China visits: a dataset of Chinese leaders’ foreign visits

Leader visits constitute an important signal in international relations. While studies of U.S. diplomacy can all use the same dataset from the Office of the Historian, IR scholars on China must make do with ad hoc datasets and often need to build their own from scratch. We contribute a novel dataset, ChinaVisits, to fill this glaring gap. Our dataset has three major advantages: (1) it covers the period from 1998 onwards so that it is widely applicable to different research agendas; (2) each recorded visit has rich auxiliary information, including its date and duration, and is accompanied by a document from official websites for verification, and the dataset in its entirety is evaluated against existing datasets; (3) it is publicly available and indexed annually with country codes and country names. To facilitate its use, we provide a detailed analysis of the patterns in leader visits.

McManus & Yarhi-Milo, 2017; Malis & Smith, 2020). 1 Like U.S. leader visits, Chinese leader visits are also often used as a proxy for China's bilateral relations (Kastner & Saunders, 2012;Bader, 2015;Lin et al., 2017;Eichenauer et al., 2021;Stone et al., 2022). 2 Unlike U.S. leader visits, however, researchers do not have a complete record of Chinese leader visits as provided by the Office of the Historian. Consequently, studies that use Chinese leader visits are often ad hoc, restricted to a relatively short time period (e.g. 1998to 2008in Kastner and Saunders (2012), 1994to 2008in Bader (2015, and 2000 to 2013 in Stone et al. 2022) and use quite diverse sources: Kastner and Saunders (2012) build their dataset on China's Foreign Affairs (Beijing: World Affairs Press); Bader (2015) uses Journal of Current Chinese Affairs; Li (2015) uses Xinhua Monthly Report; McManus (2018) partly relies on FBIS, Lexis-Nexis, and ProQuest Historical Newspapers databases; Eichenauer et al. (2021) use a combination of several publications including (Bárcena & Rosales, 2010) and (Zhu, 2010). Of these, only Kastner and Saunders (2012), Bader (2015), Li (2015), McManus (2018), and Stone et al. (2022) are publicly available. These data sets have other limitations. Unfortunately, Kastner and Saunders (2012) aggregate all visits by country, losing over-time variation. Bader (2015) aggregates Chinese leader visits with foreign leader visits to China and, because it is based exclusively on signed Chinese agreements, misses important visits including the Chinese premier's visits to the U. S. in 1999S. in , 2003S. in , to Belgium in 2000S. in , 2001S. in , 2004 and the Chinese president's visit to the U.S. in 2002. Li (2015) excludes visits to OECD countries. McManus (2018) excludes visits by the premier as well as visits to the U.S. and Russia. Stone et al. (2022) focus mostly on the first decade of the century. This lack of an established dataset on Chinese leader visits hinders research on Chinese foreign policy as well as impedes reproducibility on important IR questions.
In this paper, we introduce such a dataset on Chinese leader visits: China Visits. Our dataset has three strengths: (1) it covers the period between 1998 and 2020 so that it is widely applicable to different research agendas, and covers visits by both the president and premier; 3 (2) each recorded visit is verifiable and contains rich auxiliary information, and the dataset in its entirety is evaluated against existing datasets; (3) our dataset is publicly available and indexed annually with country codes from Correlates of War (Correlates of War Project, 2017) and from the World Bank so that researchers can easily plug the visits variable into existing datasets. Our paper is structured as follows. In Section 2, we review the literature on leader visits. In Section 3, we introduce the dataset and the data collection process, compare the dataset with existing datasets, and describe a few prominent patterns in the visits. In 1 The dataset from Office of the Historian is not without its own flaws. As pointed out by Lebovic and Saunders (2016) and Malis and Smith (2020), the dataset includes visits that are not always bilateral. For example, President Bush's visit in 2001 to Quebec, Canada to attend the third Summit of the Americas is counted as a visit to Canada (Malis & Smith, 2020). 2 Alternative and weaker proxies of bilateral ties include the maintenance of diplomatic relations (Strüver, 2016) and inter-party visit programs (Hackenesch & Bader, 2020). 3 We chose 1998 as the beginning year because this is the year Zhu Rongji succeeded Li Peng as Premier and because data before 1998 is scarce. Our data currently end in 2020, and we plan to update the dataset on an annual basis. Section 4, we analyze the determinants of leader visits using regressions. Section 5 concludes.

Literature review
Our paper builds on the burgeoning literature of leader visits in international relations. First, leader visits are frequently used as an independent variable in the study of economic diplomacy (Moons & van Bergeijk, 2017). 4 Nitsch (2007) is arguably the first major study on the impact of leader visits on exports, where the author finds that a leader visit is associated with about an 8 to 10 percent increase in bilateral exports. Lin et al. (2017) find that state visits to China by African leaders increase China's exports to Africa in capital-intensive manufacturing goods. Beaulieu et al. (2020) find that state visits by Chinese leaders have a positive effect on Chinese bilateral trade and that the effect is stronger in industries that are under stronger control by the Chinese government. Stone et al. (2022) analyze the effects of leader visits and observe that leader visits are negatively correlated with Chinese outward foreign direct investment overall and positively correlated with investments by centrally controlled firms in particular.
Second, scholars have recently been using leader visits to study public diplomacy. At the general public level, Goldsmith et al. (2021) combine high-level leader visits on the one hand and individual-level opinion polls from Gallup World Poll on the other. They observe a substantial increase in the approval level of the visiting leader. At the elite level, Custer et al. (2018) find that more visits between China and countries in the East Asia and Pacific region are correlated with higher levels of voting alignment at the United National General Assembly. Furthermore, Wang (2021) reports results showing that Chinese leaders are more likely to visit an African country when that country serves on the United Nations Security Council.
Third, leader visits have been used to study leader survival and interstate conflicts in the recipient countries. Bader (2015) analyzes the effects of Chinese leader visits and concludes that leader visits from 1993 to 2008 have no impact on the likelihood of regime survival for autocratic countries. Malis and Smith (2020), by contrast, find that U.S. leader visits between 1960 and 2013 do help foreign leaders stay in office. McManus (2018) studies leader visits from the U.S., Russia, China, Britain, and France to smaller client states and finds that visits by major powers have a significant negative effect on the probability of the recipient country being targeted in a militarized interstate dispute.
Fourth, the distribution of leader visits is often of scholarly interest. Most notably, Lebovic and Saunders (2016) and McManus and Yarhi-Milo (2017) study the empirical pattern of U.S. foreign visits and conclude that countries of high strategic interest (great powers, countries high in military spending, bilateral trade with the U.S. as well as those more politically aligned with the U.S.) are more likely to be visited by U.S. leaders. Kastner and Saunders (2012) use the pattern of Chinese leader visits to understand whether China is a status quo state or a revisionist state. The authors reason that leader visits to rising powers or rogue nations on the fringes of the existing world order would indicate that China is a revisionist state, while visits to U.S. allies would suggest that China is a status quo state. Similarly, Li (2015) uses visits to developing countries by the Chinese leaders as a proxy for strategic competition with the United States and studies whether U.S. engagement can help moderate China's competitive propensity.

Data
Our paper centers around the dataset of Chinese leader visits, China V isits. The visits consist of goodwill visits, working visits, official visits and state visits by the Chinese president, and working visits and official visits by the Chinese premier, which we collectively refer to as 'formal visits.' Similar to Kastner and Saunders (2012) and Li (2015), we do not include visits whose sole purpose is to attend a multilateral meeting, such as President Xi's visit to Japan in 2019 for 14th G20 summit. 5 We do, however, include visits whose purpose is to attend a multilateral meeting as well as pay a formal visit to the host country. 6 The dataset covers the years between 1998 and 2020. It includes visits by President Jiang and Premier Zhu made between 1998and 2002, visits by President Hu and Premier Wen between 2003and 2012, and visits made by President Xi and Premier Li from 2013 to 2020. The dataset includes all United Nations member countries, except China itself, as potential recipient countries. 7

Data collection
We collect the data set from five sources that complement and corroborate each other: (1) leader-based summaries of foreign visits, (2) year-based leaders' foreign visits, (3) summaries of China's bilateral relations, (4) keyword-based Google search, and (5) Wikipedia pages on leaders' foreign visits. Interested readers can refer to the appendix for our detailed sources and the keywords that we use. We validate each leader visit with a supporting document. Note that a few visits can refer to the same supporting document since Chinese leaders sometimes visit a group of countries in one trip (Fig. 1) Once we have collected all the visits, it is straightforward to create the binary variable P residential V isit c,t , which is 1 if country c receives a presidential visit in year t and the binary variable P remier V isit c,t , which is 1 if country c receives a visit from the Chinese premier in year t. We note that no country has received a Chinese presidential visit or premier visit more than once a year. For this reason, we code both P residential V isit c,t and P remier V isit c,t as binary variables. In addition to the fact that a country is visited, we also include information about the date of the visit, its duration, and whether a multilateral meeting is involved. In cases where the visit is part of a larger group visit, e.g., President Hu's visit to Germany in 2005, we also record the ordering of the visit. For example, the U.K. is the first host country, Germany second, Spain third, and South Korea fourth ( Fig. 1).

Dataset comparisons
Before diving deep into the dataset, we first compare it to existing datasets regarding granularity, coverage, auxiliary information, and completeness. In Table 1 we list the papers that have used leader visits as a variable, data availability, time and country coverage and data format. Six of these papers, including one dataset paper from Aid-Data, have published their data sets. Of these six, Kastner and Saunders (2012) count the number of visits by the Chinese president and the premier per country, Bader (2015) includes Chinese leaders' visits to other countries as well as foreign visits to China, Li (2015) focuses on visits to non-OECD countries, McManus (2018) focuses on presidential visits, AidData (2019) includes high-level and provincial-level bilateral visits, and Stone et al. (2022) are a subset of our dataset with coverage through 2012. Lin et al. (2017) report the cumulative number of visits to Africa by Chinese leaders.  (2015) cover both the president and the premier.

Coverage
We next compare the geographical and temporal coverage of the datasets (Table 1). AidData (2019) focuses specifically on East Asia and the Pacific and covers 25 countries. Li (2015) focuses on countries that are not in the OECD. Other datasets, including ours, largely cover all UN member countries. Some notable peculiarities include that McManus (2018) excludes leader visits to the United States and Russia and that Li (2015) and Kastner and Saunders (2012) explicitly drop countries that recognize Taiwan rather than, for example, coding them as zero. Temporally, McManus (2018) covers the longest period (58 years), and China Visits covers the second longest period (23 years). Our dataset covers the most recent years. In this regard, scholars studying recent developments, such as the Asian Infrastructure Investment Bank (Kaya & Woo, 2021;Wang, 2018), the Belt and Road Initiative (Broz et al., 2020), and the expansion of the Shanghai Cooperation Organization, will find our dataset particularly useful.

Auxiliary information
Existing studies mainly focus on whether a visit takes place, and consequently, data on Chinese leader visits are predominantly binary: one if Chinese leaders visit a country and zero otherwise. By contrast, our dataset incorporates much richer auxiliary information: (1) the date of the visit, (2) its duration, and (3) whether a multilateral meeting is involved. When a leader visits more than one country during a tour, we further record the ordering of the visits. Interested scholars may examine the significance of such orderings, as being selected as the first destination may confer honor on the destination country. For example, Prime Minister Desalegn of Ethiopia felt honored that Premier Li Keqiang visited his country first during a fourcountry tour of Africa in 2014. 8 Chinese media highlighted the significance of the order of visits when President Xi chose the United Arab Emirates as his first stop in 2018. 9 Many of our original documents include information about which foreign officials went to the airport to receive and see the Chinese leaders off. For example, it was high-ranking government officials who saw President Xi off in the airport during President Xi's visits to Argentina and Netherlands in 2014. By contrast, it was the presidents of Sri Lanka, Maldives and Tajikistan who went to the airport to see President Xi off during his respective visits in 2014. Future research could study whether such carefully choreographed diplomatic details affect Chinese public approval of the government's performance or attitude towards the host country.

Data completeness
We compare our dataset with Kastner and Saunders (2012), Li (2015), Lin et al. (2017) and McManus (2018). First, in Table 4 in the appendix, we provide a countryby-country comparison of our dataset with that of Kastner and Saunders (2012), where we aggregate our data to be consistent with the baseline and report the rows where our records differ. We find that on average, our results are the same for 184 of 192 UN member countries (96%). For the cases where our numbers do differ, we have double-checked our sources to ensure our coding is correct per definition. Second, we perform a visit-by-visit comparison with Li (2015). For a fair comparison, we exclude visits to OECD countries as is done in Li (2015). Our results are the same for 169 of 171 visits (99%). Between 1998 and 2012, both Li (2015) and we have recorded the same 169 visits. In addition, Li (2015) further includes two visits that we have not recorded: one visit to Ivory Coast and another visit to Ghana in 1999. 10 Third, in Table 5 in the appendix, we provide a leader-by-leader comparison of our dataset with that of Lin et al. (2017). Note that while Lin et al. (2017) do not have a publicly available dataset, they do report the aggregated number of visits in their paper. Therefore, we use the aggregated numbers as the basis of our comparison. For the 2001-2005 period, we have recorded two more visits by the Chinese premier than Lin et al. (2017). For the 2006-2012 period, we have recorded one fewer visit by the Chinese president and one fewer visit by the Chinese premier. For these two differences, we have provided a corroborating document to support our result. Lastly, we compare with McManus (2018) for the period of 1998 and 2006. For a fair comparison, we exclude the U.S. and Russia and focus on presidential visits only as is done in McManus (2018). Of the 84 presidential visits that McManus (2018) records over this period, we are able to match 78 of them (93%). The 6 visits missing from China Visits are invariably multilateral meetings, for example, Asia-Pacific Economic Cooperation (APEC) Summit and Conference on Interaction and Confidence-Building Measures in Asia (CICA), which do not involve a state visit according to our definition. In addition, our dataset includes 5 state visits that are missing from McManus (2018). We report the exact differences in Table 6 in the appendix. Overall, this high level of consistency with existing published datasets validates our data collection method and data quality.
In the following subsections, we analyze the leaders' visits and highlight prominent patterns that we believe may be of broad scholarly interest. These are by no means exhaustive.

Geographic distribution
In Fig. 2, we show the geographical distribution of presidential visits from 1998 to 2020. We observe that Russia and Kazakhstan are by far the most frequently visited countries by the Chinese president with 11 visits and 10 visits, respectively, in 23 years. France, South Africa, and Vietnam are a distant third with 5 visits each. The fact that Russia and Kazakhstan are visited with such a high frequency leads us to examine the role of the Shanghai Cooperation Organization in shaping presidential visits (see Section 3.6). We also note that both President Hu and President Xi chose Russia as their first stop after taking office as president. Potter (2013) refers to this type of visit as an obligatory visit; for another example, the U.S. president invariably visits the U.K. within the first few months in office.
In Fig. 3, we show the geographical distribution of premier visits. Russia is the most visited country by the Chinese premier with 11 visits, followed by Germany (9), Belgium (7), and the U.K. (6). Compared with the president, the premier is less likely to visit Africa, South America, and North America, and is more likely to visit South Asia instead.
Premiers and presidents frequently visit the same countries in different years. As we report in Table 2, 74 countries have received both presidential visits and premier visits; 81 countries, which are mainly in Africa and Western Asia, have never received a visit from either the Chinese president or the premier; 9 countries have received visits only from the premier; and 28 countries have received visits only from the president. 11 The president has visited 23% more countries than the premier. 11 Note that China holds the principle of diplomacy that all countries, big or small, are equal. While there are many countries that they have not visited in person, the Chinese leaders often meet with foreign leaders in multilateral settings. For example, President Xi held meetings with heads of state from Togo, Guinea, and Djibouti respectively during the Johannesburg Summit of the Forum on China-Africa Cooperation in 2015. None of these countries has received a visit from the Chinese president or the premier. As another example, during President Xi's state visit to Fiji in 2014, he held meetings with leaders from other Pacific Island countries, including Federated States of Micronesia, Samoa, Vanuatu, and Tonga, none of which has received a Chinese leader visit.

Temporal Distribution
Figure 4 displays the monthly distribution of leader visits from 1998 to 2020. We group them by office, president or premier, as well as by administration, Jiang, Hu, or Xi. We observe that the president and the premier exhibit similar travel patterns. Both President Jiang and Premier Zhu often visit abroad in April. They travel substantially less between December and March. Both President Hu and Premier Wen like to visit other countries in June and November. They rarely visit other countries in March, July, or August. Both President Xi and Premier Li like to travel in November. They seldom visit abroad between December and February. We believe that some of the travel patterns result from regular multilateral meetings. The Shanghai Cooperation Organization, for example, is partly responsible for 7 presidential visits in June. APEC Economic Leaders' Meeting is responsible for 6 presidential visits in December. ASEAN summits are responsible for as many as 10 premier visits in November.
In Fig. 5, we report the duration of leader visits by office and by administration. On the left, we show that the average duration of presidential visits has been steadily going down from President Jiang (3.7 days on average) to President Hu (3.2 days) and  then to President Xi (2.9 days). On the right, we observe that the duration of premier visits first decreased from Premier Zhu (4.3 days on average) to Premier Wen (2.6 days) and recently has seen some increase under Premier Li (3.3 days). In Figure 1 in the appendix, we further report the total number of days spent visiting per month by the leaders.

Shanghai Cooperation Organization
Meetings of the Shanghai Cooperation Organization (SCO), an international organization co-founded by China, Kazakhstan, Kyrgyzstan, Russia, Tajikistan, and Uzbekistan to promote economic, military and domestic security collaboration, contribute to the frequent visits to Russia and Kazakhstan as well as June being one of the most likely months for visits abroad. If the annual SCO chair country always receives a visit from the Chinese president, given that there are only 6 member countries before 2017 (8 afterwards), this would translate into a 1/6 probability of being visited for SCO membership. In Table 3 (left), we list the chair country for each year between 2003 and 2019 and whether the host country received a presidential visit. 16 We find that the Chinese president invariably pays a state visit to the host country. This suggests that SCO membership plays a role in shaping the distribution of Chinese presidential visits.
In addition to the Heads of State Council, the Heads of Government Council (HGC) also meets annually "to discuss the organisation's multilateral cooperation strategy and priority areas, to resolve current important economic and other cooperation issues, and also to approve the organisation's annual budget." 17 This lays the ground for the Chinese premier to visit the chair country. In Table 3 (right), we list the corresponding SCO chair countries between 2003 and 2019 and whether the chair country received an official visit from the Chinese premier. We observe that out of 12 http://eng.sectsco.org/about sco/. 13 For the report on Iran becoming a full member, see http://www.news.cn/english/2021-09/18/c 131019 6298.htm. For the report on Saudi Arabia, Egypt, and Qatar becoming dialogue partners, see https://www. fmprc.gov.cn/web/zyxw/t1907864.shtml. 14 For an official announcement of the state visit to Russia in 2009, please see http://www.china-un.ch/ eng/xwdt/t566908.htm. President Hu went on to visit Slovakia and Croatia after his Russian visit. 15 For an official report on this visit, please see http://www.xinhuanet.com/english/2017-06/07/c 136347953.htm. During President Xi's visit to Kazakhstan, he also attended the opening ceremony of the Expo 2017. 16 Before SCO was founded, the group was known as the Shanghai Five: China, Kazakhstan, Kyrgyzstan, Russia and Tajikistan. Even at that time, attending a leaders' meeting helped produce presidential visits, e.g., President Jiang's visit to Tajikistan in 2000. 17 http://eng.sectsco.org/about sco/.

Intervals between visits
Russia stands out for another reason: the Chinese president has visited Russia in 2001,2003,2005,2007,2009,2011,2013,2015,2017 and 2019. These visits have been occurring every two years and they all fall on odd years. The Chinese premier has visited Russia in 1999, 2001, 2004, 2007, 2008, 2010, 2012, 2014, 2016, 2019. This recurrent pattern is similar to previous observations made by Lebovic and Saunders (2016) on U.S. leaders' foreign visits in the sense that leaders tend not to visit the same country in consecutive years.
18 Iran received two presidential visits and no premier visits in the past 20 years. Once it becomes a full SCO member, leader visits likely to become more frequent. Given the fact that Russia is the most frequently visited country by the Chinese leaders and that even Russia tends not to receive visits in consecutive years, this reflects a diplomatic habit and suggests that a leader visit in one year strongly reduces the probability of a leader visit in the following year for an average country. 19 In Table 4, we show that the probability of receiving a presidential visit conditional on having a visit in the preceding year is 61% lower than the average probability and that the probability of a premier visit drops by 14%.

Association of southeast asian nations
Besides the SCO, another organization that may serve as a drawing force for premier visits is the Association of Southeast Asian Nations (ASEAN). ASEAN was established on 8 August 1967 in Bangkok, Thailand by Indonesia, Malaysia, Philippines, Singapore and Thailand. ASEAN has since expanded to include Brunei (1984), Vietnam (1995), Laos (1997), Myanmar (1997), and Cambodia (1999, making the number of members ten. ASEAN's economy collectively (2.97 trillion dollars in 2018) is larger than that of India, and has been growing at a high rate (average 5.08% among members in 2018).
China has developed multiple channels to engage with ASEAN, including ASEAN plus China (10+1) and ASEAN plus China, Japan, South Korea (10+3) and the East Asia Summit. Both ASEAN plus China and ASEAN plus China, Japan, South Korea were established in the late 1990s. 20 The East Asia Summit was established in 2005 (Emmers et al., 2010). In addition to ASEAN members, it includes Australia, China, India, Japan, New Zealand, South Korea, the United States and Russia. In spite of their varied membership, all three meetings are held annually in the ASEAN chair country at around the same time, suggesting that ASEAN plays a leading role therein.
With China being a member of all these meetings, this creates an opportunity for the Chinese premier to come to the ASEAN chair country in person. While in the late  (Fig. 6).

Li Keqiang to Attend Leaders' Meetings on East Asia Cooperation and Pay Official Visit to Laos
In Table 5, we summarize whether or not the rotating ASEAN chair has received a visit from the Chinese premier from 1998 to 2019. We observe that in the 2000s, it was still not likely that the premier would pay an official visit to the ASEAN chair. The norm of the Chinese premier visiting the host country only became entrenched in the 2010s. As a matter of fact, since 2011, all the ASEAN member countries have received an official visit from the premier when serving as the ASEAN chair. Timor Leste is the only Southeast Asian country that is not a member of ASEAN (Bertrand, 2013). The fact that it has never received a visit from the Chinese premier, while all other Southeast Asian countries have, supports the argument that ASEAN membership increases the probability of visits by the Chinese premier.
Hitherto, we have covered the SCO and ASEAN. Other prominent multilateral meetings in China Visits include APEC Economic Leaders' Meetings, the Forum on China-Africa Cooperation (FOCAC) and Cooperation between China and Central and Eastern European Countries. 22 For example, the ministerial conference of FOCAC is held every three years alternately in China and an African country. 23 The 21 See https://www.fmprc.gov.cn/mfa eng/topics 665678/wjbfw 665822/t226137.shtml for an official report on the visit. 22 Please note that strategic calculations vis-à-vis neighbors could be a factor when countries consider joining these initiatives (Wang, 2018). 23 http://www.focac.org/eng/ltjj 3/ltjz/.

Argument
Having described the distribution patterns of leader visits, in this section we test the importance of the underlying factors that shape these patterns. Broadly speaking, we argue that Chinese leader visits aim to (a) promote China's economic interests, (b) strengthen China's bilateral relations, and (c) accompany multilateral meetings. Our argument has its roots in the existing literature on U.S. leader visits (Lebovic & Saunders, 2016) and on Chinese leader visits (Kastner & Saunders, 2012;Li, 2015). Equally important, our argument also stems from the detailed documentation in China Visits. As an example of promoting economic interests, we note that during his visit to the U.K. in 2015, President Xi "presided over the signing of a raft of business deals totalling around 40 billion pounds." 24 As an example of strengthening bilateral relations, we note that during Premier Li's 4-day official visit to New Zealand in 2019, New Zealand "became the first Western developed country to sign a cooperation agreement with China on the Belt and Road Initiative." 25 For examples where official visits are bundled with IO-related multilateral meetings, we refer our readers to Section 3.

Economic incentives
Official visits abroad represent a good opportunity for Chinese leaders to promote China's economic interests in bilateral trade and investment. To estimate the strength of the economic incentives, we include a set of economic variables that characterize the recipient countries as explanatory variables. These are Log(GDP per capita), Log(Population), GDP Growth, Log(Resources) from the World Bank. We hypothesize that from an economic point of view, Chinese leaders are more likely to visit large, rich and fast-growing economies. Meanwhile, with Log(Resources), we are able to test whether these visits are resource-driven. To account for leaders' flying time, we control for geographical distance between Beijing and recipient capitals using distance data from Le Centre d'études prospectives et d'informations internationales (Mayer & Zignago, 2011).

Political incentives
Official visits abroad also represent a good opportunity for the Chinese leaders to strengthen China's bilateral political ties. We include a set of political variables widely used in the literature: Polity, which measures the level of democracy (Vreeland, 2008;Dreher et al., 2018;Kerner & Lawrence, 2014) and UN General Assembly (UNGA) voting (Bailey et al., 2015). 26 The UNGA data have been widely used as a measure of political alignment (Davis et al., 2017;Dreher et al., 2018;Clark & Dolan, 2021;Kaya & Woo, 2021). Dreher et al. (2018) use UNGA voting similarity as an explanatory variable for Chinese state financing. Kaya and Woo (2021) use UNGA voting similarity as an explanatory variable for membership shares in the Asian Infrastructure Investment Bank. In this paper, we use the percentage of voting agreement with China as a proxy for geopolitical proximity. Similar to Lebovic and Saunders (2016) or is Japan or Germany and is 0 otherwise. Given the history of Sino-Japan relationship (Davis & Meunier, 2011;Quek & Johnston, 2018), we also add a dummy variable for Japan to distinguish it from other major powers. Meeting with Dalai i,t is a dummy variable on whether country i hosts a meeting with the Dalai Lama. Existing research suggests that visits by the Dalai Lama have a detrimental effect on trade with China, particularly on the export of machinery and transport equipment to China and that such effect operates at the extensive margin and affects state-owned enterprises more than others (Fuchs & Klann, 2013;Lin et al., 2019). As leader visits represent yet another channel to voice opposition to meetings between other countries and the Dalai Lama, we expect the coefficient on Meeting with Dalai i,t to be negative. Several notable studies (Cheung et al., 2014;Dreher et al., 2018;Stone et al., 2022) have analyzed the effects of recognizing Taiwan on Chinese aid and FDI. With regard to leader visits, we observe that neither the president nor the premier has ever visited a country that recognizes Taiwan (Kastner & Saunders, 2012;Li, 2015). For this reason, we drop samples where the country recognizes Taiwan. The status of Taiwan is not a constant for all countries, as quite a few countries have broken ties with Taiwan in the past two decades ( Table 7 in the appendix).

IO Incentives
Multilateral meetings do not directly promote official visits by Chinese leaders. Instead, as they can be bundled together with official visits, they help lower the costs of these visits and thus make official visits more likely to take place. Since it is international organizations (IOs) that are responsible for hosting multilateral meetings, we test how much IO membership can contribute to leader visits. Specifically, we include two international organization variables: SCO and ASEAN. SCO is coded 1 for all SCO members after 2003 inclusive. ASEAN is coded 1 for all ASEAN members after 2011 inclusive. The years are so selected as to reflect the time when leader visits became obligatory (Potter, 2013). Given that IOs frequently yield leader visits, we hypothesize that member countries of these IOs are more likely to receive visits. In particular, SCO members should see more visits from both the Chinese president and the Chinese premier. For ASEAN, as it is mostly the premier who attends ASEAN-related summits, we hypothesize that the positive effect of ASEAN membership should lie mostly with premier visits.

Econometric modeling
We formulate our main regression equation as follows: where f is the link function and Leader Visit i,t is a binary variable and represents a leader's visit (president or premier)  We use them to control for the fact that leaders tend not to visit the same countries in consecutive years. 27 τ t stands for year fixed effects and i,t for the error term. We provide the descriptive statistics of these variables and their correlations in Table 8 and Table 9 in the appendix. We run the regressions for the president and for the premier separately so that we are able to identify not only the shared patterns but distinctions between the two as well. All standard errors are clustered by country.
In addition to probit regression, we also run OLS and bivariate probit as robustness checks (Table 10 in the appendix). Table 6 reports our main results, which are largely consistent with our initial expectations. 28 Among the economic variables, we find that countries with larger populations and higher GDP per capita are more likely to receive both presidential visits and premier visits. Specifically, when we hold all other variables at their means, the probability of the Chinese president visiting a country with a Log(Population) that is one standard deviation above its mean is 7.4%, compared with 3.9% for a country with a Log(Population) at the mean. Similarly, when we hold all other variables at their means, the probability of the Chinese president visiting a country with a Log(GDP per capita) that is one standard deviation above its mean is 8.3%, compared with 4.8% for a country with a Log(GDP per capita) at the mean. In addition, we find that the coefficient on GDP Growth is positive and significant for presidential visits. The coefficient on Resources is mostly negative and is not significant. Now moving to political variables, we find that Major Power has a positive and significant coefficient. When we hold all other variables at their means, the probability of the Chinese president visiting a major power is 10.4%, compared with 4.4% for a non-major power. This indicates that everything else being equal, the world's major powers are 136% more likely to receive leader visits than other countries. Japan, however, is an exception (Davis & Meunier, 2011;Quek & Johnston, 2018). J apan has a negative and significant coefficient. When we hold all other variables at their means, the probability of the Chinese president visiting Japan is 0.2% compared with 4.7% for other countries. Even though Japan is also coded as a major power, the magnitude of the binary variable J apan largely eclipses that of Major Power: when we hold all variables, other than J apan and Major Power, at their means, the probability of the Chinese president visiting Japan (as a major power) stands at 0.7%, compared with 4.5% for all other non-major powers. The coefficient on Polity has mixed signs

Standard errors in parentheses
Constant and year fixed effects are included in the regressions but not reported The years are between 1998 and 2019, inclusive * p < 0.10, ** p < 0.05, *** p < 0.01 and is not significant. The coefficient on UNGA Voting is positive and significant with an estimated marginal effect of 7.7%. In particular, the probability of the Chinese president visiting countries with a voting similarity score of 0.8 stands at 5.3% and is 65.6% higher than that of visiting countries with a voting similarity score of 0.5 at 3.2%. Both the president and the premier appear to avoid countries that meet with the Dalai Lama. The coefficient on Meeting with Dalai is negative for all visits and significant for premier visits, which is consistent with Fuchs (2018). 29 Based on Column 8, when we hold all other variables at their means, the probability of the Chinese premier visiting a country that meets with the Dalai Lama is 0.2% compared with 3.4% for other countries. Now moving to our IO variables, we find that the coefficient on SCO is positive and significant. When we hold all other variables at their means, the probability of the Chinese president visiting an SCO member stands at 25.8%, compared with 4.3% for non-members. The coefficient on ASEAN is positive but only significant for premier visits. This is expected as only the premier attends ASEAN-related meetings. According to Column 8, when we hold all other variables at their means, the probability of the Chinese premier visiting an ASEAN member stands at 8.7%, compared with 3.1% for non-members.
As a robustness check, we drop polity score from our regressions, as this variable tends to be missing for small countries, such as Tonga and Malta, and could cause suppression effects (Lenz & Sahn, 2021). This operation increases our sample size by 13% from 2,944 to 3,339. The regression results remain mostly the same (see Table 11 in the appendix). As a robustness check on the effects of Dalai Lama visits, we address the concern that countries that receive Dalai Lama are systematically different by including country fixed effects in our conditional logit and linear probability models (Fuchs & Klann, 2013). The coefficients remain negative but are only significant for premier visits (see Table 12 in the appendix).

Conclusion
Leader visits offer an important signal in international politics. U.S. leader visits are most often studied, partly because of the important role of the United States in the international system and partly because there exists a credible and widely used dataset. Such a dataset did not previously exist for Chinese leader visits, and China's growing status in the international system spurred us to introduce such a dataset. Our dataset is intended to be inclusive, detailed, and easy to use. It is inclusive as it covers the period from 1998 to the present. It is detailed because each recorded visit has rich auxiliary information and is accompanied by a supporting document from an official website. It is easy to use because it adopts the most widely used country code and can be immediately plugged into existing datasets.
Through descriptive statistics and regression analysis, we have uncovered quite a few patterns related to the SCO, ASEAN, and the non-consecutive nature of leader visits. We have also presented noticeable differences between presidential visits and premier visits: e.g., presidential visits are more geographically diffuse and premier visits are clustered and more focused on nearby countries. Researchers have already leveraged the China Visits dataset to study, among others, the relationship between leader visits and United Nations Security Council membership (Wang, 2021), where leader visits serve as the dependent variable, and the relationship between leader visits and Chinese FDI (Stone et al., 2022), where leader visits are an independent variable.
The China Visits dataset opens up important frontiers for future research, including (a) the division of labor between the president and the premier in Chinese leader visits, (b) whether multilateral meeting-induced formal visits differ systematically from the other, more organic, leader visits, (c) the applicability of leader visits as a measure of Chinese state preferences, and (d) the broader theorization of what constitutes the cost of leader visits, what leader visits can represent, and what they signal to the domestic audience, the host country and the wider international community.

Conflicts of Interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.