly (as nursing homes). Still, users with experience with private kindergartens and schools seem to be significantly more satisfied with these institutions compared to users with experience with public kindergartens and schools. It should also be noted that the explanatory power of the different models is weak. For example, Model II (nursing homes) in Table 7.1 explains 3% of the variation in the data.

What does it look like when we include data which allow us to separate between for-profit and nonprofit service providers? Table 7.3 reports regression results based on the SuppA dataset. As in the citizen survey, these results also suggest that the service provider is irrelevant when it comes to nursing homes. The constants reported in columns 7–9 (Model I and Model II) are virtually identical (4.13–4.20). Contrary to the citizen survey and our initial expectations, the SuppA survey does not suggest any differences in satisfaction levels between users of public and private (nonprofit) providers of primary and lower secondary schools, neither in Model I nor in Model II. The level of satisfaction is somewhat higher among users of private schools , but the difference is small and far from statistically significant.

Table 7.3 User satisfaction by public, for-profit and nonprofit service providers: Kindergartens , primary and lower secondary schools, and nursing homes . Unstandardised coefficients (OLS)
Full size table

However, turning to kindergartens, we do find some differences. Looking first at Model I, users of for-profit kindergartens report higher satisfaction than users of both public and nonprofit kindergartens. The difference between users of for-profit kindergartens and users of public/nonprofit kindergartens is 0.22 and 0.37, respectively. Interestingly, users of nonprofit kindergartens report lower levels of satisfaction than users of public kindergartens; however, this difference is not significant. These results suggest that the differences between public and private kindergartens as reported in the citizen survey in reality reflect differences between public and for-profit providers. However, in contrast to the analysis of the citizen survey, these differences do not hold when adding controls (Model II). The constants do not change substantially, but the differences between them become insignificant. This may indicate either that the results are driven by self-selection bias, i.e. that those choosing for-profit kindergartens are different from those choosing public or nonprofit kindergartens, or that the number of observations (n = 573–980) is too small to yield significant differences between public/nonprofit and for-profit providers.

Among the different control variables, we see that supporters of right-wing parties report higher satisfaction with for-profit kindergartens, and that supporters of the Christian People’s Party report as much as a scale point higher satisfaction with nonprofit kindergartens . This is not surprising and suggests that some of the selection bias is related to political ideology. Right-wing parties have traditionally been the strongest proponents of private/commercial providers, while the Christian People’s Party (together with other centrist parties) has traditionally been the strongest proponent of nonprofit providers. Finally, as in the citizen survey, we also see a somewhat higher level of satisfaction with public nursing homes over time (2014–2015).

Based on our empirical analysis of these two data sources, the overall findings are first that the provider is unrelated to user satisfaction with nursing homes; and second, that user satisfaction is higher among users of for-profit kindergartens than among users of public and nonprofit kindergartens. Third, the results are more mixed when it comes to satisfaction with primary and lower secondary schools. In both surveys, users of private schools report higher satisfaction than users of public schools; however, in the SuppA survey, this difference is small and insignificant. This may be explained by selection bias or too few observations in the SuppA dataset. However, it may also be explained by the methodological differences between the two surveys described above. In contrast to the citizen survey, in the SuppA survey, users were not restricted to evaluating only one service provider. We cannot rule out the possibility that variations are reduced when respondents evaluate more than one service provider in a single question battery.

Concluding Discussion

One of the core ideas behind offering users alternative suppliers of welfare services is to improve service quality. The argument is that quasi-markets combined with user choice empower citizens, increase participation , adapt services to the individuals actually using them, and improve service delivery. With respect to these assumptions, our results do not show a substantial for-profit or nonprofit premium over public services. Rather, the results suggest a mixed picture, where there is a weak tendency for users of for-profit kindergartens and private (nonprofit) primary and lower secondary schools to be somewhat more satisfied than users of public providers, but there was no difference when it came to nursing homes. These results must be interpreted against the backdrop of the generally high level of satisfaction with Norwegians welfare services.

The main reason why differences in user satisfaction between service providers are small in Norway is simply that the overall level of satisfaction with public welfare services is high across services and providers. The majority of respondents place themselves on the positive side of the satisfaction scale irrespective of the type of welfare services rated (see also Christensen and Midtbø 2016). For instance, in the SuppA survey, satisfaction levels for the different services and providers examined here are all close to 4 on a 5-point scale. Consequently, we are analysing variations between generally satisfied users of welfare services. In effect, the significant differences in satisfaction levels we found between service providers are indeed relatively small—the largest being 0.7 on a 7-point scale (primary and lower secondary schools in the citizen survey).

Looking more substantially at the variations we do find in our data, different mechanisms may be more or less relevant in explaining differences between welfare services. Regarding primary and lower secondary schools, private alternatives are scarce, and since such alternatives can only be approved if they provide either a religious or a pedagogic alternative, intentionally choosing either of them is a very active choice by parents compared to sending their children to a public school. It is easy to imagine, as the analysis of the data from the citizen survey indicates, that such an active choice is justified in terms of high user satisfaction. Added to this, several alternative pedagogical schools in Norway have recently been established based on parent initiatives to compensate for the closing of a public school. The latter should also result in satisfied users of such schools. As such, the observed differences in user satisfaction between private and public schools may be the result of both well-functioning quasi-markets where private initiatives improve service quality (cf. Le Grand 1997) and also the act of choosing. By being empowered to choose, some citizens may respond with higher user satisfaction. This latter mechanism cannot be ruled out because our data are vulnerable to self-selection: users were not randomly assigned to different schools, and those choosing a private alternative are different from those choosing the standard public option. Nevertheless, we only found significant differences in satisfaction levels in one of the datasets.

Kindergartens are a somewhat different story. Private providers are widespread and there is little regulation of the field, with the exception of some minimum requirements for staff. Moreover, as the overall supply of kindergartens barely meets public demands, parents rarely have the opportunity to make an intentional choice between different suppliers. As such, the differences we saw between public and for-profit kindergartens are more likely to be explained by differences in service quality. Although real choice is limited in this sector, the observed differences may be the effect of quasi-markets where different suppliers have incentives to improve quality (cf. Le Grand 1997, 2007). Again, we are hesitant about these interpretations. This is partly because the observed differences are very small: 0.2–0.4 points on either a 5- or 7-point scale, and partly because in the SuppA dataset, the differences are no longer statistically significant when including control variables. A cautious conclusion is therefore that users of for-profit kindergartens are somewhat more satisfied than users of public and nonprofit kindergartens, but that this difference is very small. This is in line with some previous studies (Børhaug and Lotsberg 2012, 31).

Finally, regarding nursing homes , the lack of variation in user satisfaction supports the findings in the case study by Feltenius (see Chap. 4). As with kindergartens, the supply of nursing homes is scarce; but in contrast to kindergartens, municipal regulation of nursing homes is tight. Through standardised quality indicators and contracts, different providers have limited options to develop distinct profiles, and users receive more or less the same service irrespective of provider. In sum, the main finding in our data is that users of the three sectors generally report a high level of satisfaction, producing limited variations between different service providers. We also found indications that users of private kindergartens and schools are somewhat more satisfied with the services compared to users of public services. This finding does not apply to nursing homes. One possible explanation for the variations between services is that real user choice is most present in the school sector and least so in the nursing home sector. Distinguishing further between for-profit and nonprofit services, we still found hardly any differences between different types of nursing homes. With regard to kindergartens, we did, however, find that users of for-profit options are somewhat more satisfied than users of nonprofit options.

In other words, based on our data and other studies from Norway (Christensen and Midtbø 2016), one should be reluctant to use evidence based on soft indicators of service satisfaction as an argument in debates about privatisation. This is partly due to general methodological concerns (cf. Van De Walle and Ryzin 2011; James 2009). The main reason, however, is that the substantial differences in user satisfaction are very small in Norway . As opposed to evidence from the US (Miller and Miller 1991), the vast majority of users are satisfied irrespective of supplier, making user satisfaction a quality indicator with limited value (see also Christensen and Midtbø 2016). This limitation is further underscored by the limited degree of freedom suppliers have in tailoring services. State regulations and user rights often limit the possibility of private as well as public suppliers to offer qualitatively different services. If soft indicators are to have any value in public debates on privatisation, future studies need to develop more accurate indicators that will measure relevant dimensions in a context with limited overall variation at the outset.

To conclude, user choice may result in increased user satisfaction. However, as our analysis shows, Norwegian welfare services already enjoy a high-level of user satisfaction. User choice may thus have more importance for active citizenship by empowering citizens. This quality, though, may have other challenging implications, as in the classical trade-off between equality and freedom to choose. Freedom to choose may increase social inequalities because resourceful citizens are better informed and take advantage of available opportunities to a higher extent than less resourceful citizens. How to balance such values is thus an important political question for a welfare state that increasingly seems to emphasise services over cash benefits.

Notes

  1. 1.

    “Political platform for a government formed by the Conservative Party and the Progress Party”. Sundvolden, 7 October 2013. Retrieved from https://www.regjeringen.no/contentassets/a93b067d9b604c5a82bd3b5590096f74/politisk_platform_eng.pdf.

  2. 2.

    A better statistical measure of the size of each provider/sector would be the number of man-years. Unfortunately, we have not been able to retrieve these numbers from Statistics Norway.