A growing literature has established the high value of legislative connections to firms. However, elected office is not the only form of government connection that firms can leverage. Firms also regularly hire civil servants—a practice commonly referred to as the bureaucratic revolving door. Despite the typically far larger number of civil servants in central government agencies compared to elected office holders,Footnote 1 examination of the value of bureaucratic connections to firms is notably lacking from the political connections literature. Are bureaucrats similarly in demand by employers? If so, which bureaucrats are in demand and what benefits do they bring to the firms that hire them?
In theory, firms have incentives to hire former bureaucrats only when those individuals can deliver tangible benefits such as technical expertise, access to government loans or contracts, influence over regulatory decisions, or other political rents. Bureaucrats without such value should hold little appeal to private employers. By contrast, bureaucratic institutions have incentives to place as many former officials as possible into postbureaucracy roles as the revolving door: (1) helps ministries maintain informal channels of influence, and (2) acts as a credible signal to prospective recruits that a relatively low-paid public sector career will be rewarded with lucrative postretirement opportunities.
Using comprehensive data on the initial postretirement placements of all former civil servants in Japan, I provide empirical evidence consistent with the theoretical predictions outlined above. A bifurcated market exists for former bureaucrats, in which the highest ranking officials are naturally in demand by large, publicly traded corporations. These officials then deliver benefits such as low-interest government loans to their new employers. However, for lower ranking bureaucrats who cannot offer these kinds of benefits to firms, alternate sources of employment exist at nonprofit organizations bolstered by government contracts. Though these connections are under-examined in the existing literature, I show that roughly half of Japanese bureaucrats are rehired by public corporations or nonprofits, and that these organizations leverage their bureaucratic connections to increase the size of government contracts they receive. In the context of a weak welfare state with relatively low pay for civil servants, the promise of such revolving door positions—regardless of private sector demand—is essential for successful bureaucratic retainment and recruitment.
I evaluate the concrete benefits that former officials provide to their new employers, and show that flows of bureaucrats to different sectors of the economy lead to sector-specific benefits. First, using a matched differences-in-differences (DiD) approach (Imai, Kim, and Wang Reference Imai, Kim and Wang2019), I show that private firms hiring senior officials from elite economic ministries receive increased volumes of government loans in subsequent years. Second, using an interrupted time series design, I find that investors respond positively to appointments of high-ranking economy ministry bureaucrats. Third, I leverage novel data on nonprofit leadership and DiD approaches that account for the negative weighting issues highlighted in recent literature (de Chaisemartin and D’Haultfœuille Reference de Chaisemartin and D’Haultfœuille2020) to show that nonprofits with ex-bureaucrats in director roles receive more lucrative government contracts. These contracts also exhibit financial irregularities according to common forensic accounting techniques, patterns not observed in comparable nonprofits without bureaucratic ties.
Most research on the bureaucratic revolving door relies on theoretical models, focusing on how postgovernment hiring shapes regulatory leniency (Che Reference Che1995; Dal Bó Reference Dal Bó2006; Salant Reference Salant1995) or the conditions under which bureaucratic connections should be expected to bring value to firms (Bils and Judd Reference Bils and Judd2020). Empirical evidence is limited and often drawn from convenience samples or single-agency case studies. Little attention has been paid to employment destinations beyond for-profit firms or to the proactive role bureaucratic agencies may play in cultivating revolving door opportunities.
This study addresses these gaps using comprehensive administrative data and a broader view of the revolving door, spanning both for-profit and nonprofit destinations. I combine newly constructed datasets of all initial revolving door hires, all government loans to private firms, stock prices of all firms that make high-level bureaucratic hires, and all government contracts with nonprofits in Japan over a period of one decade to test for benefits that accrue to organizations that hire former bureaucrats. To identify the kinds of benefits these hires may generate, I supplement the quantitative analysis with interviews with current and former bureaucrats, business leaders, and nonprofit executives. These interviews point to two recurring themes: former bureaucrats offer valuable political connections that may help secure contracts or loans, but are often perceived as lacking technical or managerial expertise.
Collectively, interviews, descriptive analysis, and causal estimates reveal a bifurcated job market for former bureaucrats in which only a select few are highly valued by for-profit firms. Top bureaucrats from ministries that control the levers of finance, industrial policy, and regulation are in high demand, and these individuals are in turn able to drive benefits to for-profit firms. However, for those less desired by the private sector, positions exist in nonprofits, where former bureaucrats appear able to leverage their connections to help these agencies secure continued government funding, as well as maintain employment opportunities for future generations of retiring bureaucrats.
These findings shed light on how the public and private sectors interact in advanced democracies, particularly those with large public-private pay gaps. While existing research on the revolving door focuses on rent-seeking firms hiring former officials for their expertise or connections, this article highlights an additional dynamic. Only a subset of bureaucrats—typically those with influence, status, or control over key sectors—are in demand by private firms. This generates a supply of would-be revolving door candidates for whom no natural market demand exists. However, to sustain elite recruitment amid declining public sector compensation and prestige, ministries must offer credible guarantees of future income and employment to all recruits. Ministries may, therefore, respond to this imbalance by actively constructing revolving door pathways—for example, outsourcing contracts and establishing informal pipelines to nonprofit or quasi-public roles—to provide employees with a credible path to long-term career value. These patterns suggest that the state is more than a passive participant in the revolving door phenomenon, which is in fact a function of both firm-driven rent seeking and state-led institutional maintenance.
THEORY AND HYPOTHESES
The Value of Political Connections to Firms
A large literature explores the economic value of legislative political connections to firms. For example, Blanes i Vidal, Draca, and Fons-Rosen (Reference Blanes i Vidal, Draca and Fons-Rosen2012) show that lobbying revenues generated by ex-congressional staffers fall sharply when their former employers leave office. In China, firms with CEOs who serve in the National People’s Congress exhibit higher stock prices and operating profits (Truex Reference Truex2014). Similarly, Faccio (Reference Faccio2006) and Faccio, Masulis, and McConnell (Reference Faccio, Masulis and McConnell2006) find that firms benefit from the political ascension of major shareholders or executives, through stock price increases and higher likelihood of bailouts, respectively. Campaign donations also appear to yield returns, as Brazilian firms that support winning candidates receive more government contracts (Boas, Hidalgo, and Richardson Reference Boas, Hidalgo and Richardson2014).
Alongside this legislative literature, growing attention has been paid to the revolving door between the bureaucracy and industry. Legislators and bureaucrats have different incentives for entering the revolving door, but both connections may be of value to firms. Classic models of regulation emphasize the role of electoral incentives, theorizing that legislators balance the preferences of voters and organized producers, who provide campaign contributions and lobbying pressure in addition to rents (Grossman and Helpman Reference Grossman and Helpman2001; Peltzman Reference Peltzman1976; Stigler Reference Stigler1971). By contrast, bureaucrats are not subject to re-election pressures, nor do they occupy dual roles at corporations that may influence their behavior while in office as elected officials at times do (see, for example, Weschle Reference Weschle2024)—their primary incentive is to secure attractive postretirement employment. This creates incentives to foster relationships with firms and to shape policy or resource allocation in ways that increase future employability. Such actions may also enable regulatory capture, where civil servants favor future employers, or signaling, where regulators act stringently to demonstrate competence to potential industry recruiters (Dal Bó Reference Dal Bó2006). Empirical studies offer evidence for both mechanisms: the capture hypothesis is supported by Cohen (Reference Cohen1986), Gormley (Reference Gormley1979), Spiller (Reference Spiller1990), and Tabakovic and Wollmann (Reference Tabakovic and Wollmann2018), while DeHaan et al. (Reference DeHaan, Kedia, Koh and Rajgopal2015) provide evidence for the signaling model.
Despite theoretical overlap, relatively few studies directly link bureaucratic connections to firm-level rents. Notable exceptions include Lee and You (Reference Lee and You2023), who show that U.S. firms with connections to the Office of the U.S. Trade Representative reduce their lobbying activity; Barbosa and Straub (Reference Barbosa and Straub2020), who find that Brazilian medical firms hiring ex-bureaucrats offer lower prices to government; and Hong and Lim (Reference Hong and Lim2016), who demonstrate that Korean universities employing former officials receive more grants. In Japan, Asai, Kawai, and Nakabayashi (Reference Asai, Kawai and Nakabayashi2021) show that firms hiring ex-infrastructure ministry officials are more likely to win public contracts, while Luechinger and Moser (Reference Luechinger and Moser2014) observe stock price boosts for U.S. defense firms hiring Department of Defense officials.
Taken together, these studies suggest that firms value former civil servants for their ability to shape regulatory outcomes, influence government decision-making, and facilitate access to public resources such as contracts, subsidies, and loans. However, only select civil servants—such as those with regulatory oversight or access to public resources—can deliver these benefits. These civil servants should, therefore, be in demand by firms, but others should be of little value. In this sense, select bureaucratic connections may yield similar benefits to legislative ones—yet empirical work remains fragmented, typically focusing on single agencies or economic sectors due to data limitations.
The Value of Post-Government Positions to Bureaucrats and Ministries
A related literature explores the “returns to office” enjoyed by politicians, demonstrating that legislators often gain significant wealth after being elected to office (e.g., Eggers and Hainmueller Reference Eggers and Hainmueller2009; Fisman, Schulz, and Vig Reference Fisman, Schulz and Vig2014). By contrast, civil servants typically experience suppressed earnings relative to their private-sector counterparts and rely on the revolving door for deferred compensation. For example, in Japan, Ramseyer and Rosenbluth (Reference Ramseyer and Rosenbluth1993) estimated that civil servant salaries were 11% lower than the monthly mean national wage in 1989. Given these bureaucrats’ elite educations, this represents a substantial gap in income when compared to both elected officials and the private sector.
Prior research and journalistic accounts document the substantial rents associated with revolving door movements in advanced economies with large pay differentials between high-skilled public and private sector workers, such as Japan, the United Kingdom, and the US (Blumenthal Reference Blumenthal1985; Kalmenovitz, Vij, and Xiao Reference Kalmenovitz, Vij and Xiao2022; Mizoguchi and Van Quyen Reference Mizoguchi and Van Quyen2012; Ramseyer and Rosenbluth Reference Ramseyer and Rosenbluth1993; The Financial Times 2024; Usui and Colignon Reference Usui and Colignon1995). Accordingly, the revolving door functions as a primary institutional mechanism through which material returns to office are (later) realized for civil servants.
Yet, little is known about the structure of the job market for former bureaucrats. Existing work has focused on politicians, offering limited insight into the broader population of civil servants. While it is intuitive that senior bureaucrats with policy influence and extensive networks would be desirable hires, the employment prospects for mid- and lower-level bureaucratic officials remain largely unexamined, in part due to data constraints.
Theoretically, firms should only seek to hire former bureaucrats who offer clear value—whether through specialized expertise, access to regulatory influence, or facilitation of rents such as contracts, loans, or bailouts. This implies that private sector hiring should be concentrated among officials from ministries with substantial regulatory or fiscal authority. In support of this view, empirical work from the UK finds that the revolving door is most active in departments with control over major policy levers, while officials from lower-capacity ministries are less likely to transition into private sector roles (Andrews and Beynon Reference Andrews and Beynon2024).
Yet, while the private sector exercises selective demand for bureaucrats, ministries face broader institutional incentives. Because earnings during public service are constrained, ministries rely on postgovernment career pathways as a form of deferred remuneration. Facilitating these placements serves two primary functions. First, it preserves ministerial influence by embedding former officials in positions across the public and private sectors. And second, it signals to prospective hires that a relatively underpaid career in the civil service will be compensated in the future with a lucrative postbureaucracy position. Indeed, contemporary policymakers have recognized this logic explicitly. For example, the head of the U.K. government’s legal department recently stated that she is “all in favor of the so-called revolving door” because of its utility for recruitment (The Financial Times 2024).
However, ministries face a problem of timing and imperfect information—it is impossible to predict ex ante who will reach senior ranks at the time of recruitment. To address this information asymmetry problem, ministries may pursue a strategy of broad-based postretirement placement, treating deferred compensation as an institution-wide guarantee rather than a reward contingent on individual career trajectory. This logic is especially salient in Japan, where civil servants retire before becoming eligible for pensions, heightening the need to secure postretirement employment. In this context, revolving door placements may function as a quasi-contractual obligation: an implicit guarantee of postcareer income and security in exchange for long-term bureaucratic loyalty. Like other informal labor institutions—such as lifetime employment in Japan, guanxi in China, or the traditional labor protections conferred on federal bureaucrats in the US—there may be severe consequences for institutional reputation and recruitment in the event that this informal contract is violated.
I generate two main predictions from these theoretical insights. First, private-sector demand for former bureaucrats should be concentrated among high-ranking officials from prestigious ministries, whose networks and influence are of most value to firms. Second, ministries should endeavor to place as many officials into postretirement roles as possible—including those who might not be in high demand from firms—in order to sustain the revolving door as a recruitment and retention tool.
Hypotheses
The revolving door offers firms access to insider knowledge, regulatory influence, and potential rents, while providing bureaucrats with deferred compensation and ministries with a mechanism for recruitment and retention.
These dynamics generate three primary hypotheses. First, private firms should be more likely to hire high-ranking officials from prestigious and economically connected ministries. Second, if ministries use the revolving door as a tool to attract and retain talent, they should endeavor to place lower-ranking officials into outside positions as well, supplementing the natural market for high-ranking hires and resulting in a bifurcated job market for former civil servants. Third, if firms derive value from political connections, hires should be associated with observable benefits to hiring organizations, such as increased government loans and contracts or favorable investor responses.
In the following sections, I (1) provide an overview of the specific context of study—the institutionalized revolving door in Japan, (2) introduce the data used for empirical analyses, and (3) test the propositions above using the described data and a variety of differences-in-differences methods.
CASE SELECTION: AMAKUDARI IN JAPAN
The revolving door is well-known in the Japanese context, where it is referred to as amakudari—literally “descent from heaven.” Amakudari is the institutionalized practice of civil servants retiring into the private or public sector at the end of their careers,Footnote 2 typically near age 60.Footnote 3 While not truly “revolving” since bureaucrats rarely return and mid-career private sector hires into the bureaucracy are uncommon, I adopt the term to align with the broader political connections literature.Footnote 4 These postretirement positions serve two primary functions: they provide deferred compensation for relatively low-paid bureaucrats (Colignon and Usui Reference Colignon and Usui2003; Mizoguchi and Van Quyen Reference Mizoguchi and Van Quyen2012),Footnote 5 and offer continued employment in a country with low old-age cash transfers and high late-life labor force participation (Estévez-Abe Reference Estévez-Abe2008). The institutionalized practice of retirement around 60 thus creates a shock that sheds light on how agencies and individuals manage sudden job insecurity.
While the supply-side motivations for amakudari are well established, less is known about the universe of opportunities available to former bureaucrats and the benefits conferred on hiring firms. Existing studies suggest that amakudari may offer firms privileged regulatory access (Calder Reference Calder1989; Colignon and Usui Reference Colignon and Usui2003; Schaede Reference Schaede1995), reduce oversight (Grimes Reference Grimes2005), or facilitate access to government loans and contracts (Blumenthal Reference Blumenthal1985; Jones Reference Jones2013; Mizoguchi and Van Quyen Reference Mizoguchi and Van Quyen2012; The Economist 2010; The Japan Times 2017; Usui and Colignon Reference Usui and Colignon1995; Woodall Reference Woodall1997). Amakudari-staffed nonprofit and public organizations have also been implicated in scandals linked to bid rigging and the receipt of public subsidies (Carlson and Reed Reference Carlson and Reed2018; Mizoguchi and Van Quyen Reference Mizoguchi and Van Quyen2012). I examine whether these suspected benefits represent systematic phenomena by testing whether amakudari leads to increased granting of government loans to for-profit firms, and increased granting of contracts to nonprofits.
Among Japan’s ministries, the Ministry of Finance (MOF) and the Ministry of Economy, Trade and Industry (METI) are widely regarded as the most prestigious and powerful (Aoki Reference Aoki1988; Calder Reference Calder1989; Mizoguchi and Van Quyen Reference Mizoguchi and Van Quyen2012; Noble Reference Noble2025; Usui and Colignon Reference Usui and Colignon1995; Vogel Reference Vogel2021). Their central roles in fiscal policy, financial regulation, and industrial planning have historically positioned them as dominant institutions in Japan’s developmental state (Johnson Reference Johnson1982; Rosenbluth and Thies Reference Rosenbluth and Thies2010). These ministries recruit heavily from Japan’s most prestigious university, the University of Tokyo, reinforcing their elite status. As a result, bureaucrats from MOF and METI are among the most desirable hires due to their close ties to powerful institutions that control the levers of financial and industrial policy.Footnote 6
Several features of amakudari make Japan a useful case for studying bureaucratic political connections more broadly. First, relatively low public sector pay creates strong incentives for deferred compensation through postretirement jobs. Second, institutionalized early retirement introduces a predictable shock to job security. Third, Japanese officials are typically generalists rather than technical experts, suggesting their value lies in information, networks, and influence rather than specialized skills. Finally, long tenures and stable careers enhance the credibility and durability of the connections they offer, as theorized by Bils and Judd (Reference Bils and Judd2020). Taken together, these conditions provide a compelling context for testing broader theories about the formation, value, and consequences of political connections between firms and the bureaucracy—particularly in settings, where bureaucrats are career generalists and connections must be valuable beyond technical expertise.
DATA
Case-Specific Interviews
In addition to reviewing existing literature, I conducted semi-structured interviews to explore how the bureaucratic revolving door is perceived to benefit both firms and bureaucrats in Japan. I draw on 19 interviews with current and former bureaucrats as well as executives involved in revolving door hiring in Tokyo between 2019 and 2022.Footnote 7 These interviews serve primarily as a source of evaluating hypothesized mechanisms in the Japanese context, offering insights into the Japanese case rather than presuming cross-national equivalence in theoretical mechanisms.
Interviews revealed that receipt of public contracts, regulatory benefits, and government financial assistance (especially in times of crisis) were viewed as potential benefits by directors of firms and bureaucrats in Japan. For example, an official in one of Japan’s ministries stated that while the revolving door “does not necessarily result explicitly in subsidies or contracts, it is definitely beneficial” (Author Interview D1a). A director in a major consulting firm noted that the revolving door “is most beneficial in industries where regulations are most strict” (Author Interview D1b). A corporate finance expert stated that bureaucratic hires tend to increase in the banking sector when banks are in trouble (Author Interview N1a). Finally, a corporate governance expert noted that “investors would definitely notice” high level appointments (Author Interview N1b). At the same time, interviewees often viewed the revolving door as more prevalent among older or weaker firms, and emphasized that the value of hires derived from connections and access rather than technical or managerial expertise.
Interviewees also echoed the logic that widening pay gaps between the public and private sectors make civil service recruitment increasingly difficult, and that the promise of postbureaucracy jobs is one of the few remaining tools to attract top talent. Current and former bureaucrats emphasized stark wage differentials between entry-level civil servants and peers entering sectors such as finance and technology, noting that graduates of Japan’s most prestigious universities are increasingly likely to join these firms or retire from the bureaucracy early citing low pay (Author Interviews J1a, N1c, S1a).Footnote 8
In sum, case-specific context and interview data supports the notion that the previously introduced theoretical predictions are applicable to amakudari in Japan. High-ranking officials from prestigious ministries with valuable connections should be most in demand by private firms, but ministries are also incentivized to ensure as many employees as possible secure outside employment as they rely on the revolving door as a tool to attract and retain talent. Finally, there may be tangible benefits to organizations that hire former bureaucrats, such as increased receipt of government loans and contracts or favorable investor responses. The administrative data described below allows for explicit testing of these hypotheses.
Administrative Data: Records of Civil Servant Re-Employment
Pressure to regulate amakudari Footnote 9 culminated in reform of the National Public Service Act (NPSA) in 2008 (Kato Reference Kato2017; Mishima Reference Mishima2013; Terada Reference Terada2019). The reform mandated that civil servants report postgovernment employment to the Cabinet Office, with appointments disclosed publicly each year (Government of Japan 1947, 106–23-25).Footnote 10
These disclosures provide the basis of a new dataset covering all amakudari placements from 2009 to 2019. Each disclosure includes the bureaucrat’s former agency and title, along with their new employer and job title (see Table A1 in the Supplementary Material). Disclosures include only the first postretirement job and omit subsequent placements.Footnote 11 As such, the strongest ties that might lead to quid pro quo exchanges may be underrepresented, since officials are barred from joining organizations they directly oversaw for two years postretirement. In addition, we do not observe bureaucrats who may have sought but failed to secure postretirement positions, but the set of bureaucrats who do not receive placements is expected to be small.Footnote 12 Unlike earlier studies that rely on convenience samples, however, this dataset captures the full universe of initial revolving door appointments. The dataset is available online as Amakudata (Incerti et al. Reference Incerti, Miyano, Stanescu and Yamagishi2024).
To evaluate the consequences of amakudari, I merge these records with outcome data across three domains: government loans to for-profit firms, stock market reactions to hires at publicly traded firms, and public contracts to nonprofit organizations. Loan data come from the NEEDS database (NIKKEI Inc. 2025a), Japan’s largest source of firm-level financial information, and are merged with firm attributes to compare amakudari and nonamakudari firms on observables. Stock price data are daily adjusted closing prices from Yahoo Finance (Yahoo, Inc. 2025). Contract data are drawn from 93 publicly released reports, comprising roughly 25,000 records of public works projects, subsidies, and contracts issued by the national government to nonprofit organizations (Incerti Reference Incerti2024). These records include agency and recipient names, contract details, award dates, values, and auction types. For noncompetitive (negotiated) contracts, the number of former civil servants on staff at the recipient organization is also reported. See Table 1 for a summary of all data sources.
I now turn to tests of each of the proposed hypotheses, along with the corresponding estimation strategies and results. Each of the upcoming sections also discusses the relevant datasets in greater detail, in relation to the specific empirical strategies employed. Replication data are available at Incerti (Reference Incerti2025).
Table 1. Overview of Data Sources
IS THERE A BIFURCATED JOB MARKET FOR FORMER CIVIL SERVANTS?
To examine the dual expectation that firms prefer high-ranking officials from prestigious ministries, while ministries also seek placements for less in-demand officials, I begin by analyzing descriptive patterns of postbureaucratic employment. These descriptive analyses confirm many insights from decades of qualitative work on amakudari, but also highlight new patterns. Notably, in line with these hypotheses, the data reveal a job market bifurcated by nonprofit versus for-profit corporations, top versus lower level officials, and ministry prestige.
Half of Former Bureaucrats Join Nonprofit Organizations
First, I examine where bureaucrats seek reemployment following retirement from the bureaucracy, and demonstrate that the job market for former bureaucrats is bifurcated by destination type and position level. 6,314—roughly one-half of—bureaucrats retired into nonprofit “public interest” (5,301) or public (1,013) corporations, compared with 6,126 bureaucrats who retired into for-profit firms (i.e., stock and nonstock corporations) as expected (see Figure 1).Footnote 13 Previous scholars discussed cases of this phenomenon (Colignon and Usui Reference Colignon and Usui2003; Carlson and Reed Reference Carlson and Reed2018; Jones Reference Jones2013), but we can now confirm that publicFootnote 14 and public interest corporation hiring of former bureaucrats is much more common as a percentage of total appointments than previously appreciated, and in fact even represents a slight majority.Footnote 15
Figure 1. Amakudari Destinations by Firm Type, All Hires 2009–2019
Nonprofit corporations (NPOs) are a Japanese legal entity that is largely analogous with the term nongovernmental organizations (NGOs) used elsewhere, or 501(c)(3) organizations in the US. As of 2021, there were approximately 51,000 NPOs in Japan (Cabinet Office 2021). These can range from grassroots civic organizations, to religious organizations, to foundations and “public interest corporations” that conduct government-sanctioned public interest projects. Movements from the bureaucracy are primarily to foundations and public interest corporations, many of which are heavily or even entirely reliant on government funding. This has led some to argue that many Japanese NPOs are “quasi-governmental organizations,” or that the government outsources public work to these organizations (Ogawa Reference Ogawa2009). Taking on former bureaucrats may therefore be viewed by NPOs as a method to ensure continued funding.
Examination of hiring by position level reveals that only the highest ranking officials (i.e., vice ministers and assistant vice ministers) retire predominantly (51%) into large, publicly traded firms, while roughly half (48%) of bureaucrats below the rank of assistant vice minister move to public interest corporations (see Figure 2). By contrast, only 39% of vice and assistant vice ministers move to public interest corporations, and only 31% of individuals below the level of assistant vice minister move to publicly traded firms. Further, top hirers in the public interest sector draw primarily from a single ministry (see Figure A4 in the Supplementary Material), suggesting that there are direct pipelines from individual ministries to public interest corporations. No vice or assistance vice minister placements exist within the top ten public interest hirers for the period observed, nor are these hirers drawing from MOF or METI.Footnote 16 Subsequent analyses will demonstrate that these individuals drive contract receipt to their new places of employment, perpetuating a system in which the government drives funds to firms that are used as revolving door placements for certain employees.
Figure 2. Distribution of Destinations (Firm Types) Among All Former bureaucrats Who Were Re-Hired, by Position Level
Higher Ranking Officials Join For-Profit and Publicly Traded Firms
Turning to for-profit firms, we again see a market bifurcated by position level. Higher ranking officials are more likely to be hired by large, publicly traded firms, while lower ranking officials are more likely to be hired by smaller, private firms (see Figure 2). Industries reliant on government contracts—such as transportation—and highly regulated industries—such as finance, banking, and insurance—are overrepresented in hiring compared to the overall economy (see Figure A2 and Table A2 in the Supplementary Material),Footnote 17 and the top for-profit hirers belong to highly regulated industries such as insurance, transportation, and finance.Footnote 18 The most common posts bureaucrats take in for-profit companies are tax advisors, consultants, auditors, lawyers, board members (internal and external), and executives, with top officials more likely to take board member or executive roles in publicly traded firms.
In contrast with public interest corporations, for-profit hirers draw from multiple ministries (see Figure A5 in the Supplementary Material). Some industries, however, draw overwhelmingly from ministries with direct connections. For example, the construction, electric power, transportation, and transport equipment sectors hire predominantly from the infrastructure and transport ministry (MLIT), the majority of banking and finance hires are from the MOF, and the majority of information and communication hires are from the Ministry of Internal Affairs and Communications (MIAC) (see Figure 3).Footnote 19 , Footnote 20
Figure 3. Private Sector Hires by Industry and Ministry (2009–2019)
Note: Top 10 industries by number of hires. Includes appointments from ministries only. Independent agencies not included.
Prestigious Ministries Place More Bureaucrats in For-Profit Firms
Next, I examine which ministries amakudari come from and whether there is variation in placements by ministry. The largest number of hires come from the largest ministries in terms of number of employees.Footnote 21 Adjusted for ministry size,Footnote 22 METI and the Ministry of Education, Culture, Sports, Science and Technology (MEXT) are the largest suppliers of former bureaucrats (see Figure A3b in the Supplementary Material).
The most prestigious ministries—the METI and MOF— place the highest percentage of total bureaucrats into for-profit firms. The MOF places the largest percentage of its retirees in publicly traded firms (59%), followed by the Ministry of Defense (MOD), Ministry of Foreign Affairs (MOFA), and MLIT (see Figure A6 in the Supplementary Material). By contrast, the MHLW, MEXT, and Ministry of Agriculture, Forestry and Fisheries (MAFF) place the largest percentages of their retirees into public interest corporations (72%, 71%, and 66%, respectively). METI’s share of employees in publicly traded firms appears low at first glance. However, this is due to the existence of the METI Patent Office, an external office (gaikyoku) of METI with a large number of employees who move to public interest corporations.Footnote 23 In fact, the Industrial Property Cooperation Center—a patent advisory firm—is the largest public interest hirer (see Figure A4 in the Supplementary Material). Excluding the Patent Office, the same percentage of METI bureaucrats retire in to publicly traded firms as MLIT.
Bureaucrats from more prestigious ministries also tend to retire at a younger age. As the mandatory retirement age is 60 for most civil service positions, the mean age at which an individual leaves the civil service is 59 and there is little variation by firm type (see Table A3 in the Supplementary Material). However, again there is variation by ministry, with younger bureaucrats more likely to leave more prestigious ministries (e.g., METI and MOF; see Figure A7 in the Supplementary Material).
These patterns are consistent with higher demand for officials from prestigious ministries—particularly given the concentration of these officials in more desirable postretirement roles (e.g., high-paying corporate board appointments) and their relative scarcity.
DO FIRMS THAT HIRE BUREAUCRATS SECURE MORE GOVERNMENT LOANS?
To investigate the hypothesis that private sector firms derive tangible value from hires of high-ranking former bureaucrats, I first examine whether private sector hires are associated with increases in government loans received by firms. I, therefore, examine the value of government loans granted to firms before and after their first amakudari hire observed in the data. The analysis shows that the value of government loans received by firms that make amakudari hires increase relative to their matched controls in the years following the hire, and that these effects are driven by hires from prestigious economic ministries.
Data Overview: Government Loans and Corporate Financials
To compare how for-profit firms that hire amakudari officials compare with those that do not, I use the universe of time-series-cross-sectional (TSCS) data on corporate attributes, financials, and loans from 2009–2019 from the NEEDS financial database. This includes data on assets, liabilities, revenue, earnings,Footnote 24 and number of employees, allowing for comparison of firms of similar size and performance. This dataset includes 5,809 unique firms across all years, and was merged with the data on bureaucratic rehires (i.e., Amakudata).
Firms that hire amakudari are different from firms that do not across a number of metrics (see Figure 4). First, amakudari hirers tend to be older and larger.Footnote 25 Second, firms that make amakudari hires have roughly 12 times the amount of debt from public sources as firms that do not make hires (see Table A5 in the Supplementary Material). While not surprising given that firms that hire former officials are on average larger, it is notable as amakudari hirers have only 6 times the amount of private sector debt as firms that do not make such hires. Third, there is empirical support for the characterization that amakudari is more often practiced by lower-performing firms—as suggested in previous research by (Horiuchi and Shimizu Reference Horiuchi and Shimizu2001; Van Rixtel Reference Van Rixtel2002) and echoed by interviewees in business and finance—with amakudari firms exhibiting less than half of the return on investment and possessing roughly half of the capital reserves when compared to firms that do not engage in bureaucratic rehiring (see Table A4 in the Supplementary Material for this data in tabular form). Finally, while only 4% of amakudari firms matched with the NEEDS financial database are missing financial information, 19% of nonamakudari firms are missing the same data,Footnote 26 once again implying that amakudari firms tend to be larger and more well-known. However, while financial data missingness is highly correlated with bureaucratic hiring, it does not vary highly across industries or years (see Figure A8 in the Supplementary Material).
Figure 4. Distributions of Financial Indicators by Amakudari Status (2009–2019)
Note: Includes all firms for which financial data exist in NEEDS.
Empirical Strategy: Time-Series Cross Sectional Matching
As noted above, the baseline financials of firms that make amakudari hires differ from those that do not. To investigate the impact of hiring former bureaucrats on government loans received, I combine a DiD approach with matching methods in order to compare firms “treated” with former bureaucrats with similar “control” firms that do not make a hire (Imai, Kim, and Wang Reference Imai, Kim and Wang2019). Given that amakudari firms differ from nonamakudari firms across their firm fundamentals, I use mahalanobis distance matching to create matches that are close on these covariate values.
I code all years prior to the first amakudari hire observed for each firm as 0 or “control,” and the year of hire and all subsequent years as 1 or “treated.”Footnote 27 As firms are considered always treated following their first amakudari hire, the percentage of firms treated is a strictly increasing function of time. As we can only observe the year in which an amakudari hire was made, not how many former bureaucrats are currently on staff at a firm at a given time, this likely underestimates the actual effect of amakudari on size of government loans overall. If a firm already possesses former bureaucrats on staff, the estimates will capture the effect of an additional hire, rather than any hire.
After matching control and treatment firms on covariates from other units with the same treatment status in the year prior to treatment (
$ {t}_{-1} $
), I apply a DiD estimator to account for a time trend. This allows me to estimate short and long-term average treatment effects (ATT) of a bureaucratic hire for treated firms. I, therefore, estimate the change in loan volume among firms that switch from no hires in the year prior to one or more hires (
$ {t}_{-1} $
) vs. the year of the hire (
$ {t}_{+0} $
) and the subsequent five years (
$ {t}_{+1} $
…
$ {t}_{+5} $
), controlling for firm fundamentals via matching. I also conduct the analysis requiring matching on covariates from other units with the same treatment status for additional years prior to treatment (e.g.,
$ {t}_{-1} $
and
$ {t}_{-2} $
). Note, however, that increasing this required number of “lags” will necessarily increase uncertainty by reducing the number of matches and effectively decreasing sample size.
Results
Figure 5 shows that the value of government loans received by firms that make amakudari hires begins to increase relative to their matched control pairs until the third year following the hire, then government loan receipts begins to decrease and returns to baseline levels by year five.Footnote 28 This increase is sizable, with the point estimate for years two and three following an amakudari hire representing an increase of roughly three billion yen in total loans held.Footnote 29
Figure 5. Estimated Effect of Bureaucratic Hires on Government Loans Received, by Year After Hire
Note: Tabular results can be found in Tables A6 and A9 in the Supplementary Material.
In keeping with their roles as the purveyors of industrial policy and domestic financing, breaking apart the results by ministry suggests that hires from the METI and MOF are the most valuable in terms of securing government loans (see Figures 5b, A10, and A11, and Tables A7–A9 in the Supplementary Material). By contrast, hires from other ministries do not appear to have any effect on the amount of government loans received in the years following an amakudari hire (see Figure A12 and Table A10 in the Supplementary Material).
Applying the same analysis to private sector loans reveals a negative—albeit not significant at conventional levels—relationship between hiring former bureaucrats and future receipt of private loans (see Figure A19 and Table A13 in the Supplementary Material). This suggests that firms may hire former bureaucrats in order to substitute away from private sector loans and toward public financing. However, it does not appear that government loans are necessarily going to Japan’s famous “zombie” firms, as these firms are primarily supported by their private sector main banking partners (Nakamura Reference Nakamura2023).Footnote 30 , Footnote 31
Robustness
To assess the robustness of the main results, I examine potential threats from violations of identification assumptions and model misspecification. Specifically, I evaluate how results vary with the choice of covariate set and matching algorithm, conduct sensitivity checks using alternative model specifications, and examine pretrends.
To examine whether results are sensitive to the choice of matching covariates, I extend the set of matching covariates to additional variables that are plausibly post treatment (leverage, reserve ratio, ROE, and ROI) (see Figure A18 in the Supplementary Material). Results remain significant at the 5% level when matching on these covariates. Patterns of loan receipt remain constant regardless of choice of matching/refinement method.Footnote
32 Increased loan receipt from METI and MOF in time period
$ {t}_{+0} $
remains significant at the 5% level across all specifications, and at either the 5% or 10% level in time period
$ {t}_{+3} $
for all specifications that improve covariate balance (see Figures A21 and A24 in the Supplementary Material).
To address the possibility that estimated effects are driven by short-term fluctuations or longer-run dynamics not captured within the originally specified window, I assess the sensitivity of the estimates to the lead time—that is, the number of years post-treatment included in the analysis. Results also remain significant at the 1% level when expanding the lead window and at the 6% level when reducing the lead window (see Figure A13 in the Supplementary Material). The results for METI and MOF at
$ {t}_{+0} $
also remain significant at the 5% level after transforming the loan outcome variable into either a binary outcome or taking the inverse hyperbolic sine (IHS).Footnote
33
An additional concern is that firms that hire bureaucrats are on a different pretreatment trajectory than those that do not. For example, a firm with declining value might hire bureaucrats to stave off further losses, or a firm might hire bureaucrats when it is already on an increasingly successful growth path. In other words, it is important to ensure that hiring firms and nonhiring firms follow parallel trends before the treatment of hiring a bureaucrat. To address this concern, I (1) require matches on additional pretreatment years, (2) conduct placebo tests on pretreatment periods, and (3) descriptively examine whether treatment and control firms’ covariates diverge in the (nonmatched) years prior to hiring ex-bureaucrats.
Requiring matches for two years prior to treatment yields a similar pattern of effects, although estimates are no longer significant at conventional levels due to increased uncertainty stemming from fewer matched pairs (see Figure A14 and Table A11 in the Supplementary Material). For METI and MOF hires only, the results remain significant at the 5% level in time period
$ {t}_{+0} $
when requiring matches for two years prior to treatment, and at the 10% level when requiring matches for three years prior to treatment, but are not significant at conventional levels for subsequent periods (see Figure A16 and Table A12 in the Supplementary Material). Once again, uncertainty increases due to the smaller number of possible matches with increased lags and fewer possible outcome years, but this analysis ensures by design that (matching) covariates do not exhibit diverging pre-trends up to and including three years prior to treatment.
Additionally, placebo tests examine the effect of treatment at time t on the difference in the outcome between the treated and control units for the pretreatment periods (i.e.,
$ {t}_{-2} $
vs.
$ {t}_{-1} $
,
$ {t}_{-3} $
vs.
$ {t}_{-1} $
, and
$ {t}_{-4} $
vs.
$ {t}_{-1} $
) for years
$ {t}_{-2} $
for all bureaucratic hires (Figure A15 in the Supplementary Material), as well as
$ {t}_{-2} $
,
$ {t}_{-3} $
,
$ {t}_{-4} $
for METI and MOF hires (Figure A17 in the Supplementary Material). In no instance is the effect of treatment on government loans at time t significantly different from zero for the treated and control units in any pretreatment period. Finally, I examine whether treatment and control firms’ covariates diverge in the years prior to hiring ex-bureaucrats for analyses with only one lag period (Figures A25and A26 in the Supplementary Material),Footnote
34 and demonstrate that loans and pretreatment covariates are not highly divergent in terms of pre-trends.Footnote
35
DO INVESTORS REWARD FIRMS FOR BUREAUCRATIC HIRES?
Next, I investigate the hypothesis that publicly traded firms may additionally derive tangible value from hires of high-ranking former bureaucrats through favorable investor responses. If top former bureaucrats are perceived as beneficial to firms, we may observe boosts in stock prices in response to their hiring as investors reward firms for their recruitment. I, therefore, conduct an interrupted time series/event study analysis in which I test for abnormal changes in stock prices on the day a high-profile hire is made. However, more prestigious ministries such as METI and MOF may also be perceived as more valuable to firms due to their abilities to secure financing and contracts, and to influence economic and financial regulations.
There may also be variation in returns in terms of the type of position a bureaucrat occupies at their new firm. Interviews revealed differential expectations for the value of amakudari hires by the type of role they occupy at a firm. A director at an executive consultancy suggested that outside directors were “probably negatively correlated with the profitability of a company” (Author Interview N1c), an executive at a major consulting firm suggested that “government outside directors have no meaning” as they lack business experience (Author Interview D1c), and analysts from a boutique investment firm claimed that government outside directors lowered return on equity (Author Interview J1a). Top bureaucrats are hired in four primary capacities according to our data: as advisors, executives, managers, and outside directors.Footnote 36 I, therefore, conduct the analysis separately for internal (advisor, manager, and executive) and corporate governance (i.e., directors) related appointments due to these different expectations of the usefulness of these positions.
Data Overview: Hiring Announcements and Daily Stock Prices
To estimate the financial value of political connections to firms that make amakudari hires, I first examine the full sample of high-profile hires (i.e., vice minister or assistant vice ministers) into publicly traded firms. I restrict the sample to vice-ministerial and assistant vice-ministerial appointments for two reasons: (1) these individuals are likely to have the largest impact due to their high level of influence, and (2) top appointments are reported in newspapers, and such announcements are necessary to identify an event day for an interrupted time-series estimation strategy. I, therefore, examine changes in stock returns on the day these hires are announced in Japan’s largest business newspaper, the Nihon Keizai Shimbun (NIKKEI Inc. 2025b). In total, I identified 47 events made public in newspaper reports. Stock price data are daily adjusted closing prices from Yahoo Finance for publicly traded firms.
Empirical Strategy: Market Model Event Study
I estimate cumulative abnormal returns (CARs) using a market model event study approach, which measures the stock valuation effects of a corporate event at the time of the event (i.e., a local average treatment effect). This is an interrupted time series model
$$ \begin{array}{rll}{R}_{it}={\alpha}_i+{\beta}_i{R}_{Mt}+{\epsilon}_{it},& & \end{array} $$
where
$ {R}_{it} $
captures the returns to firm i at time t,
$ {R}_{Mt} $
is the return on the market portfolio (here the Nikkei 225 index) at time t, and
$ {\epsilon}_{it} $
captures returns to firm i at time t that can be considered “abnormal” (above and beyond changes in the market porfolio
$ {R}_{Mt} $
). The key quantities of interest are therefore the cumulative
$ {\epsilon}_{it} $
time series, conventionally referred to as CARs, and specifically the CAR on the day of the hiring announcement. I calculate 95% confidence intervals using the bootstrap as it is free from distributional assumptions. The short time window of one day mitigates endogeneity concerns as confounding events would need to occur on the same day, and do so for a large portion of the independently tested events to influence the estimates.Footnote
37
Results
Overall, Figure 6 shows a slightly positive impact of top hiring on firm returns (event day CAR = +1.24, 95% CI = [-0.1, 2.6]). However, this aggregate analysis masks important variation by both the ministry that was the source of the hire, as well as the type of position the official was recruited for. In terms of ministries, investors also appear to react positively to recruitment from METI relative to other ministries (CAR = +2.34, 95% CI = [0.71, 4.01]; see Figure A28 and Table A18 in the Supplementary Material).
The event study results corroborate claims that different types of hires may be valued differently. Figure 6—which depicts cumulative abnormal returns on the day a hire appears in Japan’s largest financial newspaper—provides suggestive evidence that direct hires are perceived favorably by investors. However, an aggregate analysis masks a near zero and null effect of director appointments, and a larger positive effect of roughly 2.2% for direct internal roles (event day CAR = +2.16, 95% CI = [0.71, 3.5]).Footnote 38
Figure 6. CARs from Assistant Vice-Minister and Vice-Minister Appointments
Note: Tabular results can be found in Tables A14–A16.
These findings are notable as previous research has found that markets react favorably to the appointment of outside directors, especially those perceived as independent (Nguyen and Nielsen Reference Nguyen and Nielsen2010; Rosenstein and Wyatt Reference Rosenstein and Wyatt1990). Directors from the bureaucracy may therefore not be perceived as offering the same kind of effective independent oversight and industry expertise, but rather as “yes-men” for the corporation. By contrast, internal hires may bring tangible benefits such as regulatory expertise, connections to contract granting agencies, or expertise regarding loan receipt stemming from their ministerial connections.
In short, I find evidence that investors may view high level bureaucratic hires as indicative of positive future financial performance, but through the mechanism of internal connections rather than corporate governance and oversight. In addition, the most prestigious ministries such as METI again appear to be the drivers of value.
Robustness
Three potential inferential threats to the event study estimates are: (1) the CARs are driven by factors unrelated to amakudari hires, (2) the effects are underestimated as investors knew about the appointments prior to the news releases, and (3) model misspecification.
To address endogeneity concerns, I re-estimate CARs while substituting the real event dates with time-shifted placebo dates. I shift the actual event days forward and backward by the following daily increments:−200,−100,−50,−25,−10,−5, 5, 10, 25, 50, 100, 200. We should not observe significant abnormal returns when performing an identical test on dates where no hire occurred, as this would raise concerns that the abnormal returns were caused by factors other than the hires. There are no significant CARs on any shifted dates except when shifted backwards by 50 days and forward by five days. The results at−50 days are negative and sensitive to changes in the event window, and results at +5 days are at a time in which abnormal returns are still positive and volatile (see Figure A29 in the Supplementary Material and Figure 6).
There is some evidence that investors may have information about hires prior to the dates identified from newspaper reports given positive trends in the pre-event period. However, while such a priori information may call our exact point estimates into question, it would cause an underestimation of the magnitude of the effect on the event day.
To gauge the sensitivity of the estimates to changes in model specification, I recalculate all estimates using a constant mean return model (i.e., with no market index control), calculate confidence intervals using the classic t-test and the Wilcoxon rank-test,Footnote 39 and estimate effects using additional event windows.Footnote 40 Estimates remain virtually unchanged using the t-test, Wilcoxon rank test, and using different event windows (Figures A31 and A32 in the Supplementary Material), and significance levels increase using a constant mean return model (Figure A30 in the Supplementary Material).
DO NONPROFITS WITH BUREAUCRATIC CONNECTIONS RECEIVE MORE LUCRATIVE CONTRACTS?
Roughly half of civil servants initially take up posts at nonprofits after leaving the bureaucracy (see Figure 1). To investigate whether politically connected nonprofit organizations also derive tangible value from these connections, I examine whether the presence of former bureaucrats is associated with increases in the value of government contracts received by these organizations.
Data Overview: Government Contracts with Nonprofits
The Japanese Cabinet Office (CAO) collects and reports data on all subsidies and contracts granted to NPOs in a given year. These reports were scraped and compiled into a dataset of approximately 25,000 contracts and subsidies from ministries to NPOs over a 10 year period. The data include contract values and names of organizations in months in which a contract was granted. For noncompetitive-bid (i.e., negotiated) contracts, the data also include the total number of government re-employees from the ministry that granted the contract at the NPO at the time of contract receipt.Footnote 41 These data have additional benefits over the dataset of initial amakudari appointments used in the previous loan and stock price analyses, as they allow us to view the total number of former bureaucrats at the nonprofit at the time the contract was granted.Footnote 42
As noted earlier and seen in Figure A4 in the Supplementary Material, there are direct pipelines of civil servants that flow from specific ministries to specific NPOs each year. The contract data show that the same organizations that receive regular flows of bureaucrats from specific ministries often receive large volumes of contracts from those ministries. For example, the Japan Forest FoundationFootnote 43 hired 41 officials from the MAFF from 2009–2019, and in the same period received 305 contracts from MAFF totaling over 2.5 billion yen. The Japan Construction Information CenterFootnote 44 hired 21 officials from MLIT and received 67 contracts totaling over 1.15 billion yen, 48 contracts and 1.07 billion yen of which came from MLIT. I next use a DiD estimation strategy to systematically investigate if nonprofits receive higher value contracts when former officials are in director positions at their organizations.
Empirical Strategy:
$ DI{D}_M $
Estimator and Robust TWFE Approaches
The contract data take the form of an unbalanced panel in which NPO-contract dates are observed at unevenly spaced intervals. In other words, we only observe months in which a contract was granted, and contracts are not granted to all NPOs in all months. In addition, units are “treated” with former bureaucrats on staff at different points in time, and units can switch from control to treatment and from treatment to control.
As I have no covariates for NPOsFootnote 45 I cannot employ the same matching-adjusted DiD design that was used for government loans, placing us in the realm of traditional TWFE estimators. However, recent findings have shown that coefficients from traditional TWFE models may not represent an average of unit-level treatment effects when effects are heterogeneous across time or units (as in this case). de Chaisemartin and D’Haultfœuille (Reference de Chaisemartin and D’Haultfœuille2020) show that TWFE models can even lead to coefficients having the opposite sign of each of the unit-level treatment effects, as TWFE estimates are a weighted average of unit-level treatment effects and these weights can sometimes be negative due to differences in the timing of treatment or heterogeneity amongst units.
I, therefore, adjust my estimation strategy and use the
$ DI{D}_M $
Footnote
46 estimator proposed by de Chaisemartin and D’Haultfœuille (Reference de Chaisemartin and D’Haultfœuille2020) to estimate the ATT. The
$ DI{D}_M $
estimator compares outcomes among groups whose treatment status switches between time
$ t-1 $
and t, and control groups whose treatment status remains constant in time
$ t-1 $
and t. The
$ DI{D}_M $
estimator therefore accounts for the data structure as it relies on first differences only. Formally, the
$ DI{D}_M $
estimator can be described as
$$ \begin{array}{l}ATT=\unicode{x1D53C}[{Y}_{it}(1)-{Y}_{it}(0)|\\ {}({D}_{i,t-1}=0,{D}_{i,t}=1\hskip2pt \mathrm{or}\hskip2pt {D}_{i,t}=1,{D}_{i,t+1}=0)],\end{array} $$
where Y are potential outcomes and D is the treatment status of unit i in time t. The
$ DI{D}_M $
estimator is then equivalent to the average of the
$ DID $
s across all pairs of consecutive time periods and across all values of the treatment. The estimator also accommodates both binary and continuous treatments, allowing us to estimate the effect of any amakudari appointments on contract value, as well as the marginal effect of an additional appointment on contract value. The treatment effect can therefore be interpreted as the average effect of the treatment on the units that experienced a change in treatment status—specifically, “switchers in” to treatment or “switchers out” of treatment. Note that this does not necessarily imply an immediate effect within a given month, but rather the effect on contract value compared to previous periods when no bureaucrats were on staff.Footnote
47
Next, I apply Benford’s Law to the value of contracts with amakudari bureaucrats as well as those without.Footnote 48 Benford’s Law is used in forensic accounting to examine discrepancies in the natural probability of leading digits appearing in data—that is, numbers beginning with 1, 2, 3, etc. If contracts negotiated without former bureaucrats on staff conform with Benford’s Law while contracts negotiated with former bureaucrats on staff do not, this would suggest a more competitive negotiation process for nonconnected NPOs on the one hand, and evidence of contract price fixing for connected NPOs on the other. To investigate this possibility, I examine the mean absolute deviation (MAD) of leading digits in contract values compared to the predicted frequency according to Benford’s Law, for which Nigrini (Reference Nigrini2012) has proposed critical scores for conformity and nonconformity with Benford’s Law.
Results
When examining initial appointments immediately following retirement from the bureaucracy only, I find a near zero and null effect of bureaucratic rehires on government contract value for both all subsidies and contracts as well as negotiated contracts only. However, when using the total number of bureaucratic employees at the time of the contract as the dependent variable, I find a positive and statistically significant increase in the value of negotiated contracts granted to nonprofits with amakudari appointees (see Figure 7).
Figure 7. Effect of Amakudari Appointments on NPO Negotiated Contract Value
Note: Tabular results can be found in Tables A20–A21.
These results suggest that bureaucrats may be waiting until after the end of their two year cooling off period to join nonprofits, at which time they can use their connections to exhibit influence on contract negotiation. As the CAO data only include director level appointments, it is also possible that only high-level appointees have the connections needed to negotiate higher contract values.
Additional evidence of the potential influence of former bureaucrats on contract value can be found through the application of Benford’s Law. Based on Nigrini’s (Reference Nigrini2012) critical MAD scores, competitive bid contracts exhibit “close conformity” with Benford’s Law, while negotiated contracts exhibit “marginally acceptable conformity,” and negotiated contracts with former bureaucrats on staff exhibit “nonconformity.” This suggests that negotiated contracts between former officials in particular may not be subject to hard bargaining. Visual depictions of the leading digit of contract values as predicted by Benford’s Law and as observed in the NPO contract data can be found in Figure A36 in the Supplementary Material.
Robustness
To address concerns about pre-trends in contract value, I include “placebo” estimates of the ATT for −3,−2, and−1 periods before treatment as a check of whether outcomes for the treated and comparison groups move in parallel prior to the staggered treatment periods. Placebo estimates are never significantly different from zero at conventional levels.
Given recent concerns with TWFE estimators—particularly reliance on strict exogeneity, functional form assumptions, and negative weighting in the presence of treatment effect heterogeneity—I implement a number of alternative estimators which also allow for treatment status to vary over time and address the potential for negative weighting. Specifically, I use the estimators referred to by Liu, Wang, and Xu (Reference Liu, Wang and Xu2024) as: the fixed effects counterfactual estimator (FEct), the interactive FEct (IFEct), and the matrix completion (MC) estimator,Footnote 49 in addition to a traditional TWFE estimator. Finally, to assess robustness to functional form assumptions and allow for interpretation of effects in both relative (percent) and absolute terms, I run all models including both a log-transformed outcome variable and in levels.
Estimates using yearly aggregated data, alternate functional forms, and using the FEct, IFEct, MC, or traditional TWFE estimators corroborate the estimates from the
$ DI{D}_M $
estimator in terms of sign, magnitude, and pre-trends, and can be found in Figures A33–A35 and Tables A22 and A23 in the Supplementary Material.Footnote
50
DISCUSSION AND CONCLUSION
This article provides the first systematic analysis of all initial revolving door placements from the bureaucracy to postbureaucratic employment in any country, leveraging a novel, comprehensive dataset of postbureaucracy employment in Japan. By combining these data with government loan and contract records, financial outcomes, and qualitative interviews, I analyze both the structure and economic consequences of Japan’s revolving door.
Theoretically, this article puts forward a new framework for understanding the role of the revolving door in bureaucratic systems with relatively limited compensation when compared to the private sector—institutional features of bureaucratic systems in many advanced democracies. Rather than viewing postbureaucratic employment solely as a means for firms to gain influence, I argue that the revolving door is an important recruitment tool for ministries. Because information about a bureaucrat’s rank at retirement is unavailable at the time of hiring, the system virtually guarantees all recruits some form of lucrative or secure postretirement employment. This helps the civil service attract top talent who might otherwise choose more lucrative private sector careers from the outset. In support of this theory, I show that higher-ranking bureaucrats are more likely to be hired into for-profit firms that may value their prestige or influence, while lower-ranking bureaucrats are more likely to move into quasi-governmental and nonprofit organizations, many of which receive significant government contracts or funding. This makes the revolving door not only a possible postbureaucratic career benefit, but also an integral part of the bureaucratic compensation structure and recruitment pipeline.
Empirically, I find that nearly half of all revolving door hires occur in the nonprofit or public sector, and that these hires are disproportionately lower-ranking bureaucrats and those from less prestigious ministries. Ministries appear to create revolving door opportunities for lower-ranking staff by channeling funds into connected nonprofits, creating demand, where private sector demand is absent. DiD methods show that nonprofits with ex-bureaucrats on staff receive more generous government contracts. In the private sector, firms that hire high-ranking former bureaucrats—particularly from powerful economic and financial ministries such as METI and MOF—receive larger volumes of government loans, consistent with utilizing the revolving door for rent extraction. Event studies also show that these appointments generate positive stock market reactions, indicating that investors recognize their value. These findings reveal the consequences of an institutionalized bureaucratic revolving door system where individuals do not cycle back into public office.
The one way nature of Japan’s revolving door distinguishes it from other well-studied systems. For example, in France, the UK, and the US, bureaucrats and politicians may move repeatedly between the public and private sector, or hold dual roles (if they are politicians), raising concerns about regulatory capture due to socialization, future political ambition, or conflict of interest. In Japan, however, bureaucrats exit public service upon retirement, and their incentive structure while in office is driven by the need to secure stable postretirement employment—especially given early mandatory retirement and comparatively low public salaries. As a result, Japanese bureaucrats may tailor policy, resource allocations, or regulatory decisions to appeal to potential postretirement employers, not because of reelection incentives or campaign contributions, but for employment security.
Although Japan’s system is shaped by unique institutional and cultural factors, the underlying mechanisms are not likely to be unique. Countries with shrinking bureaucratic prestige, rising public-private pay gaps, and tightening retirement benefits may increasingly rely on revolving doors to sustain bureaucratic recruitment. Indeed, calls to institutionalize revolving door systems have recently emerged in countries like the UK, where officials note that without such mechanisms, government service will fail to attract top talent. Research shows that more prestigious U.K. departments are also more likely to feed into revolving door positions in the private sector (Andrews and Beynon Reference Andrews and Beynon2024). As protections for civil servants are eroded in the US, Departments may increasingly need to market the possibility of postbureaucracy positions to retain recruitment. These trends suggest that Japan’s model may foreshadow broader patterns in advanced democracies, where the state’s capacity to compete for expertise increasingly depends on informal compensation structures.
In the Japanese context, these dynamics intersect with broader issues such as the welfare state and economic stagnation. Government-connected nonprofits may serve as a public sector analog to the “zombie lending” practices of providing low interest government loans to uncompetitive firms in Japan’s private sector. Informal “contracts” such as quasi-guaranteed postretirement positions may function as a backdoor safety net and alternative to expanded programmatic welfare transfers—providing employment to former bureaucrats while channeling funds into politically connected organizations. This echoes previous arguments that Japan’s response to economic stagnation has involved sustaining employment and institutional stability through informal channels, even at the cost of efficiency (Caballero, Hoshi, and Kashyap Reference Caballero, Hoshi and Kashyap2008). Like the private sector case of providing zombie lending to ensure stable employment, it is unclear whether costs in terms of efficiency or resource misallocation outweigh the benefits of reduced unemployment, more successful bureaucratic recruitment, and increased worker motivation. Nonetheless, as the bureaucracy has lost prestige compared to the heyday of Japan’s rapid economic development and the pay gap with the private sector has grown for elite bureaucrats, theoretical expectations are that the maintenance of this system will become even more critical in order to guarantee new hires lucrative postretirement positions.
From a policy perspective, this article also highlights how transparency is undermined not only by secrecy but by fragmented and inconsistent reporting. While Japan formally discloses bureaucratic retirement data, they are dispersed across thousands of files and databases with irregular formatting. This makes it difficult for the public to trace patterns of influence or accountability. Reforming such data systems is an essential step toward meaningful transparency and oversight—not just in Japan, but in any system looking to trace the influence of institutions.
Finally, this research opens several avenues for future investigation. First, while I examine domestic outcomes such as loans and contracts, I do not study how revolving door hires affect international business strategies, regulatory alignment or language, or trade outcomes. The effects of the revolving door in international arenas and on regulatory policy decisions remain an important area for further theoretical development and empirical analysis. Second, cross-national comparisons of revolving door systems can help clarify the conditions under which revolving doors serve as tools of influence, mechanisms of state capacity, or both. I hope the public data and theoretical contributions offered here serve as a foundation for future research.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit https://doi.org/10.1017/S0003055425101263.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/JGHDYL. Limitations on data availability are discussed in the supplementary material.
ACKNOWLEDGEMENTS
I extend a special thank you to Sayumi Miyano, Diana Stanescu, and Hikaru Yamagishi, my partners in constructing a dataset that made this project possible, and to my mentor in all things Japanese politics, Frances Rosenbluth—this article is dedicated to her memory. I also thank: Ayumi Sudo, Ken Tanaka, and Junyao Zhang for excellent research assistance; Jason Anastasopoulos, P.M. Aronow, In Song Kim, Miriam Golden, Mina Pollmann, Ulrike Schaede, Kenneth Scheve, Tara Slough, Seiki Tanaka, Yuhua Wang, and Hye Young You for invaluable comments; participants at Columbia’s Horizons in Japanese Politics conference, Copenhagen Business School’s Money in Politics conference, the 2024 APSA Annual Meeting, and seminars at EUI, Georgetown, Harvard Weatherhead Center, Stanford APARC, the University of Amsterdam, the University of Toronto Munk School of Global Affairs, and Yale; the UTokyo Institute for Social Science and Waseda Institute of Political Economy for database access; and the Japan Foundation for generous financial support. Any and all errors are my own.
FUNDING STATEMENT
This research was funded by a Japan Foundation Center for Global Partnership Grant.
CONFLICT OF INTERESTS
The author declares no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The author declares the human subjects research in this article was reviewed and approved by an Institutional Review Board at Yale University and certificate numbers are provided in the text. The author affirms that this article adheres to the principles concerning research with human participants laid out in APSA’s Principles and Guidance on Human Subject Research (2020).
Open access
Comments
No Comments have been published for this article.