How Effective Are Civil Society Promotion Laws? Mixed Patterns from 3 Mexican States

1 Introduction

The question of nonprofit and civil society organizational founding is a central theme in the field of nonprofit studies and has been addressed in multiple settings in the past 30 years. Different approaches have emerged to deal with this question. The formation of nonprofit organizations has been addressed using the three-failure framework (either government, contract or market failure), and looking at supply and demand factors that determine nonprofit service provision (Grønbjerg and Paarlberg 2001; Smith and Grønbjerg 2006). Socio-economic conditions and a focus on poverty have been central concerns in the studies of nonprofit creation (Marcelli and Wolch 2003; Peck 2008; Polson 2017; Yan, Guo, and Paarlberg 2014). Consequently, the subsectors that have consistently attracted most scholarly attention are social and human services, health and education (Corbin 1999; Polson 2017; Van Puyvelde and Brown 2016; Yan, Guo, and Paarlberg 2014). Factors such as household income, education, fiscal support from government, age groups or socio-demographic diversity rank high among the nonprofit sector drivers in this literature.

Another branch of the literature has noted these same factors, but has placed them in a slightly different theoretical framework; one that is underpinned by organizational ecology. From this perspective, the socio-economic needs and characteristics of a community are variables in the environment, that lead to the founding and disbanding –or ‘birth’ and ‘death’ –of nonprofits. This approach is rooted in an open-systems view of organizations, that considers that they are in a constant exchange with their environment and adapt to the constraints around them, in order to survive (Hannan and Freeman 1989; Scott and Davis 2015).

Beyond the variables conventionally included in the study of nonprofit founding rates, policy decisions offer the opportunity to take a different approach to understand specific dynamics of organizational formation. This paper exploits the enactment of civil society promotion laws to examine the causal effect of such laws on the number of newly registered nonprofit/civil society organizations.

In particular, policies about the legal and regulatory frameworks that allow or restrict nonprofit organizations and their activities are relevant to consider here. In many places, CSOs are considered a disruptive actor, supposedly meant to destabilize local society and government. Because of this, multiple countries have passed laws to ban or severely restrain the work of CSOs and slash their ties with foreign partners (Christensen and Weinstein 2013; DeMattee 2019). At the same time, other countries have been welcoming to the dynamics of local civil society, enacting laws and regulations intended to promote it. While policies constraining nonprofit and civil society activity have been studied more in depth and have been the focus of recent concern, the effect of policies intended to promote the civil society sector, whenever they have emerged, is less clear. What happens when laws are enacted to promote these organizations? How effective are these laws?

In the last decade, a branch of literature has taken momentum as it has recognized the significance of the legal and regulatory framework in influencing the nonprofit sector (Breen et al. 2017; Gugerty 2010). Most of the studies in this area are wary about the impact created by authoritarian regimes, where governments want to stifle the sector through laws and regulations that constrain nonprofits and limit or heavily control their activities (DeMattee 2019; Toepler et al. 2020). On the other hand, countries like Mexico evolved in their stance to nonprofits and civil society and, even if for a limited time period, they seemed rather welcoming and facilitated conditions, at least formally, for civil society actors to thrive.^[1]

This paper adopts a quasi-experimental approach to the question of new nonprofit legal registration as an effect of a law passed intended to promote the sector. It studies the case of Mexico, evaluating the effect that a number of CSO promotion laws at the state level may have had on the creation of new CSOs, adjusted for population, in the period from 1990 to 2018.

The paper builds on scholarship on organizational ecology and nonprofit sector studies, and combines key insights from these research areas with an application of the synthetic control method (Abadie et al. 2015; Abadie 2021). The analysis will take us to questions of the policy process; in particular, the implementation of the law. The paper examines the effect of a civil-society promotion law in 3 Mexican states, as a different law has been passed in each of these states, in different years, since the early 2000s. The results of the 3 states investigated in this article point in fairly different directions, in terms of the causal effect of the law. These findings are worthy to explore because they emphasize the importance of a more nuanced analysis of the public policy process around CSOs, and the need to take into account larger contextual factors that exceed the scope of this paper. The analysis here is meant to prompt interest in the further dynamics and qualitative aspects of organizational decision-making. It also raises questions related to the causal mechanism at play, such as the incentives and disincentives perceived by the local actors, and the factors that make some laws more successful than others.

This paper is divided into 5 sections, in addition to this introduction. The second section presents the background to understand the case of the CSO promotion laws in Mexico. The third section introduces the data and method used in the paper. The fourth section shows the results of the synthetic control model, for each of the treated states separately, along with the standard robustness checks applicable to this model. The fifth section explores key aspects of the policy process surrounding the CSO laws as well as some areas for future research. The sixth section offers some concluding remarks.

2 Mexico – The Civil Society Landscape and Its Legal and Regulatory Framework

Mexico is a country with a long history and a valuable record of civil society activity spanning several decades. Since the early years of the Johns Hopkins Comparative Nonprofit Sector Project, Mexico was included and it proved to be a major case study, outside of the Anglo-Saxon tradition of this scholarship. The case of Mexico provides a different angle to understand how the institutional drivers shape civil society and how the sector is embedded in broader, long-term social and economic forces – what has been elaborated as the Social Origins approach, first laid out by Salamon and Anheier (1998). The Social Origins Theory has been applied to multiple countries over the last two decades. Most recently, it has been captured in the work of Salamon et al. (2017), and, specifically, a chapter on Mexico that describes the country as having a ‘persistent statist pattern’ where the PRI, the ruling political party for most of the 20th century, heavily controlled civil society (Villalobos et al. 2017). While the present paper will offer an analysis of a legal reform and the effect on the organizational ecosystem at the state level, it can be seen as complementing the major contribution that Social Origins Theory has made to our understanding of the development of civil society.^[2]

3 Data and Methods

As indicated above, the data used in this paper come from a Mexican official, federal registry of nonprofit, civil society organizations. This Federal Registry serves as the sampling frame for this exercise. The Center for Research and Studies on Civil Society (CIESC for its acronym in Spanish) and the International Center for Not-for-Profit Law (ICNL) take the Federal Registry as accurate and reliable. The dependent variable in this study is the number of newly-constituted organizations in each of the Mexican states (or ‘federal entities’) for each of the years covering the period 1990–2018. The Federal Registry uses a unique identifier called the CLUNI, and it only started in 2005. However, the Registry also includes a variable called “date of constitution”, which goes back to the early decades of the 20th century and is the key variable in this paper.

Appendix A2 provides the definition and source of the variables used in the paper. Appendix A3 shows the summary statistics of the data used. I use annual state-level panel data on 16 Mexican states for the period 1990–2018. I conduct 3 separate synthetic control exercises, one for each of the 3 ‘early-adopters’, that passed a CSO promotion law.

Since the laws were enacted in different years, the number of periods both in the pre- and post-treatment stages differs slightly among them. Appendix A4 shows in a map the Mexican states under study, both the treated states (in green) and the pool of control states (in orange). States where a CSO law was passed after 2007 lack a sufficient number of post-treatment years; because of this, those states are not included in this paper.

The effect of the enactment of the laws is assessed bearing in mind some control variables that are associated with the creation of new CSOs in a given geographic area. Previous research in the field of nonprofit studies has shed light on them. Different measures of availability of resources, both public and private, have been found to be significant to the creation of CSOs and nonprofits that serve different social needs (Da Costa 2016; Grønbjerg and Paarlberg 2001; Lecy and van Slyke 2013; Lu and Dong 2018; Marcelli and Wolch 2003). A demand-driven approach has also been explored to understand the growth of CSOs and nonprofits. This approach has examined different measures of population heterogeneity, including those based on age, race, and the presence of different minority groups, which may have an effect on the service delivery and orientation of work by the organizations in a given area. Yet, not all the measures of heterogeneity have an effect in the same direction (Kim 2015; Lu and Dong 2018; Van Puyvelde and Brown 2016). Saxton and Benson (2005) looked at measures of social capital and political engagement as another source of variation in CSO sector size. This can be construed as another type of population heterogeneity that may drive CSO creation.

Based on previous findings in the literature, a measure of income (per capita GDP), public expenditure (ratio to GDP), average schooling years, and a measure of political leanings have been included in the synthetic control modelling as control variables. The ratio of public investment to GDP remains at very low levels, similarly to other countries (Miyamoto et al. 2020). However, there is enough variation among states and over the period covered to make the variable useful. The political leanings, and a proxy for the civic orientation, are captured in the PRI’s (Partido Revolucionario Institucional) vote share in presidential elections. The strong presence of PRI was a major feature of the political culture in Mexico during the 20th century, but in the last couple decades it has faded away significantly, and people’s preferences have veered toward other political groups and orientations. The waning influence of PRI is a major factor in the evolving environment of the State-civil society relations in Mexico.

4 Synthetic Control Analysis – Effect of State-Level CSO Laws

Because of the nature and timing of the policy intervention examined in this paper, the synthetic control method is probably the most suitable tool to estimate the causal effect that a state-level CSO law may have had on the number of newly-constituted CSOs (per 10,000 residents) in a Mexican state, in the last few years. Also, as a tool to support the implementation of comparative case studies, the synthetic control method provides a first step to explore the impact that a policy decision like the enactment of a CSO law may have on the nonprofit ecosystem, and it can complement research on related topics that are more qualitative in nature (Abadie et al. 2015). All the code was written and run using Stata 16. Like other studies that use the synthetic control method, I used the synth routine to get at the standard calculations that allow the user to determine the ‘goodness of fit’ of the synthetic unit, and provide evidence to support or reject the existence of a causal effect of the intervention at hand. I used complementary routines to conduct the placebo tests to check the robustness of the causal effect modelling.

An analysis run on the seven Mexican states that implemented a CSO law found that the results for four of them did not pass the robustness tests that are standard for this type of model, indicating that the synthetic control did not provide a credible causal effect of the law under study. Because of this, these four states are not included in this paper. This is, in fact, a possible and reasonable outcome, according to Abadie (2021), who warns that there might be some units and interventions for which no appropriate synthetic control exists: “There may not exist a combination of untreated units that provide a credible approximation to the treated units, and the conventional synthetic control estimator should not be used in that case.” (Abadie 2021; p. 412).

Table 3 shows the balance of covariates in the pre-treatment period for each of the 3 states and the corresponding synthetic unit. Table 4 shows the weights that the 13 non-treated states were given as a result of the numeric algorithm in the construction of the synthetic control for each of the 3 treated states.

Figure 1 shows the outcome variable, the number of newly-constituted civil society organizations, per 10,000 residents, in each of the 3 states, and the comparison between each treated state and the corresponding synthetic-control state, both in the pre- and post-treatment periods. The gap between the treated and synthetic state in the post-treatment period provides a measure of the causal effect of the state CSO law. Figure 2 shows the gap in the prediction error of the outcome variable, between the treated and synthetic state.

Figure 1:

Outcome trends for treated and synthetic states.

Figure 2:

Gap in prediction error, between treated and synthetic unit.

The inverted-U shape in the trend of the outcome variable (number of new organizations per 10,000 residents) is a familiar feature in the organizational ecology literature, and it has been attributed to the saturation that comes as a consequence of the density-dependence effect when the carrying capacity of the organizational ecosystem reaches its peak (Carroll and Hannan 1989; Carroll and Khessina 2019; Hannan and Freeman 1989). The density-dependence effect reveals the impact of the existing CSOs on the creation of new ones, as the resource base becomes increasingly limited for new entrants and new opportunities for successful organizing efforts decrease.

As Figure 1 shows, the effects of the state CSO laws are quite varied, and in all cases, after a temporary change, the outcome variable returns to its pre-law levels. These results can be explained by the increase in the resource base and the shifting political preferences that likely were accompanied by a more open environment for new social causes. A more diversified political climate is arguably aligned with an increased willingness to formally register new organizations. In all cases, the synthetic state exhibits an increase in the outcome variable, showing that even without a CSO law an increase in the number of CSOs registered would have ensued. The CSO law, then, is an additional force that can amplify, dampen or remain neutral to the effect created by the increased resources (broadly understood) in the organizational ecosystem. Out of the seven states that had introduced a CSO law by 2007, the three states included in this paper are the ones with the most robust results, showing either a positive, negative or virtually neutral effect that the law had, compared against the synthetic-control state. Baja California exhibits a positive effect, while the effect in Zacatecas is very much insignificant, with the treated and synthetic units showing a common, overlapping trend for the entire post-treatment period. The case in Tamaulipas is quite interesting, as the law seems to have had a significantly negative effect when compared to the synthetic counterfactual.

As mentioned above, we observe that the post-treatment trend of the outcome variable exhibits an inverted-U shape. In the first years after the law was passed, the number of newly-constituted CSOs increased, but then it reached a peak and started a descending movement. This pattern makes sense if we bear in mind that the set of CSOs in each state can be treated as an organizational population that evolves according to the availability of resources in the environment and conditions that are conducive for them to grow or perish. Scholars of organizational ecology have long studied these phenomena (Carroll and Khessina 2019; Hannan and Freeman 1977, 1989; Stinchcombe 1965). Applications of this ecological tenet to nonprofit studies have translated this principle into the idea that the rate of creation of new organizations faces a marginally decreasing effect from the existing size of the population (Lecy and Van Slyke 2013; Lomi 1995; Saxton and Benson 2005; Yu 2016).

Recent studies continue to show the relevance of organizational ecology for understanding the landscape of different fields, and its significance for a number of policy areas. Abbott et al. (2016) and Lake (2021) have studied the emerging organizational forms of global governance and its changing patterns, following the principles of organizational ecology. Eilstrup-Sangiovanni (2020) has employed a similar approach to the study of international organizations and the factors surrounding the ‘death’ of many of them, on a long-term framework.

This feature of organizational ecology studies helps us understand why the effect of the CSO laws is short-lived and the outcome variable at the end of the post-treatment period under analysis returns to levels that are quite close to the average of the pre-treatment period. However, this result opens some questions about the process that unfolds during the period of increased activity, and particular aspects of organizational performance and dynamics that are triggered.

4.1 Robustness Checks–Placebo Tests

Following the standard practice in synthetic control studies, I conducted a placebo test for each of the treated states, to check the robustness of the results. Figure 3 shows the placebo tests. They display a variation in the direction and significance of the effect. Zacatecas has the least variation in gap, when compared to the placebo runs for the states in the ‘donor pool’. This conforms to the finding that this state exhibits an insignificant effect of the CSO law.^[5] In the case of Tamaulipas, the placebo test shows that the law in this state has a negative effect, and the gap in the predicted error moves toward the negative range, implying that a similar intervention in almost any other ‘donor pool’ state would have had a more favorable effect on CSO creation than in this state.

Figure 3:

Placebo test for each treated state and the control pool.

4.2 Robustness Checks–RMSPE Ratio Tests

Complementing the placebo tests, I run the tests of the post/pre-treatment root mean square prediction error (RMSPE) ratio. This ratio serves as a t-statistic that tells us how good the fit is for the post-treatment against the pre-treatment period, when we run a synthetic control analysis for each of the states (treated and those in the ‘donor pool’). As a measure of the goodness of fit for the treated unit, the post/pre ratio should be the largest (or among the largest) so that when we take the ratio of the rank of the treated unit against the total number of placebo tests, we get something similar to a sample-based p-value of 0.05 or lower. Conversely, a lower post/pre ratio for a given state signals an intervention that has an insignificant effect. Table 5 presents the post/pre RMSPE ratios for the 3 treated states, and Figure 4 depicts in a histogram the ratio for each treated state and the control pool. The red bar in each graph represents the treated state (the value may coincide with that of another state, so the frequency can be higher than 1). Consistent with the main results of the synth routine, Tamaulipas exhibits the highest ratio, confirming its significance (albeit negative); whereas Zacatecas displays the lowest ratio, as the CSO law did not have a significant effect there, and the actual and synthetic state very much follow the same path during the entire period of study, regardless of the law passed. Baja California gets a positive ratio, although it is not the largest in its group. This suggests that the robustness tests are more clear in the case of the null and negative effects (Zacatecas and Tamaulipas) than in the case of the positive effect of the law (Baja California).

Table 5:

Post/pre-treatment RMSPE ratio for the treated states.

State	Post/pre-treatment RMSPE ratio
Baja California	6.9931
Zacatecas	2.3918
Tamaulipas	15.0265

Figure 4:

Histograms of post/pre-treatment RMSPE ratios.

5 Disentangling the Causal Mechanism – The Policy Process Around the CSO Laws

The results above offer evidence of the causal effect of a state-level CSO law passed in the early and mid-2000s in each of three Mexican states. The three states included in this paper are those which most closely satisfy the robustness tests that are standard for this type of model. However, the direction of the effect exhibits a variation among them and opens some interesting questions about the process and the causal mechanism that drives these results. Even though the effects in the three states follow different patterns, they all share an inverted-U shape, and density-dependence effects that are typically found in studies of organizational ecology.

In the case of Mexico, an increased resource base and the consolidation of a more diversified political opinion during the period under study are likely behind the rise in the number of organizations, which is observed “by default” in all the scenarios. The CSO law acts as an additional factor that may reinforce, keep unaltered or detract from the effect of the organizational ecosystem. In any case, its effect is bound to keep in line with the underlying carrying capacity of the ecosystem, and eventually the annual creation of new CSOs (controlling for population) will return to its previous levels. Yet, the dynamics that is unleashed in the interim around the implementation of the law is one of the remarkable features of the process, and is worth exploring further.

Some scholars have examined the local conditions in Mexico that may help explain the performance of the CSO laws and the variability in results across states. They have looked at major aspects of the policy process that surrounded the enactment of the law and other societal factors. The policy process around the law needs a group of champions that enjoy wide representativeness and deep involvement in CSO-related debates, and who can deploy a public opinion strategy that persuade citizens. They also need to identify potential allies within the state government. Yet, since public officers tend to know fairly little about the CSO sector, its scope and role in public policy, it is critical that the discussion of the law have a strong pedagogical element that include research and ‘grey’ reports, accessible to public officers at all levels (Becker et al. 2016; Vargas González 2012).

An important aspect of the discussion of the laws is how the CSOs take part in various deliberation arenas, which are usually led by the government. This points to the representativeness of various factions within the CSO ‘sector’. García et al. (2010) have noted: “Members or groups who promote CSO laws have to bear in mind that inside the sector there are serious problems (…) In this context, rife with conflict, we have to understand the issue of representativeness and the prevailing criteria that get certain CSOs to participate in deliberative arenas. Among these criteria are the expertise about CSOs, the years of experience, public salience and institutional transparency” (Translation from García et al. 2010, p. 32).

The laws stipulate certain rights and obligations for CSOs, that act as incentives for civil society leaders to register the organizations and operate as formal bodies, or disincentives that discourage them from doing so. According to García et al. (2010): “The obligations aim to acknowledge and incentivize the “formal” or “legal” character of CSOs, to create mechanisms to promote transparency and accountability and to guarantee the professionalization of CSOs.” (Translation from García et al. 2010, p. 56) This would explain the regulations to demand records, reports and the requirement to provide training oriented towards the professionalization of the sector. Predictably, these stipulations are all the more costly and hard to implement in poorer states. In some cases, state-level governments tend to control excessively civic organizations, rather than facilitate free association.

Analysis by the CIESC (2023) has acknowledged the challenges of the implementation of the state laws: “ (…) the adequacy of their implementation depends on the one hand, on the political will of the government in power, and on the other, they depend also on policy instruments, such as budgets, consulting, and transparency mechanisms, strengthening programs, among others. As an example of this, although 27 states have legislation, only 12 consider mechanisms for CSO consultation and participation to make decisions on promotion actions; likewise, of the 27 states, only 15 have specific programs to provide economic support to CSOs.” (CIESC 2023; p. 4). Layton (2017) offers a similar analysis in regards to the federal law: “(…) a lack of consistency in the participation of governmental representatives and little or no public funding for their activities” has weakened the coordination mechanisms designed to support the implementation of the law.

Ultimately, the CSO laws in Mexico vary in the landscape of state-society relations they aim for, and the degree of commitment expected from local political actors, among other factors (Vargas González 2012). Arguably, this affects the willingness to move on to the formal ‘constitution’ of organizations that may as well serve a social purpose while running informally. A comprehensive study of the aspects of the policy process surrounding the law in each of the three states is beyond the scope of this paper, although the preceding analysis points to some hypotheses that could be useful. In the end, the causal mechanism boils down to questions about the dynamics between state actors and civil society leaders. This includes looking at which groups and interests were represented, how the discussion of the law was conducted, and how the champions of the law kept the state representatives informed and coordinated with them different actions. It also requires a detailed analysis of the incentives and disincentives for formalization created in each case.

6 Conclusions

This paper has addressed the question of the effect that a state-level CSO promotion law may have had on the number of newly-constituted CSOs, using a quasi-experimental approach based on the synthetic control method. This is the most fitting empirical strategy, given the nature of the data and the type of ‘intervention’ considered in this paper. Applied to Mexico, and the case of three states that implemented separately a law of this kind between 2000 and 2007, among other states, a synthetic control analysis shows fairly different effects across them.

However, the states share a feature that is commonly found in studies of organizational ecology; namely, an inverted-U shape pattern in the outcome variable, indicating that at the end of the post-treatment period the outcome variable returns to levels that are similar to the average of the pre-treatment period.

As the seminal literature on synthetic control has pointed out, this method contributes a great deal to comparative case studies, where it can complement nicely with qualitative methods to understand the effects and help uncover the causal mechanism of a policy change such as the CSO promotion laws in Mexico. Hopefully, the core analysis above will help address further research questions, beyond the quantitative issue at stake here. As noted in the second section, the number of new formal organizations is not the only possible outcome of the law. Other measures of the effects of the law can be devised that focus on existing organizations. Such questions will have to deal with other dynamics and trends among nonprofit and civil society actors, exploring the factors that help them thrive, and also looking at intra-organizational dynamics that make them effective instruments of civic participation and strengthened livelihoods.