Introduction

The concept of “governance” has become key in the world of public administration, aptly expressing the current characteristics of the public sector. Governance is required to manage administrative organizations by introducing private sector management methods, instead of the bureaucratic style directed by various governments. This word also describes the current situation in the public sector, which requires the management of public organizations, by replacing bureaucratic governance systems with private sector management methods. The concept of governance has attracted considerable attention in recent decades. The first issue of the journal Governance: An International Journal of Policy, Administration, and Institutions (1988) contained several articles on the changing relationship between politics and public administration, the subject of an ongoing debate in American public administration research, such as the one by Campbell and Peters (1988). Subsequent research on public administration has been divided into two major categories: (1) the use of governance as an abstract concept representing a change in the way public offices are governed, and (2) the use of governance as a specific concept in the same way as corporate governance, signifying compliance with laws and regulations and the proper operational management of organizations. Examples of the former are Compton et al. (2022) and Dickson (2016), while Brunet and Aubry (2016) and Almquist et al. (2013) represent the latter. Governance in the private sector is often used with a specific meaning related to fiscal accountability. While governance in the corporate sector implies enacting rules and strengthening the authority of shareholders and boards of directors, governance discussions in the government sector often indicate a retreat or reduction in the oversight and active role of government.

A major influence on the world of public administration in recent times has been the COVID-19 pandemic and the administrative and public sectors’ responses to it. The debate about the role of government, that has been ongoing since the 1980s, is about to change dramatically following the global pandemic. The introduction of private management methods and competitive principles into public services are not currently considered effective for addressing this public health crisis and the significant impact it can have on the lives of citizens. Hall (2021) also discusses the COVID-19 pandemic and its major impact on public administration research in the opening editorial of an issue of Public Administrative Review, a leading journal in public administration.

The objective of this study was to understand the features of Japan’s public administration research and the changes it has undergone, using statistical content analyses. A bibliometric quantitative analysis was conducted to understand the diachronic changes in administrative theories and praxis in Japan. This manuscript quantitatively explains research trends in postwar administrative studies in Japan, by statistically analyzing article titles from the Public Administration Review Quarterly from the last 40 years. A bibliometric quantitative analysis was performed through co-occurrence and correspondence analysis, a typical text mining method. Co-occurrence network and correspondence analyses revealed changes in research interests.

The characteristics of Japanese public administration that we specifically wished to clarify were those inherited from American public administration—the widespread application of business administration methods and the fact that these inherited characteristics have changed over time. The study of public administration in the United States, as a leading textbook in the field points out, began with Woodrow Wilson’s article in the Political Science Quarterly titled “The Study of Administration.” In this paper, Wilson argued for separating public administration from political phenomena and political science, stating that “the field of administration is a field of business” (1887, p. 219). American public administration has developed as a discipline since then, because of professional graduate schools that train public administration professionals for “public administration,” which is basically a parallel to “business administration,” a major in corporate management. Japanese public administration, like other modern social sciences, has been academically influenced by the United States, which assumed power after World War II. In the current context of administrative studies in Japan, some scholars like Nishio recognize administrative perspectives as one of the three important viewpoints in the discipline, which I will address later, while others like Muramatsu emphasize a political science perspective in conducting administrative studies. The quantitative content analysis in this paper attempts to reveal the characteristics of American public administration, which in Wilson’s terminology is “administration.” In other words, management-oriented methods remain strong in current discourse of Japanese public administration.

Quantitatively analyzing and understanding trends in administrative research in Japan up to 2019 has two implications. First, analyzing studies up to 2019 makes it possible to identify the characteristics of administrative research before the impact of COVID-19, which has had a significant impact on the administrative world in the short term, and when its impact subsides in the future. Thus, a pandemic’s impact on administrative research trends can be compared for both periods. In Japan, the impact of COVID-19 only began to increase from 2020, and in 2019 there were not yet many empirical administrative studies on it. Additionally, Japan has the highest ageing population in any developed country and is, facing the most challenges with respect to aging. This is called Kadai Senshin Koku in Japanese. The trends and features of administrative research largely reflect those of actual public administration. Understanding them in the context of administrative research in Japan—which has undergone economic recovery through rapid economic growth after World War II and is now facing the challenges of an ageing society due to economic and social maturity—will be helpful to administrative practitioners and researchers in other countries that are facing similar issues. For these two purposes, this study used the text mining technique, which enables a quantitative analysis of qualitative information. One of its advantages is that other researchers can perform additional tests or reproduce similar studies based on the evidence.

With the rise of online journals and mega-journals with open access, an increasing number of research papers are being published, predominantly in English, the most used language in the academic world. Research that was previously published in local languages is now written in English, and published in English language journals. Since 2019, the COVID-19 pandemic has generated a large body of academic literature on the disease and how to respond to it. This includes the status of infections in several countries, the success or failure of relevant policies, and effective medical treatment strategies. Much of the literature on COVID-19 has also been subjected to bibliometric analysis, a frequent practice in other fields since the 1990s (Akintunde et al. 2021; Tantengco 2021). Drago et al. (2023) made a bibliometric analysis of 285 studies, published in 2019–2020, on the development of telemedicine which has expanded with the spread of COVID-19. Ranjbari et al. (2023) conducted bibliometric and text-mining analyses of studies on waste disposal in the context of the COVID-19 pandemic. Utkarsh and Sigala (2021) performed a bibliometric analysis of the rich literature on the impact of the pandemic on the travel industry.

In the field of public administration and the management of public organizations, Di Vaio et al.’s study (2022) is a recent example that focuses on bibliometric analysis. They analyzed 161 articles published, from 2017 to 2021, on how data intelligence and analytics can improve the effectiveness of public sector decision-making. Among studies that have focused only on articles published in specific journals for specific periods of time, Liu et al. (2022) presented a bibliometric analysis of articles published in the Journal of Dental Sciences from 2009 to 2020. Wei and Lei (2018) focused on the New England Journal of Medicine to examine the difference between normalized citations of Harvard-based versus non-Harvard-based articles, and found it to be small. Chaudhry et al. (2021) analyzed bibliographic data over 12 years, for 1264 articles, in the Journal of Oral and Maxillofacial Surgery, to measure levels of evidence for each reference, and discuss changes in their percentages over time. In the field of pedagogy, Marchand Martella et al. (2021) conducted a bibliometric analysis of 1671 references on active learning from the Web of Science Core Collection published between 2014 and 2019, and analyzed the characteristics of the journals by field, including distribution by countries and regions. Thus, an increasing number of studies are analyzing changes in research trends and characteristics over a specific period of time by conducting bibliometric analyses or systematic reviews of the growing literature in various fields. The increase in bibliometric analysis of COVID-19 studies, as discussed here, demonstrates the importance of analyzing research trends apart from the accumulation of individual, original studies, and this may also apply to administrative studies.

Research on public administration in Japan is increasingly being published in English, particularly research conducted by young researchers. For example, Kaihara (2008) discussed the characteristics and significance of the Koizumi reforms (2001–2006) to Japan’s political and economic system in English. Similarly, Lee and Kim (2019) examined the differences in the use of welfare personnel other than civil servants, such as civic advocacy groups, in the context of the welfare system reforms in Japan and Korea. Rothacher (2010) provided an overview of the reflation policies adopted during the so-called lost decade (1992–2002), as well as their impact and characteristics, from an international perspective. A commonality between these studies is that they were conducted using literature that were predominantly published in English, which is currently the most used academic language. However, when read in isolation, the context and practical implications of many of these studies are difficult to understand. Therefore, this study aimed to understand features of Japanese public administration studies as a whole, using a bibliometric approach.

This study analyzed the titles of articles from Japan’s Public Administration Review Quarterly from its first (1978) to its 165th issue (2019) to quantitatively assess the research trends in postwar public administration. This journal is one of the most renowned in its field, and is published by the Institute of Administrative Management in Japan. The institution is closely tied to the Ministry of Internal Affairs and Communications. A similar bibliometric analysis has already been performed and published in Japanese for another leading journal—Annals of the Japanese Society for Public Administration, (Moteki 2020). By comparing the content analysis of two separate and influential journals on Japanese administrative research, the features commonly found in the results of the analysis may be followed up with different data and considered to be plausible and reliable findings. The target period is 42 years, and comprises the postwar Showa and Heisei eras. The Showa era spans from 1926 to 1989 in the Western calendar. After Japan’s defeat in World War II in 1945, various policies were established to democratize and demilitarize the nation, laying the foundations for its current political system. Public administration studies in Japan after WWII were strongly influenced by the United States, because the initial democratic reforms were carried out under the U.S.-centered Supreme Commander of the Allied Powers (called “General Headquarters” in Japan, GHQ) rule. Initially, the focus was on the introduction of American theories, but later there was an increase in studies that sought solutions to the country’s administrative issues, as a result of the accumulated research in Japan.

Qualitative analysis of research trends in Japan’s public administration

Leading public administration scholars present the characteristics of research trends in public administration, based on their experiences and insights, in their textbooks. By looking at the arguments presented in two leading Japanese public administration textbooks, we initially included them to compare and contrast the results of the quantitative analysis that this manuscript will present later. Public administration scholars at the University of Tokyo and Kyoto University have been very influential in the field before, during, and after World War II to the present day. According to Nishio, modern administrative science in Japan started in 1921 (2001, p. 43), when departments of public administration were established at Tokyo Imperial University and Kyoto Imperial University (the predecessors of the aforementioned universities). For this reason, many bureaucrats have been recruited from these two universities. They have also been a major force in the center of public administration as an academic discipline. This is why textbooks on public administration written by public administration scholars from these two universities are widely used in Japanese universities. Nishio’s textbook is also widely used by college students preparing for civil service examinations, at national and local levels. Nishio considered institutions, administration, and policies as key aspects of public administration in Japan, and these three dimensions were also observed in my correspondence analysis (2001, pp. 50–51). According to Nishio (2001) administration means management theories and methods. Professors at Japanese universities, who use Nishio’s textbook, also consider these aspects as important and teach the same to their students.

A textbook by Muramatsu (2001), a professor of Kyoto University, argued that public administration belonged to the political sciences (the subtitle of the textbook in Japanese is Political Analysis of Modern Public Administration). Muramatsu formulated a four-quadrant matrix of relevant studies, which includes foreign countries, Japan, theories, and practices (1983, p. 49). He highlighted a lack of relevant research in two quadrants: Japanese theories and practice in foreign countries. According to him, public administration research in Japan at the time focused on foreign theories and Japanese practice.

Related studies

Content analysis research started in earnest from Krippendorff (1980), although Wald (1973) attempted to quantitatively assess research trends in public administration in the United States, using the Delphi method. Content analyses are often called bibliometric analyses, and are important research evaluation instruments. Ni et al. (2017) used the bibliometric method to examine research trends in American studies on public administration, using PDF files (3934 articles) of the Public Administration Review from 1940 to 2013. Their analysis focused on the characteristics of authorship, including the most productive authors, their institutions, gender distribution, and their professions as scholars or practitioners. In addition to authorship, a word analysis was conducted to identify the 50 most frequently used title words, to explain differences between the periods 1940–1964, 1965–1989, and 1990–2013. The article referred to previous quantitative research, using bibliometric data from the Public Administrative Review (p. 505). As explained in the introduction, Ni et al. (2017) conducted a bibliometric analysis of research published over the course of 75 years, and their article introduced studies by Gooden (2015), Groeneveld et al. (2015), Jones and Doss (1977) and Perry and Kraemer (1986). Gooden (2015) found that the percentage of articles published in the Public Administration Review on social equity fell below 5% from 1940 to 2013, and that those published focused on personnel and public policy areas.

An example of a recent bibliometric analysis is the one conducted by Agrifoglio et al. (2021), based on a systematic literature review of knowledge management in public organizations, using the Scopus and Web of Science databases. Cuccurullo et al. (2016) conducted a bibliometric analysis of research trends on performance management, in both business and public administrations, using 25 years of data. The analysis of text-mining or bibliometric methods on research journals in the field of public administration, which is closely related to the analysis in this manuscript, is summarized in a list of methods, analyzed variables, names of journals analyzed, periods covered and what was found in the analysis (Table 1). The table shows that the analysis method can be seen as focusing on the title of the article.

Table 1 Characteristics of bibliometric/text mining analysis in public administration and related fields
Full size table

Among research using text mining techniques in the social sciences, Amado et al. (2018) and White et al. (2016) applied this approach to management studies. Similarly, Kwon (2022) examined the debate on transparency in the social sciences through bibliometric analysis. He used historiography and co-occurrence analyses to analyze and visually clarify the changing academic interest in transparency and relationships between themes. Following him, a co-occurrence analysis was also used in this study. Hung (2012) used refereed journal articles and proceedings obtained from the Science Citation Index and Social Science Citation Index from 2000 to 2008. In the field of public administration, however, we found no studies using this technique in English language journals. Moteki, however, examined articles from another major academic journal on Japanese public administration, the Annals of the Japanese Society for Public Administration, using the text mining method (2020).

Among Japanese academics, Saito (2011) used the same method to survey studies in the social and natural sciences, and Sakushima et al. (2012) employed it for titles published in the academic journal Clinical Neurology, obtaining text data from the Citation Information by the National Institute of Informatics (CiNii) database. Their research included a correspondence analysis. Although the research areas are different, this study refers to the methods employed by Sakushima et al. (2012). The only study that has conducted a bibliometric analysis of research trends in Japanese public administration, including text mining methods, is by Moteki (2020). It analyzed Japanese language articles from the Annals of the Japanese Society for Public Administration. There are no studies in either Japanese or English that have analyzed Public Administration Review Quarterly, the subject of this study. Public Administration Review Quarterly is the leading journal in the Japanese public administration community, along with the Annals of the Japanese Society for Public Administration, and has been published for many years since 1978. The former is published by a foundation affiliated with Japan’s Ministry of Internal Affairs and Communications, while the latter is published by the Japan Society for Public Administration. However, by analyzing the former, and verifying it together with the analysis of the latter, it will be possible to significantly overcome the issue of the lack of rigor in text mining analysis in this field.

Materials and methods

Material collection

To grasp postwar research trends in Japan’s public administration, the titles of the Public Administration Review Quarterly were gathered as target data. First, the keyword Kikan Gyosei Kanri Kenkyu (the Japanese title) was used to search the CiNii article database organized by the NII. Second, the search results were displayed and saved in the tab separated values (TSV) format. Finally, the TSV files were converted into a Microsoft Excel spreadsheet. As issues 1 to 4 were not included in the CiNii database, their hard copies were used and the data were manually entered into the Excel dataset. To enable the reproducibility of this study, machine translation from Japanese to English was conducted using Google Translate. The results were read to confirm the appropriateness of the translation. Thereafter, the titles of the articles were treated in English.

The following modifications were made for the dataset file: (1) titles (cases) indicating “book reviews” were deleted; (2) titles related to special articles on academic association conferences were excluded; (3) non-papers, such as lectures and forewords, were deleted. Furthermore, issues 1 to 4 were added, as mentioned, to the original 1356 cases. As a result of these modifications, 842 titles remained in the dataset. The advantage of using the title of the article as the object of analysis is that the theme of the article is expressed in its most condensed form in the title of the article, and the small amount of data makes it less likely that differences in procedures will arise when conducting replication studies. As shown in Table 1, representative studies in public administration and related fields have also shown encouraging results from analyses using only the title of the article. Another reason for using the title of the article as the subject of the research is the need to align the units of the research subject in order to compare it with the author’s previous research of Moteki (2020).

Methods

The author examined this text data using the KH Coder (Version 3.Alpha.17 k) developed by Dr. Higuchi Koichi, of Ritsumeikan University. This free software was developed using a Grant-in-Aid for Scientific Research from Japan’s Ministry of Education, Culture, Sports, Science, and Technology (Higuchi 2016). The Stanford POS Tagger software was used for a morphological analysis of the sentences.

Processing compound words for technical terms by using the program alone was not sufficient, because not all compound words can be grasped that way. Therefore, this study used all the words listed in the index compiled by Nishio (2001), in the standard textbook for public administration in Japan. As a result, the number of compound words based on this index amounted to 815. Following this process, the extracted words were checked. Important concepts that had not been extracted, such as governance and the United Kingdom, were added manually to the list of special designated words and compound words.

Text mining analysis is essentially an exploratory method that focuses on the frequency of extracted words and co-occurrence relationships, and attempts to quantitatively understand the content of the text by examining the relationships between extracted words using statistical tests such as the χ2 test, and visualization techniques such as correspondence analysis and co-occurrence network analysis. While text mining analysis can be positioned as part of a biometric analysis, Zhang et al. (2022), a study in the field of medicine, refers to some of the advantages of biometric analysis for specific fields of study that it could “help researchers identify a field’s trends and hotspots. Thus, bibliometric analysis is a valuable tool for researchers and clinicians to conduct relevant research and implement clinical decisions.” The analysis of research trends based on text mining techniques in the field of public administration, which this paper conducts, can also inform researchers about research trends beyond the results of individual studies and the focus of research over a period of time. It also informs practitioners about changes in research trends that have certain relevance to changes in practice, and contribute to decision making in practice. In terms of research rigor, an individual RCT study directed at a fixed research subject is ideal. However, as mentioned above, a bibliometric study of an entire individual study over a fixed period of time has research and practical advantages that are different from individual original studies.

The analysis methods used next were co-occurrence network analysis and correspondence analysis. Lebowitz et al. (2020) use co-occurrence network analysis in text mining analysis, where the size of the circle surrounding the extracted words represents the frequency of the extracted words, and a straight line between the circles indicates that the co-occurrence relationship between the extracted words is above a certain strength. The same meaning of dots and lines in the co-occurrence network analysis was also used in this analysis, with lines showing the relationship between the nominalized variable of the publication date of each data article in the analysis. This is known as a multiple correspondence analysis. Correspondence analysis in text mining analysis is often used to illustrate the relationships between extracted words, and between extracted words and variables not related to extracted words. Liu et al. (2023) proposed a new method for correspondence analysis for text mining analysis, but also included a mathematical explanation of the conventional correspondence analysis as a prerequisite for discussion.

Results

Correspondence and co-occurrence network analyses were performed to examine the relationships between the extracted words. Using 10-year time periods as external variables in deriving relationships between the extracted words led to an understanding of the transition and characteristics of each period in the target studies.

Co-occurrence network analysis

A co-occurrence network analysis was performed to understand how the focus changed from the first period (the Showa era, 1978–1988) to the second (the Heisei era, 1989–2018). In Japanese local governments, public officers use the Japanese periods (Wareki), which are named according to the reign of each emperor (Tennou). These Japanese periods were used to compare the differences between the first decade and those that followed. In the co-occurrence network analysis shown in Fig. 1, the Showa and Heisei eras were selected as external variables to draw the figures. To filter up to 60 edges (nodes), the Jaccard index was used as the selection criteria (for a mathematical explanation of the Jaccard Index, see McCormick et al. (1992)).

Fig. 1
figure 1

Co-occurrence network analysis of extracted words (the Showa and Heisei time periods)

Full size image

Words closely connected to the Showa era (on the right side of the figure) are seen in examples of article titles from the United States and the United Kingdom. Also evident are some of the specific themes of the papers that introduce each country’s administrative practices and institutions. In titles from the United Kingdom, we found mentions of the civil service system, political appointment system of officials, and industrial policy of Japan. In titles from the United States, we found mentions of the parliamentary system, Office of Management and Budget, and the approval process for new medications. After its defeat in WWII, and specifically as a result of the American occupation, Japan witnessed an evolution in social sciences, taking the United States and the United Kingdom as exemplars.

In contrast, the word Japan is located on the left side of the figure (related with the Heisei era). Japan emerged as a popular topic in administrative research in the Heisei era because the Great East Japan Earthquake of 2011 had a significant impact on the government. The journal analyzed in this study featured articles on policies and systems related to the earthquake and the crisis it caused. This indicates that research trends shifted from studies involving cases of foreign countries to those on domestic cases and solutions, partly because of the economic, human, and environmental problems caused by the earthquake. In addition, we see governance, local government, and policy evaluation on the left side of the figure (the Heisei era). These words indicate research features regarding the respective time periods. At the center of the figure, we see extracted words such as reform, administrative reform, management, theory, and administration. These are common words in both time periods.

Correspondence analysis

A correspondence analysis, which is a typical technique for graphically displaying category data, was also conducted. To grasp the changes and periodical characteristics of research trends in public administration in Japan, a 10-year time period was used as an external variable, creating five time periods.

Figure 2 shows the results of the correspondence analysis. The extracted words located near the origin, where two dimensions intersect, do not have distinctive features. Dimension 1 is an axis from left to right. We see the word Rincho at the far left of the configuration map. Rincho represents the abbreviations of government commissions for the administrative reforms from 1961–1964 (the Ikeda administration) and 1981–1983 (the Suzuki and Nakasone administrations). On the far right of the figure, we see the Great East Japan Earthquake and Governance. These extracted words differ from those related to administrative reforms. Here, Dimension 1 was interpreted as an axis of administrative reforms.

Fig. 2
figure 2

Correspondence analysis (the First and Second Dimensions)

Full size image

In Dimension 2, we see Information disclosure, Gyokakushin, and Policy evaluation in the lower part of the figure. The Act on Access to Information Held by Administrative Organs enacted in 1999 established a mechanism for government agencies to publicly disclose information. Gyokakushin (an abbreviation of Rinji Gyosei Kaikaku Suishin Shingikai [Extraordinary Administrative Reform Promotion Council]) started in July 1983, during the Nakasone Administration. The Government Policy Evaluations Act (Act No. 86 of June 29, 2001) introduced policy evaluations at the level of the national government. In the upper part of the figure, we see Great East Japan Earthquake, History, and Development. Considering the foregoing, Dimension 2 was interpreted as an axis of historical events or administrative systems.

Figure 3 is a configuration map of Dimensions 2 and 3. We see Gyokakushin and Case study at the top of the figure. In the lower part of the figure, we see Policy evaluation and improvement (Kaizen). Based on these extracted words, Dimension 3 was interpreted as an axis of evaluations and Kaizen.

Fig. 3
figure 3

Correspondence analysis (the Second and Third Dimensions)

Full size image

Discussion

Comparison of research trends based on qualitative findings presented in textbooks

This section discusses the results of text mining regarding the factors highlighted by Nishio (2001) and Muramatsu (1983). The correspondence analysis showed three dimensions: (1) administrative reforms, (2) historical events or administrative systems, and (3) evaluations and kaizen. The first dimension implies that problems related to administrative matters, which Nishio (2001) highlighted as an important perspective on administration referring to management aspects, are a strong feature of public administration research trends in Japan. The quantitative text mining methods also confirmed this. Many scholars, like Muramatsu (1983), emphasized the need to study policy practices in foreign countries and create theories based on them for Japan. Conversely, they also insisted on strengthening research in the field of Japanese empirics, and on putting forward Japanese-born theories. The co-occurrence network analysis, which focused on differences between the Showa and Heisei eras, reveal that studies on the Showa era (1978–1988) could be partly characterized by the frequent occurrence of two extracted compound words—United States and United Kingdom. Japan, as a keyword, started occurring more often in studies on public administration during the Heisei era (1989–2018). This result indicates that studies on public administration during the Heisei era shifted from importing theories from the United States and the United Kingdom, to conducting research on practices in Japan. The result of these text-mining analyses is that the shift in emphasis in empirical studies from Western countries to Japan, as argued by scholars including Muramatsu, can be observed quantitatively.

Comparison with the results of Moteki (2020)

Moteki (2020) analyzed the Annals of the Japanese Society for Public Administration by conducting a co-occurrence network analysis, cluster analysis, and correspondence analysis to grasp features of Japanese administrative studies’ trends. This study was written in Japanese, and analyzed textual data from a different journal. The characteristics identified in each of these analyses, such as the fact that there were more administrative studies on reform, evaluation, and policy evaluation in the 2000s were commonly identified. The correspondence analysis of extracted words using the decade-by-decade chronological classification of the Western calendar as an external variable revealed the characteristic words of each decade. First, it characterizes the 1970s papers are by “public participation,” “administrative planning,” “publicness,” “policy-making,” and “decision-making.” “Reform,” “policy evaluation” and “governance” were found to be characteristic terms of the 2000s. This may be related to the introduction of the policy evaluation system based on the Hashimoto Reform and Policy Evaluation Law, and the performance measurement system on programs in local governments (the pioneering prefecture on the evaluation system is Mie Prefecture) in the 2000s. Since 2010, the following extracted terms have been centrally positioned: “Great East Japan Earthquake,” “chief executive,” “Okinawa,” and “international comparison.” The Great East Japan Earthquake occurred in 2011, and in terms of time period, it was a disaster that significantly changed administrative studies research after 2010. Research on administrative reform and evaluation and policy evaluation has changed considerably from the 2000s and 2010s. The results of the correspondence analysis showed that the first component was time; the second component was “evaluation and reform;” and the third component was “policy process.” These second and third components are the major pillars of post-war public administration studies in Japan. As with the cluster analysis “identity of public administration,” “management, institutions, and policies” and “reform and local government,” formed the centers of research on post-war public administration in Japan, and were extracted from the quantitative analysis of this previous study.

Compared to the results of the analysis in this manuscript, parts of the results that are similar while other parts are different. In other words, there are features that only appear in the results of one analysis. One commonality among them is that the dimension related to administrative reform was found in the correspondence analysis. In this study, the first principal component was administrative reform. Moteki’s (2020) correspondence analysis also found “reform,” “policy evaluation” and “governance” as characteristic terms extracted in the 2000s. The distribution diagram of the second and third dimensions of the correspondence analysis in this study (Fig. 3) reveals “policy evaluation” and “improvement” (kaizen). These features may have a degree of certainty, as they could be found as features for administrative reform and policy evaluation in the analysis of two text-mining methods on separate public administration research journals.

The fact that the Great East Japan Earthquake was found as a characteristic extract term is also common to both analyses. The earthquake in 2011 had a significant impact on people’s lives and the nature of public administration, and it was confirmed that it was certainly a major feature of the research in the period under study, as highlighted in the text-mining analysis of the two journals. Great East Japan Earthquake appears at the top of the figure on the Dimension 2 axis of the configuration map (Fig. 2). This indicates that the earthquake greatly influenced the theories and practices of public administration in Japan. This historical event has significantly impacted research in this field and has been a major factor influencing research trends.

Conclusion

Results and implications

This study quantitatively examined research trends in postwar administrative studies in Japan by statistically analyzing nearly 40 years of article titles published in the Public Administration Review Quarterly, from its first (1978) to its 165th issue (2019). The two most prominent Japanese public administration journals are the Annals of the Japanese Society for Public Administration and Public Administration Review Quarterly. The former was analyzed by Moteki (2020) and the latter in the current study. The former is published by an academic society, while the latter is a research journal for practitioners, published by an organization affiliated with the Ministry of Internal Affairs and Communications. After implementing a morphological analysis, using the Stanford POS Tagger, co-occurrence network and correspondence analyses were conducted. The results revealed that research interests in the field had changed over time. Specifically, based on the co-occurrence network, there was substantially more research on administrative reforms during the Showa and Heisei eras, spanning 42 years. These results confirm that administrative (management) theories and methods are one of the three components of public administration in Japan, as pointed out by Nishio (2001). The first dimension of the correspondence analysis implies that the problems related to administrative matters are actually features of public administration in Japan. In the correspondence analysis, both the first and second, and the second and third dimensions were selected as axes of configuration, describing the relationships between extracted words. The following interpretations can be derived from each configuration—the first dimension is concerned with administrative reforms, the second with historical events or administrative systems, and the third with evaluations and kaizen. Policy evaluations are an important element of the three policy aspects suggested by Nishio (2001). The introduction of management methods from the private sector, such as kaizen, into Japanese governance was also a major feature of its administrative reforms. These methods of improving organizational management are further related to administration, one of the three public administration perspectives identified by Nishio (2001). It was quantitatively confirmed that the characteristics of American public administration, characterized by managerial methods or, in Wilson’s terminology, “administration” methods, remain a strong influence in contemporary Japanese public administration.

The results of the co-occurrence network analysis illustrated the differences in research trends between the Showa and Heisei eras. Traditional Japanese studies on public administration focused on introducing foreign theories from the United States and the United Kingdom to actual Japanese policies. However, the extracted word Japan in the co-occurrence network figure implies a research shift that led to the creation of context specific administrative theories. Featured words like governance, local government, and policy evaluation on the left side of the figure (the Heisei era) also reaffirm changes in the direction of studies. Muramatsu (1983) argued for a change in Japanese public administration studies by focusing on a political science approach. Such a change in emphasis was observed in this quantitative analysis, as evidenced by the presence of Japan in the extracted words in the co-occurrence network analysis. The appearance of extracted words such as reform, administrative reform, management, theory, and administration at the center of the figure, suggest that these words were common occurrences across both time periods.

Based on the above observations and the results of this study, it can be said that the major pillars of Japanese postwar administrative studies from 1978 to 2019 confirm the directions identified by the qualitative research conducted by Nishio (2001) and Muramatsu (1983). Furthermore, previous qualitative research, which mostly relied on the analysts’ experience and perceptions, can be comprehensively re-examined quantitatively in light of the findings of this study. The qualitative analysis in this study was based on personal experience, and was validated by quantitative analysis of the data, strengthening this study’s contribution to the research field. Muramatsu’s (1983) emphasis on the need for theories based on Japanese practices is demonstrated by the co-occurrence network analysis of this study, which focused on the differences between the Showa and Heisei eras. The changes in Japan’s public administration, which Muramatsu (1983) highlighted were necessary, have been quantitatively confirmed by the empirical evidence of the text mining method, which shows changes in the frequency of the terms United States, United Kingdom, and Japan. In Moteki’s study (2020), analyzing the titles of another journal of public administration, the three dimensions of the correspondence analysis were time, evaluation and reform, and policy processes. The fact that these characteristics were confirmed quantitatively, and with some degree of reproducibility, using text-mining methods enables practitioners reading this paper to understand the recent emphasis of research in Japanese public administration. By referring to the research results of this paper, Japanese practitioners can understand the context in which recent individual practical issues in public administration are contextualized. Meanwhile, practitioners from around the world can use this paper to understand the emphasis in Japanese public administration studies and practice, that is less known after Japan’s economic recovery, based on the involvement of the government, especially MITI. In terms of international comparisons, it can be confirmed that “reform,” “policy evaluation,” and “governance” became research features in the 2000s, similar to research trends in countries in Europe and the U.S. This is of major significance for this study, in both practical and research aspects.

It should be noted that these results were comparable with the author’s previous analysis of another journal (Moteki 2020), and both results commonly confirmed that research on administrative reform is a major feature in Japanese public administration studies. Text mining methods are quantitative methods. However, compared to a multivariate analysis such as ordinary regression analysis, they are exploratory methods and tend to lack some rigor, as the analysis procedure and the objects to be focused on differ for each author. By comparing the results of the author’s two studies, we see that the characteristics of Japanese public administration research can be shown with a certain degree of rigor and verifiability.

Limitations of this study and future agendas

This study has the following limitations. First, the period under analysis is limited to post WWII. Since the Meiji era (1868–1912), Japan adopted both natural sciences and also social sciences, including public administration from Western theories, to catch up with Western countries in terms of economic and national power development, and continued to develop research and educational infrastructure at Japanese national universities. For example, courses in public administration were established at Tokyo Imperial University (now the University of Tokyo) and other universities, and education and research in public administration were promoted on the premise that they would contribute to the development of the nation’s governance structure. The subject also became widely studied by students aiming to become national civil servants. The core content of public administration as a discipline did not change much before and after WWII. In order to create a comprehensive picture of research trends and changes in Japanese public administration, it is necessary to also consider the research trends manifested since the Meiji era. Further, to understand research trends that account for the Meiji, Taisho, and Showa periods, it is necessary to examine data available in books. Since the CiNii Books database virtually contains all books held in university libraries, the aim is to conduct a bibliometric analysis of research trends for books, using this database in the future.

Another limitation of this study derives from the fact that bibliometric analysis is a methodology limited to text mining of study titles. Some bibliometric studies in the United States for the Public Administration Review, such as Ni et al. (2017), conducted an analysis of authorship attributes. As already mentioned in this paper, the eastern University of Tokyo was the primary center of public administration studies in Japan in the beginning, and in the postwar period, the western Kyoto University also became an important research center. In the future, the research in this paper can be expanded upon and a more diverse range of analytical targets for bibliometric analysis—factors such as authors’ gender, university of origin, work experience, mono-authorship/co-authorship, and availability of external funding—introduced, thereby aiming for a more comprehensive quantitative analysis.