The communication and spread of policy ideas across territories (e.g., countries, states, and cities) is a key research niche in the field of political science, and political scientists are particularly interested in what drives policy diffusion [13]. Despite the impact of domestic attributes and policy-specific factors, various diffusion effects originating from intergovernmental relations (e.g., top-down, peer effect, and bottom-up) are more emphasized in the literature [2]. Policy diffusion can be driven by at least four processes, including norms, coercion, competition, and learning [9]. While the former three diffusion mechanisms have been extensively discussed, our understanding of policy learning and its role in policy diffusion is by and large scarce [10, 27]. Given the increasingly prominent role played by peer learning in policy diffusion, it is theoretically intriguing and policy relevant to examine the impact of policy learning on the diffusion of public service innovations. The existing studies primarily contextualize policy learning in the US and European countries, at either state or local level [4, 11, 17], while our understanding in non-Western contexts (e.g., China) is largely limited. Moreover, the current elaboration of policy learning is rather simplified [3], and its processes and mechanisms are still a black box deserving further exploration.

The creation, adoption, and diffusion of innovative ideas, practices, technologies, modes, and processes are pivotal to the resilience, legitimacy, and support of authoritarian regimes, which can partially explain why the party-state in China is proactive in policy and public service innovations [35]. The past two decades have witnessed the flourishing and spreading of policy and public service innovations across China, which have been documented by the Innovations and Excellence in Chinese Local Governance (IECLG) awards program since 2001 [35, 41, 42, 46] and many other awards and studies [36]. The generation and diffusion of government innovations in China have been examined by numerous studies (see, for instance, the recent literature [16, 19, 20, 25, 44, 48]), which help identify and understand the key factors and mechanisms through which innovations emerge from and diffuse among local governments.

Policy learning plays a pivotal role in facilitating policy diffusion in China [6, 33, 40], but its prominence is usually underestimated in the literature. In a unitary system with a single ruling party, government officials are appointed by superior party committees, which incentivizes local cadres to develop policy innovations to advance careers. It is argued that original invention is weighted more than learning and adoption in cadre evaluation, which incentivizes local officials to pursue bandwagons and fads in governance innovation [35]. Local governments are thus incentivized to create new programs and practices (e.g., pioneering or championship), but are less inclined to incorporate others’ existing experiences as followers [47]. While such arguments help to interpret some interesting phenomena in China’s local government innovations, we suggest in this study that the importance of policy learning is underestimated and should be revisited.

In this paper we use the case of public bicycle programs (PBPs) in China to empirically examine the effect of policy learning on policy diffusion. Public bicycle programs aim to bridge commuters between public transit (e.g., bus, metro), which also helps mitigate congestion, reduce private car use and carbon emission [31]. Hangzhou, the capital city of coastal Zhejiang province, is well known for its pioneering PBP since 2008 [32]. As one of the first and most prominent city-wide PBPs in China, Hangzhou has attracted many visitors from other city governments in China [45]. With professional and technological assistance from this knowledge hub, hundreds of Chinese cities have initiated their own PBPs over the past decade [21]. Hangzhou PBP is a perfect case with which to examine the effect of inter-city learning on policy diffusion, and we use this case to deepen our understanding of policy learning and diffusion. In this study, our key research questions are as follows:

  1. (1)

    Which cities are more likely to learn from others in policy formulation?

  2. (2)

    Does policy learning facilitate policy diffusion?

  3. (3)

    When is policy learning more likely to influence policy diffusion?

We draw on data from multiple sources such as archival and interview data and use mixed approaches of quantitative and qualitative methods to investigate the role played by inter-city learning in the diffusion of PBP. We find that adjacent cities characterized by similar attributes are more inclined to learn from Hangzhou’ PBP, which facilitate these cities to adopt PBPs. More frequent site visits attended by more delegates and for more days are more likely to elicit policy adoption, while the leaders’ rank of delegations does not matter. Leadership turnover, prudent decision-making (e.g., incompatibility of PBP), and alternative solutions (e.g., metro) are found to be the missing links in transforming policy learning into policy adoption. Our findings help to explain the crucial role played by policy learning in policy diffusion, and contribute to the literature on policy learning and diffusion by highlighting the underpinning mechanism of inter-city learning. In the remainder of this paper, we first discuss the relevant theories and models used in policy learning and diffusion, followed by the development of testable hypotheses. We then introduce the context of PBPs in Chinese cities, and particularly the mobilizing role played by Hangzhou’s PBP. We present the data and methods used in this study, followed by our empirical findings. We finally discuss the theoretical contributions and policy limitations of the results, and outline limitations and future research avenues.

Policy Learning and Policy Diffusion

“An innovation is an idea, practice, or object that is perceived as new by an individual or other unit of adoption.” [29] As Walker (1969) elaborates, adoption of new programs is different from their invention or creation. “An innovation will be defined simply as a program or policy which is new to the states adopting it, no matter how old the program may be or how many other states may have adopted it.” [39] As Rogers defines, “[d]iffusion is the process by which an innovation is communicated through certain channels over time among the members of a social system.” [29] For instance, PBP is not created or invented from the scratch by Hangzhou and other Chinese cities, but rather it is adopted by and diffused among them.

The transfer [34], mobility [23], lesson-drawing [30], institutional isomorphism [22], and diffusion of policies and programs [9] among jurisdictions (continents, countries, states, and cities) has been extensively examined in various domains and contexts, and several mechanisms have been identified in driving policy diffusion.Footnote 1 Policy diffusion is jointly shaped by jurisdictional attributes and interjurisdictional interactions, and the latter is primarily the focus of the literature [2]. In a recent review, Dobbin and colleagues synthesize four processes driving policy diffusion, including norms, coercion, competition, and learning. Institutional theory argues that international organizations may prompt national governments to adopt policies in line with their norms for purpose of legitimacy [37]. Policy diffusion can be facilitated by top-down mandates, which incentivize governments at lower levels to follow up in adopting the policies and programs advocated by their superior authorities [9]. In taxation and environmental regulation, forceful competition for foreign direct investment and economic growth impulses governments to “race to the bottom” to cut taxes or lower regulatory standards. Finally, successful practices of peer jurisdictions inform governments to adopt similar programs to address the same policy issues.

Social learning and economic competition are two primary forces driving policy diffusion at a horizontal level, but they are not well disentangled in empirical studies [3]. Governments can learn either from their own experiences or others’ experiments [38], and the latter plays a more important role in policy diffusion. It is useful to distinguish emulation and learning, which are often used interchangeably albeit conceptually distinct [26]. Learning is rational or bounded rational, while emulation is purely imitation without conscientious calculation of costs and benefits. “Sometimes new legislation is virtually copied from other states.” [39] Rational governments borrow ideas and adapt them to local conditions. Despite some innovations that may be applicable in virtually all contexts, governments still must align them with specific requirements, and even reinvent them for perfect match with policy circumstances. Learning can be either the direct adoption of concrete policies or indirect heuristics of ideas, and both approaches benefit policy design and implementation. Lesson-drawing could be achieved through several forms, e.g., inspiration or hybridization of multiple cases [30].

Site Visit

Policy learning and knowledge dissemination can be achieved by multiple channels, and in this study we focus on one key approach, i.e., the site visit. Business visits or site visits are face-to-face interaction and knowledge exchange among policymakers when delegations visit a successful demonstration project. Despite the fact that policy learning could be realized by indirect or second-hand approaches (e.g., online search and off-site professional training), site visits help visitors personally observe and experience the operations and effectiveness of innovative programs [5]. Particularly for more sophisticated technological innovations like PBP, site visits equip leaners with the “know-how” to implement and operate programs, which are difficult (if not impossible) to learn from other sources. On-site observations also help delegations personally discover the effectiveness of new programs, and such experiences and implications are irreplaceable in shaping policymakers’ perceptions and dispositions of adoption. A recent study suggests that abroad study visits (from South Korea to the UK) indeed facilitate the processes of policy learning, although they may not directly lead to policy change [15].

The number of potential jurisdictions policymakers can learn from is enormous and a rational decision-maker might limit the scope of search [8]. Policymakers often rely on several criteria to select their benchmarks, and the key factors include proximity (e.g., neighbors or in the same state/province), similarity (e.g., political ideology or size), and success (e.g., well-managed or satisfied by citizens). A recent survey of 52 mayors of the largest cities in the US shows that policymakers look to other cities for policy ideas, and they are more likely to target peers characterized by geographic proximity, trait similarity, and policy success. These three factors usually substitute for each other and mayors have to make trade-off in naming cities, e.g., targeting successful and proximate albeit dissimilar cities or vice versa [11]. Policies are not value-neutral, and ideological considerations are usually highlighted in policy learning [14]. For example, a survey experiment of US municipal officials reveals that policymakers are less likely to learn from others if policies are ideologically opposite, but policy success and partisan peers’ acceptance mitigate such bias [4]. A study of municipalities in Sweden also suggests that policymakers learn from proximate, similar, and successful cities [17]. While these studies focus on from whom policymakers are more inclined to learn, their findings also apply in the reverse consideration, i.e., who are more likely to learn from a hotspot of innovations, herein Hangzhou.

Governments usually compete with and learn from their adjacent peers, particularly neighbors [2]. Citizens learn about new policies and programs from neighboring jurisdictions, and governments face public pressure to adopt similar policies. In contrast to remote jurisdictions, proximate peers are more likely to be targeted due to ease of learning and long-term connections. It is thus imperative to examine whether geographic distance matters in inter-city learning. Despite the new development of air routes, bullet trains, and highways that substantially shorten travel time, geographic distance still plays a pivotal role in inter-city learning. We expect cities with a shorter geographic distance from a targeted city are more likely to visit and learn from its policies.

  • H1: Local governments are more likely to learn from geographically proximate peers than remote ones.

Governments are inclined to empathize with those similar to them in important aspects, which suggests inter-city policy learning is more likely to occur between cities sharing similarities. Innovations may not be suitable to everyone, and rational learning is different from blind imitation in how decision-makers pay attention to the right benchmark. Policy knowledge is more applicable and transferable if the learner and host are similar in important aspects. Otherwise, policy imitation may fail simply due to mismatch of policy problems and innovations. We thus expect that cities visit and learn from a targeted city more similar with it in key factors, such as population size, public finance, and other policy-specific factors.

  • H2: Local governments are more likely to learn from counterparts with a higher level of homophily in jurisdictional and policy-specific attributes.

Learning and knowledge exchange across cities play an indispensable role in facilitating the spread of innovative programs and policies [24]. Cities connect with their peers in forms of twin towns, sister cities, alliances, and networks, which are leveraged for knowledge exchange, learning, and innovation [5]. Particularly institutionalized intergovernmental relationships (e.g., inter-local contracts and professional associations) help to foster policy learning among members through formal communications and exchanges [1, 12]. Mayors and city managers usually visit peer cities to learn about new practices and programs, and such visits may challenge their long-held misassumptions or confirm their attitudes toward new programs, which are powerful in eliciting follow-up decision-making. In this regard, we expect that cities with experiences of site visits are more likely to adopt new policies. Alternatively, famous tourist destinations are more likely to be targeted for site visit, and it is also possible that visitors from other cities are primarily for leisure and entertainment. Rational policymakers may find policies of targeted cities are not suitable for their context, and site visits even remind them to copy with caveat. Policymakers can learn from others’ experiences from alternative channels, and the learning effects of site visits may not as strong as we expect. As an alternative explanation, there may be an insignificant and even negative relationship between site visit and policy adoption. Taken together, our third hypothesis is as follows.

  • H3: Local governments with site visits are more likely to adopt new policies than those without site visits.

While site visits may help to boost policy adoption, it is also interesting to examine when they matter more in eliciting policy adoption. Site visits of demonstration projects are more likely to influence policy adoption if the key policymakers are among the delegates. It is mayors and senior officials who are policy entrepreneurs with decision-making power, and their engagement in site visit may prioritize and facilitate policy adoption. Junior officials may report to policymakers about their site visits, but top managers’ direct observation and engagement can work much stronger in driving policy change. We thus expect that site visits by senior officials are more likely to prompt policy adoption.

  • H4: Site visits participated in by senior officials are more likely to generate policy adoption.

The size and composition of delegates also matter, particularly for policies requiring cross-sector collaboration in implementation. Taking the case of transport policies, they are usually not solely the responsibility of one agency, but involve multiple agencies in other domains such as land resources, park and recreation, public finance, public security, tourism, and social assistance. Without coordination with other functional departments, new policies cannot be seamlessly initiated and implemented. In other words, delegations composed of more members from these pertinent agencies are equipped with interagency coordination capacity to advocate and implement new policies.

  • H5: Site visits from larger delegations are more likely to generate policy adoption.

Finally, the period and frequency of site visits may influence policy adoption. Short periods on site may not result in enough time for a deliberate experience of the program, and such superficial policy tours would not generate policy adoption. The frequency of site visits reflects policymakers’ interest in the program, and we expect it to be positively related to policy adoption. Our two hypotheses are as follows:

  • H6: Longer site visits are more likely to generate policy adoption.

  • H7: The frequency of site visits is positively related to policy adoption, and more frequent site visits are more likely to generate policy adoption.

Public Bicycle Program

Modern urban transport is characterized by a combination of multiple traffic modes, especially wide coverage of public transit and ubiquitous use of private cars. Such transport approach is not without problem, and lots of large cities around the world suffers chronic traffic congestion and severe environmental pollution. Public transit systems are often not user-friendly and seamlessly connected, which lengthens walking distance and impedes commuting demands. Excessive use of private cars increases carbon emission, air pollution, and traffic congestion. PBP or a public bikesharing program aims to resolve “the last mile” problem in public transit by bridging commuters across different traffic modes. The adoption of PBP also help boost the use of public transit and alleviate commuters’ reliance on private cars, which helps build green urban transport systems.

The early experiments of PBP in Europe did not succeed due to various obstacles in riders’ safety, integration with urban transport systems, and financial sustainability. Public bicycles are frequently broken, stolen, or vandalized. Public bicycles are usually free or have a low charge, but intensive fixed-asset investment and high maintenance cost soon bankrupt operators. Prior failures in early trials significantly impeded the diffusion of PBPs, with few followers in advocating the adoption of PBPs. Technological advancements (e.g., fixed parking slots and electronic locks) have largely facilitated the development and spread of PBPs, although they still face challenges in many aspects [31]. Policy success plays a core role in the diffusion of innovations, since later adopters are primarily inspired by successful pioneers. No one would like to replicate failures, and policy success is among the most important factors in determining the plausibility of success and identifying best-practice cases. A recent cross-country analysis reveals that entrepreneurs in public and private sectors have facilitated the adoption of PBPs, and successful programs in Europe and North America inspired followers in accepting PBPs [28].

The global spread of PBPs can be examined from the supply-demand perspective. Urban governments demand innovative solutions to traffic challenges, which is partially addressed by the adoption of PBPs. Given the increasing demand from urban commuters, both private and public operators actively supply PBPs. The introduction of new information technologies, particularly smart phone and GPS, has been revolutionizing the PBP industry. Private entrepreneurs in China, for instance, released bikesharing services without parking slots, e.g., Mobike, ofo bicycle, UniBike and so forth. These new bikesharing services compete with traditional PBPs maintained by public operators, which needs further study.

Public Bicycle Program in Hangzhou and China

China previously was called “the Kingdom of Bicycles,” and bicycles were urban commuters’ predominant traffic mode. With the increasing purchase of private car and the rapid development of public transit (e.g., bus and metro) since the Reform and Opening-up in 1978, the bicycle has steadily faded away from people’s traffic lives [43]. Similar to their counterparts in developed countries, roads in Chinese cities are soon crowded with millions of private cars. PBP was introduced by private entrepreneurs in Beijing and other cities in early 2000 to mitigate traffic congestion, streamline commuting interchange, and prompt green transport. A few tourist destinations also piloted PBPs in limited areas to cater to tourists’ traveling needs. However, not until Hangzhou initiated its PBP in 2008, was the idea diffused and widely practiced across China [32].

Hangzhou is not only among the largest PBPs in the world, but has also sustained the program successfully for a long period of time. The operator, Hangzhou Public Bicycle Service Company, was set up in April 2008, as a wholly owned subsidiary of Hangzhou Public Bus Group. Hangzhou PBP was piloted on May 1, 2008 and formally operated by September 16, 2008. Hangzhou’s operator learned from past lessons and experiences of European PBPs, particularly that of Paris and Rotterdam. It adapted the business model to match local conditions, and underwent extensive technological R&D to address the massive demand for public bicycles in morning and afternoon peak times. It runs 3608 stations and owns 84,100 bicycles by September 2016. It supplies public bicycle service to a maximum of 473,000 riders per day, and over 96% are free rides.Footnote 2 It also incubated a company (Hangzhou GST Technology Co., Ltd.) in August 2009 dedicated to technological transfer to and operational training for PBPs in other cities. The successful story of Hangzhou’s PBP has attracted numerous visitors across China and even from other countries. Even though we could not attribute the rapid spread of PBP across China to Hangzhou’s pioneering, late adopters really benefit by learning from Hangzhou’s PBP.

According to a recent survey, there has been more than 170 PBPs in China by the end of 2016. While both private and public sectors engage in PBP, it is public sectors which dominate the industry.Footnote 3 It is thus important to examine why urban authorities are interested in adopting PBPs. A recent quantitative analysis of Chinese prefecture-level cities reveals that the diffusion of PBP is primarily driven by internal demands and government propensities (e.g., tourist destination, public transit system, road area per capita, and administrative rank), as well as external stakeholders’ advocacy and pressures (e.g., inter-city learning and competition, bottom-up diffusion, and media framing) [21]. In this paper we want to examine the learning hub effect of Hangzhou’s practices in driving the diffusion of PBP among Chinese cities.

It is possible that local governments can learn from other Chinese cities (and even oversea cities) or adopt PBP without learning from anyone. Given the exceptional popularity of Hangzhou’s PBP, we expect these alternative explanations are plausible albeit negligible. Local governments usually explicitly or implicitly refer to their pioneering peers for institutional legitimacy and policy support [37], which also works as policy endorsement to avoid blame for accountability. Foreign practices are less relevant than domestic cases in helping Chinese cities learn about new policies, particularly when the latter outperform the former in appropriateness, compatibility, and other dimensions. Despite that our analysis is based on an ego-centric network instead of a dyadic or whole network (terms used in Social Network Analysis to mean different foci of analysis), we believe it is meaningful to focus on the most central hub in the network of knowledge dissemination and policy diffusion.

Methods

Data Sources

We use the mixed-method approach to test the above propositions, and both quantitative analyses and qualitative interpretations are included. We interviewed top managers and staff in charge of Hangzhou’s PBP, and collected archives on inter-city interactions. We gleaned data from industrial sources to describe the processes by which PBP has been diffused across Chinese cities. We also interview decision-makers about their rationales in learning from and adopting PBP.

The archival data on the timing and content of visits are from Hangzhou’s PBP operators, which are primarily government documents in forms of official letters and follow-up news reports. Local governments commonly contact their peers to request official visits, which details the purposes, delegates, timing, and logistics. Official trip and visit is among the so called “three public fees” (Sangong Jingfei), and the other two are official banquets and vehicles. We did content analysis to elicit these key variables from these documents, which are used in later analyses.

The timing of PBP adoption is documented in the bike sharing database maintained by the Institute for Transportation and Development Policy (ITDP)-China (http://www.publicbike.net/). We matched adoption variables with the records of official visits, which help us identify which cities have visited Hangzhou and/or adopted PBP. We gleaned data from other sources, particularly yearbooks compiled by National Statistics Bureau (NBS), to gauge city-level attributes that may affect the adoption of PBP.

Variables and Measurements

Cities in the same province (Zhejiang) or adjacent provinces (e.g., Jiangsu, Shanghai, Anhui, and Fujian) are more likely to visit Hangzhou than cities in other remote provinces. We use the distance from the province in which visiting cities are located as a proxy of geographic distance, which is operationalized by the crow-fly distance between Hangzhou and provincial capital cities.

In comparison with jurisdictions at county- (counties, cities, districts, and industrial zones) and township-level (towns, townships, and subdistricts), prefecture-level and provincial capital cities are more inclined to learn from Hangzhou, since they share more similarities in public transportation and other circumstances.

Delegations leaded by mayors, deputy mayors or equivalent city-level officials (e.g., municipal secretaries, congress chairpersons) are more likely than those joined by agency heads and junior public managers to put PBP on municipal policy agenda, which helps adopt PBP. In the same token, county-level leaders may help to adopt PBP more than delegations led by junior officials in counties.

The size of delegations also matters, which usually reflects the amount of government attention. Site visits by more delegates from multiple agencies are more likely to facilitate the adoption of PBP. To measure delegation size, we use the means (e.g., 40 for 30 ~ 50 delegates) or lower limits (e.g., 30 for over 30 delegates). The period of site visit is measured by the number of full days the delegates spent at Hangzhou’s PBP, and the frequency is gauged by the number of official visits.

Results

Inter-City Learning

We first describe the key attributes of cities visiting Hangzhou’s PBP, and compare them with other cities. As shown in Fig. 1, the annual number of site visitors to Hangzhou’s PBP has plateaued around 40. Although it was initiated in 2008, the first batch of delegates did not arrive until the end of 2009. To measure “virtual” visits, we use “Hangzhou” and “Public bicycle” as keywords to search news reports, and the past decade has also witnessed a steady growth of media attention to Hangzhou’s PBP, from 52 in 2008 to 233 in 2016.Footnote 4 Taking together, Hangzhou’s PBP is well received by other cities, which can explain the continuous learning of its practices by peers.

Fig. 1
figure 1

The number of new PPBs in China and site visits to Hangzhou (2008–2016). Data sources: http://www.publicbike.net; Hangzhou PBP Company. The 2015 data of site visits to Hangzhou’s PBP are by October 2015

Full size image

Where do visitors come from? We can find that cities which visited Hangzhou sprawl across China, and geographic distance seems do not matter. Delegations are from cities in all provinces but Sichuan, Guizhou, and Heilongjiang. Bicycles are not user-friendly in basins and mountains in inland Sichuan and Guizhou, as well as in snowy and bitter cold Heilongjiang in northeastern China. In line with our expectation (H1), we find that cities in adjacent provinces are more likely to visit Hangzhou’s PBP than remote provinces. The distance of a province from Hangzhou is significantly and negatively related to its number of visitors (r = −0.521, p < 0.01), suggesting cities far away from Hangzhou are less likely to visit its PBP than adjacent cities. Cities in Zhejiang (58),Footnote 5 Jiangsu (25), and Guangdong (24) have frequently visited Hangzhou’s PBP, followed by Shandong (19), Fujian (15) Beijing (17), Hubei (10), Hunan (10), Shanghai (7), Anhui (7), and Henan (7). In contrast, cities in provinces far away from Hangzhou have visited occasionally. For instance, only one city in each of Ningxia, Tibet, Chongqing, and Yunnan visited Hangzhou’s PBP.

Looking at the numbers of site visits, we generate more fine-tuned results. Hangzhou (39), Beijing (17), Shanghai (7), and Suzhou (7) are among the cities sending the most delegations. Among the 330-strong prefecture-level and above jurisdictions in China, 90 or about 26.63% have visited Hangzhou’ PBP. Over three quarters of cities have visited one (54.84%) or two times (20.43%), with a median of 1, a mean of 2.72, and a standard deviation of 4.51.

The delegations are from both domestic and overseas, and the latter includes visitors from the United Nations (2014), the Philippines (2014), France (2013), and Hong Kong (2012 and 2015). Both delegations at national and local levels have visited Hangzhou’s PBP, and Ministry of Housing and Urban-Rural Development (2011) and National Tourism Administration (2014) aimed to promote it on a nationwide scale after site visits. Not only prefecture-level and above cities visited Hangzhou’s PBP, but county-level entities also admired its model. Some delegations visited Hangzhou’s PBP for several times, usually first for site-seeing and later on for implementation after adoption. In terms of government level, 64.45% of site visits are organized by prefecture-level and above government agencies, while 19.53% are county-level and below government entities. Other site visits (16.02%) were attended by delegates from nonprofit and international organizations. The results partially support our expectation (H2) that cities similar to Hangzhou (e.g., at the same administrative level) are more likely to visit, learn from and adopt its PBP.

Over half of the delegations are government agencies (56.25%), followed by 29.69% of enterprises (either private or state-owned sectors, e.g., transit operators) and 14.06% of media, nonprofits, and international organizations. With regard to the ranks of the head of a delegation, 7.81 and 54.69% of site visits were leaded by municipal leaders and agency heads (or junior officials) respectively, while 37.50% were headed by others, e.g., managers of nonprofit or profit sectors. The number of delegates ranges from one to 56, with a median of 6, a mean of 8.19, and a standard deviation of 7.77. More than half of the visits were attended by no more than six delegates (55.28%), and the majority included less than 16 delegates (91.96%).

For the number of days of each site visit, the data show that 77.34% of delegations spent one day in Hangzhou, with a median of 1, a mean of 1.267 and a standard deviation of 0.876. While one day is obviously not enough for “learning” of such technologically complicated public service program, the professional and personal connections made during site visits allow continued learning. Visitors without direct connections with Hangzhou’s PBP rely on local agents to arrange site visits, which include municipal policy research office (sister agencies in other cities), municipal reception office (mayors, vice mayors, or municipal delegations of other cities), industrial associations (foreign visitors), Hong Kong and Macao affair office (oversea visitors). Visitors contact their local agents by hierarchical or business relations to approach Hangzhou’ PBP, which help bridge the two parties and deepen interactions.

Policy Diffusion

The PBPs have been adopted by entities in 102 prefecture-level and above cities, accounting for 30.18% of total population. Among the cities adopting PBP, all but eleven used single citywide operator instead of multiple and isolated systems in one city (e.g., Guangzhou and Shanghai). Among total 174 PBPs in China, we identify the adopting period of time of 159 programs. There were only three new PBPs in each of 2008 and 2009, and then around 20 new programs were added from 2010 to 2012 (see Fig. 1). The years 2013 and 2014 have witnessed a growth of new PBPs, followed by a sharp decrease in 2015 and 2016 (Fig. 2).

Fig. 2
figure 2

The locations of PPBs in China by 2016. Data sources: http://www.publicbike.net

Full size image

It is also interesting to split cities adopting PBP into two groups, cities visiting Hangzhou’s PBP and those did not. For the cities which have not yet adopted PBP by the time of our survey, we also split them into two groups by whether they have visited Hangzhou’s PBP. The 2 × 2 matrix generates four groups, which can be compared to identify their similarities and differences (see Table 1). Among the 338 prefecture-level and above cities by 2016, 204 or 60.35% have not yet visited Hangzhou’s PBP or adopted PBP. We find that 32 or 9.47% organized delegations to visit Hangzhou’s PBP, but they have not adopted PBP so far. Still 44 or 13.02% adopted PBP without an official site visit to Hangzhou’s PBP. Finally there are 58 or 17.16% have visited Hangzhou’s PBP and adopted PBP. A simple Pearson χ2 analysis reveals that site visit and the adoption of PBP are not independent (χ2 = 68.356, p < 0.01). In other words, cities visiting Hangzhou’s PBP are more likely to adopt PBP, and vice versa. Taking together, our H3 is supported.

Table 1 The crosstab of site visits and PBPs in Chinese cities
Full size table

We calculate the period of time between inter-city visit and the adoption of PBP to analyze the relationship temporally between visits and adoption, namely to see if local governments use the policy knowledge learned from Hangzhou’s PBP, or if cities adopt first and visit later. Given that different entities in one city may visit and adopt several times, we use the overlapping span as a proxy of time lag in policy learning and adoption. The results reveal that about 61.40% of cities first visited Hangzhou’s PBP and then adopted PBP within five years, and 38.60% reversed the policy learning-adoption sequence, either visiting after adoption (15.79%) or visiting and adopting during the same year (22.81%). Given technological sophistication and operational complexity of PBP, cities commonly consider its plausibility of adoption after site visits. Cities adopting PBP may refer to Hangzhou for solutions to address their implementation obstacles, which led to the formation of an incubation company exporting Hangzhou’s PBP operational systems and technologies to operators in other cities, and so far more than 190 PBP entities have purchased its products and services.

We find that delegations led by municipal leaders have not resulted in more adoption of PBP than those organized by agency heads or junior officials. Our H4 is not supported. Part of the reason may be that PBP is a specialized and professional program, and managers in charge of transit or pertinent agencies have discretion to make decisions of adoption. Delegations headed by municipal leaders usually have multiple duties to fulfill in official and meetings and site visits, which may dilute their attention to specific programs. Another possible reason is that deputy or symbolic municipal leaders (e.g., the people’s congress chairman) are not in the positions of decision-making, and their influences in decision-making may be even weaker than agency heads.

We do find that site visits attended by more delegates and lasting more days are significantly and positively related to the adoption of PBP (p < 0.05), which generates strong the evidence of learning in policy diffusion. It is also interesting that cities dispatching more site visits are also more likely to witness the adoption of PBP (p < 0.05). Our H5–H7 are supported. These results suggest that site visits participated by more delegates from multiple sectors usually help reach a consensus across agencies in adopting new policies and programs. Longer time of learning and more frequent visits also enable policymakers to better understand the program and to adopt it.

The Missing Links in Policy Learning and Diffusion

We find that cities which visited Hangzhou are more likely to adopt PBP; that cities interested in and inclined to adopt PBP visit and learn from prior practices. However, not all cities visiting Hangzhou adopted PBP later on, which suggest that learning does not guarantee adoption. Learning facilitates adoption, and cities visiting Hangzhou adopted PBP earlier than cities without onsite learning. A recent study challenges that policy transfer is by and large policy tourism, since many travels and visits do not transform into concrete practices [15]. Our findings suggest that purposeful visits really help policy diffusion through policy learning, although their effects depend on the size, duration, and frequency of site visits.

Why did some visitors not adopt PBP? Our interviews with employees in charge of and delegates to Hangzhou’s PBP reveal three plausible reasons. First, some cities cannot implement PBP after site visits due to top leadership turnover. Given their short length of tenure (commonly less than three years) and strong impulse to quickly boost economic growth for promotion [18], local cadres are inclined to adopt new programs radically different from their predecessors [47]. The adoption of PBP usually takes one to two years, and leadership succession during this period may discontinue its initiative. As one interviewee notes, pragmatic officials would not follow in their predecessors’ footsteps, even though it makes sense for local development. Because no matter how well they perform these tasks, the credit will go to their predecessors. Furthermore, without strong support from top managers, PBP cannot be implemented with necessary human, financial, and policy resources.

Second, a few cities interested in PBP are drawn to visit Hangzhou, but later on they find it is incompatible with local conditions. While PBP sounds promising to solve urban transport problems, its successful implementation requires several preconditions. For example, local governments have to support and subsidize the program, which may not be as workable as it first seems. As a public service, PBP is often free the first hour, and the charge afterwards is also cheap, e.g., one to two RMB Yuan per hour. The operators cannot profit a lot from PBPs, and the lion’s share of revenue comes from local governments’ subsidies and other sources (e.g., advertisements). Taking together the substantial initial capital cost (e.g., Hangzhou’s 450 million Yuan, Guangzhou’ 28 million Yuan) and expensive land resources devoted to stations, less wealthy local governments have to think twice before adopting PBP. Such decisions not to adopt PBP, however, reflect the role of rational learning in policy diffusion. Although we encourage the diffusion of innovative programs, it is the adoption of compatible policies that matter in improving public services. Rapid and wide diffusion of bad and unsuitable policies work against the interests of policymakers [7].

Third, PBP and other innovations often become policy “fads” and decision-makers may be soon interested in alternative programs such as bus rapid transit (BRT) and metro. In other words, PBP can be crowded out by competing programs. Given the priorities of alternative programs, PBP fades from managerial attention. As one interviewee acknowledged, despite PBP being superior in cost, benefit, and sustainability compared with BRT or metro, local officials still admire the latter options due to financial and political considerations. The enormous fixed-asset investment, renewed urban infrastructure, and prestigious landmarks generated from BRT or metro can boost the growth of local economy and employment, which helps local officials to seek rent and pursue career advancement. Even Hangzhou is expanding its metro system and renewing its road networks and BRT, and PBP has not successfully alleviated traffic congestion. Policy learning helps to draw decision-makers’ attention to PBP, but it also might facilitate adopting alternative programs.

Theoretical and Policy Implications

In this paper we use the case of Hangzhou’s PBP to empirically examine the role of site visits in policy learning and diffusion by examining which cities are more inclined to learn from and adopt its policies. The findings support our hypotheses and are largely consistent with prior studies. We find that cities learn from innovation stars through site visits, which facilitate policy transfer and diffusion. Inter-city learning is not limited to adjacent regions, but can be across the country. However, the results show that cities are more likely to learn from their adjacent peers, suggesting geographic distance still plays an indispensable role in shaping the landscape of policy learning. While homophily prompts mutual learning, cities do not only learn from who are similar with them. Given the ubiquitous use of PBP, cities also effectively learn from heterogeneous counterparts. We find that policy learning does trigger policy transfer and diffusion, but many other factors intervene, particularly the size, duration, and frequency of site visits. We also find that leadership turnover, prudent decision-making (e.g., incompatibility of PBP), and alternative solutions (e.g., metro) are the missing links in transforming policy learning into policy adoption.

Our findings contribute to the literature on policy diffusion and the studies of government innovation in China. Our findings join the recent research of policy learning at local level [4, 11, 17], suggesting that learning matters in policy diffusion and it is primarily driven by proximity and homophily. Additionally, the learning context matters a great deal in that larger delegations visit for longer or multiple times are more likely to adopt PBP. To the best of our knowledge, this is one of the first empirical studies examining policy learning and diffusion in the field of PBP in China. By elaborating on the rationale and policy consequences of site visits in the diffusion of PBP, we contribute new evidence to the literature on policy learning and diffusion in China, identifying the mechanisms through which policies transfer and diffuse across localities. We supplement existing quantitative studies on policy diffusion by analyzing a specific form of policy learning and its role in connecting early and late adopters.

We find that policy learning is an indispensable component in policy diffusion, which helps to reframe and deepen our understanding of local government innovation in China. Several case studies in the existing literature highlight the crucial role of policy learning and institutional isomorphism in local government innovations [6, 33, 40], suggesting they are crucial in policy diffusion. Policy learning is underestimated in prior studies [35, 47], and our results help to highlight its prominent role in policy diffusion. Policy experiment and piloting is more prominent in the early stage, while policy learning and scaling up is more likely when policies develop and mature. Originality and creativeness are welcome, but what is more important is effectiveness and success, which are particularly evaluated in the cadre evaluation system in China. Instead of creating these policies, rational local governments are more inclined to revise and adapt practices of other jurisdictions to fit local conditions. To receive credit in the cadre evaluation criteria, innovations are usually reimagined through new titles in this way. Championship is more rhetoric or symbolic, while the substantive role of learning from others is institutionally embedded in the process of local government innovation.

The findings of this study also generate policy implications for practitioners interested in policy transfer and diffusion. Despite the fact that decision-makers can learn through other channels, our results suggest that site visits help to facilitate the adoption and implementation of policy innovations. Business travels and site visits play a pivotal role in bridging local governments’ policy networks, which are often used to facilitate policy transfer and diffusion. Local governments who want to export their innovative solutions to peers should consider site visit as an indispensable approach in disseminating knowledge and amplifying impacts. In comparison with official statistics and news reports, vivid experiences through site visits play a more persuasive role in facilitating policy adoption.

Limitations and Future Research Directions

Despite our findings, this study is limited in the following aspects, which suggest avenues for future research. First, the impacts of policy learning on policy diffusion could be different in other sectors and regions, and we encourage future studies to replicate and extend our findings in other contexts. Although we use the case of PBP in the context of transport policy at local level, but the implications of our results can be transferred to other policy areas [22], PBP is different from other policy innovations in many aspects, such as technological sophistication, visibility, compatibility, complexity, which may moderate the effects of policy learning on policy diffusion. Given the dearth of studies in this field, we hope future research can explore other dimensions of policy learning.

Second, given the focus and budget of this study, we cannot cover all kinds and forms of intercity learning in PBP. We use an ego-centric approach to test our hypotheses, which is different from dyadic and whole network approaches used in prior studies. While innovation stars like Hangzhou are followed by cohorts, local governments simultaneously learn from other peers. Particularly rising stars such as Wuhan, Guangzhou and Taiyuan are also frequently referred to by latter adopters. We hope future studies can examine intercity learning among other cities, which help deepen our understanding of complicated policy learning processes. For instance, how do policymakers compare and learn from multiple practices either successful, mediocre, or failed? Policy learning facilitates knowledge management in many forms apart from the site visit, including the site meeting, case study, exposition and fair, and so forth. It is meaningful to incorporate these forms of policy learning in future research.

Lastly, the mechanisms and processes of policy learning are still a black box worthy more in-depth investigation. The value of learning in policy transfer and diffusion has been recognized, but the existing evidence is primarily of anecdotes and case studies. In mainstream quantitative analyses of policy diffusion, policy learning is usually simplified by proxies such as membership of professional associations, inter-jurisdictional contracts, or symbolic ties (e.g., sister cities). We encourage future research to conceptualize new mechanisms and processes of policy learning, which help revisit its role in policy transfer and diffusion.