How does less public spending affect the motivation of citizens to contribute to nature conservation?

Abstract

Contributions to public goods, such as nature areas, are often made by private actors, as well as governmental agencies. Typically, the motivation of citizens to voluntarily contribute depends on how the distribution of such tasks has evolved between public and private actors, and wider contextual factors. Therefore, it is unclear whether decreased public spending on nature areas affects private funding. We conduct an economic experiment in The Hague Forest in the Netherlands and ask citizens to perform a task that generates funding for the forest. In the treatment condition, we describe the ongoing policy changes, which implies less state funding and greater dependency on private actors. We find that highlighting a greater need for private funding reduces individual contributions significantly, which is mostly due to a drop in motivation of frequent visitors. Overall, our study suggests that decreasing public spending may have unexpected repercussions on citizens’ motivation to contribute.

1. Introduction

Contributions to public goods, such as nature conservation, are often a combination of private and public efforts (Bergstrom et al., 1986; Rege, 2004). Typically, voluntary contributions by citizens depend on the role of the government. In particular, higher public spending may crowd in or crowd out private contributions (Chan et al., 2002; Eckel et al., 2005; Blanco et al., 2012; de Wit and Bekkers, 2017). Crowding out may occur if citizens care about having the public good, and a provision by the government will reduce the individual incentives to contribute. However, a reduction in public spending may also signal that the task is rather unimportant, lowering the willingness of private actors to contribute (Bolton and Katok, 1998). Also, less public spending may lower the motivation of reciprocally minded citizens to contribute (Levi, 1988; Nyborg and Rege, 2003; Tinkelman and Neely, 2018). Hence, the question of to what extent changes in public spending crowd in or crowd out voluntary private contributions is an empirical one.

In the Netherlands, a gradual policy change has been taking place since 2012, shifting funding of natural areas from the national budget to more private contributions. We have used this context for an economic experiment. In The Hague Forest, in the Netherlands, which is a large forest in the centre of an urban area, we test how mentioning a potential decrease in public spending for nature conservation affects voluntary efforts from citizens to generate money for the maintenance of the forest. In our study, visitors of the park (n = 574) are asked to perform a small online task at home. If the task is completed, a pre-specified amount of money will be donated for the maintenance of the forest. In the treatment condition, we mention current policy plans to decrease public spending and highlight the importance of private contributions, which is in line with how the policy is being implemented and covered in the media. In the control treatment, no such remark was made. Our findings show that mentioning a decrease in government spending reduces the probability to complete the task, and consequently decreases the amount of money donated. We find that this loss of motivation can be attributed to frequent visitors, that is citizens who visit the forest more than once a week. In the experiment, we also analyse how contributions depend on the marginal returns to the public good of participating in the task. Perhaps surprisingly, we find that the marginal return interacts with the information treatment. More participants complete the task if the returns of doing so increase, but only if they have not received the message that public funding is under pressure. If the role of the government is emphasized, a higher marginal return does not increase the probability of task completion.

This study contributes to the literature that studies whether a change in public contributions to a public good crowds in or crowds out private contributions. While the conventional assumption is that public spending crowds out private funding (Bergstrom et al., 1986), crowding in may also occur, especially when considering the agency role of public and private actors and their reciprocal relationship (Nyborg and Rege, 2003). This article uses a real-world context to provide empirical evidence on how the motivations of citizens depend on the role of the government.

2. Literature

The provision of public goods through governmental and private means has been studied at length. Early contributions pointed out that private provisions of public goods will be suboptimal from a welfare point of view, requiring governments to step in and increase public good provision (Warr, 1983; Coates, 1996). However, when a government provides public goods, there is a risk that this will crowd out private provision (Roberts, 1984, 1987). Crowding out is defined as the reduction of voluntary private contributions in response to an increase of public spending. Key theoretical contributions show that increased government spending can eventually erode all private funding for public goods (Warr, 1982; Roberts, 1984; Bergstrom et al., 1986). Andreoni (1993) was one of the first who tested experimentally whether government contributions to public goods, funded by lump-sum taxation, will crowd out voluntary contributions. One of the key findings was that crowding out is incomplete, leading Andreoni (1993) to conjecture that people are motivated to give, at least in part, because of a preference for giving, that is impure altruism. While pro-social preferences are important mechanisms to understand crowding out, it has become clear that the interaction between private and public spending is more complex with various mechanisms at place (Bolton and Katok, 1998; Chan et al., 2002; Eckel et al., 2005; Blanco et al., 2012; Rode et al., 2015; de Wit and Bekkers, 2017).

First, it has been suggested that an increase in public spending may lead to more private donations, that is crowding in, as it may signal the importance of the cause, which could encourage private individuals to give more (Bolton and Katok, 1998; Schokkaert, 2006). Conversely, if a government reduces spending, this may signal to private actors that it should not be important to them either. Secondly, reciprocity has been widely observed as an important mechanism to understand voluntary contributions to public goods, where citizens are willing to do their duties provided that others do their share (Sugden, 1984; Nyborg and Rege, 2003). In principle, reciprocal relationships may not only occur between individuals but also between private individuals and the government. The idea that good citizenship is conditional on governments doing their tasks has been termed quasi-voluntary compliance (QVC) by Levi (1988): ‘QVC will occur only if taxpayers have confidence that (1) rulers will keep their bargains and (2) the other constituents keep theirs. A perception of exploitation—that is, an unfair contract—promotes non-compliance. If individuals are not getting the gains they bargained for or if they feel they are being “suckers”, they will try to withdraw from the contract’ (Levi, 1988, p. 52). So, if private contributions are conditional on the implicit agreement that the government also does its fair share, less public spending may actually reduce private spending (Levi, 1988; Nyborg and Rege, 2003; Sokolowski, 2013; Tinkelman and Neely, 2018).

Governments often capitalize on the motivation of citizens to contribute to public goods. Examples here are tax advantages for charitable giving or a mechanism where the government matches a certain amount of privately generated funds with public funds (Brooks, 2000; Fack and Landais, 2010; Huck and Rasul, 2011; Heutel, 2014). While these mechanisms increase the marginal returns of contributing to a public good, they may also invoke reciprocity to encourage contributions and signal that it is a worthy cause. Finally, there is a rich literature pointing to the fragility of individual motivation to contribute to public goods depending on the institutional context (Ostmann, 1998; Cardenas et al., 2000; Rode et al., 2015; Cinner et al., 2021; Graafland and de Bakker, 2021). Generally, crowding out may occur, particularly if the institutional context changes, such as financial incentives, property rights, or regulation (Rode et al., 2015; 2015; Cinner et al., 2021). Crowding out is expected to be particularly strong when existing norms of reciprocity and cooperation are in place (Vollan, 2008; Gurney et al., 2016), and the policy erodes intrinsic motivation (Frey, 1997; Frey and Jegen, 2001; Frey and Stutzer, 2006). Bowles and Polania-Reyes (2012) formalized the interaction between social preferences and the external context (such as incentives) and showed that both crowding out and crowding in may occur.

So, a smaller role for the government may crowd in or crowd out private spending, making this essentially an empirical question. De Wit and Bekkers (2017) conducted a meta-study to understand better to what extent public contributions crowd in or crowd out private contributions. They found that two-thirds of the studies report crowding out (increased governmental contributions reduce private charitable donations), while one-third found crowding in. Overall, based on their meta-study on the crowding-out hypothesis, a $1 increase in governmental support is associated with a $0.17 decrease in charitable donations across all their reviewed studies, depending largely on the context. De Wit and Bekkers (2017) conclude that people in countries with smaller welfare states should be likely to compensate lack of governmental funding than people with extensive government arrangements, also supported by results from Mikołajczak and Bajak (2021). The Netherlands can be categorized as a country with more extensive government arrangements. Hence, the evidence regarding how governmental funding may crowd out private charitable contributions is inconclusive (de Wit & Bekkers, 2017). A mixed picture has also been shown by Tinkelman and Neely (2018) who review 46 empirical studies on private and public contributions to charity in the USA and find support for both crowding in and crowding out. Nikolova (2015) analyses the relationship between public and private spending with panel data from private voluntary organizations that contribute to development projects abroad implemented by the U.S. Agency for International Development. She finds a dome-shaped relationship, where low levels of public spending crowd in private contributions and high levels crowd them out. Sokolowski (2013) studies the interaction between private donations to charity and public contributions to non-profit organizations, finding a mixed picture of crowding in and crowding out. To sum up, a decrease in public spending may crowd in or crowd out voluntary contributions by citizens.

3. Experimental setup

3.1 Study region and context

The Hague Forest is situated amid the densely built city of The Hague and managed by ‘Staatsbosbeheer’, the Dutch government organization for forestry and the management of nature reserves. In the Netherlands, nature has been almost entirely financed by the national government. However, in 2012, the new coalition agreement (Rutte and Samsom, 2012) outlined a revised role for Staatsbosbeheer in the form of greater involvement of private actors and larger reliance on private financial contributions. Hence, Staatsbosbeheer must be less dependent upon state funding and must generate more income through citizens and firms. Later in 2012, the ‘governance agreement on nature’ (Bestuursakkoord natuur) was signed by national and provincial governments. In this agreement, the responsibility for implementing nature policy switched from the national government to the lower level of provinces. In the ‘National nature vision’ (Rijksnatuurvisie), developed in 2014, a further change in the funding structure of nature areas was proposed by the national government, involving a much stronger involvement of citizens and companies. Overall, the Rijksnatuurvisie envisions a role of the government that is less active, both when it comes to management and funding of public areas (Folkert and Boonstra, 2017). Though provinces did contribute more to nature conservation, this did not match the reductions from the national government. The total public contribution, that is the sum of national and provincial contributions to nature conservation, was reduced from more than one billion euros in 2010 to less than 700 million euros in 2014 (Briene et al., 2015). The policy shift goes hand in hand with the expectation that private actors will compensate for the drop of public spending, as articulated in the Rijksnatuurvisie. However, an open question is whether the motivation by citizens to contribute goes up or down if the government takes a step back.

3.2 Research questions and expected results

The purpose of the experiment is to examine the motivation of citizens to contribute to nature conservation. Specifically, the study focuses on exploring two research questions. First, how does citizens’ willingness to contribute effort to generate money for a public good depend on announcing plans from the government to decrease public spending on the same task? Secondly, how does the willingness to contribute effort depend on the marginal return of the contribution?

A reduction in public spending may increase or decrease the motivation of citizens to contribute voluntarily. Theoretically, one can distinguish three cases. First, private contributions would be unaffected by public government spending. This may happen if citizens enjoy giving for its own sake, that is have impure altruistic preferences or the incentives to contribute are absent in the first place. The latter may be the case if the marginal private costs of contributing are smaller than the marginal private benefits. In such case, the contribution to the public good would always be zero, irrespective of what the government does. Secondly, there may be the case where private contributions are crowded out by public government spending. This may happen if citizens care about the final provision of the public good, for example, because of pure altruistic preferences. Thirdly, public spending may crowd in private spending. This may happen if there is, for example, a reciprocal relationship between the government and private individuals (Levi, 1988). While we did not go into the field to test a specific behavioural theory (e.g. reciprocity), our experiment can test whether announcing a reduction in public spending crowds in or crowds out private action.

3.3 Experimental design and procedure

To answer our research questions, we conducted an economic experiment in The Hague Forest. Visitors of the forest were approached with the request to participate in a survey on the perception and use of the forest. Questions were partly socio-demographic and partially concerned with The Hague Forest itself (e.g. ‘how often do you visit the forest?’, ‘what is your prime motivation to visit the forest?’, and ‘how satisfied are you with the maintenance of the forest?’). The visitors were also asked to complete an online survey about their experience in the forest after having visited the forest and received the invitation to participate with a log-in code. If the online survey was completed, a pre-specified amount of money was earned and donated for the maintenance of the forest. We made clear that (i) the money comes from Wageningen University and (ii) will be donated to Staatsbosbeheer for the maintenance of recreational facilities, such as hiking trails and benches.

The invitation to participate in the online survey differed in terms of (i) information provided to participants regarding the role of the government and (ii) the amount of money donated upon task completion. Regarding money, the amount of money varied per treatment, as 0, 1, 10, and 20 euros were donated. So, there were four different treatments in the category ‘money’. In the information treatment arm, participants received an invitation in which the following sentence was added: ‘Government funding for the management of nature areas will be under pressure in the coming years. Therefore, it is important to involve citizens and companies more closely in its financial provision’. In the control arm, no such remark was made; see Supplementary Appendix for design of the invitation. In total, there were 4 × 2 = 8 different treatments, and each participant was given a randomly selected treatment. Our measure for contribution to the public good is whether the task at home was completed or not. Investing time and effort to generate money is certainly not the same as donating money, but we argue that it is a very useful measure for voluntary contribution for several reasons. First, it requires no contribution on the spot, so the respondent is not triggered to give any socially desirable response or serve any experimenter demand effect (Alpizar et al., 2008). Secondly, answering the questions in the online survey will take some time and effort and therefore serves as a good proxy for contribution for the public good. Thirdly, the contribution is only made at home, so the effect of our intervention must last for a long time.

Two sites located at entrances of the forest were selected for the experiment (see Figure 1). The first location (A) is situated at the ‘Boslaan’, near major office buildings and The Hague Central Station. The second location (B) is located on the ‘Bezuidenhoutseweg’ on the other side of the forest, closer to residential areas. These two sites were chosen to provide a broad representation of all visitors to the forest. The team of interviewers consisted of four research assistants from Wageningen University working in teams of two or three. On 14 different days between December 2015 and February 2016, visitors were approached and requested to participate in our study.

The survey comprised 14 questions, of which the first 10 questions were about the visit and the involvement of the respondent in The Hague Forest, and finally, there were four questions asked about the background of the respondent. Finally, the respondent was invited to answer a questionnaire online, which was the experiment. Each respondent could win out of three sums of money as an extra incentive to complete the online questionnaire. These sums were 300, 200, and 100 euros. The treatments were distributed randomly. When people arrived in a group, everyone was interviewed individually to make sure that there was no interaction or spillovers between subjects. In addition, every individual from a group received the same treatment (i.e. the invitation to participate in the online survey) to prevent people from noticing that they were part of an experiment. The text of the invitation was read aloud by the interviewer after the invitation was handed out. In this way, the interviewer ensured that all relevant information has been understood. Respondents got the chance to ask questions if something was unclear. Finally, the interviewees were thanked for their voluntary participation. In the Supplementary Appendix, we show summary statistics and a balance test between the information treatment arm (highlighting the revised role of government) and control arm (no such remark). We find that our treatments are balanced in terms of gender, residency, frequency, and timing of visits (see Supplementary Appendix Tables A1 and A2). We find that the individuals who receive the treatment arm are a bit older (51.0 versus 48.3 years) but we do control for all those observables in our regression framework.

4. Results

We analysed how the completion of the online survey depends on (i) the role of government and (ii) how much money is contributed to The Hague Forest (see Figure 2). We find that the chance of task completion is significantly lower if it was stressed in the invitation that public funding is under pressure [χ² (1,574) = 9,702, p = 0.002]. This suggests that a reduction in government spending crowds out private contributions. Furthermore, we find that the amount of money earned for the forest does not significantly affect the probability to fill in the survey [χ² (3,574) = 2.706, p = 0.439]. This suggests that people are not stimulated primarily by money to contribute to the public good. There is, however, a (weak) interaction between these two effects. Individuals are more inclined to fill in the online survey if money is paid, but not if the role of the government is emphasized [χ² (1,291) = 2.889, p = 0.089]. As a result, the difference in contribution between the treatment and control increases, as the amount of money increased.

These results carry over to a regression analysis that includes various covariates (Table 1). Indeed, we find that money has no significant effect on the likelihood of survey participation. If the respondents have received information that the governmental funding for management and maintenance of nature and landscape is under pressure in the coming years, the number of people that fill in the online survey decreased significantly (Table 1, column 1). This result is robust towards including age and gender variables (Table 1, column 2). Additionally, neither the location within The Hague Forest mattered, nor whether participants live in the Hague, or were interviewed during the week or in the weekend. However, the frequency of forest visits did make a difference. Participants who visit the forest at least once a week (about half of our sample) completed our survey significantly more often. This finding is intuitive because frequent visitors have a stronger intrinsic motivation to generate money for the forest and to participate in our study. When interacting the treatment with whether citizens have been visiting the forest frequently, we find that the lower contribution in the treatment condition can be entirely attributed to the frequent visitors. While frequent visitors are more inclined to fill in the survey, this effect is entirely wiped out if it is stressed that funding is under pressure. Hence, the crowding-out effect is entirely due to frequent visitors (Table 1, column 4). We probe into the robustness of our findings by running the analysis in various specifications, such as (i) adding a dummy variable for no monetary payment and (ii) constructing the ‘money variable’ such that only positive monetary amounts are listed (meaning no payment would not be seen as zero payment). The main results (presented in Table 1) were carried over to those robustness checks in the sense that the (non)-significance of key variables were unaltered and the effects on the size of coefficients were rather minor.

5. Conclusion and discussion

Individuals often feel responsible for maintaining public goods, while the government tends to play a central role too. What happens if the government reduces spending on tasks that have been traditionally in the public domain? Theoretical and empirical findings suggest that a reduction in public spending may crowd in or crowd out private spending, depending on the context (Nyborg and Rege, 2003; Ostrom, 2005; Richter and van Soest, 2012; de Wit and Bekkers, 2017). The key contribution of this article is to test whether announced cuts in public spending affect the motivation of visitors of a popular forest in the Netherlands to contribute to its maintenance. Our study suggests that decreasing public spending may have unexpected repercussions on citizens’ motivation to voluntarily contribute. Our results show that citizens reduce their efforts to complete the task if it is highlighted that public funding is under pressure. This finding contradicts previous literature that finds evidence for (some level of) crowding out and is therefore surprising. Three mechanisms come to mind that could explain such finding. First, highlighting the role of the government may induce an experimenter demand effect. Secondly, citizens see reduced funding as a signal that nature areas are less important than previously thought. Thirdly, citizens are reciprocal and lower their effort in response to reduced efforts by the government. An experimenter demand effect seems unlikely as one would expect a higher, not lower, task completion in the information treatment arm. The outcomes of the second and third mechanisms would give rise to the same patterns and are therefore not distinguishable by looking at the data alone. Since our results are mostly due to frequent visitors, reciprocity seems to be very plausible for two reasons. First, frequent visitors are expected to be more attached to the forest, leading to stronger frustration about the reduced funding, which may invoke a stronger reciprocal response. Secondly, frequent visitors are well informed about the value of the forest and are therefore unlikely to adjust the value downwards after having received information, compared to visitors who visit the forest infrequently. So, while our findings hint at reciprocity to be a very plausible channel, it would be a fruitful avenue for further research to test those mechanisms more directly, perhaps with a laboratory experiment.

We find that the probability increases (weakly) with the amount of money that the task delivers to the public good, but only in the control arm that does not mention the role of the government. In contrast, in the treatment arm that mentions the role of the government, a higher amount of money does not lead to higher participation. This finding suggests that the effects of the return of completing the task and the role of the government are interacting with each other. If the marginal returns of completion and the role of the government were additive arguments in the utility function of the participants, we would expect that the ‘supply function’ of task completion is just shifted down when comparing the treatment and control arms. Instead, we observe a supply function that is flat in the treatment arm. This observation is very much in line with Bowles and Polania-Reyes (2012), who suggest that monetary and social arguments in a utility function are not necessarily separable. While our study provides valuable lessons, our study design also has features that limit generalizability but also give opportunities for further research. First, our experiment is an information intervention, where we inform participants about announced cuts in public spending. While an information provision experiment can effectively research how public provision affects private giving (Horne et al., 2005), it is, of course, not identical with an actual cut in spending. In our case, the information provided was very much in line with what was known at the time about the policy, but it was not so clear how this translated into a reduction in funding for the forest where we carried out the experiment. Secondly, it may very well be that the observed loss in motivation that we observe is only transitory after an initial announcement. Over time, citizens may accept the new role, and the initial loss of motivation is not permanent. Therefore, it would be interesting to study the dynamics to see whether initial crowding out would lead to a new equilibrium in motivation and private contributions. Thirdly, it is an open question whether our results would carry over to the case where the government increases spending. These questions could be tested in a more controlled laboratory setting. Fourthly, we ask participants to complete a task at home that would generate funding for the forest. So, in that sense, it is a different setup than asking individuals to contribute money, which is often done in studies on matching donation to a public good (Brooks, 2000; Huck and Rasul, 2011; Heutel, 2014). In the Netherlands, it is very common to contribute effort and time to generate money for a public good (e.g. by going from house to house to raise money); so, our setup matches that context rather well. Also, a setup where participants had contributed money (or time) on the spot has the drawback as it tends to lead to higher contributions than doing so in private (Alpizar et al., 2008). So, our setup attenuates solicitor or experimenter demand effects, as the completion of the task occurs at a different moment than when participants were approached.

Our findings bear some important policy implications. We find that citizens will not automatically compensate for any spending cuts on nature areas by the government, as suggested by the traditional crowding-out hypothesis. This is bad news for those who hope that less public spending will stimulate higher voluntary contributions by citizens, which may partially compensate such decrease. Even more concerning, this effect is due to citizens who visit the forest most frequently and should have the highest intrinsic motivation to contribute. In fact, the Rijksnatuurvisie specifically mentions the expectation that the most committed citizens are expected to contribute more as the government contributes less. Yet, our findings come from a very specific experimental setup in the Netherlands, so an open question concerns obviously external validity. Our results show that an abrupt deviation from how tasks have been traditionally divided between the public and private domain can lead to unexpected repercussions on motivations, which has also been discussed in contexts. For example, Bouma et al. (2008) show with a field experiment in India that respondents who perceive the decentralization of natural resource management as legitimate are more likely to contribute to a public good than respondents who perceive nature provision as a government task. Overall, our findings suggest that a policy that tries to shift traditionally public tasks to citizens may backfire. Yet, further research is needed to disentangle the role of different mechanisms further and test in which contexts those undesirable losses of motivation may occur.

Supplementary material

Supplementary material is available online at the OUP website. These are the data and replication files and the Online Appendix.

Funding

This work was supported by the Ministry of Economic Affairs of the Netherlands (project BO-11–012-033 ‘Natuur inclusief consumeren en gedragseconomie’—‘Nature-inclusive consumption and behavioural economics’).

Acknowledgements

We are grateful to Robbert-Jan Schaap, Charlotte Verburg, Wiep Wissema and Puck de Jager for valuable research assistance.

References