Green Policymaking in Japanese Municipalities: An Empirical Study on External and Internal Contextual Factors

Abstract

This article examines the establishment and publication of green plans and green public procurement (GPP) policies in Japanese municipalities. The purpose of the study was to investigate these green policymaking initiatives from a contingency theory perspective. The first research question examined contextual factors for green policymaking. The second research question focused on barriers and enablers. For RQ1, through hypothesis testing and a regression analysis (n = 1663), we found that green policymaking differs by organization location, organization size, and organizational green capabilities. More specifically, we identified prefectures where municipalities score relatively higher as well as lower. Second, we found that larger (vs. smaller) municipalities undertake more (vs. less) green policymaking initiatives. Third, we observed that organizations with more (vs. less) green capabilities develop more (vs. less) green initiatives. For RQ2, through a descriptive and cluster analysis, we identified dominant barriers and enablers to establishing a GPP policy. The dominant barriers include a lack of information, lack of staff, and cost concerns, whereas manuals and example forms are important enablers. These findings are highly relevant to understanding and supporting green policymaking in Japanese municipalities.

1. Introduction

Environmental strategies are highly important in both private and public sector organizations. Public sector organizations are significant economic players and strive to reduce their environmental impact by undertaking various green policymaking initiatives. Within the scope of this research, these initiatives include the establishment of an organizational green plan or the implementation of a policy to purchase more environmental goods, services, and works (i.e., green public procurement (GPP) policy) by local governments. These are important initiatives, as public procurement represents the largest business sector in the world [1] and in the EU and Japan accounts for roughly 15% of GDP [2]. Moreover, globally, local authorities are encouraged by intergovernmental bodies to establish green plans to foster sustainable development (e.g., the OECD [3] and IPCC [4]).

The overall purpose of this study was to examine contextual factors, barriers, and enablers for green policymaking from a contingency perspective [5,6]. More concretely, we first investigated three contextual factors in Japanese municipalities. We identified the organization location and size as external variables, and organizational capabilities as internal variables. Second, our research addressed the barriers and enablers for green policymaking initiatives. Consequently, our research questions (RQs) are as follows:

• RQ1a. Do green policymaking initiatives differ among Japanese municipalities, depending on their organization location and organization size (i.e., external contextual factors)?
• RQ1b. Do green policymaking initiatives differ among Japanese municipalities, depending on their organizational green capabilities (i.e., internal contextual factors)?
• RQ2a. Which barriers to green policymaking initiatives exist among Japanese municipalities?
• RQ2b. Which enablers for green policymaking initiatives exist among Japanese municipalities?

We had access to a survey of the Ministry of Environment, Japan (MOEJ) that collected data on municipalities’ green policymaking initiatives, including green plans and GPP policies. For RQ1, we investigated the role of contextual factors by means of hypothesis testing (i.e., ANOVA or analysis of variance) and a regression analysis. As such, this work represents a search for the contextual factors that influence green policymaking initiatives. Furthermore, RQ2 examined the relationship between green policymaking initiatives, barriers, and enablers through a descriptive and cluster analysis. In sum, for RQ2, we explored how municipalities can implement more green policymaking initiatives.

We consider this work to be of utmost importance, due to three significant contributions. First, it provides a framework for empirical research on green policymaking initiatives within the context of organizational theory. This approach has the potential to inspire future studies and promote the advancement of this research field. Second, the dataset analyzed in this research comprises an unprecedented number of cases (n = 1663), which significantly enhances the external validity of the findings and corroborates earlier research. Finally, the robust external validity of the study allows for the development of practical guidelines, which could be of great value to policymakers.

This article continues with the background of green policymaking in Japan (Section 2) and the proposition development (Section 3). The research method that we applied in this study is explained in Section 4. Afterwards, the results are presented in Section 5 (RQ1) and Section 6 (RQ2). The discussion follows in Section 7, and Section 8 ends with concluding remarks.

2. Background

To elicit the research propositions for our study (Section 3), we first present the background of green public procurement (GPP) (Section 2.1), local green policymaking in Japan (Section 2.2), and previous research (Section 2.3).

2.1. Green Public Procurement in Japan

Public procurement refers to governmental actors that purchase goods, services, or works from companies [2]. When a government develops purchasing policies in response to social, environmental, and economic concerns, this is referred to as sustainable public procurement (SPP) [7]. This generic term covers, for instance, ethical trade (i.e., social criteria), green procurement (i.e., environmental criteria), and local SME procurement (i.e., economic criteria). This study focuses on green public procurement (GPP), which is defined as “a process whereby public authorities seek to procure goods, services and works with a reduced environmental impact throughout their life cycle when compared to goods, services and works with the same primary function that would otherwise be procured” [8].

GPP has an important role within environmental policies, for three key reasons. First, because of their considerable size, requests from governmental bodies for greener solutions cannot be neglected by market suppliers [9]. Second, these environmental demands from public actors may also set an example for the private sector and create markets for ecological goods, services, and works [10]. Finally, GPP might develop the general awareness of environmental issues [11].

Japan has been acknowledged as an early adopter and leading country in GPP [12]. In 1995, the Japanese national government introduced the Action Plan for the Greening of Government Organizations, requiring all public bodies to develop their own environmental procurement policy [13]. In 2001, the Japanese Green Purchasing Law obligated all national institutions to buy green products, as specified by the law. The list of designated products included 101 items in 14 categories in 2001 and evolved into 287 items in 22 categories in 2023. Besides this green purchasing obligation, the law further requires the institutions to publicly report on their green purchasing activities and to adopt a plan-do-check-act cycle for their GPP. This means national institutions must follow an approach where they develop a GPP policy (i.e., plan), implement the policy (i.e., do), analyze the policy achievements (i.e., check), and improve the policy (i.e., act) [14]. As a result, in 2017, it was estimated that green purchasing by national organizations has reduced greenhouse gas emissions by approximately 34,570 tons of CO₂ emissions [15].

2.2. Green Policymaking in Japanese Municipalities

Japan is a unitary state with three administration levels: national, prefectures, and municipalities. As of January 2023, there are 47 prefectures, 1724 municipalities, and 20 so-called “designated cities” (i.e., Yokohama, Osaka, Nagoya, Sapporo, Fukuoka, Kobe, Kawasaki, Kyoto, Saitama, Hiroshima, Sendai, Chiba, Kitakyushu, Sakai, Niigata, Hamamatsu, Kumamoto, Sagamihara, Okayama, and Shizuoka). These cities with more than 500,000 inhabitants have a special status that makes them equivalent to prefectures. They can perform many of the functions that are normally performed by prefectural governments.

Obviously, there are various interpretations of green policymaking in local governments. Green policymaking in Japanese municipalities can focus on GPP [14] as presented in Section 2.1, but also on urban regeneration [16], renewable energy [17], emissions [18], public transport [19], and bicycle planning [20], to mention a few. Therefore, it is important to discuss local green policymaking as interpreted within the scope of this study. More generally, the present study reports on green policymaking initiatives that are included in the examined MOEJ dataset. This survey collects data on municipalities’ establishment and publication of green plans and GPP policies.

In contrast to national institutions, municipalities have no obligations under the 2001 Green Purchasing Law. However, green procurement is strongly promoted across local governments [21]. The MOEJ, in collaboration with the non-governmental Green Purchasing Network (GPN), have undertaken several initiatives for local governments. This support includes training for green procurement staff and assistance in formulating and reviewing GPP policies [14].

Besides the establishment of a GPP policy, Japanese municipalities also undertake less demanding, but still environmentally beneficial, initiatives. These actions include the establishment of: (1) a “basic environmental plan” (i.e., the basis of environmental policies), (2) an environmental management system (EMS), (3) a “climate change prevention action plan” (i.e., to prevent climate change), and/or (4) a “circular society promotion plan” (i.e., to promote a circular society). Although these initiatives are less challenging than implementing a GPP policy, establishing green plans is regarded as an important step in the process of improving environmental performance. The general idea is that after raising awareness and motivation (e.g., [22]), plans should be established (e.g., [23]), which leads to the development of key performance indicators (e.g., [24]), followed by further implementation and evaluation steps (e.g., [25]).

Another important dimension in green policymaking is transparency. Similar to corporate greenwashing [26], local government greenwashing is extremely harmful, as the public might lose confidence in environmental protection [27]. Citizens have a crucial role, as they will conceivably sense the difference between reality and the publicity of municipalities [27]. To enhance social supervision, local governments are strongly advised to make information about their green policymaking initiatives publicly available [28]. Moreover, research has shown that increased transparency through online publication leads to a stronger environmental performance of cities [29].

In summary, the examined local green policymaking initiatives in this study include the establishment of green plans and the subsequent implementation in the form of GPP policies. Moreover, the publication of the green plans and GPP policies are also taken into consideration. We further elaborate on these dimensions in Section 4, where we present the research model, dataset, hypotheses, and variables.

2.3. Related Work

To the best of our knowledge, the related work on green policymaking initiatives in Japan, as interpreted in this study, is rather limited. From a national policy perspective, Hayami et al. [30] found that green procurement policies can improve both the environmental and economic performance of manufacturing supply chains. At the local level, Miyamoto et al. [14] investigated GPP product categories and the importance of a GPP policy. They concluded that air conditioners suffer from low green purchasing rates, whereas most municipalities purchase green products in the paper products and stationary categories. Moreover, they found that the presence of a GPP policy is associated with more green purchasing. Finally, Darnall et al. [21], in a technical report, presented a survey among 860 municipalities with more than 25.000 residents. The results showed that 53% had a GPP policy, 29% did not have a GPP policy, and 18% did not know if they had a GPP policy.

3. Proposition Development

3.1. Theoretical Framework

We opted to examine the MOEJ dataset by relying on a theoretical framework. This does not mean we sought to prove a theory. We rather adopted an organizational theory to structure the empirical research. More specifically, we investigated which contextual factors influenced local green policymaking initiatives by relying on contingency theory [5]. In essence, contingency theory states that organizational performance outcomes are a result of the fit between the organizations’ external environment and internal arrangements (i.e., contextual factors). As a change in any of these contextual factors can cause an adjustment in the corresponding organizational characteristics, organizations are assumed to profit more from context-aware initiatives than generic best practices [5].

The decision to structure this empirical research by relying on contingency theory had three key motivations. First, we responded to the legitimate call of Van de Ven et al. [6] to return to the frontier of organization science by reopening the study of contingency theory. Second, the distinction between external and internal contextual factors is an important area of debate in GPP research. For instance, Testa et al. [31] showed that external variables (e.g., organization size) can be offset by self-determined strategic choices, commitment, and organizational efforts (i.e., internal variables). Finally, as presented in the subsequent sections, the GPP research so far has been unclear regarding the influence of certain contextual factors. Therefore, it is highly relevant to extract, define, and further examine them within the framework offered by contingency theory.

The MOEJ dataset allowed us to examine organization location, organization size, and organizational capabilities as contextual factors. Consequently, we examined the GPP literature, in order to develop hypotheses regarding these contextual factors. The literature for proposition development is discussed in the subsequent sections.

Table 1. Overview of the examined literature.

Section	Authors and References
Organization location	Aldenius and Khan [32]; Testa et al. [33]
Organization size	Marron [34]; Michelsen and de Boer [35]; Testa et al. [33]; Testa et al. [31]; Prier et al. [36]
Organizational green capabilities	Testa et al. [31]; Testa et al. [33]; Cheng et al. [37]; Witjes and Lozano [38]; Ahsan and Rahman [39]; Günther and Scheibe [40]; Oruezabala and Rico [41]; Parikka-Alhola [42]; Clement et al. [9]; Leire and Mont [43]; Varnäs and Balfors [44]; Björklund and Gustafsson [45]; Carter and Jennings [46]; Erridge and Hennigan [47]; Grandia [48]; Preuss and Walker [49]; Sporrong and Bröchner [50]; Bratt et al. [51];
Organizational barriers and enablers	Brammer and Walker [7]; Preuss [10]; Thomson and Jackson [12]; Testa et al. [31]; Aldenius and Khan [32]; Testa et al. [33]; Carter and Jennings [46]; Erridge and Hennigan [47]; Grandia [48]; Preuss and Walker [49]; Sporrong and Bröchner [50]; Fet et al. [52]; Ho et al. [53]; Zhu et al. [54]; Hall et al. [55]; Preuss [56]; Guenther et al. [57]; Walker and Brammer [58]; Meehan and Bryde [59]

presents an overview.

It should be noted that additional factors might be at stake. However, factors that could not be examined with our dataset or that are not underpinned by peer reviewed research are not included here.

3.2. Organization Location

To the best of our knowledge, the research on organization location as a contextual factor has been limited. In a study on GPP to promote renewable fuels in public bus transport systems, Aldenius and Khan [32] found motivational differences between two Swedish regions. While, in one region, procurement was used in a strategic way to create a local market for biofuels, in the other region, procurement was used instrumentally to increase the share of biofuels in a cost-effective way. Therefore, the authors highlighted the importance of context when assessing GPP schemes [32].

Second, in an Italian study (n = 130), Testa et al. [33] found that public authorities of the Lazio region (i.e., center of Italy) were less likely to develop GPP initiatives compared to the public authorities located in the Emilia Romagna region (i.e., north of Italy). An explanation for these differences was lacking.

Although the prior research has been limited and not sufficiently underpinned, we hypothesized that organization location and green policymaking initiatives are dependent.

3.3. Organization Size

The literature provides mixed evidence for organization size as a contextual factor. First, based on economic modeling, Marron [34] argued that the potential of GPP is higher when the public sector is a large coordinated purchaser of products. Second, in a Norwegian study (n = 109), Michelsen and de Boer [35] showed that green procurement was significantly better established in large municipalities than in small ones. Moreover, in an Italian study (n = 130), the econometric model indicated that population (i.e., as a proxy of organization size) was a significant variable for explaining GPP [33].

However, a similar study by the same authors for the region of Tuscany in Italy (n = 62), found that the size of municipalities did not influence the adoption of GPP practices. Instead, differences among small and large municipalities were narrowed by providing training initiatives and guidelines on GPP [31]. Finally, in the United States (n = 94), Prier et al. [36] showed that larger municipalities did not score better than smaller municipalities for SPP efforts.

Although the previous evidence is mixed, we followed the reasoning that larger municipalities score better than smaller ones and hypothesized that organization size and green policymaking initiatives are dependent.

3.4. Organizational Green Capabilities

In contrast to external contextual factors such as organization location (Section 3.2) and organization size (Section 3.3), researchers have been more consistent on the importance of internal organizational capabilities. The lack of capabilities to deal with GPP challenges goes beyond the location and size of the organization [31,33]. Previous research offers various organizational capabilities for successful GPP implementation.

First, well-developed GPP procedures and systems should be in place. For instance, Cheng et al. [37] outlined a procedure of prerequirements, calls, selection, awarding, and contracting, while Witjes and Lozano [38] proposed a system of (non)technical product and service specifications. Other authors proposed prioritization tools to understand the trade-offs between various environmental aspects [39,40,41,42].

Second, it is recommended that these procedures and systems are well-documented. It was shown that a written purchasing strategy is a success criteria [9], written internal policies give direction [43], and that the lack of a specific formulation restricts GPP applications [44].

Another important capability concerns the collaboration and communication between actors. Testa et al. [31] concluded that only when increasing the organizational and operational links between the purchasing department and other departments, will the municipality be prepared to fruitfully apply GPP. Similarly, Björklund and Gustafsson [45] stressed the importance of exchanges during all phases of the development and implementation of the GPP initiative.

Next, training activities on GPP are widely supported as an organizational capability [31,46,47,48,49,50]. The main aim is to raise awareness among purchasers [49]. Once a “change in perspective” has been achieved, training can then focus on various key approaches that have significant implications at the managerial level [31].

Finally, this training should lead to well-informed and competent personnel that are held responsible for the GPP policy [31]. Subsequently, improving the environmental competences of the responsible purchasing team is a continuous process [51].

We conclude that the previous research is affirmative about the importance of organizational capabilities for successful GPP. Consequently, we hypothesized that organizational green capabilities and green policymaking initiatives are dependent.

3.5. Organizational Green Performance

A final important element within the contingency perspective [5,6] is organizational performance. For the present study, this concerns the environmental performance of municipalities, resulting from green policymaking initiatives (see Section 2.2). In providing evidence of organizational green performance, this study is confronted with similar difficulties as experienced by earlier research.

The main obstacle is that green policymaking initiatives need time to be completely effective and to result in stronger organizational green performance. New policy initiatives require adaptation to the overall management system, which includes organizational structure, planning activities, responsibilities, practices, procedures, processes, and resources [60,61,62,63]. By consequence, regarding our study, there might be an “unknown delay” between the establishment (and publication) of a green plan and GPP policy (see Section 2.2), and the resulting organizational green performance.

Furthermore, any statistical attempt to overcome this “unknown delay” by creating panel data (i.e., the MOEJ survey started in 2001) is prohibited by the large-scale mergers between Japanese municipalities. Following the Special Merger Law of 1995, the number of municipalities has significantly decreased. Whereas the number of municipalities was 3234 in 1995, this figure had decreased to 1724 by 2018.

Given these restrictions, we cannot examine and elaborate on organizational green performance. We acknowledge that this is an important limitation on the adoption of the contingency perspective [5,6]. However, after thorough investigation, we believe any attempt to do so with the current dataset would be based on flawed assumptions.

In conclusion, providing evidence for stronger organizational green performance is outside the scope of this study. The focus concerns contextual factors and barriers and enablers (see Section 3.6) of green policymaking initiatives. To the best of our knowledge, we are not aware of any research indicating that green policymaking initiatives might be counterproductive, i.e., leading to a weaker organizational green performance. Therefore, in line with the contingency perspective and earlier research, we simply assume that green policymaking initiatives will lead to a stronger environmental performance. However, restricted by a lack of appropriate data, we did not undertake any attempt to confirm or quantify this relationship.

3.6. Organizational Barriers and Enablers

Besides examining contextual factors (i.e., RQ1), the study will subsequently focus on municipalities that scored rather low for green policymaking initiatives (i.e., RQ2). Therefore, this section presents potential barriers and enablers for successful GPP implementation. The conducted analysis for RQ2 (see Section 6) did not require the development of hypotheses. Instead, the literature presented here provides a background for the analyzed variables.

First, a lack of information and lack of procedures are widely acknowledged as barriers to GPP establishment [12,52,53,54]. It is argued that the inclusion of environmental criteria in public tenders requires technical expertise, which is sometimes lacking in government procurement staff [54]. Moreover, it might also be the case that the staff themselves are lacking. Hall et al. [55] found that many small public organizations clearly lack the human resources for GPP establishment. Interestingly, Preuss [56] presented financial pressure as the most salient barrier to GPP implementation. Several studies support this finding, by showing that there is a common perception among public procurers that it costs more and takes longer to carry out purchasing when environmental requirements are included [7,10,32,57,58]. Finally, it was also shown that procurers are not only concerned about the costs involved, they might also doubt the effectiveness of GPP [7].

As enablers, several studies support the development of tools to assist local authorities [59]. Manuals with procedures or example forms with specifications can help municipalities in establishing GPP [12,52,53,54]. Other enablers might be expert assistance or a consultation desk. Testa et al. [33] argued that external assistance might be valuable to identify the juridical boundaries of the tender, to choose the correct and most appropriate modalities, and to articulate the corresponding technical criteria. Additional external knowledge might also come from other local governments having a successful GPP strategy [31]. Finally, GPP briefing sessions for local government officials might be another enabler, as these can produce similar benefits to those of training sessions [31,46,47,48,49,50].

4. Research Method

The research method relies on an empirical and quantitative research design. Section 4.1 shows the research model, whereas Section 4.2 elaborates on the dataset. The hypotheses and variables are presented in Section 4.3 (RQ1) and Section 4.4 (RQ2).

4.1. Research Model

Figure 1 presents the research model with a contingency perspective, which generally focuses on the notions of “fit” or “no fit” and “performance.” The idea is that organizational green performance (i.e., Box 4) results from fitting characteristics of the organization (i.e., Box 2. Green initiatives) to contingencies that reflect its external (i.e., Box 1) and internal (i.e., Box 3) context. The dependent variable concerns a green policymaking initiative scale, and the independent variables are the organization’s location and size, and the presence of organizational green capabilities (see Section 4.3). Hence, the research model builds on the proposition that organizations operating in distinctive external and internal conditions can have a different degree of organizational green performance. The notions of “fit” or “no fit” refer to the coherence between the contextual factors and the green policymaking initiatives. As outlined in Section 3.5, the dependency between green policymaking initiatives (i.e., Box 2) and organizational green performance (i.e., Box 4) is outside the scope of this work.

Finally, RQ2 focuses on the municipalities that scored low for green policymaking initiatives (i.e., Box 2). We elaborate on the barriers and enablers with a descriptive and cluster analysis (see Section 4.4).

4.2. Dataset

We had access to a survey conducted by the Japanese Ministry of the Environment (MOEJ), called the “Questionnaire survey on green procurement by local governments”. This survey collects data on municipalities’ green policymaking initiatives, including green plans and GPP policies. The survey began in 2001 and is sent to all municipalities every year. We used the dataset from the year 2020. The survey covers all municipalities in Japan, with a very high response rate of approximately 96%.

In summary, the dataset contained answers to 25 green-initiative-related questions from 1705 municipalities spread over the 47 Japanese prefectures. Hence, the dataset ranges from the smallest Japanese municipality (i.e., Utashinai with 3019 inhabitants) to the largest city (i.e., Tokyo with 13,988,129 inhabitants) (General municipality data can be retrieved from: https://www.e-stat.go.jp/en/regional-statistics/ssdsview/municipality (accessed on 15 March 2023)). For our analysis, we found 42 incomplete records. We excluded these municipalities, resulting in a dataset of 1663 observations. In subsequent sections, we elaborate further on the variables for RQ1 (Section 4.3) and RQ2 (Section 4.4).

4.3. Variables and Hypotheses for RQ1

The MOEJ survey collects data on municipalities’ establishment and publication of green plans and GPP policies. More specifically, municipalities are asked if they have established (1) a basic environmental plan, (2) an environmental management system, (3) a climate change prevention action plan, (4) a circular society promotion plan, and (5) a GPP policy, and if these have been published (i.e., online publicly accessible).

For the construction of the green policymaking initiative scale (i.e., dependent variable), we relied on two important assumptions supported by previous research and discussed in Section 2.2. First, we assumed that planning (i.e., (1) to (4)) is less challenging than the implementation (i.e., (5)) [22,23,24,25]. Second, we ranked publication of the green plans (i.e., (1) to (4)) and/or GPP policy (i.e., (5)) higher, as transparency about green policymaking initiatives counters greenwashing and results in a stronger environmental performance [27,28,29]. Although we are relying on previous research, these assumptions are supported by the distribution of the MOEJ dataset. As shown in Figure 2, an explicit downtrend can be noticed among the municipalities, in terms of the proposed grading dimensions (i.e., planning vs. implementation, and established vs. published). As we move from the more discretionary condition (i.e., green plan established) to the more demanding condition (i.e., GPP policy published), we notice a decreasing number of municipalities fulfilling the more stringent requirements.

For the construction of a green policymaking initiative scale, we proposed a grading system that relies on the dimensions of planning vs. implementation and established vs. published. The first and lowest level on the green policymaking initiative scale concerns the municipalities that do not have any green initiatives at all (i.e., lack of (1) to (5)). These municipalities are assigned 1 point. Subsequently, as summarized in

Table 2. Grading system for the construction of the green policymaking initiative scale.

		MOEJ Survey Statements: Please Indicate the Status of Your Organization’s…
		(1) Plans that form the basis of environmental policies (e.g., the basic environmental plan)	(2) EMS (ISO 14001, Eco Action 21, your own EMS, etc.)	(3) Plans that contribute to the prevention of climate change (climate change prevention action plans, etc.)	(4) Plans that contribute to the formation of a circular society (circular society promotion plans, etc.)	(5) Green purchasing policy
	Established	1 point				1 point
	Published	1 point				1 point
Grading dimensions

, a municipality obtains one additional point for each of the following requirements fulfilled: green plan (i.e., (1) to (4)) established, green plan (i.e., (1) to (4)) published, GPP policy (i.e., (5)) established, and GPP policy (i.e., (5)) published. We did not consider any rank difference between (1) and (4), as an explicit hierarchy between these plans is lacking in practice. Japanese municipalities establish these plans on a voluntary basis, and one plan is not considered more demanding than another. However, we did make a distinction between (1) to (4), and (5). A GPP policy goes beyond a green plan, as it entails concrete implementation [22,23,24,25].

In summary, we constructed a green policymaking initiative scale based on dimensions supported by previous research and observed in our dataset. Therefore, we believe this scale can act as a valid construct for the purpose of this study. Figure 3 presents the distribution of the green policymaking initiative scale, as our dependent variable.

As shown in the research model (i.e., Figure 1), we considered three distinct contextual factors as independent variables. Therefore, we also reported on three sets of hypotheses. To concretize the contextual factor propositions (see Section 3), the potential dependencies were expressed as null hypotheses and alternative hypotheses, as presented in

Table 3. Overview of the hypotheses for each contextual factor (RQ1).

Contextual Factor	Research Hypotheses
Organization location	H0: The organization location and green policymaking initiatives are independent. HA: Green policymaking initiatives significantly differ among organization locations.
Organization size	H0: The organization size and green policymaking initiatives are independent. HA: Green policymaking initiatives significantly differ with organization size.
Organizational green capabilities	H0: Organizational green capabilities and green policymaking initiatives are independent. HA: Green policymaking initiatives significantly differ among organizational green capabilities.

. The variables related to these hypotheses are given in

Table 4. Overview of the variables for RQ1.

Variables	Operationalization	Measurement Level
Green policymaking initiatives	5-point scale	Ordinal
Organization location	47 Japanese prefectures	Nominal
Organization size	“<50”, “51–100”, “101–200”, “201–500”, “501–1000”, “1001–2000”, “2001–5000”, “>5001 employees”	Ordinal
Organizational green capabilities	6-point scale	Ordinal

.

As explained above, green policymaking initiatives are represented by an ordinal 5-point scale. Organization location is a nominal variable and represents the prefecture of the municipality. Organization size is measured in total number of employees, with eight ordinal levels. Finally, organizational green capabilities has an ordinal 6-point scale that was constructed based on six statements in the survey and in line with previous research (see Section 3.4). As shown in

Table 5. Construction of the organizational green capabilities scale, linking previous research (Section 3.4), MOEJ survey statements, and grading.

Organizational Green Capability (See Section 3.4)	MOEJ Survey Statements: Please Answer the Following Question Regarding Your Organization’s Specific Initiatives:	Grading
Documents	Specific activities are defined in documents related to procurement and contracting, such as contracts, specifications, and bidding instructions.	1 point
Systems	The system and procedures are established and implemented in the operation manuals and handbooks of the persons in charge.	1 point
Notices	Sending notifications, notices, etc. regarding green purchasing or green contracts to each department.	1 point
Responsible persons	A person in charge is appointed in each department.	1 point
Training sessions	Hold training sessions for employees (including only those in charge).	1 point
Nothing	Not implemented in particular.	1 point

, a municipality was surveyed for its organizational green capabilities. In case capabilities were lacking, the municipality was assigned one point. Subsequently, a municipality obtained one point for the presence of an additional capability.

Since all context variables were categorical, we decided to supplement the hypothesis testing (Section 5.1) with a regression analysis (Section 5.2), to strengthen the results and to determine the relative influence of the context variables.

4.4. Variables for RQ2

RQ2 investigates the municipalities that scored relatively low for green policymaking initiatives. More concretely, we considered the municipalities that did not establish (and publish) a GPP policy. This concerns the municipalities that did not score higher than 3 on the green policymaking initiative scale (see Section 4.3). In practice, these municipalities were not involved in green initiatives (i.e., 1 point) or could only fulfill the rather discretionary condition of having a green plan established (i.e., 2 points) and published (i.e., 3 points). In summary, for RQ2, we examined how municipalities can move towards the more demanding condition of having a GPP policy established and published. For the municipalities without a GPP policy, the MOEJ survey provides data on potential barriers and enablers, as shown in

Table 6. Overview of GPP policy barriers, linking previous research (Section 3.6) and MOEJ survey statements.

Barriers (See Section 3.6)	MOEJ Survey Statement: What Are the Challenges You Face in Establishing a Green Procurement Policy? Please Select Two That Apply to You in Particular.
Lack of information	Lack of information on establishment
Lack of staff	Lack of staff to address establishment
Cost concerns	Concerns about increased procurement (contract) costs due to switching to green goods
Effectiveness uncertainty	Uncertainty about the effect of reducing environmental impact by switching to green goods
Lack of procedures	Establishment of a cooperative system with departments in charge of procurement
Others	…

and

Table 7. Overview of GPP policy enablers, linking previous research (Section 3.6) and MOEJ survey statements.

Enablers (See Section 3.6)	MOEJ Survey Statement: Please Select Two Statements about Government Support That You Think Are Particularly Necessary for Developing a Green Procurement Policy.
Manuals	Provision of procedures and manuals for establishing a purchasing policy.
Example forms	Provision of examples of specifications and bidding forms for green purchasing.
Information establishment	Provision of information on the establishment of the purchasing policy of other local governments.
Expert assistance	Provide support and dispatch of experts to assist in the process of establishing a purchasing policy.
Consultation desk	Establishment of a consultation desk for the formulation of a procurement policy.
Briefing session	Briefing sessions for local government officials regarding the establishment of a purchasing policy.
Effectiveness demonstration	Presentation of the effects of reducing environmental impacts through green purchasing initiatives for each item.
Others	…

, respectively.

A descriptive analysis determined the most important barriers and enablers (Section 6.1). Finally, we aimed to retrieve policy recommendations with a cluster analysis (Section 6.2).

5. Results for RQ1

We first verified the assumptions for normality, homogeneity of variance, sample size, and multicollinearity. Using Kolmogorov–Smirnov tests (p < 0.010) and Shapiro–Wilk tests (p < 0.010), we found that the dependent variable (i.e., green policymaking initiatives) was not normally distributed for each category of the independent variables (i.e., organization location, organization size, and organizational green capabilities). Regarding the homogeneity of variance, we calculated Levene’s test in its traditional version and its nonparametric version, given the fact that our data were not normally distributed [64]. Both versions of Levene’s test gave evidence for unequal variances among all variables (p < 0.001). Third, the contingency tables for each variable showed unequal sample sizes (see Appendix A). Finally, the collinearity diagnostics (correlation, tolerance, and variance inflation factor) did not show problematic values.

5.1. ANOVA-Based Hypothesis Testing

Based on the verified assumptions for normality, homogeneity of variance, and sample sizes, we opted for an ANOVA Welch’s F test to detect differences among the categories of the independent variables [65,66]. In case the ANOVA Welch’s F test showed differences among the categories of the independent variables for the green policymaking initiative variable, we performed a Games–Howell post-hoc test to identify the categories among which a difference is expected [67].

As the contingency tables and the Games–Howell post-hoc testing tables are extensive, we opted to group them in Appendix A.

5.1.1. Organization Location

The contingency table for organization location (i.e.,

Table A1. The cross-tabulation of organization location by the green policymaking initiative scale (n = 1663).

Prefecture	Green Policymaking Initiative Scale					Total
	Score 1	Score 2	Score 3	Score 4	Score 5
Aichi	3	2	26	11	11	53
Akita	7	2	11	2	1	23
Aomori	9	8	15	1	5	38
Chiba	7	3	27	7	5	49
Fukui	2	1	5	5	4	17
Fukuoka	9	5	34	6	4	58
Fukushima	21	10	20	2	2	55
Gifu	9	5	18	5	5	42
Gunma	8	4	10	3	5	30
Hiroshima	1	3	9	1	5	19
Hokkaido	39	28	80	10	12	169
Hyogo	4	4	21	5	7	41
Ibaraki	5	6	24	8	2	45
Ishikawa	1	4	10	2	3	20
Iwate	6	3	16	6	3	34
Kagawa	1	3	11	2	1	18
Kagoshima	7	11	21	1	2	42
Kanagawa	3	1	11	7	10	32
Kochi	1	8	14	3	3	29
Kumamoto	14	4	17	5	3	43
Kyoto	2	1	9	7	4	23
Mie	2	8	10	6	1	27
Miyagi	5	5	16	3	4	33
Miyagi	3	2	9	2	2	18
Miyazaki	5	10	8	1	2	26
Nagano	27	7	26	4	5	69
Nagasaki	3	4	9	5	1	22
Nara	18	5	9	3	1	36
Niigata	5	1	8	8	8	30
Oita	3	0	13	1	2	19
Okayama	5	4	12	2	2	25
Okinawa	5	12	14	2	2	35
Osaka	6	3	16	10	8	43
Saga	4	2	11	0	3	20
Saitama	1	5	29	16	9	60
Shiga	0	2	7	7	3	19
Shimane	7	2	6	1	3	19
Shizuoka	1	4	17	7	5	34
Tochigi	1	3	11	4	6	25
Tokushima	3	5	12	2	1	23
Tokyo	7	1	22	12	19	61
Tottori	4	4	2	5	1	16
Toyama	2	0	5	4	3	14
Wakayama	10	5	12	1	2	30
Yamagata	6	5	14	6	2	33
Yamaguchi	1	3	8	4	4	20
Yamanashi	5	3	11	4	3	26
Total	298	221	726	219	199	1663

) shows unequal sample sizes among the organization locations, mainly due to the larger number of municipalities in the prefectures of Hokkaido, Nagano, Tokyo, and Saitam, and the fewer municipalities in the prefectures of Toyama and Tottori.

Alternatively, the ANOVA Welch’s F test confirmed that at least one prefecture differed from another for the green policymaking initiative scale (F = 4.637; df1 = 46; df2 = 436.491; p < 0.001). In particular, the Games–Howell post-hoc test found differences between the prefectures, as presented in

Table A2. Post-hoc testing with mean differences for organization location.

	Green Policymaking Initiative Scale
		95% Confidence Interval
Games–Howell Post-Hoc Testing	Mean Difference	Lower Bound	Upper Bound
Aichi vs. Fukushima	1.308 ***	0.46	2.15
Aichi vs. Hokkaido	0.898 **	0.21	1.59
Aichi vs. Kagoshima	0.948 **	0.10	1.80
Aichi vs. Kumamoto	0.960 *	−0.03	1.95
Aichi vs. Miyazaki	1.049 *	−0.05	2.15
Aichi vs. Nagano	1.153 ***	0.30	2.00
Aichi vs. Nara	1.472 ***	0.46	2.48
Aichi vs. Okinawa	0.929 **	0.00	1.85
Aichi vs. Wakayama	1.138 **	0.05	2.23
Chiba vs. Fukushima	0.836 *	−0.05	1.72
Chiba vs. Nara	1.000 *	−0.04	2.04
Fukushima vs. Hokkaido	0.750 *	−0.04	1.54
Fukushima vs. Hyogo	−1.007 **	−1.96	−0.05
Fukushima vs. Kanagawa	−1.461 **	−2.57	−0.35
Fukushima vs. Kyoto	−1.271 **	−2.46	−0.09
Fukushima vs. Niigata	−1.270 **	−2.51	−0.03
Fukushima vs. Osaka	−1.092 **	−2.09	−0.09
Fukushima vs. Saitama	−1.286 ***	−2.06	−0.51
Fukushima vs. Shiga	−1.415 **	−2.51	−0.32
Fukushima vs. Shizuoka	−1.160 **	−2.08	−0.23
Fukushima vs. Tochigi	−1.276 **	−2.42	−0.13
Fukushima vs. Tokyo	−1.410 ***	−2.31	−0.51
Hokkaido vs. Kanagawa	−1.051 **	−2.06	−0.04
Hokkaido vs. Saitama	−0.876 ***	−1.47	−0.28
Hokkaido vs. Shiga	−1.005 *	−2.01	0.00
Hokkaido vs. Shizuoka	−0.750 *	−1.54	0.04
Hokkaido vs. Tochigi	1.005 *	0.00	2.01
Hokkaido vs. Tokyo	−1.000 ***	−1.75	−0.24
Hyogo vs. Nara	1.171 **	0.08	2.26
Kagoshima vs. Kanagawa	−1.101 *	−2.21	0.01
Kagoshima vs. Saitama	−0.926 **	−1.71	−0.14
Kagoshima vs. Shiga	−1.055 *	−2.16	0.05
Kagoshima vs. Tokyo	−1.050 **	−1.95	−0.15
Kanagawa vs. Nagano	1.306 **	0.20	2.42
Kanagawa vs. Nara	1.625 **	0.40	2.85
Kanagawa vs. Wakayama	1.292 **	0.01	2.58
Kumamoto vs. Saitama	−0.938 **	−1.87	0.00
Kumamoto vs. Tokyo	−1.062 **	−2.10	−0.03
Kyoto vs. Nara	1.435 **	0.15	2.72
Miyazaki vs. Saitama	1.027 *	−0.03	2.08
Miyazaki vs. Tokyo	−1.151 **	−2.29	−0.01
Nagano vs. Kanagawa	−1.306 **	−2.42	−0.20
Nagano vs. Kyoto	−1.116 *	−2.31	0.07
Nagano vs. Nara	1.121 *	−0.02	2.27
Nagano vs. Osaka	1.116 *	−0.07	2.31
Nagano vs. Saitama	−1.131 ***	−1.91	−0.35
Nagano vs. Shiga	−1.260 **	−2.36	−0.16
Nagano vs. Shizuoka	−1.005 **	−1.93	−0.08
Nagano vs. Tokyo	−1.255 ***	−2.16	−0.35
Nara vs. Niigata	−1.433 **	−2.77	−0.10
Nara vs. Osaka	−1.256 **	−2.39	−0.13
Nara vs. Saitama	−1.450 ***	−2.40	−0.50
Nara vs. Shiga	−1.579 **	−2.79	−0.37
Nara vs. Shizuoka	−1.324 **	−2.39	−0.25
Nara vs. Tochigi	−1.440 **	−2.69	−0.19
Nara vs. Tokyo	−1.574 ***	−2.62	−0.52
Nara vs. Yamaguchi	1.350 *	−0.02	2.72
Okinawa vs. Saitama	−0.907 **	−1.77	−0.04
Okinawa vs. Tokyo	−1.031 **	−2.00	−0.06
Saitama vs. Wakayama	1.117 **	0.07	2.16
Shiga vs. Wakayama	−1.246 *	−2.52	0.00
Tokyo vs. Wakayama	1.240 **	0.11	2.37

* p < 0.100, ** p < 0.050, *** p < 0.001.

(p < 0.100).

From the cross-tabulation and the Games–Howell post-hoc test, we learned that this difference was explained by a group of prefectures scoring relatively higher (i.e., Score 3 or higher) than a group of prefectures scoring relatively lower (i.e., Score 2 or lower).

5.1.2. Organization Size

As shown in

Table A3. The cross-tabulation of organization size by the green policymaking initiative scale (n = 1663).

Number of Employees	Green Policymaking Initiative Scale					Total
	Score 1	Score 2	Score 3	Score 4	Score 5
<50	31	16	13	1	0	61
51–100	108	72	84	5	3	272
101–200	103	74	172	20	9	378
201–500	51	48	269	62	36	466
501–1000	3	10	139	67	27	246
1001–2000	2	1	30	32	34	99
2001–5000	0	0	18	32	48	98
>5001	0	0	1	0	42	43
Total	298	221	726	219	199	1663

, the contingency table for organization size shows unequal sample sizes. The number of smaller (i.e., “<50” and “51–100”) and larger (i.e., “1001–2000”,”2001–5000” and “>5000”) municipalities are relatively low, compared to the majority of medium-sized municipalities (i.e., “101–200”, “201–500” and “501–1000”).

The ANOVA Welch’s F test confirmed that at least one organization size differed from another for the green policymaking initiative scale (F = 368.392; df1 = 7; df2 = 362.026; p < 0.001). From the Games–Howell post-hoc testing (i.e.,

Table A4. Post-hoc testing with mean differences for organization size.

	Green Policymaking Initiative Scale
		95% Confidence Interval
Games–Howell Post-Hoc Testing	Mean Difference	Lower Bound	Upper Bound
<50 vs. 101–200	−0.622 ***	−1.00	−0.25
<50 vs. 201–500	−1.228 ***	−1.60	−0.86
<50 vs. 501–1000	−1.689 ***	−2.06	−1.32
<50 vs. 1001–2000	−2.222 ***	−2.67	−1.78
<50 vs. 2001–5000	−2.568 ***	−2.98	−2.15
<50 vs. >5001	−3.216 ***	−3.59	−2.85
51–100 vs. 101–200	−0.378 ***	−0.61	−0.14
51–100 vs. 201–500	−0.984 ***	−1.21	−0.76
51–100 vs. 501–1000	−1.445 ***	−1.68	−1.21
51–100 vs. 1001–2000	−1.978 ***	−2.32	−1.64
51–100 vs. 2001–5000	−2.325 ***	−2.62	−2.03
51–100 vs. >5001	−2.972 ***	−3.20	−2.75
101–200 vs. 201–500	−0.606 ***	−0.82	−0.39
101–200 vs. 501–1000	−1.067 ***	−1.29	−0.85
101–200 vs. 1001–2000	−1.600 ***	−1.93	−1.27
101–200 vs. 2001–5000	−1.946 ***	−2.23	−1.66
101–200 vs. >5001	−2.594 ***	−2.81	−2.38
201–500 vs. 501–1000	−0.461 ***	−0.67	−0.25
201–500 vs. 1001–2000	−0.994 ***	−1.32	−0.67
201–500 vs. 2001–5000	−1.340 ***	−1.62	−1.06
201–500 vs. >5001	−1.988 ***	−2.19	−1.79
501–1000 vs. 1001–2000	−0.533 ***	−0.86	−0.21
501–1000 vs. 2001–5000	−0.879 ***	−1.16	−0.60
501–1000 vs. >5001	−1.527 ***	−1.74	−1.32
1001–2000 vs. 2001–5000	−0.347 *	−0.72	0.03
1001–2000 vs. >5001	−0.994 ***	−1.32	−0.67
2001–5000 vs. >5001	−0.647 ***	−0.93	−0.37

* p < 0.100, *** p < 0.001.

), it could be concluded that a significant difference exists between the distinct levels of organization size. Except for the “<50 vs. 51–100” comparison, we found significant differences for organization size (p < 0.001).

5.1.3. Organizational Green Capabilities

The contingency table for organizational green capabilities shows unequal sample sizes (i.e.,

Table A5. The cross-tabulation of organizational green capabilities by the green policymaking initiative scale (n = 1663).

Organizational Green Capabilities	Green Policymaking Initiative Scale					Total
	Score 1	Score 2	Score 3	Score 4	Score 5
Level 1	274	192	562	65	23	1116
Level 2	21	25	126	70	64	306
Level 3	3	3	25	37	43	111
Level 4	0	1	9	26	36	72
Level 5	0	0	3	16	21	40
Level 6	0	0	1	5	12	18
Total	298	221	726	219	199	1663

). Many municipalities have a low level of organizational green capabilities (i.e., Level 1) compared to the few municipalities that have a high level (i.e., Level 6).

Furthermore, the ANOVA Welch’s F test showed that at least one level of organizational green capabilities differed from another for the dependent variable (F = 213.179; df1 = 5; df2 = 115.314; p < 0.001). From the Games–Howell post-hoc test (i.e.,

Table A6. Post-hoc testing with mean differences for organizational capabilities.

	Green Policymaking Initiative Scale
		95% Confidence Interval
Games–Howell Post-Hoc Testing	Mean Difference	Lower Bound	Upper Bound
Level 1 vs. Level 2	−0.992 ***	−1.19	−0.79
Level 1 vs. Level 3	−1.591 ***	−1.87	−1.31
Level 1 vs. Level 4	−1.911 ***	−2.18	−1.64
Level 1 vs. Level 5	−2.014 ***	−2.33	−1.70
Level 1 vs. Level 6	−2.175 ***	−2.64	−1.71
Level 2 vs. Level 3	−0.599 ***	−0.92	−0.27
Level 2 vs. Level 4	−0.919 ***	−1.23	−0.60
Level 2 vs. Level 5	−1.022 ***	−1.37	−0.67
Level 2 vs. Level 6	−1.183 ***	−1.67	−0.70
Level 3 vs. Level 5	−0.423 **	−0.82	−0.02
Level 3 vs. Level 6	−0.584 **	−1.10	−0.07

** p < 0.050, *** p < 0.001.

), it can be concluded that a difference exists between all level comparisons except for “Level 3 vs. Level 4”, “Level 4 vs. Level 5”, “Level 4 vs. Level 6”, and “Level 5 vs. Level 6”. This indicates that a statistically significant difference in green policymaking initiatives exists among the different levels of organizational green capabilities.

5.2. Regression Analysis

A linear regression was conducted with the green policymaking initiative scale as a dependent variable. The different categories of contextual factors were recoded as dummies. As the aim was to find out which categories of the contextual factors were significant contributors to green policymaking initiatives, the stepwise regression method was used. Stepwise linear regression is a method of regressing multiple variables, while simultaneously removing those that do not contribute to the regression equation. We first conducted the regression for the categories of each contextual factor separately. Subsequently, we ran the regression for the categories of all contextual factors. The resulting regression equations, limited to the statistically significant variables (p < 0.100), were as follows:

• Green policymaking initiatives ^{organization location} = 2.770 + 0.804 × (Tokyo) − 0.606 × (Fukushima) − 0.770 × (Nara) − 0.451 × (Nagano) − 0.196 × (Hokkaido) + 0.680 × (Saitama) + 0.702 × (Aichi) + 0.855 × (Kanagawa) − 0.437 × (Wakayama) + 0.809 × (Shiga) + 0.663 × (Niigata) + 0.670 × (Tochigi) + 0.665 × (Kyoto) + 0.554 × (Shizuoka) + 0.486 × (Osaka) + 0.701 × (Fukui) + 0.580 × (Yamaguchi) + 0.401 × (Hyogo) + 0.659 × (Toyama) + 0.546 × (Hiroshima)
• Green policymaking initiatives ^{organization size} = 3.960 − 2.222 × (<50 employees) − 1.978 × (51 − 100 employees) − 1.600 × (101 − 200 employees) − 0.994 × (201 − 500 employees) − 0.533 × (501 − 1000 employees) + 0.347 × (2001 − 5000 employees) + 0.994 (>5001 employees)
• Green policymaking initiatives ^{organizational green capabilities} = 4.415 − 1.979 × (Score 1) − 0.987 × (Score 2) − 0.388 × (Score 3)
• Green policymaking initiatives ^all = 4.494 − 0.569 × (Nara) + 0.499 × (Shiga) − 0.310 × (Fukushima) − 0.239 × (Nagano) − 0.393 × (Akita) − 0.315 × (Wakayama) − 1.699 × (< 50 employees) − 1.519 × (51 − 100 employees) − 1.194 × (101 − 200 employees) − 0.740 × (201 − 500 employees) − 0.430 × (501 − 1000 employees) + 0.605 × (>5001 employees) − 0.981 × (Score 1) − 0.437 × (Score 2)

The above equations confirm or are in line with the ANOVA-based findings from Section 5.2. Regarding organization location, the differences between prefectures are similar to the findings in Section 5.1.1. The regression analysis also supported the importance of organization size (i.e., Section 5.1.2), with almost all levels presented in the equations. Finally, it seems important to have a certain level of organizational green capabilities (i.e., Score 1 to Score 3). A more advanced level (i.e., Score 4 to Score 6) seemed to contribute less. This finding was also shown in Section 5.1.3.

6. Results for RQ2

As outlined in Section 4.4, RQ2 focuses on the municipalities that did not score higher than 3 on the green policymaking initiative scale. This concerned 1206 out of 1663 municipalities. For these organizations, we investigated the potential barriers and enablers to moving towards the more demanding condition of having a GPP policy established and published (see Section 2.2). While Section 6.1 provides a descriptive analysis, Section 6.2 presents the results of a cluster analysis. Further interpretation and discussion of these results is given in Section 7.

6.1. Descriptive Analysis

While Figure 4 shows the potential barriers (i.e., RQ2a), Figure 5 depicts the potential enablers (i.e., RQ2b) with n = 2412, as municipalities were asked to stipulate two barriers and two enablers. As can be noted from Figure 4, municipalities indicated that lack of staff, lack of info, and cost concerns were the main barriers to a GPP policy, followed by establishment of a system and effectiveness uncertainty. Figure 5 shows that manuals and example forms are the main enablers, followed by information establishment, effectiveness demonstration, briefing sessions, expert assistance, and a consultation desk.

6.2. Cluster Analysis

Cluster analysis is a statistical technique used to group a set of objects in such a way that objects in the same group (i.e., a cluster) are more similar to each other than to those in the other groups (i.e., clusters) [68]. It concerns an exploratory analysis that requires a researcher’s interpretation to determine meaningful results [69].

For this study, we determined that three clusters were the only relevant outcome. More specifically, cluster 1, cluster 2, and cluster 3 should present groups of municipalities that had Score 1, Score 2, and Score 3, respectively, on the green policymaking initiative scale. Next, depending on the cluster attributes (i.e., distinct barriers and enablers), we could retrieve specific policy recommendations to assist each cluster of municipalities to obtain a GPP policy.

Subsequently, in IBM SPSS Statistics 27, we determined three clusters as outcomes and applied the k-means clustering method. The independent variables were the green policymaking initiative scale (i.e., Score 1 to Score 3), with the barriers and enablers as presented in Section 6.1. The results are shown

Table 8. The tree-cluster solution derived using the k-means method.

Attributes	Cluster 1	Cluster 2	Cluster 3
Green policymaking initiatives	Score 1	Score 2	Score 3
Lack of info		X
Lack of staff	X	X	X
Cost concerns	X		X
Manuals	X	X	X
Example forms	X

.

As intended, the clustering algorithm grouped Score 1, Score 2, and Score 3 from the green policymaking initiatives as distinct clusters. Additionally, the algorithm generated clusters based on the dominant barriers (i.e., lack of info, lack of staff, and cost concerns) and enablers (i.e., manuals and example forms) as presented in Section 6.1.

7. Discussion

The results from this study contribute to previous GPP research in Japan [14,21,30] (see Section 2.3). Moreover, valuable comparisons can be made with the examined literature in Section 3. The subsequent sections discuss the findings in light of the literature summarized in Table 1.

7.1. Organization Location

Based on the Games–Howell post-hoc testing (Section 5.1.1) and the regression analysis (Section 5.2), we rejected the null hypothesis and concluded that green policymaking initiatives differed among the organizational locations.

Table 9. Overview of prefectures that scored relatively lower and higher for green policymaking initiatives.

Green Policymaking Initiatives
Relatively lower score	Fukushima, Nara, Nagano, Hokkaido, Wakayama
Relatively higher score	Tokyo, Saitama, Aichi, Kanagawa, Shiga, Niigata, Tochigi, Kyoto, Shizuoka, Osaka, Fukui, Yamaguchi, Hyogo, Toyama, Hiroshima

presents the prefectures with municipalities that scored relatively lower and higher for green policymaking initiatives.

As presented in Section 3.2, organizational location has rarely been investigated. We contributed to the research on this contextual factor by confirming earlier results. While differences were found between two Swedish regions [32] and two Italian regions [33], our study presented differences between several Japanese prefectures. The previous research was not only confirmed but also reinforced. In contrast to the Swedish study (n = 2) [32] and the Italian study (n = 130) [33], we derived this conclusion from a significantly larger sample (n = 1663).

While Aldenius and Khan [32] found motivational differences between the Swedish regions, an explanation for the Italian differences was lacking [33]. Similarly, our study cannot provide a potential explanation for this relationship. First, the dataset only allowed fact finding. We could statistically examine the “what” but not the “why”. Second, regarding green policy, Japanese municipalities are more influenced by national than by prefectural authorities. Prefectural governments do not have the instruments to affect green policymaking in municipalities. Municipalities, instead, depend on support programs (e.g., training for green procurement staff) by the national government [14]. As a final comment, we note that the prefectures that scored relatively lower were in more rural regions, whereas prefectures that scored relatively higher were more urban areas. For now, this is solely an observation by the authors. However, it might be worthwhile to examine this in more depth, as regional differences in green policymaking are an under-investigated topic [31,32].

7.2. Organization Size

Based on Games–Howell post-hoc testing (Section 5.1.2) and the regression analysis (Section 5.2) we rejected the null hypothesis and concluded that green policymaking initiatives differ by organization size. As depicted in Figure 6, we statistically confirmed that larger (vs. smaller) organizations undertake more (vs. less) green policymaking initiatives. Smaller organizations (i.e., <50, 51–100, and 101–200 employees) were relatively more represented in Score 1 and Score 2, while larger organizations (i.e., 501–1000, 1001–2000, 2001–5000, and >5001 employees) were relatively more represented in Score 4 and Score 5.

As discussed in Section 3.3, the evidence is mixed for organization size as a contextual factor. Our research showed a dependency between green policymaking and organization size. This aligns with results from a Norwegian study (n = 109) [35] and an Italian study (n = 130) [33] and counters findings from another Italian study (n = 62) [31] and an US study (n = 94) [36]. With our research, we provide evidence that organization size is a contextual factor for Japanese municipalities.

However, as for organization location, our study does not provide an explanation for the dependency. As suggested by earlier research, the availability of more resources in larger municipalities might explain the relationship. For instance, large municipalities might have more resources for establishing a purchasing department, which can generate knowledge and develop purchasing strategies [33,35]. Although we believe this might also be a viable explanation in the Japanese context, a more in-depth analysis of this relationship should confirm this.

7.3. Organizational Green Capabilities

Based on the Games–Howell post-hoc testing (Section 5.1.3) and the regression analysis (Section 5.2), we rejected the null hypothesis and concluded that green policymaking initiatives differ by organizational green capabilities. As depicted in Figure 7, we statistically confirmed that organizations with more (vs. less) green capabilities undertake more (vs. less) green policymaking initiatives. Organizations with less green capabilities (i.e., Level 1 and Level 2) are relatively more represented in Score 1 and Score 2, while organizations with more green capabilities (i.e., Level 4, Level 5, and Level 6) are relatively more represented in Score 4 and Score 5.

These findings concur with the vast majority of previous research. It should be no surprise that well developed organizational green capabilities enhance green policymaking initiatives.

Consequently, green policymaking is not solely influenced by external variables (i.e., organization location and organization size), but it also seems to be self-determined. Obviously, the presence of green capabilities is related to the availability of human resources. However, organizational capabilities are largely self-determined by the strategic choices, the commitment, and the efforts of the administration [31]. To a large extent, it is an autonomous decision of the municipality to develop systems and procedures [37,38,39,40,41,42], documentation [9,43,44], communication [31,45], training [31,46,47,48,49,50], and responsible persons [31,51].

7.4. Barriers and Enablers

Figure 4 and Figure 5 show respectively the dominant barriers (i.e., lack of info, lack of staff, and cost concerns) and enablers (i.e., manuals and example forms). Table 8 presents the three-cluster solution derived using the k-means method. The algorithm grouped Score 1, Score 2, and Score 3 from the green policymaking initiatives as distinct clusters, supplemented by some barriers and enablers.

Although we intended to uncover distinct supporting policies for different groups of municipalities, we concluded that targeting the main barriers and enablers for all municipalities is sufficient. In terms of barriers and enablers, the municipalities do not differ enough to justify distinct supporting policies by the national government. Instead, a general policy that focuses on eliminating barriers (i.e., RQ2a) such as lack of information, lack of staff, and cost concerns, and enhancing enablers (i.e., RQ2b) such as manuals and example forms, should stimulate municipalities to establish a GPP policy.

7.5. Limitations and Future Research

This study suffers from some limitations that could be addressed in future research. First, from a contingency perspective [5,6], a major shortcoming was our inability to confirm or quantify the relationship between green policymaking initiatives and green performance, as explained in Section 3.5. Although this might be challenging because of data availability, we believe it is a highly interesting avenue for further research.

Second, the present study focused on fact finding. With the MOEJ dataset, we could retrieve some statistical relationships (i.e., what) but we were limited in explaining these relationships (i.e., why). Although we could determine that green policymaking initiatives differed by location, size, and capabilities, we could only suggest explanations based on previous literature. Therefore, we encourage researchers to examine these relationships in more depth.

Finally, the contextual factors, barriers, and enablers under consideration in this study were predetermined by the used dataset. We acknowledge that other important variables might be at stake. By incorporating additional data sources (e.g., data on population and households, natural environment, economic base, administrative base, etc.) the present contextual factors could be enriched, or novel factors could be added. We further recognize that the list of potential barriers and enablers is not complete. For instance, political support might also be an important factor [70]. We believe future research can also deliver important contributions by addressing these shortcomings.

8. Conclusions

This study considered the establishment and publication of green plans and GPP policies as green policymaking initiatives. From a contingency theory perspective, we investigated these green policymaking initiatives in Japanese municipalities. A rich dataset provided by the MOEJ enabled us to examine relevant contextual factors, barriers, and enablers for green policymaking.

Through ANOVA-based hypothesis testing and a regression analysis, we found that green policymaking differs by organization location and size (i.e., RQ1a), and organizational green capabilities (i.e., RQ1b). More specifically, we identified prefectures where municipalities scored relatively higher as well as lower. We also found that larger (vs. smaller) municipalities undertake more (vs. less) green policymaking initiatives. Finally, it was also shown that organizations with more (vs. less) green capabilities developed more (vs. less) green initiatives.

These results imply that green policymaking is not solely influenced by external variables (i.e., organization location and organization size). To a large extent it also seems to be self-determined, as the development of organizational capabilities is an autonomous choice of the municipality.

Regarding barriers (i.e., RQ2a) and enablers (i.e., RQ2b), a descriptive and cluster analysis identified some main factors. We concluded that the municipalities did not differ enough to justify distinct supporting policies from the national government. More generally, eliminating barriers such as lack of information, lack of staff, and cost concerns, and enhancing enablers such as manuals and example forms, should stimulate municipalities to establish a GPP policy.

Finally, we acknowledged some limitations that can be addressed by future research. We believe that a focus on green performance would be highly interesting. Moreover, researchers can investigate the dependencies in more depth to explain their relationship. Ultimately, future research could also aim at unveiling novel contextual factors, barriers, and enablers for local green policymaking.