Diffusion of the Micro and Small Business Law in municipalities of São Paulo

Abstract

This article analyzes the implementation of the General Law for Micro and Small Business by municipalities in São Paulo. The theory of policy diffusion was used to study the process of adopting this public policy in the different contexts of coercion. The article presents theoretical discussions on the diffusion mechanism “coercion” and answers the question: what is the influence of the vertical coercion mechanism in the diffusion of the General Law in the municipalities of São Paulo? The survival analysis technique was applied to identify explanatory factors of policy diffusion, considering institutional, political, and party design variables, neighborhood effect, internal needs, organizations/social actors, and structural factors. The results show that as the institutional rules of coercion vary, the explanatory factors of the diffusion change so that local aspects, including social actors, regional factors, and structural issues, matter for the diffusion, varying according to the coercive context.

Keywords:
diffusion of innovations; General Law for Micro and Small Business; coercion; municipalities; survival analysis

Resumo

Este artigo tem como objeto de análise a implantação da Lei Complementar nº 123, de 2006 (LC nº 123/2006), que instituiu o Estatuto Nacional da Microempresa e da Empresa de Pequeno Porte (MPE), por municípios paulistas. Utilizou-se o referencial teórico de difusão de políticas públicas tendo como objetivo estudar seu processo de adoção, observando os diferentes contextos e momentos de coerção previstos na legislação federal. Com base nas discussões teóricas relacionadas com o mecanismo de difusão chamado de coerção, respondeu-se a seguinte pergunta: qual é a influência do mecanismo de coerção vertical na difusão da Lei Geral da MPE nos municípios paulistas? A técnica “análise de sobrevivência” foi aplicada para identificar fatores explicativos da difusão da política, considerando variáveis de desenho institucional, políticas e partidárias, efeito vizinhança, necessidade da política, organizações/atores sociais e fatores estruturais. Os resultados demonstram que, conforme varia o desenho institucional de coerção, mudam os fatores explicativos da difusão dessa política na ponta, de modo que os aspectos locais, incluindo atores sociais, fatores regionais e questões estruturais, importam para a difusão, variando conforme o contexto coercitivo.

Palavras-chave:
difusão de inovações; Lei Geral da MPE; coerção; municípios; análise de sobrevivência

Resumen

Este artículo analiza la implementación de la Ley General de las Micro y Pequeñas Empresas por parte de los municipios de São Paulo. Se utilizó la teoría de la difusión de políticas, con el objetivo de estudiar el proceso de adopción de esta política pública, observando los diferentes contextos y momentos de coerción previstos en la legislación federal. A partir de las discusiones teóricas relacionadas con el mecanismo de difusión denominado coerción, se respondió a la siguiente pregunta: ¿cuál es la influencia del mecanismo de coacción vertical en la difusión de la mencionada ley en los municipios de São Paulo? Se aplicó la técnica de análisis de supervivencia para identificar factores explicativos de la difusión de políticas, considerando variables institucionales, políticas y partidarias, efecto barrio, necesidad de políticas, organizaciones/actores sociales y factores estructurales. Los resultados muestran que a medida que varía el diseño institucional de coerción, los factores explicativos de la difusión de esta política al final cambian, de modo que los aspectos locales, incluidos los actores sociales, los factores regionales y los problemas estructurales importan para la difusión, variando de acuerdo con el contexto coercitivo.

Palabras clave:
difusión de innovaciones; Ley General de las Micro y Pequeñas Empresas; coerción; municipios; análisis de supervivencia.

1. INTRODUCTION

Complementary Law No. 123 (Lei Complementar nº 123, de 14 de dezembro de 2006), which instituted the National Statute of Micro and Small Enterprises (MSESs), was passed on December 14, 2006. This landmark, also known as the General MSES Law, regulates government procurement and other public purchases among other things to include micro and small enterprises in bidding processes, instituting differentiated treatment for them in these processes, which tend to affect regional and local economic development.

In practical terms, however, for the law to become concrete in the lives of MSESs, this differentiated treatment depends on municipalities which regulate and implement the law. To accomplish this, Article 77 of the norm published at the end of 2006 stipulated that states, the Federal District, and municipalities had until the end of 2007 to alter their laws and other necessary acts to guarantee compliance with federal law.

In a complementary manner, Article 47 of the General Law envisions that in order for local entities (states or municipalities) to use their purchasing power to offer differentiated treatment to MSESs, they need to regulate federal law. That is, despite the timeframe, regulation is necessary.

On the other hand, less than ten years after the norm went into effect it was modified by Complementary Law No. 147 (Lei Complementar nº 147, de 7 de agosto de 2014), which was passed on August 7, 2014, that dispensed with local regulation.

Thus, the implementation of differentiated treatment for MSESs in government purchases by states and municipalities should follow the rules of the General Law, which have been modified over time. These rules can be divided into three different time periods: the first mandatory context which terminated at the end of 2007; a second mandatory context with an undetermined timespan, but requiring local regulation; and, as of August 8, 2014, the absence of mandatory local regulation.

Within the context of political science, speaking of rules and not speaking of institutions is practically impossible within a context like this. In sum, institutions can be understood as relatively enduring patterns of political and social life - which can have rules, norms, and/or procedures - which structure behavior and cannot be easily or instantly modified (Mahoney & Thelen, 2010).

In the case of the General MSES Law, rules structure decision-making behavior in favor of, or not in favor of, local regulation. In numbers, according to the data presented by Leão (2016), by 2012 850 municipalities had implemented Complementary Law No. 123, and by 2015, this number had jumped to 3,022. Within this context, the author studied the implementation of the General MSES Law for municipalities in 2015, based on the theoretical reference of the diffusion of public policies.

The academic literature concerning diffusion is pertinent to studies of this type. In the international literature, Graham, Shipan, and Volden (2013) pointed out how national decision makers in federations can act in a vertical manner in terms of policy diffusion, choosing to employ or not to employ coercive means. Diffusion can be understood as a process which occurs when a government decision about the adoption of a policy innovation is influenced by the previous choices made by other governments (Graham et al., 2013). This influence can permeate various aspects and mechanisms, especially vertical and horizontal influences according to Sugiyama’s study (2012) of policy diffusion in Brazil.

Within the scope of vertical influences, coercion is one of the possibilities. According to Shipan and Volden (2012), coercion is the use of incentives, force, or threats by a government to affect policy decisions taken by other governments. It can occur when, for example, the national government tries to impose the implementation of a preferential public policy together with others, utilizing incentives or imposing obligations (Graham et al., 2013). Therefore, the situation of Complementary Law No. 123 obliging local regulation can be classified as a coercion diffusion mechanism.

Even though Leão (2016) studied the diffusion of the General MSES Law, she did not make a deep analysis of the different contexts of the coercion of federal legislation related to the regulation or non-regulation of the law on the local level, complying with policy in 2015.

That being said, it is of interest to analyze General MSES Law No. 123, observing in a deeper manner the coercion used in municipal regulation or non-regulation of the law. In revisiting the studied case, it is worthwhile to reevaluate the territorial scope of the analysis, observing the diffusion phenomenon in greater detail in more delimited groups of municipalities according to Maggetti and Gilardi (2016).

In this aspect, municipalities within the State of São Paulo present an appropriate sample for analysis. Almost 30% of MSESs registered in Brazil are in the State of São Paulo (Bedê, 2006), and data from Sebrae (Brazilian Support Service for Micro and Small Enterprises) demonstrate that São Paulo is the state that had the largest number of municipalities implementing the law beginning in 2015 (Serviço Brasileiro de Apoio às Micro e Pequenas Empresas [Sebrae], 2019), a context in which local regulation was no longer mandatory.

Based on these aspects, the main objective of this article is to study the public policy adoption process, observing the various contexts and times of coercion envisaged in the federal legislation. In other words, our research question is: what has been the influence of vertical coercion mechanisms in the diffusion of the General MSES Law in the municipalities of São Paulo?

To answer this question, and considering our theoretical references which will be detailed in the following section, our research hypothesis is: to the extent that institutional design of coercion varies, the internal, external and structural explanatory factors also vary, demonstrating that coercion modifies to some extent the behavior of decision makers on a local level.

The quantitative “survival analysis” technique is appropriate to achieve our objectives. Its use has been an important methodological characteristic in diffusion studies (Graham et al., 2013) ever since the seminal American study conducted by F. S. Berry and W. D. Berry (1990).

Thus, the next section will deal with aspects related to the literature on diffusion which have guided the hypothesis of this article. Then, we will present the methodology in more detail. The fourth section presents and discusses the results of our statistical model. Last of all, we will offer our final considerations.

2. PUBLIC POLICY DIFFUSION

Considering the objective of this text, observing the dynamics of intergovernmental relations is very important. However, evaluating the literature focused on the aspects of policy decentralization is not sufficient.

In essence, the discussion of policy decentralization deals with how the center of responsibility is modified from a national authority to subnational entities (Souza, 2008), observing issues such as federative coordination and institutional design (Abrucio & Franzese, 2007; Arretche, 2012). However, diffusion studies, in addition to considering vertical coercive elements, traditionally permit the observation of broader aspects, such as horizontal connections.

Studying federative contexts also requires observing policy and local social dynamics. The institutional design of federations is important, however local dynamics are equally relevant to understanding the formation of a public policy agenda (Pierson, 1995).

Generally, studies focused on the formation of public policy agenda processes need to observe: the actions of actors and organizations which employ them for the entrance of a subject on the policy agenda (Kingdon, 2011); an understanding of aspects related to the limited rationality of decision makers (Baumgarter & Jones, 2010); and the role of values, ideas, and interactions between actors and organized groups, as well as institutions and external factors (Sabatier & Weible, 2019).

The diffusion literature contemplates all of these aspects in some way or another. A seminal study by Dolowitz and Marsh (2000) proposes key questions for the understanding of the diffusion process: Why do actors diffuse policies? Who are the main actors? What is diffused? From where are the lessons drawn? What restricts or facilitates the diffusion of public policies? How is the public policy diffusion process related to the results of its adoption (success or failure)?

In dialogue with this work, Marsh and Sharman (2009) highlight the importance of diffusion studies observing policy complexity, the media, experiences with previous policies, institutions, ideology, cultural proximity, technology, economics, and bureaucracy, etc.

Observing all of these elements helps us understand a complex process that ranges from adherence to a perfectly rational initiative, to the fruit of a learning process, and adherence through a process of coercion (Dolowitz & Marsh, 2000).

Specifically considering the “coercion” mechanism, the key issue is how can some vertical coercion, such as a financial incentive, subsidies, or a linked law be used by a national government, for example, to pressure subnational entities to implement a given initiative (Graham et al., 2013). In this way, coercion is associated with the notion of the use of power and sanctions by one agent to stimulate the adoption of an innovation by other agents (Shipan & Volden, 2012; Coêlho, 2016). It is an external factor of influence (F. S. Berry & W. D. Berry, 2018), in which a government implements a public policy in a constringed manner, which it otherwise - if there was no coercion promoted by another actor with greater power - would not do (Dobbin, Simmons, & Garrett, 2007).

Dobbin et al. (2007) further argue that the greatest challenge of a “coercion” mechanism is to concretely associate the adoption of a policy to the existence of a coercive process that is underway. It has to be demonstrated that the actions of the government that is acting coercively to promote a public policy affect the probability of its adoption by the party under duress (Dobbin et al., 2007).

Under these conditions, coercion studies need to also evaluate the local context of the implementation of a policy, given that the same policy could be implemented for different reasons. Some governments implement a program because they have learned from the experiences of others, while other governments just copy others (Meseguer & Gilardi, 2008). This is why the comprehension of diffusion processes consists of understanding the various elements which suffuse internal, external, and structural determinants which dialogue with the debate concerning the formation of the public policy agenda.

For example, the interactions among actors should be considered. Wampler (2016) has already demonstrated that policy convergence in diffusion may be relevant, especially when it is motivated by alignment between government parties and coalitions. In this aspect, political-party ideology stands out. It would be the occurrence of a phenomenon of isomorphism, in which a government is more likely to consider an innovation that has already been adopted by other governments with similar ideologies as a reference (F. S. Berry & W. D. Berry, 2018). In terms of Brazil, Sugiyama (2012) as well as Coêlho, Cavalcante, and Turgeon (2016), have already demonstrated that leftist party ideology leads to greater adoption of social policies and programs.

Besides political-party convergence, the diffusion literature also emphasizes convergence within a regional context (F. S. Berry & W. D. Berry, 2018), which may occur through formal networks, such as associations (Walker, 1969), or informal networks such as interaction between neighboring federal entities (Mooney, 2001).

In terms of internal effects, given the specific aspects of local dynamics, a decision maker may be driven to adhere to a public policy for political reasons at the time. For example, public policies which are not controversial have greater diffusion during election years, as has been observed in the United States (F. S. Berry & W. D. Berry, 1990). In the case of Brazil, this electoral aspect has not been widely explored, but a study by Coêlho (2021) has already indicated the relevance of this subject.

Another local political aspect which can also have an impact is political competition (F. S. Berry & W. D. Berry, 1990). In Brazil, a study applied to the diffusion of the Family Health Program has demonstrated the importance of this factor (Coêlho et al., 2016).

In addition to local dynamics, the perception of the social needs of a program also often matters (F. S. Berry & W. D. Berry, 2018). Leão (2016), for example, has studied the demand for the implementation of the General MSES Law observing the number of MSESs located in municipalities.

Moreover, the construction of the perception of the need for a public policy also involves the actions of local actors and organizations. Evans (2009) exemplifies which agents have the capacity to influence diffusion: politicians, bureaucrats, political entrepreneurs, think tanks, academics, and pressure groups, etc. In the case of the General MSES Law, Leão (2016) observes the importance of the role of Sebrae locally especially before 2011, in which the main strategy of Sebrae, called a mobilizing target, was to ensure that Brazilian municipalities would regulate this law (Sebrae, 2017).

In a complementary manner, we should mention structural economic aspects which can facilitate or impede the actions of an organization like Sebrae in the diffusion process. For example, the level of economic development recurs in diffusion studies (Batista, 2017; F. S. Berry & W. D. Berry, 2018). To be more specific, aspects such as financial and administrative capacity should be considered (Dolowitz & Marsh, 2000; Marsh & Sharman, 2009). For example, a study of the municipality of Ituverava in the State of São Paulo observed that the capacity of public spending has become a relevant element in the implementation of the General MSES Law in a scenario considering a municipality with fewer than 50 thousand inhabitants where the government is the largest buyer (Cardoso, 2017).

Finally, this aspect of population or other structural characteristics may be relevant in the sense of the profile of programs that are compatible with the structural characteristics of the municipality, as indicated by Dolowitz and Marsh (2000), as well as Marsh and Sharman (2009).

In sum, various aspects can explain the diffusion of a public policy, Within the Brazilian context, even though there are valid coercive rules for all, one has to consider these external, internal, and structural aspects. In numerical terms, consulting the monitoring database of Sebrae (2019), we identified 340 municipalities in the State of São Paulo which regulated the law between 2007 and 2015. That is, the coercion mechanism by itself, is not able to achieve total adhesion due to aspects of the autonomy of Brazilian municipalities, which is a subject that has been widely explored by the literature on federalism (Abrucio & Franzese, 2007).

On the other hand, the case we are analyzing has an important characteristic. Until 2007, there was a more coercive context in which the regulation was envisaged happening by a year after the legislation’s passage, that is by the end of 2007. In addition, for the differentiated treatment of MSES in governmental purchases, local regulation was necessary, because without a fixed end date it could be understood as a less coercive context - even though it remained coercive. Finally, this rule was modified by law, and since August 8, 2014, there has no longer been a need for local regulation, which denotes a context without coercion.

That being said, the distinct elements inherent in the local political and social dynamics of municipalities and other possible aspects of political-party and regional convergence can have greater or lesser effects on the diffusion of the General MSES Law, depending on different legal institutional contexts which may be more coercive or less coercive.

In other words, our research hypothesis is: to the extent that the institutional design of coercion varies, the internal, external, and structural explanatory factors can also vary, which demonstrates that coercion, to some extent, modifies the behavior of decision makers on the local level.

3. METHODOLOGY

From a quantitative perspective, survival analysis is the method that is used most often in diffusion studies of this type. A seminal study by F. S. Berry and W. D. Berry (1990) in the United States has become a great reference on this subject (Graham et al., 2013).

Its main aspect is that the dependent variable measures the length of time that units remain in a given situation before experiencing a new situation (Box-Steffensmeier & Jones, 2004) - in other words, the time that a municipality which has not implemented the General MSES Law takes to implement it.

Survival analyses use the Survival package of the R software (Therneau, 2015), and the central point is the statistical regression which measures the covariate effect (independent variables) in relation to the “answer” variable (the time to adopt the analyzed policy).

In diffusion studies, Cox’s semi-parametric regression is often used, because it is the most flexible technique (Batista, 2017). Given its versatility, it is possible to study cases in which the covariates change over time. This consists of a Cox model with covariates that are dependent on time or Cox’s extended model (Carvalho et al., 2011; Colosimo & Giolo, 2006; Pereira, 2004), which will be used in this work, because it is possible to imagine, for example, that the party ideology of a city hall changes over time, and it may be governed by a left-wing party at one moment and a right-wing party at another.

We will perform the first regression with all of the covariates that will be tested in accordance with Box 1, which systemizes the group of variables which dialogue with the theoretical references presented above. Next, we will use the stepwise procedure to improve the variable selection, making the model more appropriate to measure the effects of the covariates which present statistical significance.

The covariates which explain diffusion will be those in the final model which were significant at a level of 10% and which passed the proportionality assumption evaluation based on an analysis of the Schoenfeld residuals. This last analysis is relevant, given that in using the Cox model, we assume that the relative risk of a covariate remains stable during the observation period (Carvalho et al., 2011).

Moreover, the dependent variable will be considered dependent on the date of the regulation of the General MSES Law in municipalities in the State of São Paulo - and in consulting Sebrae’s monitoring database (2019), we identified 340 which regulated the law between 2007 and 2015. Thus, those which did not regulate it will be right-censored, that is the other 305 municipalities. The analysis will consider the exact date of the publication of the local norm, and the time will be counted in terms of days, with the start being the publication of the federal law (12/15/2006).

Given the mandatory nature of municipal regulation of the General MSES Law in 2007, we will run a second specific model considering this year. This will involve an analysis of 645 municipalities, with 60 of them considered adopting municipalities. The data will be truncated on December 31, 2007, which means the 585 municipalities which did not regulate the law by this date will be right-censored.

In addition, we will run a third model beginning with January 2008 and ending on August 7, 2014, given that on the following day a new rule went into effect dispensing the need for local regulation. Thus, the 60 municipalities which regulated the law in 2007 will be left-censored, and the data will be truncated on August 7, 2014, thereby right-censoring 340 municipalities, considering 245 of them adopting municipalities.

Finally, a fourth model will be run for the legislation that went into effect beginning on August 8, 2014. The 305 municipalities which already adopted the policy before will be left-censored, while the other 340 will be right-censored, because they did not regulate the General MSES Law during the observation period. Thus, 35 municipalities will be considered adopting municipalities.

Box1presents the covariates which will be tested, including the operationalization of each one.

Box 1
Independent Variables, Sources and Operationalizations

Box 1
continuation

4. RESULTS AND ANALYSIS OF THE DIFFUSION OF THE MSES LAW

First of all, it should be registered that the details of the primary data of the performed analyses in this section are in a certified and public repository.¹

To begin, Table 1 summarizes the number of municipalities which regulated the General Law from the publication of the federal law in December 2006 until December 2015.

Table 1
Number of Municipalities Regulating the General MSES Law

Semester	Nº of municipalities per semester	Cumulative total
Dec./Dec. 2006	0	0
Jan./Jun. 2007	3	3
Jul./Dec. 2007	57	60
Jan./Jun. 2008	18	78
Jul./Dec. 2008	7	85
Jan./Jun. 2009	1	86
Jul./Dec. 2009	22	108
Jan./Jun. 2010	13	121
Jul./Dec. 2010	24	145
Jan./Jun. 2011	16	161
Jul./Dec. 2011	72	233
Jan./Jun. 2012	13	246
Jul./Dec. 2012	11	257
Jan./Jun. 2013	9	266
Jul./Dec. 2013	26	292
Jan./Jun. 2014	12	304
Jul./Dec. 2014	14	318
Jan./Jun. 2015	8	326
Jul./Dec. 2015	14	340

• Source: Research data.

The second semesters of 2007 and 2011 stand out from the other periods with a considerable number of policy adoptions.

The second semester of 2011 was Sebrae’s last mobilization target for the regulation of the General MSES Law (Sebrae, 2017). Perhaps this explains the elevated number of cases during this period. In terms of the second semester of 2007, according to the federal legislation, the states and municipalities had one year to adjust their legislation beginning with the passage of the federal law in 2006, which explains the notable number of adoptions during this period.

4.1. Survival analysis (extended Cox model): initial estimates

Table 2 presents the estimates obtained in the first analysis prepared with the Cox model, with dependent time covariates for the four study models: M1, the entire study period; M2, the period envisaged for regulation by the General MSES Law (2007); M3, from January 2008 to August 7, 2014, when there was a requirement to regulate the General Law but there was no definite timeframe; and M4, beginning with August 8, 2014 when there was no longer any need for local regulation.

According to Table 2, the first information that stands out refers to the variables “instit.design.1”, “instit.design.2”, “nat.elect”, “munic.elect” and “sebrae.act”. Even though they changed over time, these variables did not present variations among the municipalities on a specific date. Thus, their estimates cannot be observed in this extended Cox model analysis for any of the four performed analyses. On the other hand, especially for the first two covariates, this does not affect them adversely, given that models M2, M3 and M4 make it possible to understand the dynamics of various institutional designs in detail.

Table 2
Initial Estimates: Cox Model with Dependent Time Covariates - Regulation of the General MSES Law

Independent variable	M1 The entire analysis period¹			M2(instit.design.1=1; instit.design.2=1) until 12/31/2007²			M3 (instit.design.1=0; instit.design.2=1) From 1/1/2008 to 8/7/2014³			M4 (instit.design.1=0; instit.design.2=0) From 8/8/2014 to 12/31/20154
	Coef	RR	s.e.	Coef	RR	s.e.	Coef	RR	s.e.	Coef	RR
instit.design 1	NA	NA	0.000	-	-	-	-	-	-	-	-	-
instit.design 2	NA	NA	0.000	-	-	-	-	-	-	-	-	-
presid.party	0.332	1.394 *	0.200	-0.171	0.843	0.477	0.370	1.447	0.247	1.374	3.951 *	0.622
nat.elect	NA	NA	0.000	NA	NA	0.000	NA	NA	0.000	NA	NA	0.000
munic.elect	NA	NA	0.000	NA	NA	0.000	NA	NA	0.000	NA	NA	0.000
dif.vot.elect	-0.111	0.895	0.243	-0.112	0.894	0.712	-0.156	0.855	0.283	0.338	1.402	0.638
ideol	0.081	1.084	0.179	-0.233	0.792	0.384	0.237	1.267	0.216	0.117	1.124	0.634
bigs.neigh.reg	0.053	1.055	0.122	NA	NA	0.000	0.019	1.019	0.131	0.245	1.277	0.362
gov.neigh.reg	-0.005	0.995	0.149	NA	NA	0.000	-0.011	0.989	0.159	-0.089	0.915	0.507
sebrae.neigh.reg	0.421	1.524 *	0.142	NA	NA	0.000	0.395	1.484 **	0.151	0.593	1.809	0.479
unemployed.pop	0.023	1.023	0.015	0.043	1.043	0.034	0.014	1.015	0.018	0.016	1.016	0.046
log.inc.percap	0.039	1.040	0.255	-1.268	0.281*	0.692	0.255	1.291	0.295	-0.105	0.900	0.745
num.mses	0.169	1.184	0.150	0.701	2.016 **	0.356	0.176	1.192	0.178	-0.488	0.614	0.447
sebrae.office	1.034	2.812 ***	0.140	1.171	3.224 ***	0,335	1.044	2.841 ***	0.164	0.683	1.980	0.506
sebrae.act	NA	NA	0.000	NA	NA	0.000	NA	NA	0.000	NA	NA	0.000
autonomy	0.213	1.237	0.138	0.581	1.788	0.325	0.193	1.213	0.162	-0.653	0.521	0.528
small.pop	0.197	1.218	0.170	0.244	1.276	0.399	0.271	1.311	0.202	-1.364	0.256 *	0.636
log.gdp.percap	0.012	1.013	0.114	-0.050	0.951	0.277	0.056	1.057	0.132	-0.227	0.797	0.380

• Obs.: Coef: coefficient; RR: relative risk = exp(Coef); s.e.: standard error. *p<0.1; **p<0.05; ***p<0.01.
• Notes:
• ¹ Correlation = 0.684 (se = 0.014). Likelihood ratio = 137.6 on 13 df; p=<2e-16. Wald test = 138.5 on 14 df; p=<2e-16. Score test (logrank) = 158.9 on 14 df; p=<2e-16; Robust = 125.4 p=<2e-16. Proportional risk analysis: Schoenfeld (global) residuals p=0.168.
• ² Correlation = 0.694 (se = 0.036). Likelihood ratio = 29.65 on 10 df; p=0.001. Wald test = 38.55 on 10 df; p=3e-05. Score testlogrank = 32.31 on 10 df; p=4e-04; Robust = 26.11 p=0.004. Proportional risk analysis: Schoenfeld (global) residuals p=0.97.
• ³ Correlation = 0.69 (se = 0.016). Likelihood ratio = 102.7 on 13 df; p=5e-16. Wald test = 113.6 on 13 df; p=<2e-16. Score testlogrank = 119.6 on 13 df; p=<2e-16; Robust = 86.57 p=6e-13. Proportional risk analysis: Schoenfeld (global) residuals p=0.138.
• ⁴ Correlation = 0.77 (se = 0.039). Likelihood ratio = 33.49 on 13 df; p=0.001. Wald test = 51.98 on 13 df; p=1e-06. Score testlogrank = 43 on 13 df; p=4e-05; Robust = 21.26 p=0.07. Proportional risk analysis: Schoenfeld (global) residuals p=0.57.
• Source: Research data.

In general, for each tested model, the results are quite distinct. For example, the vertical influence variable “presid.party” has a relevant effect on Models 1 and 4. The neighboring effect, associated with the Sebrae regions, has relevance for Models 1 and 3. However, these are initial estimates. As indicated in the methodology, we conducted a stepwise procedure to arrive at our results which are displayed in Table 3.

4.2. Survival analysis (extended Cox model): final estimates

Table 3 presents the final results which are a better representation of the effect of the relevant covariates in explaining the adoption of this policy, which includes only those covariates which do not violate the assumption of proportionality - the “presid.party” variable in M1 had to be excluded from the final model, because it presented a p value less than 0.05 presenting a violation in the proportional risk analysis, which makes sense given the elevated level of relative risk observed in M4.

In addition, the four models have positive results in their likelihood ratios, Wald tests, and score tests. Overall, the level of correlation was around 0.7, which indicates that our findings are not by chance.

Table 3
Final Estimates: Cox with Time Dependent Covariates - Regulation of the General MSES Law

Independent variable	M1 The entire analysis period¹			M2 (instit.design.1=1; instit.design.2=1) until 12/31/2007²			M3 (instit.design.1=0; instit.design.2=1) From 1/1/2008 to 8/7/2014³			M4(instit.design.1=0; instit.design.2=0) From 8/8/2014 to 12/31/20154
	Coef	RR	s.e.	Coef	RR	s.e.	Coef	RR	s.e.	Coef	RR	s.e.
presid.party	-	-	-	-	-	-	-	-	-	1.290	3.631 ***	0.393
sebrae.neigh.reg	0.434	1.543 ***	0.121	-	-	-	0.413	1.511 ***	0.129	0.589	1.801	0.354
log.inc.percap	-	-	-	-1.503	0.223 **	0.617	-	-	-	-	-	-
num.mses	-	-	-	0.628	1.873 **	0.351	-	-	-	-0.636	0.530	0.362
sebrae.office	1.055	2.873 ***	0.122	1.283	3.608 ***	0.325	1.041	2.831 ***	0.143	0.697	2.008 *	0.476
autonomy	0.224	1.251 *	0.123	0.629	1.876 *	0.313	0.253	1.288 *	0.144	-0.732	0.481	0.476
small.pop	-	-	-	-	-	-	-	-	-	-1.280	0.278 **	0.537

• Obs.: Coef: coefficient; RR: relative risk = exp (Coef); s.e: standard error. *p<0.1; **p<0.05; ***p<0.01.
• Notes:
• ¹ Correlation = 0.676 (se = 0.014). Likelihood ratio = 130.3 on 3 df; p=<2e-16. Wald test = 133,8 on 3 df; p=<2e-16. Score testlogrank = 151.7 on 3 df; p=<2e-16; Robust = 117.7 p=<2e-16. Proportional risk analysis: Schoenfeld (global) residuals p=0.92.
• ² Correlation = 0.687 (se = 0.037). Likelihood ratio = 27.68 on 4 df; p=1e-05. Wald test = 26.22 on 4 df; p=3e-05. Score testlogrank = 29.7 on 4 df; p=6e-06; Robust = 19.24 p=7e-04. Proportional risk analysis: Schoenfeld (global) residuals p=0.94.
• ³ Correlation = 0.677 (se = 0.016). Likelihood ratio = 95.47 on 3 df; p=<2e-16. Wald test = 104.9 on 3 df; p=<2e-16. Score testlogrank = 111.5 on 3 df; p=<2e-16; Robust = 80.05 p=<2e-16. Proportional risk analysis: Schoenfeld (global) residuals p=0.36.
• ⁴ Correlation = 0.757 (se = 0.04). Likelihood ratio = 32.23 on 6 df; p=1e-05. Wald test = 47.62 on 6 df; p=1e-08. Score testlogrank = 41.67 on 6 df; p=2e-07; Robust = 19.3 p=0.004. Proportional risk analysis: Schoenfeld (global) residuals p = 0.70.
• Source: Research data.

In sum, when we observe the entire analysis period from December 2006 to December 2015 (M1), that is, independent of the institutional design of coercion, we may conclude that the factors which are relevant to explaining diffusion are: the neighboring effect related to regional Sebrae offices (“sebrae.neigh.reg”), Sebrae actions measured by the presence of one or more offices in the municipality (“sebrae.office”) and the administrative and financial capacity of the municipality measured by the Firjan “autonomy” index.

On the other hand, when we specifically observe the first year of the diffusion of General Law regulation (M2), a more coercive period, we may conclude that what explains diffusion is the local need for this public policy based on two covariates - income per capita (“log.inc.percapita”) and the number of MSESs (“num.mses”), the actions of Sebrae (“sebrae,office”), and the administrative and financial capacity of the municipality (“autonomy”).

When we observe the second period highlighted in the diffusion of the regulation of the General MSES Law (M3) - a coercive context without a timeframe, that is one less coercive than the previous period - what explains this diffusion is: the neighboring effect (“sebrae.neigh.reg”), the actions of Sebrae (“sebrae.office”) and the “autonomy” of the municipality, however with relative risk values considerably lower than those for M2.

Finally, when we observe the last highlighted period (M4), during which there were no longer coercive rules, diffusion can be explained by the mayor being a member of the same party as the country’s president (“presid.party”), the municipality not being a city with less than 50 thousand inhabitants (“small.pop”) and the actions of Sebrae (“sebrae.office”), with relatively less risk for this covariate in the four models.

Given these findings, the different institutional designs envisaged in the federal legislation are relevant and explain these distinctions in terms of the explanatory factors. This therefore confirms the research hypothesis that coercion to some extent modifies decision-maker behavior on the local level.

In the absence of coercion (M4), and only in this context, the political aspect of party alignment between the mayor and the country’s president is most relevant, confirming the relevance of policy convergence indicated by Wampler (2016).

On the other hand, it is within the more coercive context (M2), and only in this context, that the internal determinants of the need for this public policy - evaluated by the number of MSESs and the population’s income per capita - explain diffusion, which is one aspect which confirms the importance of local dynamics already observed by F. S. Berry and W. D. Berry (2018) as well as Pierson (1995).

Looking at it a different way, only one explanatory factor stands out in all of the institutional designs: the presence of Sebrae in municipalities as the acting organization in this area, which indicates the importance of the “agency” factor pointed out by Evans (2009), noting that the Sebrae effect diminishes with time, varying in accordance with changes in the context of coercion.

Finally, the level of autonomy of the municipality is relevant in three models (M1, M2 and M3), having a greater effect within the most coercive institutional design (M2), which indicates the importance of considering structural aspects in diffusion, as indicated by Dolowitz and Marsh (2000), as well as Marsh and Sharman (2009). The neighboring effect is significant in Models 1 and 3, which indicates the importance of the interaction between federal entities in a regional context, thus reinforcing the understanding of F. S. Berry and W. D. Berry (2018), as well as Mooney (2001).

5. FINAL CONSIDERATIONS

This work has analyzed the factors which imply that municipalities in the State of São Paulo have a greater probability of implementing the General MSES Law. Our analysis has made clear that the explanatory factors for this policy diffusion changed in keeping with variations in the institutional design of coercion on the federal level.

Thus, we conclude that the coercive process makes a difference, and it is important to highlight the role of the diffusion literature in studies which observe aspects of federalism in Brazil. Beyond the issues which permeate federative coordination, we should consider local aspects, such as the need for a public policy; political-party issues; the regional effect and role of Sebrae; as well as structural elements, such as the financial autonomy of these municipalities. Therefore, while local, regional, and structural aspects matter in terms of diffusion, it has been demonstrated that a coercive institutional design is capable of influencing the explanatory factors of diffusion.

In this manner, the results of this work indicate that coercive processes need to be examined more closely by students of this subject, especially considering the diffusion literature.

Based on these conclusions, there is room for a wide variety of studies. For example, considering adherence to new initiatives which is motivated by coercion, do the actions of relevant organizations or political-party aspects lead to a pro-forma implementation? Thus, qualifying the analysis of adherence to initiatives induced by the federal sphere from this perspective could be of interest. In addition, new case studies observing coercion in other Brazilian states would also be welcome, as well as similar studies to this one dealing with other public policies.

To a certain extent, even though this study has been restricted to the municipalities of the State of São Paulo, this work’s findings make us reflect, because understanding the possible effects of local social, economic, and political dynamics is relevant to the definition of national government ideas and policies which depend on local implementation. In this way, this work’s findings permit a discussion of the limits of the diffusion of public policies based on a variety of local and institutional contexts.

REFERÊNCIAS

• Abrucio, F. L., & Franzese, C. (2007). Federalismo e políticas públicas: o impacto das relações intergovernamentais no Brasil. In M. Araújo, & L. Beira(Orgs.), Tópicos da economia paulista para gestores públicos (Vol. 1, pp. 13-31). São Paulo, SP: Fundap.
• Arretche, M. (2012). Democracia, federalismo e centralização no Brasil Rio de Janeiro, RJ: SciELO-Editora Fiocruz.
• Batista, M. (2017). A difusão da Lei de Acesso à Informação nos municípios brasileiros: determinantes internos e externos(Cadernos Enap, n. 53). Brasília, DF: Fundação Escola Nacional de Administração Pública.
• Baumgartner, F. R., & Jones, B. D. (2010). Agendas and instability in American politics Chicago, IL: University of Chicago Press.
• Bedê, M. A. (2006). Onde estão as micro e pequenas empresas no Brasil São Paulo, SP: Sebrae.
• Berry, F. S., & Berry, W. D. (1990). State lottery adoptions as policy innovations: an event history analysis. American Political Science Review, 84(2), 395-415. Recuperado de https://doi.org/10.2307/1963526
  » https://doi.org/10.2307/1963526
• Berry, F. S., & Berry, W. D. (2018). Innovation and diffusion models in policy research. In C. M. Weible, & P. A. Sabatier (Eds.), Theories of the policy process (4a ed., pp. 253-297). New York, NY: Routledge.
• Box-Steffensmeier, J. M., & Jones, B. S. (2004). Event history modeling: a guide for social scientists Cambridge, UK: Cambridge University Press.
• Cardoso, M. S. (2017). O poder de compras e o desenvolvimento econômico local: a avaliação da política pública de compras governamentais no município de Ituverava (SP) (Dissertação de Mestrado). Universidade Estadual Paulista, Franca, SP.
• Carvalho, M. S., Andreozzi, V. L., Codeço, C. T., Campos, D. P., Barbosa, M. T. S., & Shimakura, S. E. (2011). Análise de sobrevivência: teoria e aplicações em saúde Rio de Janeiro, RJ: SciELO-Editora Fiocruz.
• Coêlho, D. B. (2016). Mecanismos políticos e institucionais da difusão de políticas. In C. A. P. Faria, D. B. Coêlho, & S. D. Silva (Orgs.), Difusão de políticas públicas Santo André, SP: Editora da UFABC.
• Coêlho, D. B. (2021). Determinants of policy diffusion in Brazil and the U.S. International Review of Public Policy, 3(2), 194-215. Recuperado de https://doi.org/10.4000/irpp.2254
  » https://doi.org/10.4000/irpp.2254
• Coêlho, D. B., Cavalcante, P., & Turgeon, M. (2016). Mecanismos de difusão de políticas sociais no Brasil: uma análise do Programa Saúde da Família. Revista de Sociologia e Política, 24(58), 145-165. Recuperado de https://doi.org/10.1590/1678-987316245807
  » https://doi.org/10.1590/1678-987316245807
• Colosimo, E. A., & Giolo, S. R. (2006). Análise de sobrevivência aplicada São Paulo, SP: Editora Blucher.
• Dobbin, F., Simmons, B., & Garrett, G. (2007). The global diffusion of public policies: social construction, coercion, competition, or learning? Annual Review of Sociology, 33, 449-472. Recuperado de https://doi.org/10.1146/annurev.soc.33.090106.142507
  » https://doi.org/10.1146/annurev.soc.33.090106.142507
• Dolowitz, D. P., & Marsh, D. (2000). Learning from abroad: the role of policy transfer in contemporary policy‐making. Governance, 13(1), 5-23. Recuperado de https://doi.org/10.1111/0952-1895.00121
  » https://doi.org/10.1111/0952-1895.00121
• Evans, M. (2009). Policy transfer in critical perspective. Policy Studies, 30(3), 243-268. Recuperado de https://doi.org/10.1080/01442870902863828
  » https://doi.org/10.1080/01442870902863828
• Graham, E. R., Shipan, C. R., & Volden, C. (2013). The diffusion of policy diffusion research in political science. British Journal of Political Science, 43(3), 673-701. Recuperado de https://doi.org/10.1017/S0007123412000415
  » https://doi.org/10.1017/S0007123412000415
• Kingdon, J. W. (2011). Agendas, alternatives, and public policies (2a ed.). Boston, MA: Longman Classics in Political Science.
• Leão, A. (2016). Difusão da lei de incentivos à participação das microempresas e empresas de pequeno porte nas compras públicas municipais brasileiras (Dissertação de Mestrado). Universidade Federal de Pernambuco, Recife, PE.
• Lei Complementar Federal nº 123, de 14 de dezembro de 2006 (2006). Institui o Estatuto Nacional da Microempresa e da Empresa de Pequeno Porte e dá outras providências. Brasília, DF. Recuperado de http://www.planalto.gov.br/ccivil_03/leis/lcp/lcp123.htm
  » http://www.planalto.gov.br/ccivil_03/leis/lcp/lcp123.htm
• Lei Complementar nº 147, de 7 de agosto de 2014 (2014). Altera a Lei Complementar no 123, de 14 de dezembro de 2006, e dá outras providências. Brasília, DF. Recuperado de http://www.planalto.gov.br/ccivil_03/leis/lcp/lcp147.htm
  » http://www.planalto.gov.br/ccivil_03/leis/lcp/lcp147.htm
• Maggetti, M., & Gilardi, F. (2016). Problems (and solutions) in the measurement of policy diffusion mechanisms. Journal of Public Policy, 36(S1), 87-107. Recuperado de https://doi.org/10.1017/S0143814X1400035X
  » https://doi.org/10.1017/S0143814X1400035X
• Mahoney, J., & Thelen, K. (2010). A theory of gradual institutional change. In J. Mahoney, & K. Thelen (Eds.), Explaining institutional change: ambiguity, agency, and power Cambridge, UK: Cambridge University Press.
• Marsh, D., & Sharman, J. C. (2009). Policy diffusion and policy transfer. Policy Studies, 30(3), 269-288. Recuperado de https://doi.org/10.1080/01442870902863851
  » https://doi.org/10.1080/01442870902863851
• Meseguer, C., & Gilardi, F. (2008). Reflexiones sobre el debate de la difusión de políticas. Política y Gobierno, 15(2), 315-351. Recuperado de http://www.politicaygobierno.cide.edu/index.php/pyg/article/view/221
  » http://www.politicaygobierno.cide.edu/index.php/pyg/article/view/221
• Mooney, C. Z. (2001). Modeling regional effects on state policy diffusion. Political Research Quarterly, 54(1), 103-124. Recuperado de https://doi.org/10.1177/106591290105400106
  » https://doi.org/10.1177/106591290105400106
• Pereira, T. L. (2004). Modelos de riscos proporcionais e aditivos para o tratamento de covariáveis dependentes do tempo (Dissertação de Mestrado). Universidade Federal de Pernambuco, Recife, PE.
• Pierson, P. (1995). Fragmented welfare states: Federal institutions and the development of social policy. Governance, 8(4), 449-478. Recuperado de https://doi.org/10.1111/j.1468-0491.1995.tb00223.x
  » https://doi.org/10.1111/j.1468-0491.1995.tb00223.x
• Sabatier, P. A., & Weible, C. M. (2019). The advocacy coalition framework: Innovations and clarifications. In P. A. Sabatier , & C. M. Weible (Eds.), Theories of the policy process(pp. 189-220). London, UK: Routledge.
• Serviço Brasileiro de Apoio às Micro e Pequenas Empresas. (2013). Anuário do trabalho na micro e pequena empresa São Paulo, SP: Autor.
• Serviço Brasileiro de Apoio às Micro e Pequenas Empresas. (2017). Manual do indicador: municípios com políticas de desenvolvimento implantadas Brasília, DF: Autor.
• Serviço Brasileiro de Apoio às Micro e Pequenas Empresas. (2019). DataSebrae Recuperado de http://sistema.datasebrae.com.br
  » http://sistema.datasebrae.com.br
• Shipan, C. R., & Volden, C. (2012). Policy diffusion: Seven lessons for scholars and practitioners. Public Administration Review, 72(6), 788-796. Recuperado de ttps://doi.org/10.1111/j.1540-6210.2012.02610.x
  » https://doi.org/10.1111/j.1540-6210.2012.02610.x
• Souza, C. (2008). Federalismo: teorias e conceitos revisitados. Revista Brasileira de Informação Bibliográfica em Ciências Sociais, 65, 27-48. Recuperado de https://bibanpocs.emnuvens.com.br/revista/article/view/311
  » https://bibanpocs.emnuvens.com.br/revista/article/view/311
• Sugiyama, N. B. (2012). Diffusion of good government: social sector reforms in Brazil Notre Dame, IN: University of Notre Dame Press.
• Therneau, T. (2015). A package for survival analysis in R Version 2.38. Recuperado de https://CRAN.R-project.org/package=survival
  » https://CRAN.R-project.org/package=survival
• Walker, J. L. (1969). The diffusion of innovations among the American States. American Political Science Review, 63(3), 880-899. Recuperado de https://doi.org/10.2307/1954434
  » https://doi.org/10.2307/1954434
• Wampler, B. (2016). The spread and impact of participatory budgeting: Brazil, Peru, China, South Korea, and the Philippines. In C. A. P. Faria, D. B. Coêlho, & S. D. Silva (Orgs.), Difusão de políticas públicas Santo André, SP: Editora da UFABC .
• Zucco, C. Jr., & Power, T. J. (2021). Fragmentation without cleavages? Endogenous fractionalization in the Brazilian party system. Comparative Politics, 53(3), 477-500. Recuperado de https://doi.org/10.5129/001041521X15941508069585
  » https://doi.org/10.5129/001041521X15941508069585

• 1
  Primary data. Diffusion - Micro and Small Enterprises (MSESs). Retrieved from https://doi.org/10.6084/m9.figshare.21075865

• [Translated version] Note: All quotes in English translated by this article’s translator.

Reviewers:

• 14
  Sergio Bulgacov (Universidade Positivo, Curitiba / PR - Brazil) https://orcid.org/0000-0002-9596-814X
• 15
  Maria Clara Oliveira (Universidade de Coimbra, Coimbra - Portugal; Laboratório Colaborativo para o Trabalho, Emprego e Proteção Social, Lisboa - Portugal) https://orcid.org/0000-0003-3045-6300
• 16
  One of the reviewers did not authorize the disclosure of their identity.

• Peer review report: the peer review report is available at this URL. https://bibliotecadigital.fgv.br/ojs/index.php/rap/article/view/88614/83349

Edited by

Alketa Peci (Fundação Getulio Vargas, Rio de Janeiro / RJ - Brazil) https://orcid.org/0000-0002-0488-1744

Gabriela Spanghero Lotta (Fundação Getulio Vargas, São Paulo / SP - Brazil) https://orcid.org/0000-0003-2801-1628

Publication Dates

• Publication in this collection
  17 July 2023
• Date of issue
  May-Jun 2023

History

• Received
  09 Sept 2022
• Accepted
  28 Apr 2023